From 77bb1c5616b8d1cd2bf3b22cb9f251f960b237ba Mon Sep 17 00:00:00 2001 From: foreman Date: Fri, 4 Jul 2014 16:17:05 -0400 Subject: [PATCH] initial commit --- opencl/api/opencl/amdocl/amdocl.def.in | 208 + opencl/api/opencl/amdocl/amdocl.map.in | 245 + opencl/api/opencl/amdocl/amdocl.rc | 75 + opencl/api/opencl/amdocl/cl_agent_amd.h | 312 + opencl/api/opencl/amdocl/cl_command.cpp | 356 + opencl/api/opencl/amdocl/cl_common.hpp | 402 + opencl/api/opencl/amdocl/cl_context.cpp | 612 + opencl/api/opencl/amdocl/cl_counter.cpp | 130 + opencl/api/opencl/amdocl/cl_d3d10.cpp | 1900 +++ opencl/api/opencl/amdocl/cl_d3d10_amd.hpp | 391 + opencl/api/opencl/amdocl/cl_d3d11.cpp | 2043 +++ opencl/api/opencl/amdocl/cl_d3d11_amd.hpp | 402 + opencl/api/opencl/amdocl/cl_d3d9.cpp | 927 ++ opencl/api/opencl/amdocl/cl_d3d9_amd.hpp | 333 + opencl/api/opencl/amdocl/cl_device.cpp | 797 + opencl/api/opencl/amdocl/cl_event.cpp | 393 + opencl/api/opencl/amdocl/cl_execute.cpp | 1043 ++ opencl/api/opencl/amdocl/cl_gl.cpp | 2601 ++++ opencl/api/opencl/amdocl/cl_gl_amd.hpp | 348 + opencl/api/opencl/amdocl/cl_icd.cpp | 235 + opencl/api/opencl/amdocl/cl_icd_amd.h | 981 ++ opencl/api/opencl/amdocl/cl_kernel.h | 121 + .../api/opencl/amdocl/cl_kernel_info_amd.cpp | 136 + opencl/api/opencl/amdocl/cl_kernel_info_amd.h | 165 + opencl/api/opencl/amdocl/cl_memobj.cpp | 4936 ++++++ opencl/api/opencl/amdocl/cl_object.cpp | 72 + opencl/api/opencl/amdocl/cl_open_video.cpp | 829 + opencl/api/opencl/amdocl/cl_open_video_amd.h | 362 + opencl/api/opencl/amdocl/cl_pipe.cpp | 187 + opencl/api/opencl/amdocl/cl_platform_amd.cpp | 28 + opencl/api/opencl/amdocl/cl_platform_amd.h | 117 + opencl/api/opencl/amdocl/cl_profile_amd.cpp | 382 + opencl/api/opencl/amdocl/cl_profile_amd.h | 247 + opencl/api/opencl/amdocl/cl_program.cpp | 1895 +++ opencl/api/opencl/amdocl/cl_sampler.cpp | 328 + opencl/api/opencl/amdocl/cl_sdi_amd.cpp | 259 + opencl/api/opencl/amdocl/cl_sdi_amd.h | 48 + opencl/api/opencl/amdocl/cl_semaphore_amd.h | 118 + opencl/api/opencl/amdocl/cl_svm.cpp | 1091 ++ .../api/opencl/amdocl/cl_thread_trace_amd.cpp | 498 + .../api/opencl/amdocl/cl_thread_trace_amd.h | 439 + opencl/api/opencl/amdocl/gl_functions.hpp | 56 + opencl/api/opencl/config/amdocl32.icd | 1 + opencl/api/opencl/config/amdocl64.icd | 1 + opencl/api/opencl/khronos/headers/EGL/egl.h | 329 + .../api/opencl/khronos/headers/EGL/eglext.h | 645 + .../opencl/khronos/headers/EGL/eglplatform.h | 125 + opencl/api/opencl/khronos/headers/GL/glext.h | 9253 ++++++++++++ .../opencl/khronos/headers/KHR/khrplatform.h | 282 + .../opencl/khronos/headers/opencl1.2/CL/cl.h | 1216 ++ .../khronos/headers/opencl1.2/CL/cl.hpp | 12305 +++++++++++++++ .../khronos/headers/opencl1.2/CL/cl_d3d10.h | 126 + .../khronos/headers/opencl1.2/CL/cl_d3d11.h | 126 + .../opencl1.2/CL/cl_dx9_media_sharing.h | 127 + .../khronos/headers/opencl1.2/CL/cl_ext.h | 489 + .../khronos/headers/opencl1.2/CL/cl_gl.h | 162 + .../khronos/headers/opencl1.2/CL/cl_gl_ext.h | 69 + .../headers/opencl1.2/CL/cl_platform.h | 1171 ++ .../khronos/headers/opencl1.2/CL/opencl.h | 54 + .../khronos/headers/opencl1.2/CL/spir.h | 118 + .../opencl/khronos/headers/opencl2.0/CL/cl.h | 1380 ++ .../khronos/headers/opencl2.0/CL/cl.hpp | 12305 +++++++++++++++ .../khronos/headers/opencl2.0/CL/cl_d3d10.h | 126 + .../khronos/headers/opencl2.0/CL/cl_d3d11.h | 126 + .../opencl2.0/CL/cl_dx9_media_sharing.h | 127 + .../khronos/headers/opencl2.0/CL/cl_egl.h | 133 + .../khronos/headers/opencl2.0/CL/cl_ext.h | 492 + .../khronos/headers/opencl2.0/CL/cl_gl.h | 162 + .../khronos/headers/opencl2.0/CL/cl_gl_ext.h | 69 + .../headers/opencl2.0/CL/cl_platform.h | 1216 ++ .../khronos/headers/opencl2.0/CL/opencl.h | 54 + .../khronos/headers/opencl2.0/CL/spir.h | 118 + opencl/api/opencl/khronos/icd/OpenCL.rc | 74 + opencl/api/opencl/khronos/icd/icd.c | 215 + opencl/api/opencl/khronos/icd/icd.h | 272 + opencl/api/opencl/khronos/icd/icd_dispatch.c | 2524 ++++ opencl/api/opencl/khronos/icd/icd_dispatch.h | 1411 ++ opencl/api/opencl/khronos/icd/icd_linux.c | 191 + opencl/api/opencl/khronos/icd/icd_windows.c | 163 + opencl/init.txt | 1 - opencl/tools/clinfo/cl.hpp | 12474 ++++++++++++++++ opencl/tools/clinfo/clinfo.cpp | 837 ++ 82 files changed, 88116 insertions(+), 1 deletion(-) create mode 100644 opencl/api/opencl/amdocl/amdocl.def.in create mode 100644 opencl/api/opencl/amdocl/amdocl.map.in create mode 100644 opencl/api/opencl/amdocl/amdocl.rc create mode 100644 opencl/api/opencl/amdocl/cl_agent_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_command.cpp create mode 100644 opencl/api/opencl/amdocl/cl_common.hpp create mode 100644 opencl/api/opencl/amdocl/cl_context.cpp create mode 100644 opencl/api/opencl/amdocl/cl_counter.cpp create mode 100644 opencl/api/opencl/amdocl/cl_d3d10.cpp create mode 100644 opencl/api/opencl/amdocl/cl_d3d10_amd.hpp create mode 100644 opencl/api/opencl/amdocl/cl_d3d11.cpp create mode 100644 opencl/api/opencl/amdocl/cl_d3d11_amd.hpp create mode 100644 opencl/api/opencl/amdocl/cl_d3d9.cpp create mode 100644 opencl/api/opencl/amdocl/cl_d3d9_amd.hpp create mode 100644 opencl/api/opencl/amdocl/cl_device.cpp create mode 100644 opencl/api/opencl/amdocl/cl_event.cpp create mode 100644 opencl/api/opencl/amdocl/cl_execute.cpp create mode 100644 opencl/api/opencl/amdocl/cl_gl.cpp create mode 100644 opencl/api/opencl/amdocl/cl_gl_amd.hpp create mode 100644 opencl/api/opencl/amdocl/cl_icd.cpp create mode 100644 opencl/api/opencl/amdocl/cl_icd_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_kernel.h create mode 100644 opencl/api/opencl/amdocl/cl_kernel_info_amd.cpp create mode 100644 opencl/api/opencl/amdocl/cl_kernel_info_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_memobj.cpp create mode 100644 opencl/api/opencl/amdocl/cl_object.cpp create mode 100644 opencl/api/opencl/amdocl/cl_open_video.cpp create mode 100644 opencl/api/opencl/amdocl/cl_open_video_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_pipe.cpp create mode 100644 opencl/api/opencl/amdocl/cl_platform_amd.cpp create mode 100644 opencl/api/opencl/amdocl/cl_platform_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_profile_amd.cpp create mode 100644 opencl/api/opencl/amdocl/cl_profile_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_program.cpp create mode 100644 opencl/api/opencl/amdocl/cl_sampler.cpp create mode 100644 opencl/api/opencl/amdocl/cl_sdi_amd.cpp create mode 100644 opencl/api/opencl/amdocl/cl_sdi_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_semaphore_amd.h create mode 100644 opencl/api/opencl/amdocl/cl_svm.cpp create mode 100644 opencl/api/opencl/amdocl/cl_thread_trace_amd.cpp create mode 100644 opencl/api/opencl/amdocl/cl_thread_trace_amd.h create mode 100644 opencl/api/opencl/amdocl/gl_functions.hpp create mode 100644 opencl/api/opencl/config/amdocl32.icd create mode 100644 opencl/api/opencl/config/amdocl64.icd create mode 100644 opencl/api/opencl/khronos/headers/EGL/egl.h create mode 100644 opencl/api/opencl/khronos/headers/EGL/eglext.h create mode 100644 opencl/api/opencl/khronos/headers/EGL/eglplatform.h create mode 100644 opencl/api/opencl/khronos/headers/GL/glext.h create mode 100644 opencl/api/opencl/khronos/headers/KHR/khrplatform.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.hpp create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d10.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d11.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_dx9_media_sharing.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_ext.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl_ext.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_platform.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/opencl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl1.2/CL/spir.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.hpp create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d10.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d11.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_dx9_media_sharing.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_egl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_ext.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl_ext.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_platform.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/opencl.h create mode 100644 opencl/api/opencl/khronos/headers/opencl2.0/CL/spir.h create mode 100644 opencl/api/opencl/khronos/icd/OpenCL.rc create mode 100644 opencl/api/opencl/khronos/icd/icd.c create mode 100644 opencl/api/opencl/khronos/icd/icd.h create mode 100644 opencl/api/opencl/khronos/icd/icd_dispatch.c create mode 100644 opencl/api/opencl/khronos/icd/icd_dispatch.h create mode 100644 opencl/api/opencl/khronos/icd/icd_linux.c create mode 100644 opencl/api/opencl/khronos/icd/icd_windows.c delete mode 100644 opencl/init.txt create mode 100644 opencl/tools/clinfo/cl.hpp create mode 100644 opencl/tools/clinfo/clinfo.cpp diff --git a/opencl/api/opencl/amdocl/amdocl.def.in b/opencl/api/opencl/amdocl/amdocl.def.in new file mode 100644 index 0000000000..62a59dd2c7 --- /dev/null +++ b/opencl/api/opencl/amdocl/amdocl.def.in @@ -0,0 +1,208 @@ +EXPORTS +clBuildProgram +clCreateBuffer +clCreateCommandQueue +clCreateContext +clCreateContextFromType +clCreateFromGLBuffer +clCreateFromGLRenderbuffer +clCreateFromGLTexture2D +clCreateFromGLTexture3D +clCreateImage2D +clCreateImage3D +clCreateKernel +clCreateKernelsInProgram +clCreateProgramWithBinary +clCreateProgramWithSource +clCreateSampler +clEnqueueAcquireGLObjects +clEnqueueBarrier +clEnqueueCopyBuffer +clEnqueueCopyBufferToImage +clEnqueueCopyImage +clEnqueueCopyImageToBuffer +clEnqueueMapBuffer +clEnqueueMapImage +clEnqueueMarker +clEnqueueNDRangeKernel +clEnqueueNativeKernel +clEnqueueReadBuffer +clEnqueueReadImage +clEnqueueReleaseGLObjects +clEnqueueTask +clEnqueueUnmapMemObject +clEnqueueWaitForEvents +clEnqueueWriteBuffer +clEnqueueWriteImage +clFinish +clFlush +clGetCommandQueueInfo +clGetContextInfo +clGetDeviceIDs +clGetDeviceInfo +clGetEventInfo +clGetEventProfilingInfo +clGetExtensionFunctionAddress +clGetGLObjectInfo +clGetGLTextureInfo +clGetImageInfo +clGetKernelInfo +clGetKernelWorkGroupInfo +clGetMemObjectInfo +clGetPlatformIDs +clGetPlatformInfo +clGetProgramBuildInfo +clGetProgramInfo +clGetSamplerInfo +clGetSupportedImageFormats +clReleaseCommandQueue +clReleaseContext +clReleaseEvent +clReleaseKernel +clReleaseMemObject +clReleaseProgram +clReleaseSampler +clRetainCommandQueue +clRetainContext +clRetainEvent +clRetainKernel +clRetainMemObject +clRetainProgram +clRetainSampler +clSetCommandQueueProperty +clSetKernelArg +clUnloadCompiler +clWaitForEvents +clIcdGetPlatformIDsKHR +clCreateUserEvent +clSetUserEventStatus +clSetEventCallback +clSetMemObjectDestructorCallback +clCreateSubBuffer +clEnqueueReadBufferRect +clEnqueueWriteBufferRect +clEnqueueCopyBufferRect + +#if (OPENCL_MAJOR > 1) || (OPENCL_MAJOR == 1 && OPENCL_MINOR >= 2) +clCompileProgram +clCreateFromGLTexture +clCreateImage +clCreateProgramWithBuiltInKernels +clCreateSubDevices +clEnqueueBarrierWithWaitList +clEnqueueFillBuffer +clEnqueueFillImage +clEnqueueMarkerWithWaitList +clEnqueueMigrateMemObjects +clGetExtensionFunctionAddressForPlatform +clGetKernelArgInfo +clLinkProgram +clReleaseDevice +clRetainDevice +clUnloadPlatformCompiler +#endif + +#if (OPENCL_MAJOR >= 2) +clCreateCommandQueueWithProperties +clCreateSamplerWithProperties +clCreatePipe +clGetPipeInfo +clSVMAlloc +clSVMFree +clSetKernelArgSVMPointer +clSetKernelExecInfo +clEnqueueSVMFree +clEnqueueSVMMemcpy +clEnqueueSVMMemFill +clEnqueueSVMMap +clEnqueueSVMUnmap +#endif + +aclCompilerInit +aclCompilerFini +aclCompilerVersion +aclVersionSize +aclGetErrorString +aclGetArchInfo +aclGetDeviceInfo +aclGetTargetInfo +aclGetArchitecture +aclGetFamily +aclGetChip +aclBinaryInit +aclBinaryFini +aclReadFromFile +aclReadFromMem +aclWriteToFile +aclWriteToMem +aclCreateFromBinary +aclBinaryVersion +aclInsertSection +aclRemoveSection +aclExtractSection +aclInsertSymbol +aclRemoveSymbol +aclExtractSymbol +aclDbgAddArgument +aclDbgRemoveArgument +aclQueryInfo +aclCompile +aclLink +aclGetCompilerLog +aclRetrieveType +aclSetType +aclConvertType +aclDisassemble +aclGetDeviceBinary +aclDumpBinary + +oclCompilerInit +oclCompilerFini +oclCompilerVersion +oclVersionSize +oclCompileSource +oclCompileSourceToIR +oclCompileIRToIL +oclCompileILToISA +oclCompileBinary +oclGetCompilerLog +oclGetArchInfo +oclGetFamilyInfo +oclGetDeviceInfo +oclGetAsicInfo +oclGetTargetInfo +oclGetArchitecture +oclGetFamily +oclGetChip +oclGetAsic +elfInsertSection +elfExtractSection +elfInsertSymbol +elfExtractSymbol +oclirSetType +oclirRetrieveType +oclirLink +oclhsaCompileSource +oclhsaCompileBinary +oclhsaSetType +oclhsaRetrieveType +oclhsaConvertType +oclilCompileSource +oclilCompileBinary +oclilSetType +oclilRetrieveType +oclilConvertType +oclisaDisassemble +oclisaGetBinaryBlob +constructOclElf +destructOclElf +readOclElfFromFile +readOclElfFromMem +writeOclElfToFile +writeOclElfToMem +createElfFromElf +getBIFVersion +rtGetInfo + +#if !defined(OPENCL_MAINLINE) +#endif diff --git a/opencl/api/opencl/amdocl/amdocl.map.in b/opencl/api/opencl/amdocl/amdocl.map.in new file mode 100644 index 0000000000..e05d43d93a --- /dev/null +++ b/opencl/api/opencl/amdocl/amdocl.map.in @@ -0,0 +1,245 @@ +OPENCL_1.0 { +global: + clBuildProgram; + clCreateBuffer; + clCreateCommandQueue; + clCreateContext; + clCreateContextFromType; + clCreateFromD3D10Buffer; + clCreateFromGLBuffer; + clCreateFromGLRenderbuffer; + clCreateFromGLTexture2D; + clCreateFromGLTexture3D; + clCreateImage2D; + clCreateImage3D; + clCreateImageFromD3D10Resource; + clCreateKernel; + clCreateKernelsInProgram; + clCreateProgramWithBinary; + clCreateProgramWithSource; + clCreateSampler; + clEnqueueAcquireExternalObjects; + clEnqueueAcquireGLObjects; + clEnqueueBarrier; + clEnqueueCopyBuffer; + clEnqueueCopyBufferToImage; + clEnqueueCopyImage; + clEnqueueCopyImageToBuffer; + clEnqueueMapBuffer; + clEnqueueMapImage; + clEnqueueMarker; + clEnqueueNDRangeKernel; + clEnqueueNativeKernel; + clEnqueueReadBuffer; + clEnqueueReadImage; + clEnqueueReleaseExternalObjects; + clEnqueueReleaseGLObjects; + clEnqueueTask; + clEnqueueUnmapMemObject; + clEnqueueWaitForEvents; + clEnqueueWriteBuffer; + clEnqueueWriteImage; + clFinish; + clFlush; + clGetCommandQueueInfo; + clGetContextInfo; + clGetDeviceIDs; + clGetDeviceInfo; + clGetEventInfo; + clGetEventProfilingInfo; + clGetExtensionFunctionAddress; + clGetGLObjectInfo; + clGetGLTextureInfo; + clGetImageInfo; + clGetKernelInfo; + clGetKernelWorkGroupInfo; + clGetMemObjectInfo; + clGetPlatformIDs; + clGetPlatformInfo; + clGetProgramBuildInfo; + clGetProgramInfo; + clGetSamplerInfo; + clGetSupportedImageFormats; + clReleaseCommandQueue; + clReleaseContext; + clReleaseEvent; + clReleaseKernel; + clReleaseMemObject; + clReleaseProgram; + clReleaseSampler; + clRetainCommandQueue; + clRetainContext; + clRetainEvent; + clRetainKernel; + clRetainMemObject; + clRetainProgram; + clRetainSampler; + clSetCommandQueueProperty; + clSetKernelArg; + clUnloadCompiler; + clWaitForEvents; + clIcdGetPlatformIDsKHR; +#if defined(OPENCL_MAINLINE) +local: + *; +#endif +}; + +#if (OPENCL_MAJOR > 1) || (OPENCL_MAJOR == 1 && OPENCL_MINOR >= 1) +OPENCL_1.1 { +global: + clCreateUserEvent; + clSetUserEventStatus; + clSetEventCallback; + clSetMemObjectDestructorCallback; + clCreateSubBuffer; + clEnqueueReadBufferRect; + clEnqueueWriteBufferRect; + clEnqueueCopyBufferRect; + + aclGetTargetInfo; + aclCompilerInit; + aclCompilerFini; + aclReadFromMem; + aclReadFromFile; + aclBinaryInit; + aclBinaryFini; + aclWriteToMem; + aclInsertSection; + aclExtractSection; + aclRemoveSection; + aclQueryInfo; + aclDbgAddArgument; + aclExtractSymbol; + aclInsertSymbol; + aclRemoveSymbol; + aclCompile; + aclDisassemble; +} OPENCL_1.0; +#endif + +#if (OPENCL_MAJOR > 1) || (OPENCL_MAJOR == 1 && OPENCL_MINOR >= 2) +OPENCL_1.2 { +global: + clCompileProgram; + clCreateFromGLTexture; + clCreateImage; + clCreateProgramWithBuiltInKernels; + clCreateSubDevices; + clEnqueueBarrierWithWaitList; + clEnqueueFillBuffer; + clEnqueueFillImage; + clEnqueueMarkerWithWaitList; + clEnqueueMigrateMemObjects; + clGetExtensionFunctionAddressForPlatform; + clGetKernelArgInfo; + clLinkProgram; + clReleaseDevice; + clRetainDevice; + clUnloadPlatformCompiler; +} OPENCL_1.1; +#endif + +#if (OPENCL_MAJOR >= 2) +OPENCL_2.0 { +global: + clCreateCommandQueueWithProperties; + clCreateSamplerWithProperties; + clCreatePipe; + clGetPipeInfo; + clSVMAlloc; + clSVMFree; + clSetKernelArgSVMPointer; + clSetKernelExecInfo; + clEnqueueSVMFree; + clEnqueueSVMMemcpy; + clEnqueueSVMMemFill; + clEnqueueSVMMap; + clEnqueueSVMUnmap; +} OPENCL_1.2; +#endif + +ACL_0.8 { +global: + oclCompilerInit; + oclCompilerFini; + oclCompilerVersion; + oclVersionSize; + oclCompileSource; + oclCompileSourceToIR; + oclCompileIRToIL; + oclCompileILToISA; + oclCompileBinary; + oclGetCompilerLog; + oclGetArchInfo; + oclGetDeviceInfo; + oclGetTargetInfo; + oclGetArchitecture; + oclGetFamily; + oclGetChip; + readOclElfFromFile; + readOclElfFromMem; + writeOclElfToFile; + writeOclElfToMem; + createElfFromElf; + getBIFVersion; + constructOclElf; + destructOclElf; + elfInsertSection; + elfExtractSection; + elfInsertSymbol; + elfExtractSymbol; + oclirSetType; + oclirRetrieveType; + oclirLink; + oclhsaCompileSource; + oclhsaCompileBinary; + oclhsaSetType; + oclhsaRetrieveType; + oclhsaConvertType; + oclilCompileSource; + oclilCompileBinary; + oclilSetType; + oclilRetrieveType; + oclilConvertType; + oclisaDisassemble; + oclisaGetBinaryBlob; + rtGetInfo; + aclCompilerInit; + aclCompilerFini; + aclCompilerVersion; + aclVersionSize; + aclGetErrorString; + aclGetArchInfo; + aclGetDeviceInfo; + aclGetTargetInfo; + aclGetArchitecture; + aclGetFamily; + aclGetChip; + aclBinaryInit; + aclBinaryFini; + aclReadFromFile; + aclReadFromMem; + aclWriteToFile; + aclWriteToMem; + aclCreateFromBinary; + aclBinaryVersion; + aclInsertSection; + aclRemoveSection; + aclExtractSection; + aclInsertSymbol; + aclRemoveSymbol; + aclExtractSymbol; + aclDbgAddArgument; + aclDbgRemoveArgument; + aclQueryInfo; + aclCompile; + aclLink; + aclGetCompilerLog; + aclRetrieveType; + aclSetType; + aclConvertType; + aclDisassemble; + aclGetDeviceBinary; + aclDumpBinary; +}; diff --git a/opencl/api/opencl/amdocl/amdocl.rc b/opencl/api/opencl/amdocl/amdocl.rc new file mode 100644 index 0000000000..3e437d09b1 --- /dev/null +++ b/opencl/api/opencl/amdocl/amdocl.rc @@ -0,0 +1,75 @@ +#define STR(__macro__) #__macro__ +#define XSTR(__macro__) STR(__macro__) + +#if defined(_DEBUG) +#define DEBUG_ONLY(x) x +#else +#define DEBUG_ONLY(x) +#endif + +#define VERSION_PREFIX_MAJOR 1 +#define VERSION_PREFIX_MINOR 2 + + +#define APSTUDIO_READONLY_SYMBOLS +///////////////////////////////////////////////////////////////////////////// +// +// Generated from the TEXTINCLUDE 2 resource. +// +#include "winresrc.h" +#include "utils/versions.hpp" + +///////////////////////////////////////////////////////////////////////////// +#undef APSTUDIO_READONLY_SYMBOLS + +///////////////////////////////////////////////////////////////////////////// +// English (U.S.) resources + +#if !defined(AFX_RESOURCE_DLL) || defined(AFX_TARG_ENU) +#ifdef _WIN32 +LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US +#pragma code_page(1252) +#endif //_WIN32 + + +///////////////////////////////////////////////////////////////////////////// +// +// Version +// + +VS_VERSION_INFO VERSIONINFO + FILEVERSION 10,0,AMD_PLATFORM_BUILD_NUMBER,AMD_PLATFORM_REVISION_NUMBER + PRODUCTVERSION 10,0,AMD_PLATFORM_BUILD_NUMBER,AMD_PLATFORM_REVISION_NUMBER + FILEFLAGSMASK 0x3fL +#ifdef _DEBUG + FILEFLAGS 0x1L +#else + FILEFLAGS 0x0L +#endif + FILEOS 0x40004L + FILETYPE 0x2L + FILESUBTYPE 0x0L +BEGIN + BLOCK "StringFileInfo" + BEGIN + BLOCK "040904b0" + BEGIN + VALUE "Comments", " \0" + VALUE "CompanyName", "Advanced Micro Devices Inc.\0" + VALUE "FileDescription", AMD_PLATFORM_NAME " OpenCL " XSTR(VERSION_PREFIX_MAJOR) "." XSTR(VERSION_PREFIX_MINOR) " Runtime\0" + VALUE "FileVersion", "10, 0, " XSTR(AMD_PLATFORM_BUILD_NUMBER) ", " XSTR(AMD_PLATFORM_REVISION_NUMBER) + VALUE "InternalName", "OpenCL" + VALUE "LegalCopyright", "Copyright (C) 2011 Advanced Micro Devices Inc.\0" + VALUE "OriginalFilename", "OpenCL.dll" + VALUE "ProductName", "OpenCL " XSTR(VERSION_PREFIX_MAJOR) "." XSTR(VERSION_PREFIX_MINOR) " " AMD_PLATFORM_INFO "\0" + VALUE "ProductVersion", "10, 0, " XSTR(AMD_PLATFORM_BUILD_NUMBER) ", " XSTR(AMD_PLATFORM_REVISION_NUMBER) + END + END + BLOCK "VarFileInfo" + BEGIN + VALUE "Translation", 0x409, 1200 + END +END + +#endif // English (U.S.) resources +///////////////////////////////////////////////////////////////////////////// diff --git a/opencl/api/opencl/amdocl/cl_agent_amd.h b/opencl/api/opencl/amdocl/cl_agent_amd.h new file mode 100644 index 0000000000..7967750df9 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_agent_amd.h @@ -0,0 +1,312 @@ +/* ============================================================ + +Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +#ifndef __OPENCL_CL_AGENT_AMD_H +#define __OPENCL_CL_AGENT_AMD_H + +#include +#include "cl_icd_amd.h" + +#define cl_amd_agent 1 + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +typedef const struct _cl_agent cl_agent; + +#define CL_AGENT_VERSION_1_0 100 + +/* Context Callbacks */ + +typedef void (CL_CALLBACK * acContextCreate_fn)( + cl_agent * /* agent */, + cl_context /* context */); + +typedef void (CL_CALLBACK * acContextFree_fn)( + cl_agent * /* agent */, + cl_context /* context */); + +/* Command Queue Callbacks */ + +typedef void (CL_CALLBACK * acCommandQueueCreate_fn)( + cl_agent * /* agent */, + cl_command_queue /* queue */); + +typedef void (CL_CALLBACK * acCommandQueueFree_fn)( + cl_agent * /* agent */, + cl_command_queue /* queue */); + +/* Event Callbacks */ + +typedef void (CL_CALLBACK * acEventCreate_fn)( + cl_agent * /* agent */, + cl_event /* event */, + cl_command_type /* type */); + +typedef void (CL_CALLBACK * acEventFree_fn)( + cl_agent * /* agent */, + cl_event /* event */); + +typedef void (CL_CALLBACK * acEventStatusChanged_fn)( + cl_agent * /* agent */, + cl_event /* event */, + cl_int /* execution_status */, + cl_long /* epoch_time_stamp */); + +/* Memory Object Callbacks */ + +typedef void (CL_CALLBACK * acMemObjectCreate_fn)( + cl_agent * /* agent */, + cl_mem /* memobj */); + +typedef void (CL_CALLBACK * acMemObjectFree_fn)( + cl_agent * /* agent */, + cl_mem /* memobj */); + +typedef void (CL_CALLBACK * acMemObjectAcquired_fn)( + cl_agent * /* agent */, + cl_mem /* memobj */, + cl_device_id /* device */, + cl_long /* elapsed_time */); + +/* Sampler Callbacks */ + +typedef void (CL_CALLBACK * acSamplerCreate_fn)( + cl_agent * /* agent */, + cl_sampler /* sampler */); + +typedef void (CL_CALLBACK * acSamplerFree_fn)( + cl_agent * /* agent */, + cl_sampler /* sampler */); + +/* Program Callbacks */ + +typedef void (CL_CALLBACK * acProgramCreate_fn)( + cl_agent * /* agent */, + cl_program /* program */); + +typedef void (CL_CALLBACK * acProgramFree_fn)( + cl_agent * /* agent */, + cl_program /* program */); + +typedef void (CL_CALLBACK * acProgramBuild_fn)( + cl_agent * /* agent */, + cl_program /* program */); + +/* Kernel Callbacks */ + +typedef void (CL_CALLBACK * acKernelCreate_fn)( + cl_agent * /* agent */, + cl_kernel /* kernel */); + +typedef void (CL_CALLBACK * acKernelFree_fn)( + cl_agent * /* agent */, + cl_kernel /* kernel */); + +typedef void (CL_CALLBACK * acKernelSetArg_fn)( + cl_agent * /* agent */, + cl_kernel /* kernel */, + cl_int /* arg_index */, + size_t /* size */, + const void * /* value_ptr */); + +typedef struct _cl_agent_callbacks +{ + /* Context Callbacks */ + acContextCreate_fn ContextCreate; + acContextFree_fn ContextFree; + + /* Command Queue Callbacks */ + acCommandQueueCreate_fn CommandQueueCreate; + acCommandQueueFree_fn CommandQueueFree; + + /* Event Callbacks */ + acEventCreate_fn EventCreate; + acEventFree_fn EventFree; + acEventStatusChanged_fn EventStatusChanged; + + /* Memory Object Callbacks */ + acMemObjectCreate_fn MemObjectCreate; + acMemObjectFree_fn MemObjectFree; + acMemObjectAcquired_fn MemObjectAcquired; + + /* Sampler Callbacks */ + acSamplerCreate_fn SamplerCreate; + acSamplerFree_fn SamplerFree; + + /* Program Callbacks */ + acProgramCreate_fn ProgramCreate; + acProgramFree_fn ProgramFree; + acProgramBuild_fn ProgramBuild; + + /* Kernel Callbacks */ + acKernelCreate_fn KernelCreate; + acKernelFree_fn KernelFree; + acKernelSetArg_fn KernelSetArg; + +} cl_agent_callbacks; + +typedef cl_uint cl_agent_capability_action; + +#define CL_AGENT_ADD_CAPABILITIES 0x0 +#define CL_AGENT_RELINQUISH_CAPABILITIES 0x1 + +typedef struct _cl_agent_capabilities +{ + cl_bitfield canGenerateContextEvents : 1; + cl_bitfield canGenerateCommandQueueEvents : 1; + cl_bitfield canGenerateEventEvents : 1; + cl_bitfield canGenerateMemObjectEvents : 1; + cl_bitfield canGenerateSamplerEvents : 1; + cl_bitfield canGenerateProgramEvents : 1; + cl_bitfield canGenerateKernelEvents : 1; + +} cl_agent_capabilities; + +struct _cl_agent +{ + cl_int (CL_API_CALL * GetVersionNumber)( + cl_agent * /* agent */, + cl_int * /* version_ret */); + + cl_int (CL_API_CALL * GetPlatform)( + cl_agent * /* agent */, + cl_platform_id * /* platform_id_ret */); + + cl_int (CL_API_CALL * GetTime) ( + cl_agent * /* agent */, + cl_long * /* time_nanos */); + + cl_int (CL_API_CALL * SetCallbacks)( + cl_agent * /* agent */, + const cl_agent_callbacks * /* callbacks */, + size_t /* size */); + + + cl_int (CL_API_CALL * GetPotentialCapabilities)( + cl_agent * /* agent */, + cl_agent_capabilities * /* capabilities */); + + cl_int (CL_API_CALL * GetCapabilities)( + cl_agent * /* agent */, + cl_agent_capabilities * /* capabilities */); + + cl_int (CL_API_CALL * SetCapabilities)( + cl_agent * /* agent */, + const cl_agent_capabilities * /* capabilities */, + cl_agent_capability_action /* action */); + + + cl_int (CL_API_CALL * GetICDDispatchTable)( + cl_agent * /* agent */, + cl_icd_dispatch_table * /* table */, + size_t /* size */); + + cl_int (CL_API_CALL * SetICDDispatchTable)( + cl_agent * /* agent */, + const cl_icd_dispatch_table * /* table */, + size_t /* size */); + + /* add Kernel/Program helper functions, etc... */ +}; + +extern cl_int CL_CALLBACK +clAgent_OnLoad(cl_agent * /* agent */); + +extern void CL_CALLBACK +clAgent_OnUnload(cl_agent * /* agent */); + +#ifdef __cplusplus +} +#endif /* __cplusplus */ + +#endif /* __OPENCL_CL_AGENT_AMD_H */ diff --git a/opencl/api/opencl/amdocl/cl_command.cpp b/opencl/api/opencl/amdocl/cl_command.cpp new file mode 100644 index 0000000000..1b4cb5fd63 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_command.cpp @@ -0,0 +1,356 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/object.hpp" +#include "platform/context.hpp" +#include "platform/command.hpp" +#include "platform/agent.hpp" + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_Queues + * + * OpenCL objects such as memory objects, program and kernel objects are + * created using a context. Operations on these objects are performed using + * a command-queue. The command-queue can be used to queue a set of operations + * (referred to as commands) in order. Having multiple command-queues allows + * applications to queue multiple independent commands without requiring + * synchronization. Note that this should work as long as these objects are + * not being shared. Sharing of objects across multiple command-queues will + * require the application to perform appropriate synchronization. + * + * @{ + */ + +/*! \brief Create a command-queue on a specific device. + * + * \param context must be a valid OpenCL context. + * + * \param device must be a device associated with context. It can either be + * in the list of devices specified when context is created using + * clCreateContext or have the same device type as device type specified wheni + * context is created using clCreateContextFromType. + * + * \param properties specifies a list of properties for the command-queue. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero command-queue and \a errcode_ret is set to + * CL_SUCCESS if the command-queue is created successfully or a NULL value + * with one of the following error values returned \a in errcode_ret: + * - CL_INVALID_CONTEXT if context is not a valid. + * - CL_INVALID_DEVICE if device is not a valid device or is not associated + * with context + * - CL_INVALID_VALUE if values specified in properties are not valid. + * - CL_INVALID_QUEUE_PROPERTIES if values specified in properties are valid + * but are not supported by the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_command_queue, clCreateCommandQueueWithProperties, ( + cl_context context, + cl_device_id device, + const cl_queue_properties *queue_properties, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return (cl_command_queue) 0; + } + + amd::Context& amdContext = *as_amd(context); + amd::Device& amdDevice = *as_amd(device); + + if (!is_valid(device) || + !amdContext.containsDevice(&amdDevice)) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + return (cl_command_queue) 0; + } + + cl_command_queue_properties properties = 0; + const struct QueueProperty { + cl_queue_properties name; + union { + cl_queue_properties raw; + //FIXME_lmoriche: Check with Khronos. cl_queue_properties is an intptr, + //but cl_command_queue_properties is a bitfield (truncate?). + //cl_command_queue_properties properties; + cl_uint size; + } value; + } *p = reinterpret_cast(queue_properties); + + uint queueSize = amdDevice.info().queueOnDevicePreferredSize_; + if (p != NULL) while(p->name != 0) { + switch(p->name) { + case CL_QUEUE_PROPERTIES: + //FIXME_lmoriche: See comment above. + //properties = p->value.properties; + properties = static_cast(p->value.raw); + break; + case CL_QUEUE_SIZE: // Unimplemented + queueSize = p->value.size; + break; + default: + *not_null(errcode_ret) = CL_INVALID_QUEUE_PROPERTIES; + LogWarning("invalid property name"); + return (cl_command_queue) 0; + } + ++p; + } + + amd::CommandQueue* queue = NULL; + { + amd::ScopedLock lock(amdContext.lock()); + + // Check if the app creates a host queue + if (!(properties & CL_QUEUE_ON_DEVICE)) { + queue = new amd::HostQueue(amdContext, amdDevice, properties); + } + else { + // Is it a device default queue + if (properties & CL_QUEUE_ON_DEVICE_DEFAULT) { + queue = amdContext.defDeviceQueue(amdDevice); + // If current context has one already then retunr it + if (NULL != queue) { + queue->retain(); + return as_cl(queue); + } + } + // Check if runtime can allocate a new device queue on this context + if (amdContext.isDevQueuePossible(amdDevice)) { + queue = new amd::DeviceQueue( + amdContext, amdDevice, properties, queueSize); + } + } + + if (queue == NULL || !queue->create()) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + delete queue; + return (cl_command_queue) 0; + } + } + + if (amd::Agent::shouldPostCommandQueueEvents()) { + amd::Agent::postCommandQueueCreate(as_cl(queue->asCommandQueue())); + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(queue); +} +RUNTIME_EXIT + +RUNTIME_ENTRY_RET(cl_command_queue, clCreateCommandQueue, ( + cl_context context, + cl_device_id device, + cl_command_queue_properties properties, + cl_int *errcode_ret)) +{ + const cl_queue_properties cprops[] = { + CL_QUEUE_PROPERTIES, + static_cast(properties), + 0 }; + return clCreateCommandQueueWithProperties( + context, device, properties ? cprops : NULL, errcode_ret); +} +RUNTIME_EXIT + +/*! \brief Increment the \a command_queue reference count. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid + * command-queue. + * + * clCreateCommandQueue performs an implicit retain. This is very helpful for + * 3rd party libraries, which typically get a command-queue passed to them + * by the application. However, it is possible that the application may delete + * the command-queue without informing the library. Allowing functions to + * attach to (i.e. retain) and release a command-queue solves the problem of a + * command-queue being used by a library no longer being valid. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainCommandQueue, (cl_command_queue command_queue)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + as_amd(command_queue)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the \a command_queue reference count. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid + * command-queue. + * + * After the command_queue reference count becomes zero and all commands queued + * to \a command_queue have finished (eg. kernel executions, memory object + * updates etc.), the command-queue is deleted. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseCommandQueue, (cl_command_queue command_queue)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + as_amd(command_queue)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Query information about a command-queue. + * + * \param command_queue specifies the command-queue being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * If param_value is NULL, it is ignored. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by \a param_value. If \a param_value_size_ret is NULL, + * it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid + * command-queue. + * - CL_INVALID_VALUE if \a param_name is not one of the supported + * values or if size in bytes specified by \a param_value_size is < size of + * return type and \a param_value is not a NULL value. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetCommandQueueInfo, ( + cl_command_queue command_queue, + cl_command_queue_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + switch (param_name) { + case CL_QUEUE_CONTEXT: { + cl_context context = const_cast( + as_cl(&as_amd(command_queue)->context())); + return amd::clGetInfo( + context, param_value_size, param_value, param_value_size_ret); + } + case CL_QUEUE_DEVICE: { + cl_device_id device = const_cast( + as_cl(&as_amd(command_queue)->device())); + return amd::clGetInfo( + device, param_value_size, param_value, param_value_size_ret); + } + case CL_QUEUE_PROPERTIES: { + cl_command_queue_properties properties + = as_amd(command_queue)->properties().value_; + return amd::clGetInfo( + properties, param_value_size, param_value, param_value_size_ret); + } + case CL_QUEUE_REFERENCE_COUNT: { + cl_uint count = as_amd(command_queue)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_QUEUE_SIZE: { + const amd::DeviceQueue* deviceQueue = as_amd(command_queue)->asDeviceQueue(); + if (NULL == deviceQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + cl_uint size = deviceQueue->size(); + return amd::clGetInfo( + size, param_value_size, param_value, param_value_size_ret); + } + case CL_QUEUE_THREAD_HANDLE_AMD: { + const amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + const void* handle = hostQueue->thread().handle(); + return amd::clGetInfo( + handle, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! \brief Enable or disable the properties of a command-queue. + * + * \param command_queue specifies the command-queue being queried. + * + * \param properties specifies the new command-queue properties to be applied + * to \a command_queue . + * + * \param enable determines whether the values specified by properties are + * enabled (if enable is CL_TRUE) or disabled (if enable is CL_FALSE) for the + * command-queue . + * + * \param old_properties returns the command-queue properties before they were + * changed by clSetCommandQueueProperty. If \a old_properties is NULL, + * it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the command-queue properties are successfully updated. + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_VALUE if the values specified in properties are not valid. + * - CL_INVALID_QUEUE_PROPERTIES if values specified in properties are + * not supported by the device. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clSetCommandQueueProperty, ( + cl_command_queue command_queue, + cl_command_queue_properties properties, + cl_bool enable, + cl_command_queue_properties *old_properties)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + *not_null(old_properties) + = as_amd(command_queue)->properties().value_; + + if (properties & CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE) { + clFinish(command_queue); + } + + bool success; + if (enable == CL_TRUE) { + success = as_amd(command_queue)->properties().set(properties); + } + else { + success = as_amd(command_queue)->properties().clear(properties); + } + + return success ? CL_SUCCESS : CL_INVALID_QUEUE_PROPERTIES; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_common.hpp b/opencl/api/opencl/amdocl/cl_common.hpp new file mode 100644 index 0000000000..4ba956a8e7 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_common.hpp @@ -0,0 +1,402 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifndef CL_COMMON_HPP_ +#define CL_COMMON_HPP_ + +#include "top.hpp" +#include "platform/runtime.hpp" +#include "platform/command.hpp" +#include "platform/memory.hpp" +#include "platform/video_session.hpp" +#include "thread/thread.hpp" + +#include +#include + +//! \cond ignore +namespace amd { + +template +class NotNullWrapper +{ +private: + T* const ptrOrNull_; + +protected: + explicit NotNullWrapper(T* ptrOrNull) + : ptrOrNull_(ptrOrNull) + { } + +public: + void operator = (T value) const + { + if (ptrOrNull_ != NULL) { + *ptrOrNull_ = value; + } + } +}; + +template +class NotNullReference : protected NotNullWrapper +{ +public: + explicit NotNullReference(T* ptrOrNull) + : NotNullWrapper(ptrOrNull) + { } + + const NotNullWrapper& operator * () const { return *this; } +}; + +} // namespace amd + +template +inline amd::NotNullReference +not_null(T* ptrOrNull) +{ + return amd::NotNullReference(ptrOrNull); +} + +#define CL_CHECK_THREAD(thread) \ + (thread != NULL || ((thread = new amd::HostThread()) != NULL \ + && thread == amd::Thread::current())) + +#define RUNTIME_ENTRY_RET(ret, func, args) \ +CL_API_ENTRY ret CL_API_CALL \ +func args \ +{ \ + amd::Thread* thread = amd::Thread::current(); \ + if (!CL_CHECK_THREAD(thread)) { \ + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; \ + return (ret) 0; \ + } + +#define RUNTIME_ENTRY_RET_NOERRCODE(ret, func, args) \ +CL_API_ENTRY ret CL_API_CALL \ +func args \ +{ \ + amd::Thread* thread = amd::Thread::current(); \ + if (!CL_CHECK_THREAD(thread)) { \ + return (ret) 0; \ + } + +#define RUNTIME_ENTRY(ret, func, args) \ +CL_API_ENTRY ret CL_API_CALL \ +func args \ +{ \ + amd::Thread* thread = amd::Thread::current(); \ + if (!CL_CHECK_THREAD(thread)) { \ + return CL_OUT_OF_HOST_MEMORY; \ + } + +#define RUNTIME_ENTRY_VOID(ret, func, args) \ +CL_API_ENTRY ret CL_API_CALL \ +func args \ +{ \ + amd::Thread* thread = amd::Thread::current(); \ + if (!CL_CHECK_THREAD(thread)) { \ + return; \ + } + +#define RUNTIME_EXIT \ + /* FIXME_lmoriche: we should check to thread->lastError here! */ \ +} + +//! Helper function to check "properties" parameter in various functions +int checkContextProperties( + const cl_context_properties *properties, + bool* offlineDevices); + +namespace amd { + +namespace detail { + +template +struct ParamInfo +{ + static inline std::pair get(const T& param) { + return std::pair(¶m, sizeof(T)); + } +}; + +template <> +struct ParamInfo +{ + static inline std::pair get(const char* param) { + return std::pair(param, strlen(param) + 1); + } +}; + +template +struct ParamInfo +{ + static inline std::pair get(const char* param) { + return std::pair(param, strlen(param) + 1); + } +}; + +} // namespace detail + +template +static inline cl_int +clGetInfo( + T& field, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret) +{ + const void *valuePtr; + size_t valueSize; + + std::tie(valuePtr, valueSize) + = detail::ParamInfo::type>::get(field); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + ::memcpy(param_value, valuePtr, valueSize); + if (param_value_size > valueSize) { + ::memset(static_cast
(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + + return CL_SUCCESS; +} + +static inline cl_int +clSetEventWaitList( + Command::EventWaitList& eventWaitList, + const Context& context, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list) +{ + if ((num_events_in_wait_list == 0 && event_wait_list != NULL) + || (num_events_in_wait_list != 0 && event_wait_list == NULL)) { + return CL_INVALID_EVENT_WAIT_LIST; + } + + while (num_events_in_wait_list-- > 0) { + cl_event event = *event_wait_list++; + Event* amdEvent = as_amd(event); + if (!is_valid(event)) { + return CL_INVALID_EVENT_WAIT_LIST; + } + if (&context != &amdEvent->context()) { + return CL_INVALID_CONTEXT; + } + eventWaitList.push_back(amdEvent); + } + return CL_SUCCESS; +} + +//! Common function declarations for CL-external graphics API interop +cl_int clEnqueueAcquireExtObjectsAMD(cl_command_queue command_queue, + cl_uint num_objects, const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, const cl_event* event_wait_list, + cl_event* event, cl_command_type cmd_type); +cl_int clEnqueueReleaseExtObjectsAMD(cl_command_queue command_queue, + cl_uint num_objects, const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, const cl_event* event_wait_list, + cl_event* event, cl_command_type cmd_type); + +// This may need moving somewhere tidier... + +struct PlatformIDS { const struct KHRicdVendorDispatchRec* dispatch_; }; +class PlatformID { +public: + static PlatformIDS Platform; +}; +#define AMD_PLATFORM (reinterpret_cast(&amd::PlatformID::Platform)) + +#if cl_amd_open_video +cl_int clEnqueueVideoDecodeAMD( + VideoSession& session, + cl_video_decode_data_amd* video_data, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event); + +cl_int clEnqueueVideoEncodeAMD( + VideoSession& session, + cl_video_encode_data_amd* video_data, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event); + +#endif // cl_amd_open_video + +} // namespace amd + +extern "C" { + +#ifdef cl_ext_device_fission + +extern CL_API_ENTRY cl_int CL_API_CALL +clCreateSubDevicesEXT( + cl_device_id in_device, + const cl_device_partition_property_ext * partition_properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices); + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainDeviceEXT(cl_device_id device); + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseDeviceEXT(cl_device_id device); + +#else // cl_ext_device_fission + +#define clCreateSubDevicesEXT ((void (*)(void))0) +#define clRetainDeviceEXT ((void (*)(void))0) +#define clReleaseDeviceEXT ((void (*)(void))0) + +#endif // cl_ext_device_fission + +extern CL_API_ENTRY cl_key_amd CL_API_CALL +clCreateKeyAMD( + cl_platform_id platform, + void (CL_CALLBACK * destructor)( void * ), + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_int CL_API_CALL +clObjectGetValueForKeyAMD( + void * object, + cl_key_amd key, + void ** ret_val); + +extern CL_API_ENTRY cl_int CL_API_CALL +clObjectSetValueForKeyAMD( + void * object, + cl_key_amd key, + void * value); + +#if defined(CL_VERSION_1_1) +extern CL_API_ENTRY cl_int CL_API_CALL +clSetCommandQueueProperty( + cl_command_queue command_queue, + cl_command_queue_properties properties, + cl_bool enable, + cl_command_queue_properties *old_properties) CL_API_SUFFIX__VERSION_1_0; +#endif // CL_VERSION_1_1 + +#if cl_amd_open_video +extern CL_API_ENTRY cl_video_session_amd CL_API_CALL +clCreateVideoSessionAMD( + cl_context context, + cl_device_id device, + cl_video_session_flags_amd flags, + cl_video_config_type_amd config_buffer_type, + cl_uint config_buffer_size, + void* config_buffer, + cl_int* errcode_ret); + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainVideoSessionAMD( + cl_video_session_amd video_session); + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseVideoSessionAMD( + cl_video_session_amd video_session); + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetVideoSessionInfoAMD( + cl_video_session_amd video_session, + cl_video_session_info_amd param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret); + + +extern CL_API_ENTRY cl_video_session_amd CL_API_CALL +clCreateVideoEncSessionAMD( + cl_context context, + cl_device_id device, + cl_video_session_flags_amd flags, + cl_video_config_type_amd config_buffer_type, + cl_uint config_buffer_size, + void* config_buffer, + cl_int* errcode_ret); + + +extern CL_API_ENTRY cl_int CL_API_CALL +clDestroyVideoEncSessionAMD( + cl_video_session_amd video_session); + + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetVideoSessionEncInfoAMD( + cl_video_session_amd video_session, + cl_video_session_enc_info_amd param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret); + +extern CL_API_ENTRY cl_int CL_API_CALL +clSendEncodeConfigInfoAMD( + cl_video_session_amd video_session, + size_t numBuffers, + void* pConfigBuffers); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueRunVideoProgramAMD( + cl_video_session_amd video_session, + void* video_data_struct, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEncodeGetDeviceCapAMD( + cl_device_id device_id, + cl_uint encode_mode, + cl_uint encode_cap_total_size, + cl_uint* num_encode_cap, + void* pEncodeCAP); + +#if 1 +extern CL_API_ENTRY cl_int CL_API_CALL +clEncodePictureAMD( + cl_video_session_amd video_session, + cl_uint number_of_encode_task_input_buffers, + void* encode_task_input_buffer_list, + void* picture_parameter, + cl_uint* pTaskID); +#endif +extern CL_API_ENTRY cl_int CL_API_CALL +clEncodeQueryTaskDescriptionAMD( + cl_video_session_amd session, + cl_uint num_of_task_description_request, + cl_uint* num_of_task_description_return, + void * task_description_list); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEncodeReleaseOutputResourceAMD( + cl_video_session_amd session, + cl_uint task_id); + +#endif // cl_amd_open_video + +extern CL_API_ENTRY cl_mem CL_API_CALL +clConvertImageAMD( + cl_context context, + cl_mem image, + const cl_image_format * image_format, + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateBufferFromImageAMD( + cl_context context, + cl_mem image, + cl_int * errcode_ret); + +} // extern "C" + +//! \endcond + +#endif /*CL_COMMON_HPP_*/ diff --git a/opencl/api/opencl/amdocl/cl_context.cpp b/opencl/api/opencl/amdocl/cl_context.cpp new file mode 100644 index 0000000000..5133863c12 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_context.cpp @@ -0,0 +1,612 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/context.hpp" +#include "device/device.hpp" +#include "platform/runtime.hpp" +#include "platform/agent.hpp" +#ifdef _WIN32 +#include +#include "CL/cl_d3d10.h" +#include "CL/cl_d3d11.h" +#include "cl_d3d9_amd.hpp" +#include "cl_d3d10_amd.hpp" +#include "cl_d3d11_amd.hpp" +#endif // _WIN32 +#include "cl_kernel_info_amd.h" +#include "cl_profile_amd.h" +#include "cl_platform_amd.h" +#include "cl_sdi_amd.h" +#include "cl_thread_trace_amd.h" + +#include +#include +#include "CL/cl_gl.h" + +#ifndef WITH_GPU_DEVICE +# undef cl_amd_open_video +#endif // WITH_GPU_DEVICE + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_Contexts + * @{ + */ + +/*! \brief Create an OpenCL context. + * + * An OpenCL context is created with one or more devices. Contexts are used by + * the OpenCL runtime for managing objects such as command-queues, memory, + * program and kernel objects and for executing kernels on one or more devices + * specified in the context. + * + * \param properties is reserved and must be zero. + * + * \param num_devices is the number of devices specified in the \a devices + * argument. + * + * \param devices is a pointer to a list of unique devices returned by + * clGetDevices. If more than one device is specified in devices, + * a selection criteria may be applied to determine if the list of devices + * specified can be used together to create a context. + * + * \param pfn_notify is a callback function that can be registered by the + * application. This callback function will be used by the runtime to report + * information on errors that occur in this context. This callback function + * may be called asynchronously by the runtime. If \a pfn_notify is NULL, + * no callback function is registered. + * + * \param user_data will be passed as the user_data argument when \a pfn_notify + * is called. \a user_data can be NULL. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero context and errcode_ret is set to CL_SUCCESS + * if the context is created successfully or NULL with the following + * error values stored in \a errcode_ret: + * - CL_INVALID_VALUE if \a properties is not zero. + * - CL_INVALID_VALUE if \a devices is NULL. + * - CL_INVALID_VALUE if \a num_devices is equal to zero. + * - CL_INVALID_DEVICE if \a devices contains an invalid device. + * - CL_INVALID_DEVICE_LIST if more than one device is specified in + * \a devices and the list of devices specified cannot be used together + * to create a context. + * - CL_DEVICE_NOT_AVAILABLE if a device in \a devices is currently not + * available even though the device was returned by clGetDevices. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_context, clCreateContext, ( + const cl_context_properties *properties, + cl_uint num_devices, + const cl_device_id *devices, + void (CL_CALLBACK * pfn_notify)( + const char *, + const void *, + size_t, + void *), + void *user_data, + cl_int *errcode_ret)) +{ + cl_int errcode; + amd::Context::Info info; + + errcode = amd::Context::checkProperties(properties, &info); + if (CL_SUCCESS != errcode) { + *not_null(errcode_ret) = errcode; + return (cl_context) 0; + } + + if (num_devices == 0 || devices == NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_context) 0; + } + + std::vector devices_; + for (cl_uint i = 0; i < num_devices; ++i) { + // FIXME_lmoriche: Set errcode_ret to CL_DEVICE_NOT_AVAILABLE if a + // device in devices is no longer available. + cl_device_id device = devices[i]; + + if (!is_valid(device)) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + return (cl_context) 0; + } + devices_.push_back(as_amd(device)); + } + + amd::Context* context = new amd::Context(devices_, info); + if (context == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_context) 0; + } + + if (CL_SUCCESS != (errcode = context->create(properties))) { + context->release(); + *not_null(errcode_ret) = errcode; + return (cl_context) 0; + } + + if (amd::Agent::shouldPostContextEvents()) { + amd::Agent::postContextCreate(as_cl(context)); + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(context); +} +RUNTIME_EXIT + +/*! \brief Create an OpenCL context from a device type that identifies the + * specific device(s) to use. + * + * \param properties is reserved and must be zero. + * + * \param device_type is a bit-field that identifies the type of device. + * + * \param pfn_notify described in clCreateContext. + * + * \param user_data described in clCreateContext. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero context and errcode_ret is set to CL_SUCCESS + * if the context is created successfully or NULL with the following error + * values stored in errcode_ret: + * - CL_INVALID_VALUE if \a properties is not zero. + * - CL_INVALID_DEVICE_TYPE if \a device_type is not a valid value. + * - CL_DEVICE_NOT_AVAILABLE if no devices that match \a device_type + * are currently available. + * - CL_DEVICE_NOT_FOUND if no devices that match \a device_type were found. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_context, clCreateContextFromType, ( + const cl_context_properties *properties, + cl_device_type device_type, + void (CL_CALLBACK * pfn_notify)( + const char *, + const void *, + size_t, + void *), + void *user_data, + cl_int *errcode_ret)) +{ + amd::Context::Info info; + cl_int errcode = amd::Context::checkProperties(properties, &info); + if (errcode != CL_SUCCESS) { + *not_null(errcode_ret) = errcode; + return (cl_context) 0; + } + + if (device_type & CL_HSA_ENABLED_AMD && + device_type & CL_HSA_DISABLED_AMD) { + device_type ^= (CL_HSA_ENABLED_AMD | CL_HSA_DISABLED_AMD); + } + + // Get the devices of the given type. + cl_uint num_devices; + bool offlineDevices = + (info.flags_ & amd::Context::OfflineDevices) ? true : false; + if (!amd::Device::getDeviceIDs(device_type, 0, NULL, + &num_devices, offlineDevices)) { + *not_null(errcode_ret) = CL_DEVICE_NOT_FOUND; + return (cl_context) 0; + } + + assert(num_devices > 0 && "Should have returned an error!"); + cl_device_id* devices = (cl_device_id *) + alloca(num_devices * sizeof(cl_device_id)); + + if (!amd::Device::getDeviceIDs(device_type, num_devices, + devices, NULL, offlineDevices)) { + *not_null(errcode_ret) = CL_DEVICE_NOT_FOUND; + return (cl_context) 0; + } + + // Create a new context with the devices + cl_context context = clCreateContext( + properties, num_devices, devices, pfn_notify, user_data, errcode_ret); + + return context; +} +RUNTIME_EXIT + +/*! \brief Increment the context reference count. + * + * \return One of the following values: + * - CL_INVALID_CONTEXT if context is not a valid OpenCL context. + * - CL_SUCCESS if the function is executed successfully. + * + * clCreateContext and clCreateContextFromType perform an implicit retain. + * This is very helpful for 3rd party libraries, which typically get a context + * passed to them by the application. + * However, it is possible that the application may delete the context without + * informing the library. Allowing functions to attach to (i.e. retain) and + * release a context solves the problem of a context being used by a library + * no longer being valid. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainContext, (cl_context context)) +{ + if (!is_valid(context)) { + return CL_INVALID_CONTEXT; + } + as_amd(context)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the context reference count. + * + * \return One of the following values: + * - CL_INVALID_CONTEXT if context is not a valid OpenCL context. + * - CL_SUCCESS if the function is executed successfully. + * + * After the context reference count becomes zero and all the objects attached + * to context (such as memory objects, command-queues) are released, + * the context is deleted. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseContext, (cl_context context)) +{ + if (!is_valid(context)) { + return CL_INVALID_CONTEXT; + } + as_amd(context)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Query information about a context. + * + * \param context specifies the OpenCL context being queried. + * + * \param param_name is an enum that specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size specifies the size in bytes of memory pointed to by + * \a param_value. This size must be greater than or equal to the size of + * return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by \a param_value. If \a param_value_size_ret is NULL, + * it is ignored. + * + * \return One of the following values: + * - CL_INVALID_CONTEXT if context is not a valid context. + * - CL_INVALID_VALUE if \a param_name is not one of the supported values + * or if size in bytes specified by \a param_value_size is < size of return + * type and \a param_value is not a NULL value. + * - CL_SUCCESS if the function is executed successfully. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetContextInfo, ( + cl_context context, + cl_context_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(context)) { + return CL_INVALID_CONTEXT; + } + + switch (param_name) { + case CL_CONTEXT_REFERENCE_COUNT: { + cl_uint count = as_amd(context)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_CONTEXT_NUM_DEVICES: { + cl_uint numDevices = (cl_uint)as_amd(context)->devices().size(); + return amd::clGetInfo( + numDevices, param_value_size, param_value, param_value_size_ret); + } + case CL_CONTEXT_DEVICES: { + const std::vector& devices = as_amd(context)->devices(); + size_t numDevices = devices.size(); + size_t valueSize = numDevices * sizeof(cl_device_id*); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + cl_device_id* device_list = (cl_device_id*) param_value; + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + *device_list++ = const_cast(as_cl(*it)); + } + } + return CL_SUCCESS; + } + case CL_CONTEXT_PROPERTIES: { + const amd::Context* amdContext = as_amd(context); + size_t valueSize = amdContext->info().propertiesSize_; + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if ((param_value != NULL) && (valueSize != 0)) { + ::memcpy(param_value, amdContext->properties(), valueSize); + } + return CL_SUCCESS; + } +#ifdef _WIN32 + case CL_CONTEXT_D3D10_DEVICE_KHR: { + // Not defined in the ext.spec, but tested in the conf.test + // Guessing functionality from the test... + if (param_value != NULL && param_value_size < sizeof(void*)) { + return CL_INVALID_VALUE; + } + const amd::Context* amdContext = as_amd(context); + if (!(amdContext->info().flags_ & amd::Context::D3D10DeviceKhr)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(intptr_t); + if (param_value != NULL) { + *(intptr_t*) param_value = + reinterpret_cast(amdContext->info().hDev_); + } + return CL_SUCCESS; + } + case CL_CONTEXT_D3D10_PREFER_SHARED_RESOURCES_KHR: { + if (param_value != NULL && param_value_size < sizeof(cl_bool)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(cl_bool); + if (param_value != NULL) { + *(cl_bool*) param_value = CL_TRUE; + } + return CL_SUCCESS; + } + case CL_CONTEXT_D3D11_DEVICE_KHR: { + // Not defined in the ext.spec, but tested in the conf.test + // Guessing functionality from the test... + if (param_value != NULL && param_value_size < sizeof(void*)) { + return CL_INVALID_VALUE; + } + const amd::Context* amdContext = as_amd(context); + if (!(amdContext->info().flags_ & amd::Context::D3D11DeviceKhr)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(intptr_t); + if (param_value != NULL) { + *(intptr_t*) param_value = + reinterpret_cast(amdContext->info().hDev_); + } + return CL_SUCCESS; + } + case CL_CONTEXT_D3D11_PREFER_SHARED_RESOURCES_KHR: { + if (param_value != NULL && param_value_size < sizeof(cl_bool)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(cl_bool); + if (param_value != NULL) { + *(cl_bool*) param_value = CL_TRUE; + } + return CL_SUCCESS; + } + case CL_CONTEXT_ADAPTER_D3D9_KHR: { + if (param_value != NULL && param_value_size < sizeof(void*)) { + return CL_INVALID_VALUE; + } + const amd::Context* amdContext = as_amd(context); + if (!(amdContext->info().flags_ & amd::Context::D3D9DeviceKhr)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(intptr_t); + if (param_value != NULL) { + *(intptr_t*) param_value = + reinterpret_cast(amdContext->info().hDev_); + } + return CL_SUCCESS; + } + case CL_CONTEXT_ADAPTER_D3D9EX_KHR: { + if (param_value != NULL && param_value_size < sizeof(void*)) { + return CL_INVALID_VALUE; + } + const amd::Context* amdContext = as_amd(context); + if (!(amdContext->info().flags_ & amd::Context::D3D9DeviceEXKhr)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(intptr_t); + if (param_value != NULL) { + *(intptr_t*) param_value = + reinterpret_cast(amdContext->info().hDev_); + } + return CL_SUCCESS; + } + case CL_CONTEXT_ADAPTER_DXVA_KHR: { + if (param_value != NULL && param_value_size < sizeof(void*)) { + return CL_INVALID_VALUE; + } + const amd::Context* amdContext = as_amd(context); + if (!(amdContext->info().flags_ & amd::Context::D3D9DeviceVAKhr)) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = sizeof(intptr_t); + if (param_value != NULL) { + *(intptr_t*) param_value = + reinterpret_cast(amdContext->info().hDev_); + } + return CL_SUCCESS; + } +#endif //_WIN32 + default: + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! \brief returns the address of the extension function named by + * funcname for a given platform. The pointer returned should be cast + * to a function pointer type matching the extension function’s definition + * defined in the appropriate extension specification and header file. + * A return value of NULL indicates that the specified function does not + * exist for the implementation or platform is not a valid platform. + * A non-NULL return value for \a clGetExtensionFunctionAddressForPlatform + * does not guarantee that an extension function is actually supported by + * the platform. The application must also make a corresponding query using + * \a clGetPlatformInfo(platform, CL_PLATFORM_EXTENSIONS, ... ) or + * \a clGetDeviceInfo(device, CL_DEVICE_EXTENSIONS, ... ) to determine if + * an extension is supported by the OpenCL implementation. + * + * \version 1.2r07 + */ +CL_API_ENTRY void* CL_API_CALL clGetExtensionFunctionAddressForPlatform( + cl_platform_id platform, + const char *funcname) +{ + if (platform != NULL && platform != AMD_PLATFORM) { + return NULL; + } + + return clGetExtensionFunctionAddress(funcname); +} + +CL_API_ENTRY void* CL_API_CALL +clGetExtensionFunctionAddress(const char* func_name) +{ +#define CL_EXTENSION_ENTRYPOINT_CHECK(name) \ + if (!strcmp(func_name, #name)) return reinterpret_cast(name); +#define CL_EXTENSION_ENTRYPOINT_CHECK2(name1,name2) \ + if (!strcmp(func_name, #name1)) return reinterpret_cast(name2); + + switch (func_name[2]) { + case 'C': + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateEventFromGLsyncKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreatePerfCounterAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateThreadTraceAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromGLBuffer); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromGLTexture2D); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromGLTexture3D); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromGLRenderbuffer); +#ifdef _WIN32 + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromD3D10BufferKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromD3D10Texture2DKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromD3D10Texture3DKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateFromDX9MediaSurfaceKHR); +#endif //_WIN32 +#ifdef cl_ext_device_fission + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateSubDevicesEXT); +#endif // cl_ext_device_fission +#if cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateVideoSessionAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateVideoEncSessionAMD); +#endif // cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateKeyAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clConvertImageAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clCreateBufferFromImageAMD); + break; + case 'D': +#if cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clDestroyVideoEncSessionAMD); +#endif // cl_amd_open_video + break; + case 'E': + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueBeginPerfCounterAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueEndPerfCounterAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueAcquireGLObjects); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueReleaseGLObjects); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueBindThreadTraceBufferAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueThreadTraceCommandAMD); +#ifdef _WIN32 + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueAcquireD3D10ObjectsKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueReleaseD3D10ObjectsKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueAcquireDX9MediaSurfacesKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueReleaseDX9MediaSurfacesKHR); +#endif //_WIN32 +#if cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueRunVideoProgramAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEncodeGetDeviceCapAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEncodePictureAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEncodeQueryTaskDescriptionAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEncodeReleaseOutputResourceAMD); +#endif // cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueWaitSignalAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueWriteSignalAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clEnqueueMakeBuffersResidentAMD); + break; + case 'G': + CL_EXTENSION_ENTRYPOINT_CHECK(clGetKernelInfoAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetPerfCounterInfoAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetGLObjectInfo); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetGLTextureInfo); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetGLContextInfoKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetThreadTraceInfoAMD); +#if cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clGetVideoSessionInfoAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetVideoSessionEncInfoAMD); +#endif // cl_amd_open_video +#ifdef _WIN32 + CL_EXTENSION_ENTRYPOINT_CHECK(clGetDeviceIDsFromD3D10KHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetDeviceIDsFromDX9MediaAdapterKHR); + CL_EXTENSION_ENTRYPOINT_CHECK(clGetPlaneFromImageAMD); +#endif //_WIN32 + CL_EXTENSION_ENTRYPOINT_CHECK(clGetKernelSubGroupInfoKHR); + break; + case 'I': + CL_EXTENSION_ENTRYPOINT_CHECK(clIcdGetPlatformIDsKHR); + break; + case 'O': + CL_EXTENSION_ENTRYPOINT_CHECK(clObjectGetValueForKeyAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clObjectSetValueForKeyAMD); + break; + case 'R': + CL_EXTENSION_ENTRYPOINT_CHECK(clReleasePerfCounterAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clRetainPerfCounterAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clReleaseThreadTraceAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clRetainThreadTraceAMD); +#ifdef cl_ext_device_fission + CL_EXTENSION_ENTRYPOINT_CHECK(clRetainDeviceEXT); + CL_EXTENSION_ENTRYPOINT_CHECK(clReleaseDeviceEXT); +#endif // cl_ext_device_fission +#if cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clRetainVideoSessionAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clReleaseVideoSessionAMD); +#endif // cl_amd_open_video + break; + case 'S': +#if cl_amd_open_video + CL_EXTENSION_ENTRYPOINT_CHECK(clSendEncodeConfigInfoAMD); + CL_EXTENSION_ENTRYPOINT_CHECK(clSetThreadTraceParamAMD); +#endif // cl_amd_open_video + break; + case 'U': + CL_EXTENSION_ENTRYPOINT_CHECK(clUnloadPlatformAMD); + default: + break; + } + + return NULL; +} + +RUNTIME_ENTRY(cl_int, clTerminateContextKHR, (cl_context context)) +{ + return CL_INVALID_CONTEXT; +} +RUNTIME_EXIT + + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_counter.cpp b/opencl/api/opencl/amdocl/cl_counter.cpp new file mode 100644 index 0000000000..3153055f60 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_counter.cpp @@ -0,0 +1,130 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" +#include + +#include "platform/object.hpp" +#include "platform/context.hpp" +#include "platform/command.hpp" +#include "platform/counter.hpp" + +#ifdef cl_amd_atomic_counters + +/*! \addtogroup API + * @{ + * \addtogroup CL_Counters + * + * Counter objects ... + * + * @{ + */ + +/*! \brief + * + * \version 1.1r18 + */ +RUNTIME_ENTRY_RET(cl_counter_amd, clCreateCounterAMD, ( + cl_context context, + cl_counter_flags_amd flags, + cl_uint value, + cl_int * errcode_ret)) +{ + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return (cl_counter_amd)0; +} +RUNTIME_EXIT + +/*! \brief + * + * \version 1.1r18 + */ +RUNTIME_ENTRY(cl_int, clGetCounterInfoAMD, ( + cl_counter_amd counter, + cl_counter_info_amd param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret)) +{ + return CL_INVALID_COUNTER_AMD; +} +RUNTIME_EXIT + +/*! \brief Increment the counter reference count. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_COUNTER if \a counter is not a valid counter object. + * + * The OpenCL commands that return a counter perform an implicit retain. + * + * \version 1.1r18 + */ +RUNTIME_ENTRY(cl_int, clRetainCounterAMD, (cl_counter_amd counter)) +{ + if (!is_valid(counter)) { + return CL_INVALID_COUNTER_AMD; + } + as_amd(counter)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the counter reference count. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_EVENT if \a counter is not a valid counter object. + * + * The counter object is deleted once the reference count becomes zero. + * + * \version 1.1r18 + */ +RUNTIME_ENTRY(cl_int, clReleaseCounterAMD, (cl_counter_amd counter)) +{ + if (!is_valid(counter)) { + return CL_INVALID_COUNTER_AMD; + } + as_amd(counter)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief + * + * \version 1.1r18 + */ +RUNTIME_ENTRY(cl_int, clEnqueueReadCounterAMD, ( + cl_command_queue command_queue, + cl_counter_amd counter, + cl_bool blocking_read, + cl_uint * value, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event)) +{ + return CL_INVALID_COUNTER_AMD; +} +RUNTIME_EXIT + +/*! \brief + * + * \version 1.1r18 + */ +RUNTIME_ENTRY(cl_int, clEnqueueWriteCounterAMD, ( + cl_command_queue command_queue, + cl_counter_amd counter, + cl_bool blocking_write, + cl_uint value, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event)) +{ + return CL_INVALID_COUNTER_AMD; +} +RUNTIME_EXIT + +/*! @} + * @} + */ + +#endif // cl_amd_atomic_counters diff --git a/opencl/api/opencl/amdocl/cl_d3d10.cpp b/opencl/api/opencl/amdocl/cl_d3d10.cpp new file mode 100644 index 0000000000..389a0d88b6 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_d3d10.cpp @@ -0,0 +1,1900 @@ +// +// Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifdef _WIN32 + +#include +#include + +#include "cl_common.hpp" +#include "cl_d3d10_amd.hpp" +#include "runtime/platform/command.hpp" + +#include +#include + + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_D3D10_Interops + * + * This section discusses OpenCL functions that allow applications to use Direct3D 10 + * resources (buffers/textures) as OpenCL memory objects. This allows efficient sharing of + * data between OpenCL and Direct3D 10. The OpenCL API can be used to execute kernels that + * read and/or write memory objects that are also the Direct3D resources. + * An OpenCL image object can be created from a D3D10 texture object. An + * OpenCL buffer object can be created from a D3D10 buffer object (index/vertex). + * + * @} + * \addtogroup clGetDeviceIDsFromD3D10KHR + * @{ + */ + +RUNTIME_ENTRY(cl_int, clGetDeviceIDsFromD3D10KHR, ( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices)) +{ + cl_int errcode; + ID3D10Device* d3d10_device = NULL; + cl_device_id* gpu_devices; + cl_uint num_gpu_devices = 0; + bool create_d3d10Device = false; + static const bool VALIDATE_ONLY = true; + HMODULE d3d10Module = NULL; + + if (platform != NULL && platform != AMD_PLATFORM) { + LogWarning("\"platrform\" is not a valid AMD platform"); + return CL_INVALID_PLATFORM; + } + if(((num_entries > 0 || num_devices == NULL) && devices == NULL) + || (num_entries == 0 && devices != NULL)) { + return CL_INVALID_VALUE; + } + // Get GPU devices + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, 0, NULL, &num_gpu_devices); + if (errcode != CL_SUCCESS && errcode != CL_DEVICE_NOT_FOUND) { + return CL_INVALID_VALUE; + } + + if (!num_gpu_devices) { + *not_null(num_devices) = 0; + return CL_DEVICE_NOT_FOUND; + } + + switch(d3d_device_source) + { + case CL_D3D10_DEVICE_KHR: + d3d10_device = static_cast(d3d_object); + break; + case CL_D3D10_DXGI_ADAPTER_KHR: + { + typedef HRESULT (WINAPI* LPD3D10CREATEDEVICE)(IDXGIAdapter*, D3D10_DRIVER_TYPE, + HMODULE, UINT, UINT32, ID3D10Device**); + static LPD3D10CREATEDEVICE dynamicD3D10CreateDevice = NULL; + + d3d10Module = LoadLibrary("D3D10.dll"); + if (d3d10Module == NULL) { + return CL_INVALID_PLATFORM; + } + + dynamicD3D10CreateDevice = (LPD3D10CREATEDEVICE)GetProcAddress(d3d10Module, "D3D10CreateDevice"); + + IDXGIAdapter* dxgi_adapter = static_cast(d3d_object); + HRESULT hr = dynamicD3D10CreateDevice(dxgi_adapter, D3D10_DRIVER_TYPE_HARDWARE, + NULL, 0, D3D10_SDK_VERSION, &d3d10_device); + if (SUCCEEDED(hr) && (NULL != d3d10_device)) { + create_d3d10Device = true; + } else { + FreeLibrary(d3d10Module); + return CL_INVALID_VALUE; + } + } + break; + default: + LogWarning("\"d3d_device_source\" is invalid"); + return CL_INVALID_VALUE; + } + + switch(d3d_device_set) { + case CL_PREFERRED_DEVICES_FOR_D3D10_KHR: + case CL_ALL_DEVICES_FOR_D3D10_KHR: + { + gpu_devices = (cl_device_id *) alloca(num_gpu_devices * sizeof(cl_device_id)); + + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, num_gpu_devices, gpu_devices, NULL); + if (errcode != CL_SUCCESS) { + break; + } + + std::vector compatible_devices; + for (cl_uint i = 0; i < num_gpu_devices; ++i) { + cl_device_id device = gpu_devices[i]; + if (is_valid(device) && + as_amd(device)->bindExternalDevice(CL_CONTEXT_D3D10_DEVICE_KHR, d3d10_device, NULL, VALIDATE_ONLY)) { + compatible_devices.push_back(as_amd(device)); + } + } + if (compatible_devices.size() == 0) { + *not_null(num_devices) = 0; + errcode = CL_DEVICE_NOT_FOUND; + break; + } + + std::vector::iterator it = compatible_devices.begin(); + cl_uint compatible_count = std::min(num_entries, (cl_uint)compatible_devices.size()); + + while (compatible_count--) { + *devices++ = as_cl(*it++); + --num_entries; + } + while (num_entries--) { + *devices++ = (cl_device_id) 0; + } + + *not_null(num_devices) = (cl_uint)compatible_devices.size(); + } + break; + + default: + LogWarning("\"d3d_device_set\" is invalid"); + errcode = CL_INVALID_VALUE; + } + + if (create_d3d10Device) { + d3d10_device->Release(); + FreeLibrary(d3d10Module); + } + return errcode; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromD3D10BufferKHR + * @{ + */ + +/*! \brief Creates an OpenCL buffer object from a Direct3D 10 resource. + * + * \param context is a valid OpenCL context. + * + * \param flags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param pD3DResource is a valid pointer to a D3D10 resource of type ID3D10Buffer. + * + * \return valid non-zero OpenCL buffer object and \a errcode_ret is set + * to CL_SUCCESS if the buffer object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context or if Direct3D 10 + * interoperatbility has not been initialized between context and the ID3D10Device + * from which pD3DResource was created. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_D3D_RESOURCE if \a pD3DResource is not of type ID3D10Buffer. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33? + */ + +RUNTIME_ENTRY_RET(cl_mem, clCreateFromD3D10BufferKHR, ( + cl_context context, + cl_mem_flags flags, + ID3D10Buffer* pD3DResource, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + if(!flags) flags = CL_MEM_READ_WRITE; + if(!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + if(!pD3DResource) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("parameter \"pD3DResource\" is a NULL pointer"); + return clMemObj; + } + return(amd::clCreateBufferFromD3D10ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + errcode_ret)); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateImageFromD3D10Resource + * @{ + */ + +/*! \brief Create an OpenCL 2D or 3D image object from a D3D10 resource. + * + * \param context is a valid OpenCL context. + * + * \param flags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param pD3DResource is a valid pointer to a D3D10 resource of type + * ID3D10Texture2D, ID3D10Texture2D, or ID3D10Texture3D. + * If pD3DResource is of type ID3D10Texture1D then the created image object + * will be a 1D mipmapped image object. + * If pD3DResource is of type ID3D10Texture2D and was not created with flag + * D3D10_RESOURCE_MISC_TEXTURECUBE then the created image object will be a + * 2D mipmapped image object. + * If pD3DResource is of type ID3D10Texture2D and was created with flag + * D3D10_RESOURCE_MISC_TEXTURECUBE then the created image object will be + * a cubemap mipmapped image object. + * errocde_ret returns CL_INVALID_D3D_RESOURCE if an OpenCL memory object has + * already been created from pD3DResource in context. + * If pD3DResource is of type ID3D10Texture3D then the created image object will + * be a 3D mipmapped imageobject. + * + * \return valid non-zero OpenCL image object and \a errcode_ret is set + * to CL_SUCCESS if the image object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context or if Direct3D 10 + * interoperatbility has not been initialized between context and the ID3D10Device + * from which pD3DResource was created. + * - CL_INVALID_VALUE if values specified in \a flags are not valid. + * - CL_INVALID_D3D_RESOURCE if \a pD3DResource is not of type ID3D10Texture1D, + * ID3D10Texture2D, or ID3D10Texture3D. + * - CL_INVALID_D3D_RESOURCE if an OpenCL memory object has already been created + * from \a pD3DResource in context. + * - CL_INVALID_IMAGE_FORMAT if the Direct3D 10 texture format does not map + * to an appropriate OpenCL image format. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r48? + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateImageFromD3D10Resource, ( + cl_context context, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret, + UINT dimension)) +{ + cl_mem clMemObj = NULL; + + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + if(!flags) flags = CL_MEM_READ_WRITE; + if(!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + if(!pD3DResource) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("parameter \"pD3DResource\" is a NULL pointer"); + return clMemObj; + } + + // Verify context init'ed for interop + ID3D10Device* pDev; + pD3DResource->GetDevice(&pDev); + if(pDev == NULL) { + *not_null(errcode_ret) = CL_INVALID_D3D10_DEVICE_KHR; + LogWarning("Cannot retrieve D3D10 device from D3D10 resource"); + return (cl_mem) 0; + } + pDev->Release(); + if (!((*as_amd(context)).info().flags_ & amd::Context::D3D10DeviceKhr)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from D3D10 device"); + return (cl_mem) 0; + } + + // Check for image support + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if((*it)->info().imageSupport_) { + supportPass = true; + } + } + if(!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return (cl_mem) 0; + } + + switch(dimension) { +#if 0 + case 1: + return(amd::clCreateImage1DFromD3D10ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + subresource, + errcode_ret)); +#endif //0 + case 2: + return(amd::clCreateImage2DFromD3D10ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + subresource, + errcode_ret)); + case 3: + return(amd::clCreateImage3DFromD3D10ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + subresource, + errcode_ret)); + default: + break; + } + + *not_null(errcode_ret) = CL_INVALID_D3D10_RESOURCE_KHR; + return (cl_mem) 0; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromD3D10Texture2DKHR + * @{ + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromD3D10Texture2DKHR, ( + cl_context context, + cl_mem_flags flags, + ID3D10Texture2D* resource, + UINT subresource, + cl_int* errcode_ret)) +{ + return clCreateImageFromD3D10Resource(context, flags, resource, + subresource, errcode_ret, 2); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromD3D10Texture3DKHR + * @{ + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromD3D10Texture3DKHR, ( + cl_context context, + cl_mem_flags flags, + ID3D10Texture3D* resource, + UINT subresource, + cl_int* errcode_ret)) +{ + return clCreateImageFromD3D10Resource(context, flags, resource, + subresource, errcode_ret, 3); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clEnqueueAcquireD3D10ObjectsKHR + * @{ + */ +RUNTIME_ENTRY(cl_int, clEnqueueAcquireD3D10ObjectsKHR, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + return amd::clEnqueueAcquireExtObjectsAMD(command_queue, num_objects, + mem_objects, num_events_in_wait_list, event_wait_list, event, + CL_COMMAND_ACQUIRE_D3D10_OBJECTS_KHR); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clEnqueueReleaseD3D10ObjectsKHR + * @{ + */ +RUNTIME_ENTRY(cl_int, clEnqueueReleaseD3D10ObjectsKHR, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + return amd::clEnqueueReleaseExtObjectsAMD(command_queue, num_objects, + mem_objects, num_events_in_wait_list, event_wait_list, event, + CL_COMMAND_RELEASE_D3D10_OBJECTS_KHR); +} +RUNTIME_EXIT + + +// +// +// namespace amd +// +// +namespace amd +{ +/*! @} + * \addtogroup CL-D3D10 interop helper functions + * @{ + */ + + + +//******************************************************************* +// +// Internal implementation of CL API functions +// +//******************************************************************* +// +// clCreateBufferFromD3D10ResourceAMD +// +cl_mem clCreateBufferFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + int* errcode_ret) +{ + // Verify pD3DResource is a buffer + D3D10_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D10_RESOURCE_DIMENSION_BUFFER) { + *not_null(errcode_ret) = CL_INVALID_D3D10_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D10Object obj; + int errcode = D3D10Object::initD3D10Object(amdContext, pD3DResource, 0, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + BufferD3D10 *pBufferD3D10 = new (amdContext) + BufferD3D10(amdContext, flags, obj); + if(!pBufferD3D10) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pBufferD3D10->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pBufferD3D10->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pBufferD3D10); +} +#if 0 +// There is no support for 1D images in the base imagee code +// +// clCreateImage1DFromD3D10ResourceAMD +// +cl_mem clCreateImage1DFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret) +{ + + // Verify the resource is a 1D texture + D3D10_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D10_RESOURCE_DIMENSION_TEXTURE1D) { + *not_null(errcode_ret) = CL_INVALID_D3D10_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D10Object obj; + int errcode = D3D10Object::initD3D10Object(pD3DResource, subresource, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + Image1DD3D10 *pImage1DD3D10 = new Image1DD3D10(amdContext, flags, obj); + if(!pImage1DD3D10) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage1DD3D10->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage1DD3D10->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage1DD3D10); +} +#endif + +// +// clCreateImage2DFromD3D10ResourceAMD +// +cl_mem clCreateImage2DFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret) +{ + // Verify the resource is a 2D texture + D3D10_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D10_RESOURCE_DIMENSION_TEXTURE2D) { + *not_null(errcode_ret) = CL_INVALID_D3D10_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D10Object obj; + int errcode = D3D10Object::initD3D10Object(amdContext, pD3DResource, subresource, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + Image2DD3D10 *pImage2DD3D10 = new (amdContext) + Image2DD3D10(amdContext, flags, obj); + if(!pImage2DD3D10) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage2DD3D10->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage2DD3D10->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage2DD3D10); +} + +// +// clCreateImage2DFromD3D10ResourceAMD +// +cl_mem clCreateImage3DFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret) +{ + // Verify the resource is a 2D texture + D3D10_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D10_RESOURCE_DIMENSION_TEXTURE3D) { + *not_null(errcode_ret) = CL_INVALID_D3D10_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D10Object obj; + int errcode = D3D10Object::initD3D10Object(amdContext, pD3DResource, subresource, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + Image3DD3D10 *pImage3DD3D10 = new (amdContext) + Image3DD3D10(amdContext, flags, obj); + if(!pImage3DD3D10) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage3DD3D10->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage3DD3D10->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage3DD3D10); +} + +// +// Helper function SyncD3D10Objects +// +void SyncD3D10Objects(std::vector& memObjects) +{ + Memory*& mem = memObjects.front(); + if(!mem) { + LogWarning("\nNULL memory object\n"); + return; + } + InteropObject* interop = mem->getInteropObj(); + if(!interop) { + LogWarning("\nNULL interop object\n"); + return; + } + D3D10Object* d3d10Obj = interop->asD3D10Object(); + if(!d3d10Obj) { + LogWarning("\nNULL D3D10 object\n"); + return; + } + ID3D10Query* query = d3d10Obj->getQuery(); + if(!query) { + LogWarning("\nNULL ID3D10Query\n"); + return; + } + query->End(); + BOOL data; + while(S_OK != query->GetData(&data, sizeof(BOOL), 0) && data != TRUE) + { + } +} + +// +// Class D3D10Object implementation +// +size_t +D3D10Object::getElementBytes(DXGI_FORMAT dxgiFmt) +{ + size_t bytesPerPixel; + + switch(dxgiFmt) + { + case DXGI_FORMAT_R32G32B32A32_TYPELESS: + case DXGI_FORMAT_R32G32B32A32_FLOAT: + case DXGI_FORMAT_R32G32B32A32_UINT: + case DXGI_FORMAT_R32G32B32A32_SINT: + bytesPerPixel = 16; + break; + + case DXGI_FORMAT_R32G32B32_TYPELESS: + case DXGI_FORMAT_R32G32B32_FLOAT: + case DXGI_FORMAT_R32G32B32_UINT: + case DXGI_FORMAT_R32G32B32_SINT: + bytesPerPixel = 12; + break; + + case DXGI_FORMAT_R16G16B16A16_TYPELESS: + case DXGI_FORMAT_R16G16B16A16_FLOAT: + case DXGI_FORMAT_R16G16B16A16_UNORM: + case DXGI_FORMAT_R16G16B16A16_UINT: + case DXGI_FORMAT_R16G16B16A16_SNORM: + case DXGI_FORMAT_R16G16B16A16_SINT: + case DXGI_FORMAT_R32G32_TYPELESS: + case DXGI_FORMAT_R32G32_FLOAT: + case DXGI_FORMAT_R32G32_UINT: + case DXGI_FORMAT_R32G32_SINT: + case DXGI_FORMAT_R32G8X24_TYPELESS: + case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: + case DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS: + case DXGI_FORMAT_X32_TYPELESS_G8X24_UINT: + bytesPerPixel = 8; + break; + + case DXGI_FORMAT_R10G10B10A2_TYPELESS: + case DXGI_FORMAT_R10G10B10A2_UNORM: + case DXGI_FORMAT_R10G10B10A2_UINT: + case DXGI_FORMAT_R11G11B10_FLOAT: + case DXGI_FORMAT_R8G8B8A8_TYPELESS: + case DXGI_FORMAT_R8G8B8A8_UNORM: + case DXGI_FORMAT_R8G8B8A8_UNORM_SRGB: + case DXGI_FORMAT_R8G8B8A8_UINT: + case DXGI_FORMAT_R8G8B8A8_SNORM: + case DXGI_FORMAT_R8G8B8A8_SINT: + case DXGI_FORMAT_R16G16_TYPELESS: + case DXGI_FORMAT_R16G16_FLOAT: + case DXGI_FORMAT_R16G16_UNORM: + case DXGI_FORMAT_R16G16_UINT: + case DXGI_FORMAT_R16G16_SNORM: + case DXGI_FORMAT_R16G16_SINT: + case DXGI_FORMAT_R32_TYPELESS: + case DXGI_FORMAT_D32_FLOAT: + case DXGI_FORMAT_R32_FLOAT: + case DXGI_FORMAT_R32_UINT: + case DXGI_FORMAT_R32_SINT: + case DXGI_FORMAT_R24G8_TYPELESS: + case DXGI_FORMAT_D24_UNORM_S8_UINT: + case DXGI_FORMAT_R24_UNORM_X8_TYPELESS: + case DXGI_FORMAT_X24_TYPELESS_G8_UINT: + + case DXGI_FORMAT_R9G9B9E5_SHAREDEXP: + case DXGI_FORMAT_R8G8_B8G8_UNORM: + case DXGI_FORMAT_G8R8_G8B8_UNORM: + + case DXGI_FORMAT_B8G8R8A8_UNORM: + case DXGI_FORMAT_B8G8R8X8_UNORM: + bytesPerPixel = 4; + break; + + case DXGI_FORMAT_R8G8_TYPELESS: + case DXGI_FORMAT_R8G8_UNORM: + case DXGI_FORMAT_R8G8_UINT: + case DXGI_FORMAT_R8G8_SNORM: + case DXGI_FORMAT_R8G8_SINT: + case DXGI_FORMAT_R16_TYPELESS: + case DXGI_FORMAT_R16_FLOAT: + case DXGI_FORMAT_D16_UNORM: + case DXGI_FORMAT_R16_UNORM: + case DXGI_FORMAT_R16_UINT: + case DXGI_FORMAT_R16_SNORM: + case DXGI_FORMAT_R16_SINT: + + case DXGI_FORMAT_B5G6R5_UNORM: + case DXGI_FORMAT_B5G5R5A1_UNORM: + bytesPerPixel = 2; + break; + + case DXGI_FORMAT_R8_TYPELESS: + case DXGI_FORMAT_R8_UNORM: + case DXGI_FORMAT_R8_UINT: + case DXGI_FORMAT_R8_SNORM: + case DXGI_FORMAT_R8_SINT: + case DXGI_FORMAT_A8_UNORM: + case DXGI_FORMAT_R1_UNORM: + bytesPerPixel = 1; + break; + + + case DXGI_FORMAT_BC1_TYPELESS: + case DXGI_FORMAT_BC1_UNORM: + case DXGI_FORMAT_BC1_UNORM_SRGB: + case DXGI_FORMAT_BC2_TYPELESS: + case DXGI_FORMAT_BC2_UNORM: + case DXGI_FORMAT_BC2_UNORM_SRGB: + case DXGI_FORMAT_BC3_TYPELESS: + case DXGI_FORMAT_BC3_UNORM: + case DXGI_FORMAT_BC3_UNORM_SRGB: + case DXGI_FORMAT_BC4_TYPELESS: + case DXGI_FORMAT_BC4_UNORM: + case DXGI_FORMAT_BC4_SNORM: + case DXGI_FORMAT_BC5_TYPELESS: + case DXGI_FORMAT_BC5_UNORM: + case DXGI_FORMAT_BC5_SNORM: + // Less than 1 byte per pixel - needs special consideration + bytesPerPixel = 0; + break; + + default: + bytesPerPixel = 0; + _ASSERT(FALSE); + break; + } + return bytesPerPixel; +} + +cl_image_format +D3D10Object::getCLFormatFromDXGI(DXGI_FORMAT dxgiFmt) +{ + cl_image_format fmt; + + //! @todo [odintsov]: add real fmt conversion from DXGI to CL + fmt.image_channel_order = 0;//CL_RGBA; + fmt.image_channel_data_type = 0;//CL_UNSIGNED_INT8; + + switch(dxgiFmt) + { + case DXGI_FORMAT_R32G32B32A32_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R32G32B32A32_FLOAT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32G32B32A32_UINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32B32A32_SINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32B32_TYPELESS: + fmt.image_channel_order = CL_RGB; + break; + + case DXGI_FORMAT_R32G32B32_FLOAT: + fmt.image_channel_order = CL_RGB; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32G32B32_UINT: + fmt.image_channel_order = CL_RGB; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32B32_SINT: + fmt.image_channel_order = CL_RGB; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R16G16B16A16_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R16G16B16A16_FLOAT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case DXGI_FORMAT_R16G16B16A16_UNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16G16B16A16_UINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNSIGNED_INT16; + break; + + case DXGI_FORMAT_R16G16B16A16_SNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SNORM_INT16; + break; + + case DXGI_FORMAT_R16G16B16A16_SINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SIGNED_INT16; + break; + + case DXGI_FORMAT_R32G32_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_R32G32_FLOAT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32G32_UINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32_SINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R32G8X24_TYPELESS: + break; + + case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: + break; + + case DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS: + break; + + case DXGI_FORMAT_X32_TYPELESS_G8X24_UINT: + break; + + case DXGI_FORMAT_R10G10B10A2_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R10G10B10A2_UNORM: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R10G10B10A2_UINT: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R11G11B10_FLOAT: + fmt.image_channel_order = CL_RGB; + break; + + case DXGI_FORMAT_R8G8B8A8_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R8G8B8A8_UNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_UNORM_SRGB: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_UINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_SNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SNORM_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_SINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SIGNED_INT8; + break; + + case DXGI_FORMAT_R16G16_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_R16G16_FLOAT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case DXGI_FORMAT_R16G16_UNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16G16_UINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNSIGNED_INT16; + break; + + case DXGI_FORMAT_R16G16_SNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SNORM_INT16; + break; + + case DXGI_FORMAT_R16G16_SINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SIGNED_INT16; + break; + + case DXGI_FORMAT_R32_TYPELESS: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_D32_FLOAT: + break; + + case DXGI_FORMAT_R32_FLOAT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32_UINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32_SINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R24G8_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_D24_UNORM_S8_UINT: + break; + + case DXGI_FORMAT_R24_UNORM_X8_TYPELESS: + break; + + case DXGI_FORMAT_X24_TYPELESS_G8_UINT: + break; + + case DXGI_FORMAT_R9G9B9E5_SHAREDEXP: + break; + + case DXGI_FORMAT_R8G8_B8G8_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_G8R8_G8B8_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_B8G8R8A8_UNORM: + fmt.image_channel_order = CL_BGRA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_B8G8R8X8_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_R8G8_UNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8_UINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + break; + + case DXGI_FORMAT_R8G8_SNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SNORM_INT8; + break; + + case DXGI_FORMAT_R8G8_SINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SIGNED_INT8; + break; + + case DXGI_FORMAT_R16_TYPELESS: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_R16_FLOAT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case DXGI_FORMAT_D16_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16_UNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16_UINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT16; + break; + + case DXGI_FORMAT_R16_SNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SNORM_INT16; + break; + + case DXGI_FORMAT_R16_SINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SIGNED_INT16; + break; + + case DXGI_FORMAT_B5G6R5_UNORM: + fmt.image_channel_data_type = CL_UNORM_SHORT_565; + break; + + case DXGI_FORMAT_B5G5R5A1_UNORM: + fmt.image_channel_order = CL_BGRA; + break; + + case DXGI_FORMAT_R8_TYPELESS: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_R8_UNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8_UINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + break; + + case DXGI_FORMAT_R8_SNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SNORM_INT8; + break; + + case DXGI_FORMAT_R8_SINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SIGNED_INT8; + break; + + case DXGI_FORMAT_A8_UNORM: + fmt.image_channel_order = CL_A; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R1_UNORM: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_BC1_TYPELESS: + case DXGI_FORMAT_BC1_UNORM: + case DXGI_FORMAT_BC1_UNORM_SRGB: + case DXGI_FORMAT_BC2_TYPELESS: + case DXGI_FORMAT_BC2_UNORM: + case DXGI_FORMAT_BC2_UNORM_SRGB: + case DXGI_FORMAT_BC3_TYPELESS: + case DXGI_FORMAT_BC3_UNORM: + case DXGI_FORMAT_BC3_UNORM_SRGB: + case DXGI_FORMAT_BC4_TYPELESS: + case DXGI_FORMAT_BC4_UNORM: + case DXGI_FORMAT_BC4_SNORM: + case DXGI_FORMAT_BC5_TYPELESS: + case DXGI_FORMAT_BC5_UNORM: + case DXGI_FORMAT_BC5_SNORM: + break; + + default: + _ASSERT(FALSE); + break; + } + + return fmt; +} + +size_t +D3D10Object::getResourceByteSize() +{ + size_t bytes = 1; + + //! @todo [odintsov]: take into consideration the mip level?! + + switch(objDesc_.objDim_) + { + case D3D10_RESOURCE_DIMENSION_BUFFER: + bytes = objDesc_.objSize_.ByteWidth; + break; + + case D3D10_RESOURCE_DIMENSION_TEXTURE3D: + bytes = objDesc_.objSize_.Depth; + + case D3D10_RESOURCE_DIMENSION_TEXTURE2D: + bytes *= objDesc_.objSize_.Height; + + case D3D10_RESOURCE_DIMENSION_TEXTURE1D: + bytes *= objDesc_.objSize_.Width * getElementBytes(); + break; + + default: + LogError("getResourceByteSize: unknown type of D3D10 resource"); + bytes = 0; + break; + } + return bytes; +} + +int +D3D10Object::initD3D10Object(const Context& amdContext, ID3D10Resource* pRes, UINT subres, D3D10Object& obj) +{ + ID3D10Device *pDev; + HRESULT hr; + ScopedLock sl(resLock_); + + // Check if this ressource has already been used for interop + std::vector>::iterator it; + for(it = resources_.begin(); it != resources_.end(); ++it) { + if((*it).first == (void*) pRes && (*it).second == subres) { + return CL_INVALID_D3D10_RESOURCE_KHR; + } + } + + (obj.pD3D10Res_ = pRes)->GetDevice(&pDev); + + if(!pDev) { + return CL_INVALID_D3D10_DEVICE_KHR; + } + + D3D10_QUERY_DESC desc = {D3D10_QUERY_EVENT, 0}; \ + pDev->CreateQuery(&desc, &obj.pQuery_); \ + +#define SET_SHARED_FLAGS() \ + { \ + obj.pD3D10ResOrig_ = obj.pD3D10Res_; \ + memcpy(&obj.objDescOrig_, &obj.objDesc_, sizeof(D3D10ObjDesc_t)); \ + /* @todo - Check device type and select right usage for resource */ \ + /* For now get only DPU path, CPU path for buffers */ \ + /* will not worl on DEFAUL resources */ \ + /*desc.Usage = D3D10_USAGE_STAGING;*/ \ + desc.Usage = D3D10_USAGE_DEFAULT; \ + desc.MiscFlags = D3D10_RESOURCE_MISC_SHARED; \ + desc.CPUAccessFlags = 0; \ + } + +#define STORE_SHARED_FLAGS(restype) \ + { \ + if(S_OK == hr && obj.pD3D10Res_) { \ + obj.objDesc_.objFlags_.d3d10Usage_ = desc.Usage; \ + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; \ + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; \ + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; \ + } \ + else { \ + LogError("\nCannot create shared " #restype "\n"); \ + return CL_INVALID_D3D10_RESOURCE_KHR; \ + } \ + } + +#define SET_BINDING() \ + { \ + switch(desc.Format) { \ + case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: \ + case DXGI_FORMAT_D32_FLOAT: \ + case DXGI_FORMAT_D24_UNORM_S8_UINT: \ + case DXGI_FORMAT_D16_UNORM: \ + desc.BindFlags = D3D10_BIND_DEPTH_STENCIL; \ + break; \ + default: \ + desc.BindFlags = D3D10_BIND_SHADER_RESOURCE | D3D10_BIND_RENDER_TARGET; \ + break; \ + } \ + } + + pRes->GetType(&obj.objDesc_.objDim_); + + // Init defaults + obj.objDesc_.objSize_.Height = 1; + obj.objDesc_.objSize_.Depth = 1; + obj.objDesc_.mipLevels_ = 1; + obj.objDesc_.arraySize_ = 1; + obj.objDesc_.dxgiFormat_ = DXGI_FORMAT_UNKNOWN; + obj.objDesc_.dxgiSampleDesc_ = dxgiSampleDescDefault; + + switch(obj.objDesc_.objDim_) { + case D3D10_RESOURCE_DIMENSION_BUFFER: // = 1, + { + D3D10_BUFFER_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + obj.objDesc_.objSize_.ByteWidth = desc.ByteWidth; + obj.objDesc_.objFlags_.d3d10Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + // Handle D3D10Buffer without shared handle - create + // a duplicate with shared handle to provide for CAL + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D10_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + desc.BindFlags = D3D10_BIND_SHADER_RESOURCE | D3D10_BIND_RENDER_TARGET; + hr = pDev->CreateBuffer(&desc, NULL, + (ID3D10Buffer**) &obj.pD3D10Res_); + STORE_SHARED_FLAGS(ID3D10Buffer); + } + } + break; + + case D3D10_RESOURCE_DIMENSION_TEXTURE1D: // = 2, + { + D3D10_TEXTURE1D_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + + if(subres) { + // Calculate correct size of the subresource + UINT miplevel = subres; + if(desc.ArraySize > 1) { + miplevel = subres % desc.ArraySize; + } + if(miplevel >= desc.MipLevels) { + LogWarning("\nMiplevel >= number of miplevels\n"); + } + if(subres >= desc.MipLevels*desc.ArraySize) { + return CL_INVALID_VALUE; + } + desc.Width >>= miplevel; + if(!desc.Width) { + desc.Width = 1; + } + } + obj.objDesc_.objSize_.Width = desc.Width; + obj.objDesc_.mipLevels_ = desc.MipLevels; + obj.objDesc_.arraySize_ = desc.ArraySize; + obj.objDesc_.dxgiFormat_ = desc.Format; + obj.objDesc_.objFlags_.d3d10Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + // Handle D3D10Texture1D without shared handle - create + // a duplicate with shared handle and provide it for CAL + // Workaround for subresource > 0 in shared resource + if(subres) + obj.objDesc_.objFlags_.miscFlags_ &= + ~(D3D10_RESOURCE_MISC_SHARED); + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D10_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + SET_BINDING(); + obj.objDesc_.mipLevels_ = desc.MipLevels = 1; + obj.objDesc_.arraySize_ = desc.ArraySize = 1; + hr = pDev->CreateTexture1D(&desc, NULL, + (ID3D10Texture1D**) &obj.pD3D10Res_); + STORE_SHARED_FLAGS(ID3D10Texture1D); + } + } + break; + + case D3D10_RESOURCE_DIMENSION_TEXTURE2D: // = 3, + { + D3D10_TEXTURE2D_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + + if(subres) { + // Calculate correct size of the subresource + UINT miplevel = subres; + if(desc.ArraySize > 1) { + miplevel = subres % desc.MipLevels; + } + if(miplevel >= desc.MipLevels) { + LogWarning("\nMiplevel >= number of miplevels\n"); + } + if(subres >= desc.MipLevels*desc.ArraySize) { + return CL_INVALID_VALUE; + } + desc.Width >>= miplevel; + if(!desc.Width) { + desc.Width = 1; + } + desc.Height >>= miplevel; + if(!desc.Height) { + desc.Height = 1; + } + } + obj.objDesc_.objSize_.Width = desc.Width; + obj.objDesc_.objSize_.Height = desc.Height; + obj.objDesc_.mipLevels_ = desc.MipLevels; + obj.objDesc_.arraySize_ = desc.ArraySize; + obj.objDesc_.dxgiFormat_ = desc.Format; + obj.objDesc_.dxgiSampleDesc_ = desc.SampleDesc; + obj.objDesc_.objFlags_.d3d10Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + // Handle D3D10Texture2D without shared handle - create + // a duplicate with shared handle and provide it for CAL + // Workaround for subresource > 0 in shared resource + if(subres) + obj.objDesc_.objFlags_.miscFlags_ &= + ~(D3D10_RESOURCE_MISC_SHARED); + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D10_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + SET_BINDING(); + obj.objDesc_.mipLevels_ = desc.MipLevels = 1; + obj.objDesc_.arraySize_ = desc.ArraySize = 1; + hr = pDev->CreateTexture2D(&desc, NULL, + (ID3D10Texture2D**) &obj.pD3D10Res_); + STORE_SHARED_FLAGS(ID3D10Texture2D); + } + } + break; + + case D3D10_RESOURCE_DIMENSION_TEXTURE3D: // = 4 + { + D3D10_TEXTURE3D_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + + if(subres) { + // Calculate correct size of the subresource + UINT miplevel = subres; + if(miplevel >= desc.MipLevels) { + LogWarning("\nMiplevel >= number of miplevels\n"); + } + if(subres >= desc.MipLevels) { + return CL_INVALID_VALUE; + } + desc.Width >>= miplevel; + if(!desc.Width) { + desc.Width = 1; + } + desc.Height >>= miplevel; + if(!desc.Height) { + desc.Height = 1; + } + desc.Depth >>= miplevel; + if(!desc.Depth) { + desc.Depth = 1; + } + } + obj.objDesc_.objSize_.Width = desc.Width; + obj.objDesc_.objSize_.Height = desc.Height; + obj.objDesc_.objSize_.Depth = desc.Depth; + obj.objDesc_.mipLevels_ = desc.MipLevels; + obj.objDesc_.dxgiFormat_ = desc.Format; + obj.objDesc_.objFlags_.d3d10Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + // Handle D3D10Texture3D without shared handle - create + // a duplicate with shared handle and provide it for CAL + // Workaround for subresource > 0 in shared resource + if(obj.objDesc_.mipLevels_ > 1) + obj.objDesc_.objFlags_.miscFlags_ &= + ~(D3D10_RESOURCE_MISC_SHARED); + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D10_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + SET_BINDING(); + obj.objDesc_.mipLevels_ = desc.MipLevels = 1; + hr = pDev->CreateTexture3D(&desc, NULL, + (ID3D10Texture3D**) &obj.pD3D10Res_); + STORE_SHARED_FLAGS(ID3D10Texture3D); + } + } + break; + + default: + LogError("unknown type of D3D10 resource"); + return CL_INVALID_D3D10_RESOURCE_KHR; + } + obj.subRes_ = subres; + pDev->Release(); + // Check for CL format compatibilty + if(obj.objDesc_.objDim_ != D3D10_RESOURCE_DIMENSION_BUFFER) { + cl_image_format clFmt = obj.getCLFormatFromDXGI(obj.objDesc_.dxgiFormat_); + amd::Image::Format imageFormat(clFmt); + if(!imageFormat.isSupported(amdContext)) { + return CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + } + } + resources_.push_back(std::make_pair(pRes, subres)); + return CL_SUCCESS; +} + +bool +D3D10Object::copyOrigToShared() +{ + // Don't copy if there is no orig + if (NULL == getD3D10ResOrig()) return true; + + ID3D10Device *d3dDev; + pD3D10Res_->GetDevice(&d3dDev); + if(!d3dDev) { + LogError("\nCannot get D3D10 device from D3D10 resource\n"); + return false; + } + // Any usage source can be read by GPU + d3dDev->CopySubresourceRegion(pD3D10Res_, 0, 0, 0, 0, + pD3D10ResOrig_, subRes_, NULL); + + // Flush D3D queues and make sure D3D stuff is finished + pQuery_->End(); + BOOL data; + while(S_OK != pQuery_->GetData(&data, sizeof(BOOL), 0) && data != TRUE) + { + } + + d3dDev->Release(); + return true; +} + +bool +D3D10Object::copySharedToOrig() +{ + // Don't copy if there is no orig + if (NULL == getD3D10ResOrig()) return true; + + ID3D10Device *d3dDev; + pD3D10Res_->GetDevice(&d3dDev); + if(!d3dDev) { + LogError("\nCannot get D3D10 device from D3D10 resource\n"); + return false; + } + + d3dDev->CopySubresourceRegion(pD3D10ResOrig_, subRes_, 0, 0, 0, + pD3D10Res_, 0, NULL); + + d3dDev->Release(); + return true; +} + +std::vector> D3D10Object::resources_; +Monitor D3D10Object::resLock_; + +// +// Class BufferD3D10 implementation +// +void +BufferD3D10::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(BufferD3D10)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +BufferD3D10::mapExtObjectInCQThread() +{ + void* pCpuMem = NULL; + HRESULT hr; + D3D10_MAP gpuMap; + UINT cpuAccess; + + + if (getMemFlags() & CL_MEM_READ_WRITE) { + gpuMap = D3D10_MAP_READ_WRITE; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + gpuMap = D3D10_MAP_READ; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + gpuMap = D3D10_MAP_WRITE; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + + if(getUsage() == D3D10_USAGE_STAGING) { + // Can map directly + hr = reinterpret_cast( + getD3D10Resource())->Map(gpuMap, 0, &pCpuMem); + if(hr != S_OK || !pCpuMem) { + LogError("Cannot map ID3D10Buffer object to CPU memory"); + return false; + } + } + else { + // The buffer need to be mapped indirectly + // Create auxiliary buffer + ID3D10Device* pD3D10Dev; + getD3D10Resource()->GetDevice(&pD3D10Dev); + if(!pD3D10Dev) { + LogError("\nCannot get D3D10 device"); + return false; + } + pD3D10Dev->Release(); + D3D10_BUFFER_DESC bufDesc = { + getResourceByteSize(), + D3D10_USAGE_STAGING, + 0, + cpuAccess, + 0}; + ID3D10Buffer* pAuxBuf; + hr = pD3D10Dev->CreateBuffer(&bufDesc, NULL, &pAuxBuf); + if(hr != S_OK || !pAuxBuf) { + LogError("\nCannot create auxiliary buffer"); + return false; + } + setD3D10AuxRes(pAuxBuf); + // Copy contents of original buffer to auxiliary + pD3D10Dev->CopyResource(pAuxBuf, getD3D10Resource()); + // Now map the aux buffer + hr = pAuxBuf->Map(gpuMap, 0, &pCpuMem); + if(hr != S_OK || !pCpuMem) { + LogError("Cannot map D3D10 auxiliary buffer to CPU memory"); + return false; + } + } + + setHostMem(pCpuMem); + return true; +} + +bool +BufferD3D10::unmapExtObjectInCQThread() +{ + if(getMemFlags() & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY)) { + if(getD3D10AuxRes()) { + // Need to copy data from aux to original + reinterpret_cast(getD3D10AuxRes())->Unmap(); + ID3D10Device* pD3D10Dev; + getD3D10AuxRes()->GetDevice(&pD3D10Dev); + if(!pD3D10Dev) { + LogError("\nCannot get D3D10 device"); + return false; + } + pD3D10Dev->Release(); + pD3D10Dev->CopyResource(getD3D10Resource(), getD3D10AuxRes()); + getD3D10AuxRes()->Release(); + setD3D10AuxRes(NULL); + } + else { + reinterpret_cast(getD3D10Resource())->Unmap(); + } + } + else { + // Just unmap everything, no need to copy contents + if(getD3D10AuxRes()) { + reinterpret_cast(getD3D10AuxRes())->Unmap(); + getD3D10AuxRes()->Release(); + setD3D10AuxRes(NULL); + } + else { + reinterpret_cast(getD3D10Resource())->Unmap(); + } + } + setHostMem(NULL); + return true; +} + +// +// Class Image1DD3D10 implementation +// + +void +Image1DD3D10::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image1DD3D10)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +Image1DD3D10::mapExtObjectInCQThread() +{ + LogError("\nImage1DD3D10::mapExtObjectInCQThread() is not implemented yet\n"); + return false; +} + +bool +Image1DD3D10::unmapExtObjectInCQThread() +{ + LogError("\nImage1DD3D10::unmapExtObjectInCQThread() is not implemented yet\n"); + return false; +} + +// +// Class Image2DD3D10 implementation +// + +void +Image2DD3D10::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image2DD3D10)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +Image2DD3D10::mapExtObjectInCQThread() +{ + D3D10_MAPPED_TEXTURE2D texture2D; + HRESULT hr; + D3D10_MAP gpuMap; + UINT cpuAccess; + + + if (getMemFlags() & CL_MEM_READ_WRITE) { + gpuMap = D3D10_MAP_READ_WRITE; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + gpuMap = D3D10_MAP_READ; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + gpuMap = D3D10_MAP_WRITE; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + + if(getUsage() == D3D10_USAGE_STAGING) { + // Can map directly + hr = reinterpret_cast(getD3D10Resource()) + ->Map(getSubresource(), gpuMap, 0, &texture2D); + if(hr != S_OK || !texture2D.pData) { + LogError("Cannot map ID3D10Texture2D object to CPU memory"); + return false; + } + } + else { + // The texture needs to be mapped indirectly. + // Create auxiliary texture. + ID3D10Device* pD3D10Dev; + getD3D10Resource()->GetDevice(&pD3D10Dev); + if(!pD3D10Dev) { + LogError("\nCannot get D3D10 device"); + return false; + } + pD3D10Dev->Release(); + D3D10_TEXTURE2D_DESC texDesc; + reinterpret_cast(getD3D10Resource()) + ->GetDesc(&texDesc); + texDesc.Usage = D3D10_USAGE_STAGING; + texDesc.MipLevels = 1; + texDesc.BindFlags = 0; + texDesc.CPUAccessFlags = cpuAccess; + texDesc.MiscFlags = 0; + ID3D10Texture2D* pAuxTex; + hr = pD3D10Dev->CreateTexture2D(&texDesc, NULL, &pAuxTex); + if(hr != S_OK) { + LogError("\nCannot create auxiliary 2D texture"); + return false; + } + setD3D10AuxRes(pAuxTex); + // Copy contents of original texture to auxiliary + pD3D10Dev->CopyResource(pAuxTex, getD3D10Resource()); + // Now map the aux texture + hr = pAuxTex->Map(0, gpuMap, 0, &texture2D); + if(hr != S_OK || !texture2D.pData) { + LogError("Cannot map D3D10 auxiliary 2D texture to CPU memory"); + return false; + } + } + + setHostMem(texture2D.pData); + return true; +} + +bool +Image2DD3D10::unmapExtObjectInCQThread() +{ + if(getMemFlags() & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY)) { + if(getD3D10AuxRes()) { + // Need to copy data from aux to original + reinterpret_cast(getD3D10AuxRes())->Unmap(0); + ID3D10Device* pD3D10Dev; + getD3D10AuxRes()->GetDevice(&pD3D10Dev); + if(!pD3D10Dev) { + LogError("\nCannot get D3D10 device"); + return false; + } + pD3D10Dev->Release(); + pD3D10Dev->CopyResource(getD3D10Resource(), getD3D10AuxRes()); + getD3D10AuxRes()->Release(); + setD3D10AuxRes(NULL); + } + else { + reinterpret_cast(getD3D10Resource()) + ->Unmap(getSubresource()); + } + } + else { + // Just unmap everything, no need to copy contents + if(getD3D10AuxRes()) { + reinterpret_cast(getD3D10AuxRes())->Unmap(0); + getD3D10AuxRes()->Release(); + setD3D10AuxRes(NULL); + } + else { + reinterpret_cast(getD3D10Resource()) + ->Unmap(getSubresource()); + } + } + setHostMem(NULL); + return true; +} + +// +// Class Image3DD3D10 implementation +// +void +Image3DD3D10::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image3DD3D10)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + + +bool +Image3DD3D10::mapExtObjectInCQThread() +{ + D3D10_MAPPED_TEXTURE3D texture3D; + HRESULT hr; + D3D10_MAP gpuMap; + UINT cpuAccess; + + + if (getMemFlags() & CL_MEM_READ_WRITE) { + gpuMap = D3D10_MAP_READ_WRITE; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + gpuMap = D3D10_MAP_READ; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + gpuMap = D3D10_MAP_WRITE; + cpuAccess = D3D10_CPU_ACCESS_READ | D3D10_CPU_ACCESS_WRITE; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + + if(getUsage() == D3D10_USAGE_STAGING) { + // Can map directly + hr = reinterpret_cast(getD3D10Resource()) + ->Map(getSubresource(), gpuMap, 0, &texture3D); + if(hr != S_OK || !texture3D.pData) { + LogError("Cannot map ID3D10Texture3D object to CPU memory"); + return false; + } + } + else { + // The texture needs to be mapped indirectly. + // Create auxiliary texture. + ID3D10Device* pD3D10Dev; + getD3D10Resource()->GetDevice(&pD3D10Dev); + if(!pD3D10Dev) { + LogError("\nCannot get D3D10 device"); + return false; + } + pD3D10Dev->Release(); + D3D10_TEXTURE3D_DESC texDesc; + reinterpret_cast(getD3D10Resource()) + ->GetDesc(&texDesc); + texDesc.Usage = D3D10_USAGE_STAGING; + texDesc.MipLevels = 1; + texDesc.BindFlags = 0; + texDesc.CPUAccessFlags = cpuAccess; + texDesc.MiscFlags = 0; + ID3D10Texture3D* pAuxTex; + hr = pD3D10Dev->CreateTexture3D(&texDesc, NULL, &pAuxTex); + if(hr != S_OK) { + LogError("\nCannot create auxiliary 3D texture"); + return false; + } + setD3D10AuxRes(pAuxTex); + // Copy contents of original texture to auxiliary + pD3D10Dev->CopyResource(pAuxTex, getD3D10Resource()); + // Now map the aux texture + hr = pAuxTex->Map(0, gpuMap, 0, &texture3D); + if(hr != S_OK || !texture3D.pData) { + LogError("Cannot map D3D10 auxiliary 3D texture to CPU memory"); + return false; + } + } + + setHostMem(texture3D.pData); + return true; +} + +bool +Image3DD3D10::unmapExtObjectInCQThread() +{ + if(getMemFlags() & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY)) { + if(getD3D10AuxRes()) { + // Need to copy data from aux to original + reinterpret_cast(getD3D10AuxRes())->Unmap(0); + ID3D10Device* pD3D10Dev; + getD3D10AuxRes()->GetDevice(&pD3D10Dev); + if(!pD3D10Dev) { + LogError("\nCannot get D3D10 device"); + return false; + } + pD3D10Dev->Release(); + pD3D10Dev->CopyResource(getD3D10Resource(), getD3D10AuxRes()); + getD3D10AuxRes()->Release(); + setD3D10AuxRes(NULL); + } + else { + reinterpret_cast(getD3D10Resource()) + ->Unmap(getSubresource()); + } + } + else { + // Just unmap everything, no need to copy contents + if(getD3D10AuxRes()) { + reinterpret_cast(getD3D10AuxRes())->Unmap(0); + getD3D10AuxRes()->Release(); + setD3D10AuxRes(NULL); + } + else { + reinterpret_cast(getD3D10Resource()) + ->Unmap(getSubresource()); + } + } + setHostMem(NULL); + return true; +} + +} //namespace amd + +#endif //_WIN32 + diff --git a/opencl/api/opencl/amdocl/cl_d3d10_amd.hpp b/opencl/api/opencl/amdocl/cl_d3d10_amd.hpp new file mode 100644 index 0000000000..fdacf1a5d3 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_d3d10_amd.hpp @@ -0,0 +1,391 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifndef CL_D3D10_AMD_HPP_ +#define CL_D3D10_AMD_HPP_ + +#include "CL/cl_d3d10.h" + +#include "platform/context.hpp" +#include "platform/memory.hpp" + +#include + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromD3D10KHR( + cl_platform_id /*platform*/, + cl_d3d10_device_source_khr /*d3d_device_source*/, + void * /*d3d_object*/, + cl_d3d10_device_set_khr /*d3d_device_set*/, + cl_uint /*num_entries*/, + cl_device_id * /*devices*/, + cl_uint * /*num_devices*/); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10BufferKHR( + cl_context /* context */, + cl_mem_flags /* flags */, + ID3D10Buffer * /* buffer */, + cl_int * /* errcode_ret */); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10Texture2DKHR( + cl_context /* context */, + cl_mem_flags /* flags */, + ID3D10Texture2D * /* resource */, + UINT /* subresource */, + cl_int * /* errcode_ret */); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10Texture3DKHR( + cl_context /* context */, + cl_mem_flags /* flags */, + ID3D10Texture3D * /* resource */, + UINT /* subresource */, + cl_int * /* errcode_ret */); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireD3D10ObjectsKHR( + cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseD3D10ObjectsKHR( + cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */); + +namespace amd +{ + +typedef struct +{ + union + { + UINT ByteWidth; + UINT Width; + }; + UINT Height; + UINT Depth; +} D3D10ObjSize_t; + +typedef struct +{ + D3D10_USAGE d3d10Usage_; + UINT bindFlags_; + UINT cpuAccessFlags_; + UINT miscFlags_; +} D3D10Flags_t; + +typedef struct +{ + D3D10_RESOURCE_DIMENSION objDim_; + D3D10ObjSize_t objSize_; + D3D10Flags_t objFlags_; + UINT mipLevels_; + UINT arraySize_; + DXGI_FORMAT dxgiFormat_; + DXGI_SAMPLE_DESC dxgiSampleDesc_; +} D3D10ObjDesc_t; + +const DXGI_SAMPLE_DESC dxgiSampleDescDefault = {1, 0}; + +//! Class D3D10Object keeps all the info about the D3D10 object +//! from which the CL object is created +class D3D10Object : public InteropObject +{ +private: + ID3D10Resource* pD3D10Aux_; + + // @todo: TBD: Do we need to sync data after access + // or it'll be done by the D3D driver? + cl_int cliChecksum_; + bool releaseResources_; + + static bool createSharedResource(D3D10Object& obj); + static std::vector> resources_; + //! Global lock. + static Monitor resLock_; + +protected: + ID3D10Resource* pD3D10Res_; + ID3D10Resource* pD3D10ResOrig_; + ID3D10Query* pQuery_; + D3D10ObjDesc_t objDesc_; + D3D10ObjDesc_t objDescOrig_; + UINT subRes_; + +public: + // Default constructor + D3D10Object() + :pD3D10Aux_(NULL) + ,cliChecksum_(0) + ,releaseResources_(false) + ,pD3D10Res_(NULL) + ,pD3D10ResOrig_(NULL) + ,pQuery_(NULL) + ,subRes_(0) + { + memset(&objDesc_,0,sizeof(objDesc_)); + memset(&objDescOrig_,0,sizeof(objDescOrig_)); + } + // Copy constructor + D3D10Object(D3D10Object& d3d10obj) + : pQuery_(NULL) + { + *this = d3d10obj; + this->releaseResources_ = true; + // Add reference to the D3D10 resource to prevent its disappearance + if(pD3D10ResOrig_) { + pD3D10ResOrig_->AddRef(); + } + else if(pD3D10Res_) { + pD3D10Res_->AddRef(); + } + } + + //! Virtual destructor + virtual ~D3D10Object() + { + ScopedLock sl(resLock_); + if(releaseResources_) { + // Decrement reference to the D3D10 objects + if(pD3D10Res_) pD3D10Res_->Release(); + if(pD3D10Aux_) pD3D10Aux_->Release(); + if(pD3D10ResOrig_) pD3D10ResOrig_->Release(); + if(pQuery_) pQuery_->Release(); + // Check if this resource has already been used for interop + std::vector>::iterator it; + if(resources_.size()) { + for(it = resources_.begin(); it != resources_.end(); ++it) { + if(((pD3D10ResOrig_ && (*it).first == (void*) pD3D10ResOrig_) + || ((*it).first == (void*) pD3D10Res_)) + && (*it).second == subRes_) { + resources_.erase(it); + break; + } + } + } + } + } + + static int initD3D10Object(const Context& amdContext, ID3D10Resource* pRes, UINT subresource, + D3D10Object& obj); + + D3D10Object* asD3D10Object() { return this; } + +//! D3D10Object query functions to get D3D10 info from member variables + ID3D10Resource* getD3D10Resource() const {return pD3D10Res_;} + ID3D10Resource* getD3D10ResOrig() const {return pD3D10ResOrig_;} + D3D10_USAGE getUsage() const { return objDesc_.objFlags_.d3d10Usage_; } + void setD3D10AuxRes(ID3D10Resource* pAux) {pD3D10Aux_ = pAux;} + ID3D10Resource* getD3D10AuxRes() const {return pD3D10Aux_;} + ID3D10Query* getQuery() const {return pQuery_;} + + UINT getWidth() const {return objDesc_.objSize_.Width;} + UINT getHeight() const {return objDesc_.objSize_.Height;} + UINT getDepth() const {return objDesc_.objSize_.Depth;} + size_t getElementBytes(DXGI_FORMAT dxgiFomat); + size_t getElementBytes() {return getElementBytes(objDesc_.dxgiFormat_);} + DXGI_FORMAT getDxgiFormat() {return objDesc_.dxgiFormat_;} + UINT getSubresource() const {return subRes_;} + const D3D10ObjDesc_t* getObjDesc() const { return &objDesc_; } + + //! Returns bytes per pixel > 0 if conversion successful, 0 otherwise; + //! if formats are not compatible, cl format channel + //! order and type are set to 0 + cl_image_format getCLFormatFromDXGI(DXGI_FORMAT dxgiFmt); + cl_image_format getCLFormatFromDXGI() + { + return getCLFormatFromDXGI(objDesc_.dxgiFormat_); + } + size_t getResourceByteSize(); + + // On acquire copy data from original resource to shared resource + virtual bool copyOrigToShared(); + // On release copy data from shared copy to the original resource + virtual bool copySharedToOrig(); +}; + +//! Class BufferD3D10 is derived from classes Buffer and D3D10Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D10 object +class BufferD3D10 : public D3D10Object, public Buffer +{ +protected: + //! Initializes the device memory array which is nested + // after 'BufferD3D10' object in memory layout. + virtual void initDeviceMemory(); +public: + //! BufferD3D10 constructor just calls constructors of base classes + //! to pass down the parameters + BufferD3D10( + Context& amdContext, + cl_mem_flags clFlags, + D3D10Object& d3d10obj) + : // Call base classes constructors + D3D10Object(d3d10obj), + Buffer( + amdContext, + clFlags, + d3d10obj.getResourceByteSize()) + { + setInteropObj(this); + } + virtual ~BufferD3D10() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Class Image1DD3D10 is derived from classes Image1D and D3D10Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D10 object +class Image1DD3D10 : public D3D10Object, public Image +{ +protected: + //! Initializes the device memory array which is nested + // after'Image1DD3D10' object in memory layout. + virtual void initDeviceMemory(); +public: + //! Image1DD3D10 constructor just calls constructors of base classes + //! to pass down the parameters + Image1DD3D10( + Context& amdContext, + cl_mem_flags clFlags, + D3D10Object& d3d10obj) + : // Call base classes constructors + D3D10Object(d3d10obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE1D, + clFlags, + getCLFormatFromDXGI(d3d10obj.getDxgiFormat()), //format, + d3d10obj.getWidth(), + 1, + 1, + d3d10obj.getWidth() * d3d10obj.getElementBytes(), //rowPitch), + 0) + { + setInteropObj(this); + } + virtual ~Image1DD3D10() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Class Image2DD3D10 is derived from classes Image2D and D3D10Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D10 object +class Image2DD3D10 : public D3D10Object, public Image +{ +protected: + //! Initializes the device memory array which is nested + // after'Image2DD3D10' object in memory layout. + virtual void initDeviceMemory(); +public: + //! Image2DD3D10 constructor just calls constructors of base classes + //! to pass down the parameters + Image2DD3D10( + Context& amdContext, + cl_mem_flags clFlags, + D3D10Object& d3d10obj) + : // Call base classes constructors + D3D10Object(d3d10obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE2D, + clFlags, + getCLFormatFromDXGI(d3d10obj.getDxgiFormat()), //format, + d3d10obj.getWidth(), + d3d10obj.getHeight(), + 1, + d3d10obj.getWidth() * d3d10obj.getElementBytes(), //rowPitch), + 0) + { + setInteropObj(this); + } + virtual ~Image2DD3D10() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Class Image3DD3D10 is derived from classes Image3D and D3D10Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D10 object +class Image3DD3D10 : public D3D10Object, public Image +{ +protected: + //! Initializes the device memory array which is nested + // after'Image3DD3D10' object in memory layout. + virtual void initDeviceMemory(); +public: +//! Image2DD3D10 constructor just calls constructors of base classes +//! to pass down the parameters + Image3DD3D10( + Context& amdContext, + cl_mem_flags clFlags, + D3D10Object& d3d10obj) + : // Call base classes constructors + D3D10Object(d3d10obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE3D, + clFlags, + getCLFormatFromDXGI(d3d10obj.getDxgiFormat()), //format, + d3d10obj.getWidth(), + d3d10obj.getHeight(), + d3d10obj.getDepth(), + d3d10obj.getWidth() * d3d10obj.getElementBytes(), //rowPitch), + d3d10obj.getWidth() * d3d10obj.getHeight() * d3d10obj.getElementBytes()) + { + setInteropObj(this); + } + virtual ~Image3DD3D10() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Functions for executing the D3D10 related stuff +cl_mem clCreateBufferFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + int* errcode_ret); +cl_mem clCreateImage1DFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret); +cl_mem clCreateImage2DFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret); +cl_mem clCreateImage3DFromD3D10ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D10Resource* pD3DResource, + UINT subresource, + int* errcode_ret); +void SyncD3D10Objects(std::vector& memObjects); +} //namespace amd + +#endif //CL_D3D10_AMD_HPP_ diff --git a/opencl/api/opencl/amdocl/cl_d3d11.cpp b/opencl/api/opencl/amdocl/cl_d3d11.cpp new file mode 100644 index 0000000000..d0970004e2 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_d3d11.cpp @@ -0,0 +1,2043 @@ +// +// Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifdef _WIN32 + +#include +#include + +#include "cl_common.hpp" +#include "cl_d3d11_amd.hpp" +#include "runtime/platform/command.hpp" + +#include +#include + +#define DXGI_FORMAT_NV12 103 + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_D3D11_Interops + * + * This section discusses OpenCL functions that allow applications to use Direct3D 11 + * resources (buffers/textures) as OpenCL memory objects. This allows efficient sharing of + * data between OpenCL and Direct3D 11. The OpenCL API can be used to execute kernels that + * read and/or write memory objects that are also the Direct3D resources. + * An OpenCL image object can be created from a D3D11 texture object. An + * OpenCL buffer object can be created from a D3D11 buffer object (index/vertex). + * + * @} + * \addtogroup clGetDeviceIDsFromD3D11KHR + * @{ + */ + +RUNTIME_ENTRY(cl_int, clGetDeviceIDsFromD3D11KHR, ( + cl_platform_id platform, + cl_d3d11_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d11_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices)) +{ + cl_int errcode; + ID3D11Device* d3d11_device = NULL; + cl_device_id* gpu_devices; + cl_uint num_gpu_devices = 0; + bool create_d3d11Device = false; + static const bool VALIDATE_ONLY = true; + HMODULE d3d11Module = NULL; + + if (platform != NULL && platform != AMD_PLATFORM) { + LogWarning("\"platrform\" is not a valid AMD platform"); + return CL_INVALID_PLATFORM; + } + if(((num_entries > 0 || num_devices == NULL) && devices == NULL) + || (num_entries == 0 && devices != NULL)) { + return CL_INVALID_VALUE; + } + // Get GPU devices + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, 0, NULL, &num_gpu_devices); + if (errcode != CL_SUCCESS && errcode != CL_DEVICE_NOT_FOUND) { + return CL_INVALID_VALUE; + } + + if (!num_gpu_devices) { + *not_null(num_devices) = 0; + return CL_DEVICE_NOT_FOUND; + } + + switch(d3d_device_source) + { + case CL_D3D11_DEVICE_KHR: + d3d11_device = static_cast(d3d_object); + break; + case CL_D3D11_DXGI_ADAPTER_KHR: + { + static PFN_D3D11_CREATE_DEVICE dynamicD3D11CreateDevice = NULL; + + d3d11Module = LoadLibrary("D3D11.dll"); + if (d3d11Module == NULL) { + return CL_INVALID_PLATFORM; + } + + dynamicD3D11CreateDevice = (PFN_D3D11_CREATE_DEVICE)GetProcAddress(d3d11Module, "D3D11CreateDevice"); + + IDXGIAdapter* dxgi_adapter = static_cast(d3d_object); + D3D_FEATURE_LEVEL requestedFeatureLevels[] = {D3D_FEATURE_LEVEL_10_0}; + D3D_FEATURE_LEVEL featureLevel = D3D_FEATURE_LEVEL_11_0; + HRESULT hr = dynamicD3D11CreateDevice(dxgi_adapter, D3D_DRIVER_TYPE_UNKNOWN, + NULL, 0, requestedFeatureLevels, 1, + D3D11_SDK_VERSION, &d3d11_device, &featureLevel, NULL); + if (SUCCEEDED(hr) && (NULL != d3d11_device)) { + create_d3d11Device = true; + } else { + FreeLibrary(d3d11Module); + return CL_INVALID_VALUE; + } + } + break; + default: + LogWarning("\"d3d_device_source\" is invalid"); + return CL_INVALID_VALUE; + } + + switch(d3d_device_set) { + case CL_PREFERRED_DEVICES_FOR_D3D11_KHR: + case CL_ALL_DEVICES_FOR_D3D11_KHR: + { + gpu_devices = (cl_device_id *) alloca(num_gpu_devices * sizeof(cl_device_id)); + + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, num_gpu_devices, gpu_devices, NULL); + if (errcode != CL_SUCCESS) { + break; + } + + std::vector compatible_devices; + for (cl_uint i = 0; i < num_gpu_devices; ++i) { + cl_device_id device = gpu_devices[i]; + if (is_valid(device) && + as_amd(device)->bindExternalDevice(CL_CONTEXT_D3D11_DEVICE_KHR, d3d11_device, NULL, VALIDATE_ONLY)) { + compatible_devices.push_back(as_amd(device)); + } + } + if (compatible_devices.size() == 0) { + *not_null(num_devices) = 0; + errcode = CL_DEVICE_NOT_FOUND; + break; + } + + std::vector::iterator it = compatible_devices.begin(); + cl_uint compatible_count = std::min(num_entries, (cl_uint)compatible_devices.size()); + + while (compatible_count--) { + *devices++ = as_cl(*it++); + --num_entries; + } + while (num_entries--) { + *devices++ = (cl_device_id) 0; + } + + *not_null(num_devices) = (cl_uint)compatible_devices.size(); + } + break; + + default: + LogWarning("\"d3d_device_set\" is invalid"); + errcode = CL_INVALID_VALUE; + } + + if (create_d3d11Device) { + d3d11_device->Release(); + FreeLibrary(d3d11Module); + } + return errcode; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromD3D11BufferKHR + * @{ + */ + +/*! \brief Creates an OpenCL buffer object from a Direct3D 10 resource. + * + * \param context is a valid OpenCL context. + * + * \param flags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param pD3DResource is a valid pointer to a D3D11 resource of type ID3D11Buffer. + * + * \return valid non-zero OpenCL buffer object and \a errcode_ret is set + * to CL_SUCCESS if the buffer object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context or if Direct3D 10 + * interoperatbility has not been initialized between context and the ID3D11Device + * from which pD3DResource was created. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_D3D_RESOURCE if \a pD3DResource is not of type ID3D11Buffer. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33? + */ + +RUNTIME_ENTRY_RET(cl_mem, clCreateFromD3D11BufferKHR, ( + cl_context context, + cl_mem_flags flags, + ID3D11Buffer* pD3DResource, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + if(!flags) flags = CL_MEM_READ_WRITE; + if(!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + if(!pD3DResource) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("parameter \"pD3DResource\" is a NULL pointer"); + return clMemObj; + } + return(amd::clCreateBufferFromD3D11ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + errcode_ret)); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateImageFromD3D11Resource + * @{ + */ + +/*! \brief Create an OpenCL 2D or 3D image object from a D3D11 resource. + * + * \param context is a valid OpenCL context. + * + * \param flags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param pD3DResource is a valid pointer to a D3D11 resource of type + * ID3D11Texture2D, ID3D11Texture2D, or ID3D11Texture3D. + * If pD3DResource is of type ID3D11Texture1D then the created image object + * will be a 1D mipmapped image object. + * If pD3DResource is of type ID3D11Texture2D and was not created with flag + * D3D11_RESOURCE_MISC_TEXTURECUBE then the created image object will be a + * 2D mipmapped image object. + * If pD3DResource is of type ID3D11Texture2D and was created with flag + * D3D11_RESOURCE_MISC_TEXTURECUBE then the created image object will be + * a cubemap mipmapped image object. + * errocde_ret returns CL_INVALID_D3D_RESOURCE if an OpenCL memory object has + * already been created from pD3DResource in context. + * If pD3DResource is of type ID3D11Texture3D then the created image object will + * be a 3D mipmapped imageobject. + * + * \return valid non-zero OpenCL image object and \a errcode_ret is set + * to CL_SUCCESS if the image object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context or if Direct3D 11 + * interoperatbility has not been initialized between context and the ID3D11Device + * from which pD3DResource was created. + * - CL_INVALID_VALUE if values specified in \a flags are not valid. + * - CL_INVALID_D3D_RESOURCE if \a pD3DResource is not of type ID3D11Texture1D, + * ID3D11Texture2D, or ID3D11Texture3D. + * - CL_INVALID_D3D_RESOURCE if an OpenCL memory object has already been created + * from \a pD3DResource in context. + * - CL_INVALID_IMAGE_FORMAT if the Direct3D 11 texture format does not map + * to an appropriate OpenCL image format. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r48? + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateImageFromD3D11Resource, ( + cl_context context, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + UINT subresource, + int* errcode_ret, + UINT dimension)) +{ + cl_mem clMemObj = NULL; + + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + if(!flags) flags = CL_MEM_READ_WRITE; + if(!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + if(!pD3DResource) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("parameter \"pD3DResource\" is a NULL pointer"); + return clMemObj; + } + + // Verify context init'ed for interop + ID3D11Device* pDev; + pD3DResource->GetDevice(&pDev); + if(pDev == NULL) { + *not_null(errcode_ret) = CL_INVALID_D3D11_DEVICE_KHR; + LogWarning("Cannot retrieve D3D11 device from D3D11 resource"); + return (cl_mem) 0; + } + pDev->Release(); + if (!((*as_amd(context)).info().flags_ & amd::Context::D3D11DeviceKhr)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from D3D11 device"); + return (cl_mem) 0; + } + + // Check for image support + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if((*it)->info().imageSupport_) { + supportPass = true; + } + } + if(!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return (cl_mem) 0; + } + + switch(dimension) { +#if 0 + case 1: + return(amd::clCreateImage1DFromD3D11ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + subresource, + errcode_ret)); +#endif //0 + case 2: + return(amd::clCreateImage2DFromD3D11ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + subresource, + errcode_ret)); + case 3: + return(amd::clCreateImage3DFromD3D11ResourceAMD( + *as_amd(context), + flags, + pD3DResource, + subresource, + errcode_ret)); + default: + break; + } + + *not_null(errcode_ret) = CL_INVALID_D3D11_RESOURCE_KHR; + return (cl_mem) 0; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromD3D11Texture2DKHR + * @{ + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromD3D11Texture2DKHR, ( + cl_context context, + cl_mem_flags flags, + ID3D11Texture2D* resource, + UINT subresource, + cl_int* errcode_ret)) +{ + return clCreateImageFromD3D11Resource(context, flags, resource, + subresource, errcode_ret, 2); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromD3D11Texture3DKHR + * @{ + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromD3D11Texture3DKHR, ( + cl_context context, + cl_mem_flags flags, + ID3D11Texture3D* resource, + UINT subresource, + cl_int* errcode_ret)) +{ + return clCreateImageFromD3D11Resource(context, flags, resource, + subresource, errcode_ret, 3); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clEnqueueAcquireD3D11ObjectsKHR + * @{ + */ +RUNTIME_ENTRY(cl_int, clEnqueueAcquireD3D11ObjectsKHR, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + return amd::clEnqueueAcquireExtObjectsAMD(command_queue, num_objects, + mem_objects, num_events_in_wait_list, event_wait_list, event, + CL_COMMAND_ACQUIRE_D3D11_OBJECTS_KHR); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clEnqueueReleaseD3D11ObjectsKHR + * @{ + */ +RUNTIME_ENTRY(cl_int, clEnqueueReleaseD3D11ObjectsKHR, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + return amd::clEnqueueReleaseExtObjectsAMD(command_queue, num_objects, + mem_objects, num_events_in_wait_list, event_wait_list, event, + CL_COMMAND_RELEASE_D3D11_OBJECTS_KHR); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clGetPlaneFromImageAMD + * @{ + */ +RUNTIME_ENTRY_RET(cl_mem, clGetPlaneFromImageAMD, ( + cl_context context, + cl_mem mem, + cl_uint plane, + cl_int* errcode_ret)) +{ + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return 0; + } + if (mem == 0) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return 0; + } + if (!is_valid(mem)) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return 0; + } + amd::Memory* amdMem = as_amd(mem); + amd::Context& amdContext = *as_amd(context); + if (amdMem->getInteropObj() == NULL) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return 0; + } + amd::Image2DD3D11 *pImage = reinterpret_cast(amdMem); + ID3D11Resource* pD3DResource = pImage->getD3D11Resource(); + // Verify the resource is a 2D texture + D3D11_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D11_RESOURCE_DIMENSION_TEXTURE2D) { + *not_null(errcode_ret) = CL_INVALID_D3D11_RESOURCE_KHR; + return (cl_mem) 0; + } + + amd::D3D11Object obj; + int errcode = amd::D3D11Object::initD3D11Object(amdContext, pD3DResource, 0, obj, plane); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + amd::Image2DD3D11 *pImage2DD3D11 = new (amdContext) + amd::Image2DD3D11(amdContext, pImage->getMemFlags(), obj); + if(!pImage2DD3D11) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage2DD3D11->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage2DD3D11->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage2DD3D11); +} +RUNTIME_EXIT + +// +// +// namespace amd +// +// +namespace amd +{ +/*! @} + * \addtogroup CL-D3D11 interop helper functions + * @{ + */ + + + +//******************************************************************* +// +// Internal implementation of CL API functions +// +//******************************************************************* +// +// clCreateBufferFromD3D11ResourceAMD +// +cl_mem clCreateBufferFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + int* errcode_ret) +{ + // Verify pD3DResource is a buffer + D3D11_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D11_RESOURCE_DIMENSION_BUFFER) { + *not_null(errcode_ret) = CL_INVALID_D3D11_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D11Object obj; + int errcode = D3D11Object::initD3D11Object(amdContext, pD3DResource, 0, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + BufferD3D11 *pBufferD3D11 = new (amdContext) + BufferD3D11(amdContext, flags, obj); + if(!pBufferD3D11) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pBufferD3D11->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pBufferD3D11->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pBufferD3D11); +} + +// +// clCreateImage2DFromD3D11ResourceAMD +// +cl_mem clCreateImage2DFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + UINT subresource, + int* errcode_ret) +{ + // Verify the resource is a 2D texture + D3D11_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D11_RESOURCE_DIMENSION_TEXTURE2D) { + *not_null(errcode_ret) = CL_INVALID_D3D11_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D11Object obj; + int errcode = D3D11Object::initD3D11Object(amdContext, pD3DResource, subresource, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + Image2DD3D11 *pImage2DD3D11 = new (amdContext) + Image2DD3D11(amdContext, flags, obj); + if(!pImage2DD3D11) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage2DD3D11->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage2DD3D11->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage2DD3D11); +} + +// +// clCreateImage2DFromD3D11ResourceAMD +// +cl_mem clCreateImage3DFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + UINT subresource, + int* errcode_ret) +{ + // Verify the resource is a 2D texture + D3D11_RESOURCE_DIMENSION rType; + pD3DResource->GetType(&rType); + if(rType != D3D11_RESOURCE_DIMENSION_TEXTURE3D) { + *not_null(errcode_ret) = CL_INVALID_D3D11_RESOURCE_KHR; + return (cl_mem) 0; + } + + D3D11Object obj; + int errcode = D3D11Object::initD3D11Object(amdContext, pD3DResource, subresource, obj); + if(CL_SUCCESS != errcode) + { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + Image3DD3D11 *pImage3DD3D11 = new (amdContext) + Image3DD3D11(amdContext, flags, obj); + if(!pImage3DD3D11) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage3DD3D11->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage3DD3D11->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage3DD3D11); +} + +size_t +D3D11Object::getResourceByteSize() +{ + size_t bytes = 1; + + //! @todo [odintsov]: take into consideration the mip level?! + + switch(objDesc_.objDim_) + { + case D3D11_RESOURCE_DIMENSION_BUFFER: + bytes = objDesc_.objSize_.ByteWidth; + break; + + case D3D11_RESOURCE_DIMENSION_TEXTURE3D: + bytes = objDesc_.objSize_.Depth; + + case D3D11_RESOURCE_DIMENSION_TEXTURE2D: + bytes *= objDesc_.objSize_.Height; + + case D3D11_RESOURCE_DIMENSION_TEXTURE1D: + bytes *= objDesc_.objSize_.Width * getElementBytes(); + break; + + default: + LogError("getResourceByteSize: unknown type of D3D11 resource"); + bytes = 0; + break; + } + return bytes; +} + +cl_uint +D3D11Object::getMiscFlag() +{ + if (objDesc_.dxgiFormat_ == DXGI_FORMAT_NV12) + { + return 1; + } + return 0; +} + +int +D3D11Object::initD3D11Object(const Context& amdContext, ID3D11Resource* pRes, UINT subres, D3D11Object& obj, INT plane) +{ + ID3D11Device *pDev; + HRESULT hr; + ScopedLock sl(resLock_); + + // Check if this ressource has already been used for interop + std::vector >>::iterator it; + for(it = resources_.begin(); it != resources_.end(); ++it) { + if((*it).first == (void*) pRes && (*it).second.first == subres && (*it).second.second == plane) { + return CL_INVALID_D3D11_RESOURCE_KHR; + } + } + + (obj.pD3D11Res_ = pRes)->GetDevice(&pDev); + + if(!pDev) { + return CL_INVALID_D3D11_DEVICE_KHR; + } + + D3D11_QUERY_DESC desc = {D3D11_QUERY_EVENT, 0}; \ + pDev->CreateQuery(&desc, &obj.pQuery_); \ + +#define SET_SHARED_FLAGS() \ + { \ + obj.pD3D11ResOrig_ = obj.pD3D11Res_; \ + /* @todo - Check device type and select right usage for resource */ \ + /* For now get only DPU path, CPU path for buffers */ \ + /* will not worl on DEFAUL resources */ \ + /*desc.Usage = D3D11_USAGE_STAGING;*/ \ + desc.Usage = D3D11_USAGE_DEFAULT; \ + desc.MiscFlags = D3D11_RESOURCE_MISC_SHARED; \ + desc.CPUAccessFlags = 0; \ + } + +#define STORE_SHARED_FLAGS_BUFFER(restype) \ + { \ + if(S_OK == hr && obj.pD3D11Res_) { \ + obj.objDesc_.objFlags_.d3d11Usage_ = desc.Usage; \ + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; \ + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; \ + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; \ + obj.objDesc_.objFlags_.structureByteStride_ = desc.StructureByteStride; \ + } \ + else { \ + LogError("\nCannot create shared " #restype "\n"); \ + return CL_INVALID_D3D11_RESOURCE_KHR; \ + } \ + } + +#define STORE_SHARED_FLAGS(restype) \ + { \ + if(S_OK == hr && obj.pD3D11Res_) { \ + obj.objDesc_.objFlags_.d3d11Usage_ = desc.Usage; \ + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; \ + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; \ + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; \ + } \ + else { \ + LogError("\nCannot create shared " #restype "\n"); \ + return CL_INVALID_D3D11_RESOURCE_KHR; \ + } \ + } + +#define SET_BINDING() \ + { \ + switch(desc.Format) { \ + case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: \ + case DXGI_FORMAT_D32_FLOAT: \ + case DXGI_FORMAT_D24_UNORM_S8_UINT: \ + case DXGI_FORMAT_D16_UNORM: \ + desc.BindFlags = D3D11_BIND_DEPTH_STENCIL; \ + break; \ + default: \ + desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; \ + break; \ + } \ + } + + pRes->GetType(&obj.objDesc_.objDim_); + + // Init defaults + obj.objDesc_.objSize_.Height = 1; + obj.objDesc_.objSize_.Depth = 1; + obj.objDesc_.mipLevels_ = 1; + obj.objDesc_.arraySize_ = 1; + obj.objDesc_.dxgiFormat_ = DXGI_FORMAT_UNKNOWN; + obj.objDesc_.dxgiSampleDesc_ = dxgiSampleDescDefault; + + switch(obj.objDesc_.objDim_) { + case D3D11_RESOURCE_DIMENSION_BUFFER: // = 1, + { + D3D11_BUFFER_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + obj.objDesc_.objSize_.ByteWidth = desc.ByteWidth; + obj.objDesc_.objFlags_.d3d11Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + obj.objDesc_.objFlags_.structureByteStride_ = desc.StructureByteStride; + // Handle D3D11Buffer without shared handle - create + // a duplicate with shared handle to provide for CAL + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D11_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + desc.BindFlags = D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_RENDER_TARGET; + hr = pDev->CreateBuffer(&desc, NULL, + (ID3D11Buffer**) &obj.pD3D11Res_); + STORE_SHARED_FLAGS_BUFFER(ID3D11Buffer); + } + } + break; + + case D3D11_RESOURCE_DIMENSION_TEXTURE1D: // = 2, + { + D3D11_TEXTURE1D_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + + if(subres) { + // Calculate correct size of the subresource + UINT miplevel = subres; + if(desc.ArraySize > 1) { + miplevel = subres % desc.ArraySize; + } + if(miplevel >= desc.MipLevels) { + LogWarning("\nMiplevel >= number of miplevels\n"); + } + if(subres >= desc.MipLevels*desc.ArraySize) { + return CL_INVALID_VALUE; + } + desc.Width >>= miplevel; + if(!desc.Width) { + desc.Width = 1; + } + } + obj.objDesc_.objSize_.Width = desc.Width; + obj.objDesc_.mipLevels_ = desc.MipLevels; + obj.objDesc_.arraySize_ = desc.ArraySize; + obj.objDesc_.dxgiFormat_ = desc.Format; + obj.objDesc_.objFlags_.d3d11Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + // Handle D3D11Texture1D without shared handle - create + // a duplicate with shared handle and provide it for CAL + // Workaround for subresource > 0 in shared resource + if(subres) + obj.objDesc_.objFlags_.miscFlags_ &= + ~(D3D11_RESOURCE_MISC_SHARED); + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D11_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + SET_BINDING(); + obj.objDesc_.mipLevels_ = desc.MipLevels = 1; + obj.objDesc_.arraySize_ = desc.ArraySize = 1; + hr = pDev->CreateTexture1D(&desc, NULL, + (ID3D11Texture1D**) &obj.pD3D11Res_); + STORE_SHARED_FLAGS(ID3D11Texture1D); + } + } + break; + + case D3D11_RESOURCE_DIMENSION_TEXTURE2D: // = 3, + { + D3D11_TEXTURE2D_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + + if(subres) { + // Calculate correct size of the subresource + UINT miplevel = subres; + if(desc.ArraySize > 1) { + miplevel = subres % desc.MipLevels; + } + if(miplevel >= desc.MipLevels) { + LogWarning("\nMiplevel >= number of miplevels\n"); + } + if(subres >= desc.MipLevels*desc.ArraySize) { + return CL_INVALID_VALUE; + } + desc.Width >>= miplevel; + if(!desc.Width) { + desc.Width = 1; + } + desc.Height >>= miplevel; + if(!desc.Height) { + desc.Height = 1; + } + } + obj.objDesc_.objSize_.Width = desc.Width; + obj.objDesc_.objSize_.Height = desc.Height; + obj.objDesc_.mipLevels_ = desc.MipLevels; + obj.objDesc_.arraySize_ = desc.ArraySize; + obj.objDesc_.dxgiFormat_ = desc.Format; + obj.objDesc_.dxgiSampleDesc_ = desc.SampleDesc; + obj.objDesc_.objFlags_.d3d11Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + + // Handle D3D11Texture2D without shared handle - create + // a duplicate with shared handle and provide it for CAL + // Workaround for subresource > 0 in shared resource + if(subres) + obj.objDesc_.objFlags_.miscFlags_ &= + ~(D3D11_RESOURCE_MISC_SHARED); + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D11_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + SET_BINDING(); + obj.objDesc_.mipLevels_ = desc.MipLevels = 1; + obj.objDesc_.arraySize_ = desc.ArraySize = 1; + hr = pDev->CreateTexture2D(&desc, NULL, + (ID3D11Texture2D**) &obj.pD3D11Res_); + STORE_SHARED_FLAGS(ID3D11Texture2D); + } + + if (desc.Format == DXGI_FORMAT_NV12) + { + if (plane == -1) { + obj.objDesc_.objSize_.Height += obj.objDesc_.objSize_.Height/2; + } + if (plane == 1) { + obj.objDesc_.objSize_.Width /= 2; + obj.objDesc_.objSize_.Height /= 2; + } + } + } + break; + + case D3D11_RESOURCE_DIMENSION_TEXTURE3D: // = 4 + { + D3D11_TEXTURE3D_DESC desc; + (reinterpret_cast(pRes))->GetDesc(&desc); + + if(subres) { + // Calculate correct size of the subresource + UINT miplevel = subres; + if(miplevel >= desc.MipLevels) { + LogWarning("\nMiplevel >= number of miplevels\n"); + } + if(subres >= desc.MipLevels) { + return CL_INVALID_VALUE; + } + desc.Width >>= miplevel; + if(!desc.Width) { + desc.Width = 1; + } + desc.Height >>= miplevel; + if(!desc.Height) { + desc.Height = 1; + } + desc.Depth >>= miplevel; + if(!desc.Depth) { + desc.Depth = 1; + } + } + obj.objDesc_.objSize_.Width = desc.Width; + obj.objDesc_.objSize_.Height = desc.Height; + obj.objDesc_.objSize_.Depth = desc.Depth; + obj.objDesc_.mipLevels_ = desc.MipLevels; + obj.objDesc_.dxgiFormat_ = desc.Format; + obj.objDesc_.objFlags_.d3d11Usage_ = desc.Usage; + obj.objDesc_.objFlags_.bindFlags_ = desc.BindFlags; + obj.objDesc_.objFlags_.cpuAccessFlags_ = desc.CPUAccessFlags; + obj.objDesc_.objFlags_.miscFlags_ = desc.MiscFlags; + // Handle D3D11Texture3D without shared handle - create + // a duplicate with shared handle and provide it for CAL + // Workaround for subresource > 0 in shared resource + if(obj.objDesc_.mipLevels_ > 1) + obj.objDesc_.objFlags_.miscFlags_ &= + ~(D3D11_RESOURCE_MISC_SHARED); + if(!(obj.objDesc_.objFlags_.miscFlags_ & D3D11_RESOURCE_MISC_SHARED)) { + SET_SHARED_FLAGS(); + SET_BINDING(); + obj.objDesc_.mipLevels_ = desc.MipLevels = 1; + hr = pDev->CreateTexture3D(&desc, NULL, + (ID3D11Texture3D**) &obj.pD3D11Res_); + STORE_SHARED_FLAGS(ID3D11Texture3D); + } + } + break; + + default: + LogError("unknown type of D3D11 resource"); + return CL_INVALID_D3D11_RESOURCE_KHR; + } + obj.subRes_ = subres; + obj.plane_ = plane; + pDev->Release(); + // Check for CL format compatibilty + if(obj.objDesc_.objDim_ != D3D11_RESOURCE_DIMENSION_BUFFER) { + cl_image_format clFmt = obj.getCLFormatFromDXGI(obj.objDesc_.dxgiFormat_, plane); + amd::Image::Format imageFormat(clFmt); + if(!imageFormat.isSupported(amdContext)) { + return CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + } + } + resources_.push_back(std::make_pair(pRes, std::make_pair(subres, plane))); + return CL_SUCCESS; +} + +bool +D3D11Object::copyOrigToShared() +{ + // Don't copy if there is no orig + if (NULL == getD3D11ResOrig()) return true; + + ID3D11Device *d3dDev; + pD3D11Res_->GetDevice(&d3dDev); + if(!d3dDev) { + LogError("\nCannot get D3D11 device from D3D11 resource\n"); + return false; + } + ID3D11DeviceContext *pImmediateContext = NULL; + d3dDev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + assert(pD3D11ResOrig_!=NULL); + // Any usage source can be read by GPU + pImmediateContext->CopySubresourceRegion(pD3D11Res_, 0, 0, 0, 0, + pD3D11ResOrig_, subRes_, NULL); + pImmediateContext->Release(); + + // Flush D3D queues and make sure D3D stuff is finished + { + ScopedLock sl(resLock_);//protect from multiple + pImmediateContext->End(pQuery_); + BOOL data; + while(S_OK != pImmediateContext->GetData(pQuery_, &data, sizeof(BOOL), 0) && data != TRUE) + { + } + } + + d3dDev->Release(); + return true; +} + +bool +D3D11Object::copySharedToOrig() +{ + // Don't copy if there is no orig + if (NULL == getD3D11ResOrig()) return true; + + ID3D11Device *d3dDev; + pD3D11Res_->GetDevice(&d3dDev); + if(!d3dDev) { + LogError("\nCannot get D3D11 device from D3D11 resource\n"); + return false; + } + ID3D11DeviceContext *pImmediateContext = NULL; + d3dDev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + assert(pD3D11ResOrig_); + pImmediateContext->CopySubresourceRegion(pD3D11ResOrig_, subRes_, 0, 0, 0, + pD3D11Res_, 0, NULL); + pImmediateContext->Release(); + + d3dDev->Release(); + return true; +} + +std::vector >> D3D11Object::resources_; +Monitor D3D11Object::resLock_; + +// +// Class BufferD3D11 implementation +// +void +BufferD3D11::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(BufferD3D11)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +BufferD3D11::mapExtObjectInCQThread() +{ + D3D11_MAPPED_SUBRESOURCE mappedResource; + HRESULT hr; + D3D11_MAP gpuMap; + UINT cpuAccess; + + + if (getMemFlags() & CL_MEM_READ_WRITE) { + gpuMap = D3D11_MAP_READ_WRITE; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + gpuMap = D3D11_MAP_READ; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + gpuMap = D3D11_MAP_WRITE; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + + ID3D11Device* pD3D11Dev; + getD3D11Resource()->GetDevice(&pD3D11Dev); + if(!pD3D11Dev) { + LogError("\nCannot get D3D11 device"); + return false; + } + pD3D11Dev->Release(); + ID3D11DeviceContext *pImmediateContext = NULL; + pD3D11Dev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + pImmediateContext->Release(); + if(getUsage() == D3D11_USAGE_STAGING) { + // XXX Christophe: Use DeviceContext to map + //// Can map directly + hr = pImmediateContext->Map(getD3D11Resource(), 0, gpuMap, 0, &mappedResource); + if(hr != S_OK || !mappedResource.pData) { + LogError("Cannot map ID3D11Buffer object to CPU memory"); + return false; + } + } + else { + // The buffer need to be mapped indirectly + // Create auxiliary buffer + D3D11_BUFFER_DESC bufDesc = { + getResourceByteSize(), + D3D11_USAGE_STAGING, + 0, + cpuAccess, + 0}; + ID3D11Buffer* pAuxBuf; + hr = pD3D11Dev->CreateBuffer(&bufDesc, NULL, &pAuxBuf); + if(hr != S_OK || !pAuxBuf) { + LogError("\nCannot create auxiliary buffer"); + return false; + } + setD3D11AuxRes(pAuxBuf); + // Copy contents of original buffer to auxiliary + pImmediateContext->CopyResource(pAuxBuf, getD3D11Resource()); + // Now map the aux buffer + hr = pImmediateContext->Map(pAuxBuf, 0, gpuMap, 0, &mappedResource); + if(hr != S_OK || !mappedResource.pData) { + LogError("Cannot map D3D11 auxiliary buffer to CPU memory"); + return false; + } + } + + setHostMem(mappedResource.pData); + return true; +} + +bool +BufferD3D11::unmapExtObjectInCQThread() +{ + ID3D11Device* pD3D11Dev; + getD3D11AuxRes()->GetDevice(&pD3D11Dev); + if(!pD3D11Dev) { + LogError("\nCannot get D3D11 device"); + return false; + } + pD3D11Dev->Release(); + ID3D11DeviceContext *pImmediateContext = NULL; + pD3D11Dev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + pImmediateContext->Release(); + if(getMemFlags() & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY)) { + if(getD3D11AuxRes()) { + // Need to copy data from aux to original + pImmediateContext->Unmap(getD3D11AuxRes(), 0); + pImmediateContext->CopyResource(getD3D11Resource(), getD3D11AuxRes()); + getD3D11AuxRes()->Release(); + setD3D11AuxRes(NULL); + } + else { + pImmediateContext->Unmap(getD3D11Resource(), 0); + } + } + else { + // Just unmap everything, no need to copy contents + if(getD3D11AuxRes()) { + pImmediateContext->Unmap(getD3D11AuxRes(), 0); + getD3D11AuxRes()->Release(); + setD3D11AuxRes(NULL); + } + else { + pImmediateContext->Unmap(getD3D11Resource(), 0); + } + } + setHostMem(NULL); + return true; +} + +// +// Class Image1DD3D11 implementation +// +void +Image1DD3D11::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image1DD3D11)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +Image1DD3D11::mapExtObjectInCQThread() +{ + LogError("\nImage1DD3D11::mapExtObjectInCQThread() is not implemented yet\n"); + return false; +} + +bool +Image1DD3D11::unmapExtObjectInCQThread() +{ + LogError("\nImage1DD3D11::unmapExtObjectInCQThread() is not implemented yet\n"); + return false; +} + +// +// Class Image2DD3D11 implementation +// + +void +Image2DD3D11::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image2DD3D11)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +Image2DD3D11::mapExtObjectInCQThread() +{ + D3D11_MAPPED_SUBRESOURCE texture2D; + HRESULT hr; + D3D11_MAP gpuMap; + UINT cpuAccess; + + + if (getMemFlags() & CL_MEM_READ_WRITE) { + gpuMap = D3D11_MAP_READ_WRITE; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + gpuMap = D3D11_MAP_READ; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + gpuMap = D3D11_MAP_WRITE; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + + ID3D11Device* pD3D11Dev; + getD3D11Resource()->GetDevice(&pD3D11Dev); + if(!pD3D11Dev) { + LogError("\nCannot get D3D11 device"); + return false; + } + pD3D11Dev->Release(); + ID3D11DeviceContext *pImmediateContext = NULL; + pD3D11Dev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + pImmediateContext->Release(); + if(getUsage() == D3D11_USAGE_STAGING) { + // Can map directly + hr = pImmediateContext->Map(getD3D11Resource(), getSubresource(), gpuMap, 0, &texture2D); + if(hr != S_OK || !texture2D.pData) { + LogError("Cannot map ID3D11Texture2D object to CPU memory"); + return false; + } + } + else { + // The texture needs to be mapped indirectly. + // Create auxiliary texture. + D3D11_TEXTURE2D_DESC texDesc; + reinterpret_cast(getD3D11Resource()) + ->GetDesc(&texDesc); + texDesc.Usage = D3D11_USAGE_STAGING; + texDesc.MipLevels = 1; + texDesc.BindFlags = 0; + texDesc.CPUAccessFlags = cpuAccess; + texDesc.MiscFlags = 0; + ID3D11Texture2D* pAuxTex; + hr = pD3D11Dev->CreateTexture2D(&texDesc, NULL, &pAuxTex); + if(hr != S_OK) { + LogError("\nCannot create auxiliary 2D texture"); + return false; + } + setD3D11AuxRes(pAuxTex); + // Copy contents of original texture to auxiliary + pImmediateContext->CopyResource(pAuxTex, getD3D11Resource()); + // Now map the aux texture + hr = pImmediateContext->Map(pAuxTex, 0, gpuMap, 0, &texture2D); + if(hr != S_OK || !texture2D.pData) { + LogError("Cannot map D3D11 auxiliary 2D texture to CPU memory"); + return false; + } + } + + setHostMem(texture2D.pData); + return true; +} + +bool +Image2DD3D11::unmapExtObjectInCQThread() +{ + ID3D11Device* pD3D11Dev; + getD3D11AuxRes()->GetDevice(&pD3D11Dev); + if(!pD3D11Dev) { + LogError("\nCannot get D3D11 device"); + return false; + } + pD3D11Dev->Release(); + ID3D11DeviceContext *pImmediateContext = NULL; + pD3D11Dev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + pImmediateContext->Release(); + if(getMemFlags() & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY)) { + if(getD3D11AuxRes()) { + // Need to copy data from aux to original + pImmediateContext->Unmap(getD3D11AuxRes(), 0); + pImmediateContext->CopyResource(getD3D11Resource(), getD3D11AuxRes()); + getD3D11AuxRes()->Release(); + setD3D11AuxRes(NULL); + } + else { + pImmediateContext->Unmap(getD3D11Resource(), getSubresource()); + } + } + else { + // Just unmap everything, no need to copy contents + if(getD3D11AuxRes()) { + pImmediateContext->Unmap(getD3D11AuxRes(), 0); + getD3D11AuxRes()->Release(); + setD3D11AuxRes(NULL); + } + else { + pImmediateContext->Unmap(getD3D11Resource(), getSubresource()); + } + } + setHostMem(NULL); + return true; +} + +// +// Class Image3DD3D11 implementation +// +void +Image3DD3D11::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image3DD3D11)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +Image3DD3D11::mapExtObjectInCQThread() +{ + D3D11_MAPPED_SUBRESOURCE texture3D; + HRESULT hr; + D3D11_MAP gpuMap; + UINT cpuAccess; + + + if (getMemFlags() & CL_MEM_READ_WRITE) { + gpuMap = D3D11_MAP_READ_WRITE; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + gpuMap = D3D11_MAP_READ; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + gpuMap = D3D11_MAP_WRITE; + cpuAccess = D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + + ID3D11Device* pD3D11Dev; + getD3D11AuxRes()->GetDevice(&pD3D11Dev); + if(!pD3D11Dev) { + LogError("\nCannot get D3D11 device"); + return false; + } + pD3D11Dev->Release(); + ID3D11DeviceContext *pImmediateContext = NULL; + pD3D11Dev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + pImmediateContext->Release(); + if(getUsage() == D3D11_USAGE_STAGING) { + // Can map directly + hr = pImmediateContext->Map(getD3D11Resource(), getSubresource(), gpuMap, 0, &texture3D); + if(hr != S_OK || !texture3D.pData) { + LogError("Cannot map ID3D11Texture3D object to CPU memory"); + return false; + } + } + else { + // The texture needs to be mapped indirectly. + // Create auxiliary texture. + D3D11_TEXTURE3D_DESC texDesc; + reinterpret_cast(getD3D11Resource()) + ->GetDesc(&texDesc); + texDesc.Usage = D3D11_USAGE_STAGING; + texDesc.MipLevels = 1; + texDesc.BindFlags = 0; + texDesc.CPUAccessFlags = cpuAccess; + texDesc.MiscFlags = 0; + ID3D11Texture3D* pAuxTex; + hr = pD3D11Dev->CreateTexture3D(&texDesc, NULL, &pAuxTex); + if(hr != S_OK) { + LogError("\nCannot create auxiliary 3D texture"); + return false; + } + setD3D11AuxRes(pAuxTex); + // Copy contents of original texture to auxiliary + pImmediateContext->CopyResource(pAuxTex, getD3D11Resource()); + // Now map the aux texture + hr = pImmediateContext->Map(pAuxTex, 0, gpuMap, 0, &texture3D); + if(hr != S_OK || !texture3D.pData) { + LogError("Cannot map D3D11 auxiliary 3D texture to CPU memory"); + return false; + } + } + + setHostMem(texture3D.pData); + return true; +} + +bool +Image3DD3D11::unmapExtObjectInCQThread() +{ + ID3D11Device* pD3D11Dev; + getD3D11AuxRes()->GetDevice(&pD3D11Dev); + if(!pD3D11Dev) { + LogError("\nCannot get D3D11 device"); + return false; + } + pD3D11Dev->Release(); + ID3D11DeviceContext *pImmediateContext = NULL; + pD3D11Dev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return false; + } + pImmediateContext->Release(); + if(getMemFlags() & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY)) { + if(getD3D11AuxRes()) { + // Need to copy data from aux to original + pImmediateContext->Unmap(getD3D11AuxRes(), 0); + pImmediateContext->CopyResource(getD3D11Resource(), getD3D11AuxRes()); + getD3D11AuxRes()->Release(); + setD3D11AuxRes(NULL); + } + else { + pImmediateContext->Unmap(getD3D11Resource(), getSubresource()); + } + } + else { + // Just unmap everything, no need to copy contents + if(getD3D11AuxRes()) { + pImmediateContext->Unmap(getD3D11AuxRes(), 0); + getD3D11AuxRes()->Release(); + setD3D11AuxRes(NULL); + } + else { + pImmediateContext->Unmap(getD3D11Resource(), getSubresource()); + } + } + setHostMem(NULL); + return true; +} + +// +// Helper function SyncD3D11Objects +// +void SyncD3D11Objects(std::vector& memObjects) +{ + Memory*& mem = memObjects.front(); + if(!mem) { + LogWarning("\nNULL memory object\n"); + return; + } + InteropObject* interop = mem->getInteropObj(); + if(!interop) { + LogWarning("\nNULL interop object\n"); + return; + } + D3D11Object* d3dObj = interop->asD3D11Object(); + if(!d3dObj) { + LogWarning("\nNULL D3D11 object\n"); + return; + } + ID3D11Query* query = d3dObj->getQuery(); + if(!query) { + LogWarning("\nNULL ID3D11Query\n"); + return; + } + ID3D11Device *d3dDev; + query->GetDevice(&d3dDev); + if(!d3dDev) { + LogError("\nCannot get D3D11 device from D3D11 resource\n"); + return; + } + ID3D11DeviceContext *pImmediateContext = NULL; + d3dDev->GetImmediateContext(&pImmediateContext); + if(!pImmediateContext) { + LogError("\nCannot get D3D11 device context"); + return; + } + pImmediateContext->Release(); + + // Flush D3D queues and make sure D3D stuff is finished + { + ScopedLock sl(d3dObj->getResLock()); + pImmediateContext->End(query); + BOOL data; + while(S_OK != pImmediateContext->GetData(query, &data, sizeof(BOOL), 0) && data != TRUE) + { + } + } + + d3dDev->Release(); +} + +// +// Class D3D11Object implementation +// +size_t +D3D11Object::getElementBytes(DXGI_FORMAT dxgiFmt, cl_uint plane) +{ + size_t bytesPerPixel; + + switch(dxgiFmt) + { + case DXGI_FORMAT_R32G32B32A32_TYPELESS: + case DXGI_FORMAT_R32G32B32A32_FLOAT: + case DXGI_FORMAT_R32G32B32A32_UINT: + case DXGI_FORMAT_R32G32B32A32_SINT: + bytesPerPixel = 16; + break; + + case DXGI_FORMAT_R32G32B32_TYPELESS: + case DXGI_FORMAT_R32G32B32_FLOAT: + case DXGI_FORMAT_R32G32B32_UINT: + case DXGI_FORMAT_R32G32B32_SINT: + bytesPerPixel = 12; + break; + + case DXGI_FORMAT_R16G16B16A16_TYPELESS: + case DXGI_FORMAT_R16G16B16A16_FLOAT: + case DXGI_FORMAT_R16G16B16A16_UNORM: + case DXGI_FORMAT_R16G16B16A16_UINT: + case DXGI_FORMAT_R16G16B16A16_SNORM: + case DXGI_FORMAT_R16G16B16A16_SINT: + case DXGI_FORMAT_R32G32_TYPELESS: + case DXGI_FORMAT_R32G32_FLOAT: + case DXGI_FORMAT_R32G32_UINT: + case DXGI_FORMAT_R32G32_SINT: + case DXGI_FORMAT_R32G8X24_TYPELESS: + case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: + case DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS: + case DXGI_FORMAT_X32_TYPELESS_G8X24_UINT: + bytesPerPixel = 8; + break; + + case DXGI_FORMAT_R10G10B10A2_TYPELESS: + case DXGI_FORMAT_R10G10B10A2_UNORM: + case DXGI_FORMAT_R10G10B10A2_UINT: + case DXGI_FORMAT_R11G11B10_FLOAT: + case DXGI_FORMAT_R8G8B8A8_TYPELESS: + case DXGI_FORMAT_R8G8B8A8_UNORM: + case DXGI_FORMAT_R8G8B8A8_UNORM_SRGB: + case DXGI_FORMAT_R8G8B8A8_UINT: + case DXGI_FORMAT_R8G8B8A8_SNORM: + case DXGI_FORMAT_R8G8B8A8_SINT: + case DXGI_FORMAT_R16G16_TYPELESS: + case DXGI_FORMAT_R16G16_FLOAT: + case DXGI_FORMAT_R16G16_UNORM: + case DXGI_FORMAT_R16G16_UINT: + case DXGI_FORMAT_R16G16_SNORM: + case DXGI_FORMAT_R16G16_SINT: + case DXGI_FORMAT_R32_TYPELESS: + case DXGI_FORMAT_D32_FLOAT: + case DXGI_FORMAT_R32_FLOAT: + case DXGI_FORMAT_R32_UINT: + case DXGI_FORMAT_R32_SINT: + case DXGI_FORMAT_R24G8_TYPELESS: + case DXGI_FORMAT_D24_UNORM_S8_UINT: + case DXGI_FORMAT_R24_UNORM_X8_TYPELESS: + case DXGI_FORMAT_X24_TYPELESS_G8_UINT: + + case DXGI_FORMAT_R9G9B9E5_SHAREDEXP: + case DXGI_FORMAT_R8G8_B8G8_UNORM: + case DXGI_FORMAT_G8R8_G8B8_UNORM: + + case DXGI_FORMAT_B8G8R8A8_UNORM: + case DXGI_FORMAT_B8G8R8X8_UNORM: + bytesPerPixel = 4; + break; + + case DXGI_FORMAT_R8G8_TYPELESS: + case DXGI_FORMAT_R8G8_UNORM: + case DXGI_FORMAT_R8G8_UINT: + case DXGI_FORMAT_R8G8_SNORM: + case DXGI_FORMAT_R8G8_SINT: + case DXGI_FORMAT_R16_TYPELESS: + case DXGI_FORMAT_R16_FLOAT: + case DXGI_FORMAT_D16_UNORM: + case DXGI_FORMAT_R16_UNORM: + case DXGI_FORMAT_R16_UINT: + case DXGI_FORMAT_R16_SNORM: + case DXGI_FORMAT_R16_SINT: + + case DXGI_FORMAT_B5G6R5_UNORM: + case DXGI_FORMAT_B5G5R5A1_UNORM: + bytesPerPixel = 2; + break; + + case DXGI_FORMAT_R8_TYPELESS: + case DXGI_FORMAT_R8_UNORM: + case DXGI_FORMAT_R8_UINT: + case DXGI_FORMAT_R8_SNORM: + case DXGI_FORMAT_R8_SINT: + case DXGI_FORMAT_A8_UNORM: + case DXGI_FORMAT_R1_UNORM: + bytesPerPixel = 1; + break; + + + case DXGI_FORMAT_BC1_TYPELESS: + case DXGI_FORMAT_BC1_UNORM: + case DXGI_FORMAT_BC1_UNORM_SRGB: + case DXGI_FORMAT_BC2_TYPELESS: + case DXGI_FORMAT_BC2_UNORM: + case DXGI_FORMAT_BC2_UNORM_SRGB: + case DXGI_FORMAT_BC3_TYPELESS: + case DXGI_FORMAT_BC3_UNORM: + case DXGI_FORMAT_BC3_UNORM_SRGB: + case DXGI_FORMAT_BC4_TYPELESS: + case DXGI_FORMAT_BC4_UNORM: + case DXGI_FORMAT_BC4_SNORM: + case DXGI_FORMAT_BC5_TYPELESS: + case DXGI_FORMAT_BC5_UNORM: + case DXGI_FORMAT_BC5_SNORM: + // Less than 1 byte per pixel - needs special consideration + bytesPerPixel = 0; + break; + case DXGI_FORMAT_NV12: + bytesPerPixel = 1; + if(plane == 1) { + bytesPerPixel = 2; + } + break; + default: + bytesPerPixel = 0; + _ASSERT(FALSE); + break; + } + return bytesPerPixel; +} + +cl_image_format +D3D11Object::getCLFormatFromDXGI(DXGI_FORMAT dxgiFmt, cl_uint plane) +{ + cl_image_format fmt; + + //! @todo [odintsov]: add real fmt conversion from DXGI to CL + fmt.image_channel_order = 0;//CL_RGBA; + fmt.image_channel_data_type = 0;//CL_UNSIGNED_INT8; + + switch(dxgiFmt) + { + case DXGI_FORMAT_R32G32B32A32_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R32G32B32A32_FLOAT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32G32B32A32_UINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32B32A32_SINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32B32_TYPELESS: + fmt.image_channel_order = CL_RGB; + break; + + case DXGI_FORMAT_R32G32B32_FLOAT: + fmt.image_channel_order = CL_RGB; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32G32B32_UINT: + fmt.image_channel_order = CL_RGB; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32B32_SINT: + fmt.image_channel_order = CL_RGB; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R16G16B16A16_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R16G16B16A16_FLOAT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case DXGI_FORMAT_R16G16B16A16_UNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16G16B16A16_UINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNSIGNED_INT16; + break; + + case DXGI_FORMAT_R16G16B16A16_SNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SNORM_INT16; + break; + + case DXGI_FORMAT_R16G16B16A16_SINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SIGNED_INT16; + break; + + case DXGI_FORMAT_R32G32_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_R32G32_FLOAT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32G32_UINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32G32_SINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R32G8X24_TYPELESS: + break; + + case DXGI_FORMAT_D32_FLOAT_S8X24_UINT: + break; + + case DXGI_FORMAT_R32_FLOAT_X8X24_TYPELESS: + break; + + case DXGI_FORMAT_X32_TYPELESS_G8X24_UINT: + break; + + case DXGI_FORMAT_R10G10B10A2_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R10G10B10A2_UNORM: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R10G10B10A2_UINT: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R11G11B10_FLOAT: + fmt.image_channel_order = CL_RGB; + break; + + case DXGI_FORMAT_R8G8B8A8_TYPELESS: + fmt.image_channel_order = CL_RGBA; + break; + + case DXGI_FORMAT_R8G8B8A8_UNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_UNORM_SRGB: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_UINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_SNORM: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SNORM_INT8; + break; + + case DXGI_FORMAT_R8G8B8A8_SINT: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_SIGNED_INT8; + break; + + case DXGI_FORMAT_R16G16_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_R16G16_FLOAT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case DXGI_FORMAT_R16G16_UNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16G16_UINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNSIGNED_INT16; + break; + + case DXGI_FORMAT_R16G16_SNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SNORM_INT16; + break; + + case DXGI_FORMAT_R16G16_SINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SIGNED_INT16; + break; + + case DXGI_FORMAT_R32_TYPELESS: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_D32_FLOAT: + break; + + case DXGI_FORMAT_R32_FLOAT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case DXGI_FORMAT_R32_UINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT32; + break; + + case DXGI_FORMAT_R32_SINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SIGNED_INT32; + break; + + case DXGI_FORMAT_R24G8_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_D24_UNORM_S8_UINT: + break; + + case DXGI_FORMAT_R24_UNORM_X8_TYPELESS: + break; + + case DXGI_FORMAT_X24_TYPELESS_G8_UINT: + break; + + case DXGI_FORMAT_R9G9B9E5_SHAREDEXP: + break; + + case DXGI_FORMAT_R8G8_B8G8_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_G8R8_G8B8_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_B8G8R8A8_UNORM: + fmt.image_channel_order = CL_BGRA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_B8G8R8X8_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8_TYPELESS: + fmt.image_channel_order = CL_RG; + break; + + case DXGI_FORMAT_R8G8_UNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8G8_UINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + break; + + case DXGI_FORMAT_R8G8_SNORM: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SNORM_INT8; + break; + + case DXGI_FORMAT_R8G8_SINT: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_SIGNED_INT8; + break; + + case DXGI_FORMAT_R16_TYPELESS: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_R16_FLOAT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case DXGI_FORMAT_D16_UNORM: + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16_UNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case DXGI_FORMAT_R16_UINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT16; + break; + + case DXGI_FORMAT_R16_SNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SNORM_INT16; + break; + + case DXGI_FORMAT_R16_SINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SIGNED_INT16; + break; + + case DXGI_FORMAT_B5G6R5_UNORM: + fmt.image_channel_data_type = CL_UNORM_SHORT_565; + break; + + case DXGI_FORMAT_B5G5R5A1_UNORM: + fmt.image_channel_order = CL_BGRA; + break; + + case DXGI_FORMAT_R8_TYPELESS: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_R8_UNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R8_UINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + break; + + case DXGI_FORMAT_R8_SNORM: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SNORM_INT8; + break; + + case DXGI_FORMAT_R8_SINT: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_SIGNED_INT8; + break; + + case DXGI_FORMAT_A8_UNORM: + fmt.image_channel_order = CL_A; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case DXGI_FORMAT_R1_UNORM: + fmt.image_channel_order = CL_R; + break; + + case DXGI_FORMAT_BC1_TYPELESS: + case DXGI_FORMAT_BC1_UNORM: + case DXGI_FORMAT_BC1_UNORM_SRGB: + case DXGI_FORMAT_BC2_TYPELESS: + case DXGI_FORMAT_BC2_UNORM: + case DXGI_FORMAT_BC2_UNORM_SRGB: + case DXGI_FORMAT_BC3_TYPELESS: + case DXGI_FORMAT_BC3_UNORM: + case DXGI_FORMAT_BC3_UNORM_SRGB: + case DXGI_FORMAT_BC4_TYPELESS: + case DXGI_FORMAT_BC4_UNORM: + case DXGI_FORMAT_BC4_SNORM: + case DXGI_FORMAT_BC5_TYPELESS: + case DXGI_FORMAT_BC5_UNORM: + case DXGI_FORMAT_BC5_SNORM: + break; + case DXGI_FORMAT_NV12: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNSIGNED_INT8; + + if(plane == 1) { + fmt.image_channel_order = CL_RG; + } + break; + default: + _ASSERT(FALSE); + break; + } + + return fmt; +} + +} //namespace amd + +#endif //_WIN32 + diff --git a/opencl/api/opencl/amdocl/cl_d3d11_amd.hpp b/opencl/api/opencl/amdocl/cl_d3d11_amd.hpp new file mode 100644 index 0000000000..2cccd12c96 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_d3d11_amd.hpp @@ -0,0 +1,402 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifndef CL_D3D11_AMD_HPP_ +#define CL_D3D11_AMD_HPP_ + +#include "CL/cl_d3d11.h" + +#include "cl_d3d10_amd.hpp" +#include "platform/context.hpp" +#include "platform/memory.hpp" + +#include + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromD3D11KHR( + cl_platform_id /*platform*/, + cl_d3d11_device_source_khr /*d3d_device_source*/, + void * /*d3d_object*/, + cl_d3d11_device_set_khr /*d3d_device_set*/, + cl_uint /*num_entries*/, + cl_device_id * /*devices*/, + cl_uint * /*num_devices*/); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11BufferKHR( + cl_context /* context */, + cl_mem_flags /* flags */, + ID3D11Buffer * /* buffer */, + cl_int * /* errcode_ret */); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11Texture2DKHR( + cl_context /* context */, + cl_mem_flags /* flags */, + ID3D11Texture2D * /* resource */, + UINT /* subresource */, + cl_int * /* errcode_ret */); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11Texture3DKHR( + cl_context /* context */, + cl_mem_flags /* flags */, + ID3D11Texture3D * /* resource */, + UINT /* subresource */, + cl_int * /* errcode_ret */); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireD3D11ObjectsKHR( + cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseD3D11ObjectsKHR( + cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clGetPlaneFromImageAMD( + cl_context /* context */, + cl_mem /* mem */, + cl_uint /* plane */, + cl_int* /* errcode_ret */); + +namespace amd +{ + +typedef struct +{ + union + { + UINT ByteWidth; + UINT Width; + }; + UINT Height; + UINT Depth; +} D3D11ObjSize_t; + +typedef struct +{ + D3D11_USAGE d3d11Usage_; + UINT bindFlags_; + UINT cpuAccessFlags_; + UINT miscFlags_; + UINT structureByteStride_; +} D3D11Flags_t; + +typedef struct +{ + D3D11_RESOURCE_DIMENSION objDim_; + D3D11ObjSize_t objSize_; + D3D11Flags_t objFlags_; + UINT mipLevels_; + UINT arraySize_; + DXGI_FORMAT dxgiFormat_; + DXGI_SAMPLE_DESC dxgiSampleDesc_; +} D3D11ObjDesc_t; + +//! Class D3D11Object keeps all the info about the D3D11 object +//! from which the CL object is created +class D3D11Object : public InteropObject +{ +private: + ID3D11Resource* pD3D11Aux_; + + // @todo: TBD: Do we need to sync data after access + // or it'll be done by the D3D driver? + cl_int cliChecksum_; + bool releaseResources_; + + static bool createSharedResource(D3D11Object& obj); + static std::vector>> resources_; +protected: + //! Global lock. + static Monitor resLock_; + + ID3D11Resource* pD3D11Res_; + ID3D11Resource* pD3D11ResOrig_; + ID3D11Query* pQuery_; + D3D11ObjDesc_t objDesc_; + UINT subRes_; + INT plane_; + +public: + // Default constructor + D3D11Object() + :pD3D11Aux_(NULL) + ,cliChecksum_(0) + ,releaseResources_(false) + ,pD3D11Res_(NULL) + ,pD3D11ResOrig_(NULL) + ,pQuery_(NULL) + ,subRes_(NULL) + ,plane_(NULL) + { + memset(&objDesc_,0,sizeof(objDesc_)); + } + // Copy constructor + D3D11Object(D3D11Object& d3d11obj) + : pQuery_(NULL) + { + *this = d3d11obj; + this->releaseResources_ = true; + // Add reference to the D3D11 resource to prevent its disappearance + if(pD3D11ResOrig_) { + pD3D11ResOrig_->AddRef(); + } + else if(pD3D11Res_) { + pD3D11Res_->AddRef(); + } + assert(pD3D11Res_ != pD3D11ResOrig_); + } + + //! Virtual destructor + virtual ~D3D11Object() + { + ScopedLock sl(resLock_); + if(releaseResources_) { + // Decrement reference to the D3D11 objects + if(pD3D11Res_) pD3D11Res_->Release(); + if(pD3D11Aux_) pD3D11Aux_->Release(); + if(pD3D11ResOrig_) pD3D11ResOrig_->Release(); + if(pQuery_) pQuery_->Release(); + // Check if this resource has already been used for interop + std::vector>>::iterator it; + if(resources_.size()) { + for(it = resources_.begin(); it != resources_.end(); ++it) { + if(((pD3D11ResOrig_ && (*it).first == (void*) pD3D11ResOrig_) + || ((*it).first == (void*) pD3D11Res_)) + && (*it).second.first == subRes_ + && (*it).second.second == plane_) { + resources_.erase(it); + break; + } + } + } + } + } + + static int initD3D11Object(const Context& amdContext, ID3D11Resource* pRes, UINT subresource, + D3D11Object& obj, INT plane = -1); + + D3D11Object* asD3D11Object() { return this; } + +//! D3D11Object query functions to get D3D11 info from member variables + ID3D11Resource* getD3D11Resource() const {return pD3D11Res_;} + ID3D11Resource* getD3D11ResOrig() const {return pD3D11ResOrig_;} + D3D11_USAGE getUsage() const { return objDesc_.objFlags_.d3d11Usage_; } + void setD3D11AuxRes(ID3D11Resource* pAux) {pD3D11Aux_ = pAux;} + ID3D11Resource* getD3D11AuxRes() const {return pD3D11Aux_;} + ID3D11Query* getQuery() const {return pQuery_;} + Monitor& getResLock() { return resLock_;} + UINT getWidth() const {return objDesc_.objSize_.Width;} + UINT getHeight() const {return objDesc_.objSize_.Height;} + UINT getDepth() const {return objDesc_.objSize_.Depth;} + size_t getElementBytes(DXGI_FORMAT dxgiFomat, cl_uint plane); + size_t getElementBytes() {return getElementBytes(objDesc_.dxgiFormat_, plane_);} + DXGI_FORMAT getDxgiFormat() {return objDesc_.dxgiFormat_;} + UINT getSubresource() const {return subRes_;} + INT getPlane() const {return plane_;} + const D3D11ObjDesc_t* getObjDesc() const { return &objDesc_; } + + cl_uint getMiscFlag(void); + //! Returns bytes per pixel > 0 if conversion successful, 0 otherwise; + //! if formats are not compatible, cl format channel + //! order and type are set to 0 + cl_image_format getCLFormatFromDXGI(DXGI_FORMAT dxgiFmt, cl_uint plane); + cl_image_format getCLFormatFromDXGI() + { + return getCLFormatFromDXGI(objDesc_.dxgiFormat_, plane_); + } + size_t getResourceByteSize(); + + // On acquire copy data from original resource to shared resource + virtual bool copyOrigToShared(); + // On release copy data from shared copy to the original resource + virtual bool copySharedToOrig(); +}; + +//! Class BufferD3D11 is derived from classes Buffer and D3D11Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D11 object +class BufferD3D11 : public D3D11Object, public Buffer +{ +protected: + //! Initializes the device memory array which is nested + // after'BufferD3D11' object in memory layout. + virtual void initDeviceMemory(); +public: +//! BufferD3D11 constructor just calls constructors of base classes +//! to pass down the parameters + BufferD3D11( + Context& amdContext, + cl_mem_flags clFlags, + D3D11Object& d3d11obj) + : // Call base classes constructors + D3D11Object(d3d11obj), + Buffer( + amdContext, + clFlags, + d3d11obj.getResourceByteSize()) + { + setInteropObj(this); + } + virtual ~BufferD3D11() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Class Image1DD3D11 is derived from classes Image1D and D3D11Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D11 object +class Image1DD3D11 : public D3D11Object, public Image +{ +protected: + //! Initializes the device memory array which is nested + // after'Image1DD3D11' object in memory layout. + virtual void initDeviceMemory(); +public: +//! Image1DD3D11 constructor just calls constructors of base classes +//! to pass down the parameters + Image1DD3D11( + Context& amdContext, + cl_mem_flags clFlags, + D3D11Object& d3d11obj) + : // Call base classes constructors + D3D11Object(d3d11obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE1D, + clFlags, + getCLFormatFromDXGI(d3d11obj.getDxgiFormat(), d3d11obj.getPlane()), //format, + d3d11obj.getWidth(), + 1, + 1, + d3d11obj.getWidth() * d3d11obj.getElementBytes(), //rowPitch), + 0) + { + setInteropObj(this); + } + virtual ~Image1DD3D11() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Class Image2DD3D11 is derived from classes Image2D and D3D11Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D11 object +class Image2DD3D11 : public Image, public D3D11Object +{ +protected: + //! Initializes the device memory array which is nested + // after'Image2DD3D11' object in memory layout. + virtual void initDeviceMemory(); +public: +//! Image2DD3D11 constructor just calls constructors of base classes +//! to pass down the parameters + Image2DD3D11( + Context& amdContext, + cl_mem_flags clFlags, + D3D11Object& d3d11obj) + : // Call base classes constructors + D3D11Object(d3d11obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE2D, + clFlags, + getCLFormatFromDXGI(d3d11obj.getDxgiFormat(), d3d11obj.getPlane()), //format, + d3d11obj.getWidth(), + d3d11obj.getHeight(), + 1, + d3d11obj.getWidth() * d3d11obj.getElementBytes(), //rowPitch), + 0) + { + setInteropObj(this); + } + virtual ~Image2DD3D11() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Class Image3DD3D11 is derived from classes Image3D and D3D11Object +//! where the former keeps all data for CL object and +//! the latter keeps all data for D3D11 object +class Image3DD3D11 : public D3D11Object, public Image +{ +protected: + //! Initializes the device memory array which is nested + // after'Image3DD3D11' object in memory layout. + virtual void initDeviceMemory(); +public: +//! Image2DD3D11 constructor just calls constructors of base classes +//! to pass down the parameters + Image3DD3D11( + Context& amdContext, + cl_mem_flags clFlags, + D3D11Object& d3d11obj) + : // Call base classes constructors + D3D11Object(d3d11obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE3D, + clFlags, + getCLFormatFromDXGI(d3d11obj.getDxgiFormat(), d3d11obj.getPlane()), //format, + d3d11obj.getWidth(), + d3d11obj.getHeight(), + d3d11obj.getDepth(), + d3d11obj.getWidth() * d3d11obj.getElementBytes(), //rowPitch), + d3d11obj.getWidth() * d3d11obj.getHeight() * d3d11obj.getElementBytes()) + { + setInteropObj(this); + } + virtual ~Image3DD3D11() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +//! Functions for executing the D3D11 related stuff +cl_mem clCreateBufferFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + int* errcode_ret); +cl_mem clCreateImage1DFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + UINT subresource, + int* errcode_ret); +cl_mem clCreateImage2DFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + UINT subresource, + int* errcode_ret); +cl_mem clCreateImage3DFromD3D11ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + ID3D11Resource* pD3DResource, + UINT subresource, + int* errcode_ret); +void SyncD3D11Objects(std::vector& memObjects); +} //namespace amd + +#endif //CL_D3D11_AMD_HPP_ diff --git a/opencl/api/opencl/amdocl/cl_d3d9.cpp b/opencl/api/opencl/amdocl/cl_d3d9.cpp new file mode 100644 index 0000000000..ddbad29c25 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_d3d9.cpp @@ -0,0 +1,927 @@ +// +// Copyright (c) 2012 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifdef _WIN32 + +#include +#include "cl_common.hpp" +#include "cl_d3d9_amd.hpp" +#include "runtime/platform/command.hpp" + +#include +#include + +#define D3DFMT_NV_12 static_cast(MAKEFOURCC('N', 'V', '1', '2')) +#define D3DFMT_YV_12 static_cast(MAKEFOURCC('Y', 'V', '1', '2')) + + +RUNTIME_ENTRY(cl_int, clGetDeviceIDsFromDX9MediaAdapterKHR, ( + cl_platform_id platform, + cl_uint num_media_adapters, + cl_dx9_media_adapter_type_khr * media_adapters_type, + void * media_adapters, + cl_dx9_media_adapter_set_khr media_adapter_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices)) +{ + cl_int errcode; + //Accept an array of DX9 devices here as the spec mention of array of num_media_adapters size. + IDirect3DDevice9Ex** d3d9_device = static_cast(media_adapters); + cl_device_id* gpu_devices = NULL; + cl_uint num_gpu_devices = 0; + static const bool VALIDATE_ONLY = true; + + if (platform != NULL && platform != AMD_PLATFORM) { + LogWarning("\"platrform\" is not a valid AMD platform"); + return CL_INVALID_PLATFORM; + } + //check if input parameter are correct + if ( (num_media_adapters == 0) + ||(media_adapters_type == NULL) + ||(media_adapters == NULL) + ||(media_adapter_set != CL_PREFERRED_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR && media_adapter_set != CL_ALL_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR) + ||(num_entries == 0 && devices != NULL) + ) { + return CL_INVALID_VALUE; + } + // Get GPU devices + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, 0, NULL, &num_gpu_devices); + if (errcode != CL_SUCCESS && errcode != CL_DEVICE_NOT_FOUND) { + return CL_INVALID_VALUE; + } + + if (!num_gpu_devices) { + *not_null(num_devices) = 0; + return CL_DEVICE_NOT_FOUND; + } + + switch(media_adapter_set) { + case CL_PREFERRED_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR: + case CL_ALL_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR: { + gpu_devices = new cl_device_id[num_gpu_devices]; + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, num_gpu_devices, gpu_devices, NULL); + if (errcode != CL_SUCCESS) { + break; + } + + std::vector compatible_devices; + for (cl_uint i = 0; i < num_gpu_devices; ++i) { + cl_device_id device = gpu_devices[i]; + intptr_t context_type; + switch (media_adapters_type[i]) { + case CL_ADAPTER_D3D9_KHR: + context_type = CL_CONTEXT_ADAPTER_D3D9_KHR; + break; + case CL_ADAPTER_D3D9EX_KHR: + context_type = CL_CONTEXT_ADAPTER_D3D9EX_KHR; + break; + case CL_ADAPTER_DXVA_KHR: + context_type = CL_CONTEXT_ADAPTER_DXVA_KHR; + break; + } + + for (cl_uint j = 0; j < num_media_adapters; ++j) { + //Since there can be multiple DX9 adapters passed in the array we need to validate interopability with each. + if (is_valid(device) && (media_adapters_type[j] == CL_ADAPTER_D3D9EX_KHR) && + as_amd(device)->bindExternalDevice(context_type, + d3d9_device[j], NULL, VALIDATE_ONLY)) { + compatible_devices.push_back(as_amd(device)); + } + } + } + if (compatible_devices.size() == 0) { + *not_null(num_devices) = 0; + errcode = CL_DEVICE_NOT_FOUND; + break; + } + + std::vector::iterator it = compatible_devices.begin(); + cl_uint compatible_count = std::min(num_entries, (cl_uint)compatible_devices.size()); + + while (compatible_count--) { + *devices++ = as_cl(*it++); + --num_entries; + } + while (num_entries--) { + *devices++ = (cl_device_id) 0; + } + + *not_null(num_devices) = (cl_uint)compatible_devices.size(); + } + break; + + default: + LogWarning("\"d3d9_device_set\" is invalid"); + errcode = CL_INVALID_VALUE; + } + + delete[] gpu_devices; + return errcode; +} +RUNTIME_EXIT + +CL_API_ENTRY cl_mem CL_API_CALL \ +clCreateFromDX9MediaSurfaceKHR( + cl_context context, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + void * surface_info, + cl_uint plane, + cl_int * errcode_ret) +{ + cl_mem clMemObj = NULL; + + cl_dx9_surface_info_khr * cl_surf_info = NULL; + + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + + if(!flags) flags = CL_MEM_READ_WRITE; + if(!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + + if ((adapter_type != CL_ADAPTER_D3D9_KHR) + && (adapter_type != CL_ADAPTER_D3D9EX_KHR) + && (adapter_type != CL_ADAPTER_DXVA_KHR)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return clMemObj; + } + + if (!surface_info) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("parameter \"pD3DResource\" is a NULL pointer"); + return clMemObj; + } + + cl_surf_info = (cl_dx9_surface_info_khr *) surface_info; + IDirect3DSurface9 * pD3D9Resource = cl_surf_info->resource; + HANDLE shared_handle = cl_surf_info->shared_handle; + + if (!pD3D9Resource) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("parameter \"surface_info\" is a NULL pointer"); + return clMemObj; + } + + D3DSURFACE_DESC Desc; + pD3D9Resource->GetDesc(&Desc); + + if ((Desc.Format != D3DFMT_NV_12) && (Desc.Format != D3DFMT_YV_12) && + (plane != 0)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("The plane has to be Zero if the surface format is non-planar !"); + return clMemObj; + } + + // Check for image support + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().imageSupport_) { + supportPass = true; + } + } + if (!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return (cl_mem) 0; + } + //Verify the resource is a 2D image + return amd::clCreateImage2DFromD3D9ResourceAMD( + *as_amd(context), + flags, + adapter_type, + cl_surf_info, + plane, + errcode_ret); +} + +RUNTIME_ENTRY(cl_int, clEnqueueAcquireDX9MediaSurfacesKHR, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event)) +{ + return amd::clEnqueueAcquireExtObjectsAMD(command_queue, num_objects, + mem_objects, num_events_in_wait_list, event_wait_list, event, + CL_COMMAND_ACQUIRE_DX9_MEDIA_SURFACES_KHR); +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clEnqueueReleaseDX9MediaSurfacesKHR, ( + cl_command_queue command_queue, + cl_uint num_objects, + cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event)) +{ + return amd::clEnqueueReleaseExtObjectsAMD(command_queue, num_objects, + mem_objects, num_events_in_wait_list, event_wait_list, event, + CL_COMMAND_RELEASE_DX9_MEDIA_SURFACES_KHR); +} +RUNTIME_EXIT + +// +// +// namespace amd +// +// +namespace amd +{ +/*! @} + * \addtogroup CL-D3D9 interop helper functions + * @{ + */ +// +// Class D3D9Object implementation +// +std::vector> D3D9Object::resources_; +Monitor D3D9Object::resLock_; + +// +// clCreateImage2DFromD3D9ResourceAMD +// +cl_mem clCreateImage2DFromD3D9ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + cl_dx9_surface_info_khr* surface_info, + cl_uint plane, + int* errcode_ret) +{ + cl_dx9_surface_info_khr* cl_surf_info = + reinterpret_cast(surface_info); + IDirect3DSurface9 * pD3D9Resource = cl_surf_info->resource; + HANDLE shared_handle = cl_surf_info->shared_handle; + + D3D9Object obj; + cl_int errcode = D3D9Object::initD3D9Object(amdContext, + adapter_type, surface_info, plane, obj); + if (CL_SUCCESS != errcode) { + *not_null(errcode_ret) = errcode; + return (cl_mem) 0; + } + + Image2DD3D9 *pImage2DD3D9 = new (amdContext) + Image2DD3D9(amdContext, flags, obj); + if (!pImage2DD3D9) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem) 0; + } + if (!pImage2DD3D9->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImage2DD3D9->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImage2DD3D9); +} + +// +// Helper function SyncD3D9Objects +// +void SyncD3D9Objects(std::vector& memObjects) +{ + Memory*& mem = memObjects.front(); + if(!mem) { + LogWarning("\nNULL memory object\n"); + return; + } + InteropObject* interop = mem->getInteropObj(); + if(!interop) { + LogWarning("\nNULL interop object\n"); + return; + } + D3D9Object* d3d9Obj = interop->asD3D9Object(); + if(!d3d9Obj) { + LogWarning("\nNULL D3D9 object\n"); + return; + } + IDirect3DQuery9* query = d3d9Obj->getQuery(); + if(!query) { + LogWarning("\nNULL IDirect3DQuery9\n"); + return; + } + ScopedLock sl(d3d9Obj->getResLock()); + query->Issue(D3DISSUE_END); + BOOL data; + while(S_OK != query->GetData(&data, sizeof(BOOL), D3DGETDATA_FLUSH) && data != TRUE) + { + } +} + +// +// Class D3D10Object implementation +// +size_t +D3D9Object::getElementBytes(D3DFORMAT d3d9Format, cl_uint plane) +{ + size_t bytesPerPixel; + + switch(d3d9Format) { + case D3DFMT_UNKNOWN: + case D3DFMT_UYVY: + case D3DFMT_YUY2: + case D3DFMT_DXT1: + case D3DFMT_DXT2: + case D3DFMT_DXT3: + case D3DFMT_DXT4: + case D3DFMT_DXT5: + case D3DFMT_VERTEXDATA: + case D3DFMT_D32: + case D3DFMT_D15S1: + case D3DFMT_D24S8: + case D3DFMT_D24X8: + case D3DFMT_D24X4S4: + case D3DFMT_D16: + case D3DFMT_INDEX16: + case D3DFMT_INDEX32: + case D3DFMT_MULTI2_ARGB8: + case D3DFMT_CxV8U8: + // Less than 1 byte per pixel - needs special consideration + bytesPerPixel = 0; + break; + + case D3DFMT_R3G3B2: + case D3DFMT_P8: + case D3DFMT_A8: + case D3DFMT_L8: + case D3DFMT_A4L4: + bytesPerPixel = 1; + break; + + case D3DFMT_R16F: + case D3DFMT_R5G6B5: + case D3DFMT_X1R5G5B5: + case D3DFMT_A1R5G5B5: + case D3DFMT_A4R4G4B4: + case D3DFMT_A8R3G3B2: + case D3DFMT_X4R4G4B4: + case D3DFMT_A8P8: + case D3DFMT_A8L8: + case D3DFMT_V8U8: + case D3DFMT_L6V5U5: + case D3DFMT_D16_LOCKABLE: + case D3DFMT_L16: + bytesPerPixel = 2; + break; + + case D3DFMT_R8G8B8: + case D3DFMT_D24FS8: + bytesPerPixel = 3; + break; + + case D3DFMT_D32F_LOCKABLE: + case D3DFMT_A8R8G8B8: + case D3DFMT_R32F: + case D3DFMT_X8R8G8B8: + case D3DFMT_A2B10G10R10: + case D3DFMT_A8B8G8R8: + case D3DFMT_X8B8G8R8: + case D3DFMT_G16R16: + case D3DFMT_A2R10G10B10: + case D3DFMT_Q8W8V8U8: + case D3DFMT_X8L8V8U8: + case D3DFMT_V16U16: + case D3DFMT_A2W10V10U10: + case D3DFMT_R8G8_B8G8: + case D3DFMT_G8R8_G8B8: + case D3DFMT_G16R16F: + bytesPerPixel = 4; + break; + + case D3DFMT_G32R32F: + case D3DFMT_A16B16G16R16: + case D3DFMT_A16B16G16R16F: + case D3DFMT_Q16W16V16U16: + bytesPerPixel = 8; + break; + case D3DFMT_A32B32G32R32F: + bytesPerPixel = 16; + break; +//#if !defined(D3D_DISABLE_9EX) + //case D3DFMT_D32_LOCKABLE: + //case D3DFMT_S8_LOCKABLE: +//#endif // !D3D_DISABLE_9EX + case D3DFMT_NV_12: + if (plane == 0) { + bytesPerPixel = 1; + } + else if(plane == 1) { + bytesPerPixel = 2; + } //plane != 0 or != 1 shouldn't happen here + break; + case D3DFMT_YV_12: + bytesPerPixel = 1; + break; + + default: + bytesPerPixel = 0; + _ASSERT(FALSE); + break; + } + return bytesPerPixel; +} + +void setObjDesc(amd::D3D9ObjDesc_t& objDesc, D3DSURFACE_DESC& resDesc, cl_uint plane) +{ + objDesc.d3dPool_ = resDesc.Pool; + objDesc.resType_ = resDesc.Type; + objDesc.usage_ = resDesc.Usage; + objDesc.d3dFormat_ = resDesc.Format; + switch (resDesc.Format) { + case D3DFMT_NV_12: + objDesc.surfRect_.left = 0; + objDesc.surfRect_.top = 0; + if (plane == 0) { + objDesc.objSize_.Height = resDesc.Height; + objDesc.objSize_.Width = resDesc.Width; + objDesc.surfRect_.right = resDesc.Width; //resDesc.Width/2-1; + objDesc.surfRect_.bottom = 3*resDesc.Height/2;; //3*resDesc.Height/2-1; + } + else if(plane == 1) { + objDesc.objSize_.Height = resDesc.Height/2; + objDesc.objSize_.Width = resDesc.Width/2; + objDesc.surfRect_.right = resDesc.Width; //resDesc.Width/2-1; + objDesc.surfRect_.bottom = 3*resDesc.Height/2;; //3*resDesc.Height/2-1; + } //plane != 0 or != 1 shouldn't happen here + break; + case D3DFMT_YV_12: + objDesc.surfRect_.left = 0; + if (plane == 0) { + objDesc.objSize_.Height = resDesc.Height; + objDesc.objSize_.Width = resDesc.Width; + objDesc.surfRect_.top = 0; + objDesc.surfRect_.right = resDesc.Width-1; + objDesc.surfRect_.bottom = resDesc.Height-1; + } + else if(plane == 1) { + objDesc.objSize_.Height = resDesc.Height/2; + objDesc.objSize_.Width = resDesc.Width/2; + objDesc.surfRect_.top = resDesc.Height; + objDesc.surfRect_.right = resDesc.Width/2-1; + objDesc.surfRect_.bottom = 3*resDesc.Height/2-1; + } + else if(plane == 2) { + objDesc.objSize_.Height = resDesc.Height/2; + objDesc.objSize_.Width = resDesc.Width/2; + objDesc.surfRect_.top = 3*resDesc.Height/2; + objDesc.surfRect_.right = resDesc.Width/2-1; + objDesc.surfRect_.bottom = 2*resDesc.Height-1; + } //plane > 0 or > 2 shouldn't happen here + break; + default: + objDesc.objSize_.Height = resDesc.Height; + objDesc.objSize_.Width = resDesc.Width; + objDesc.surfRect_.left = 0; + objDesc.surfRect_.top = 0; + objDesc.surfRect_.right = resDesc.Width-1; + objDesc.surfRect_.bottom = resDesc.Height-1; + break; + } +} + +int +D3D9Object::initD3D9Object(const Context& amdContext, cl_dx9_media_adapter_type_khr adapter_type, + cl_dx9_surface_info_khr* cl_surf_info, + cl_uint plane, D3D9Object& obj) +{ + ScopedLock sl(resLock_); + + IDirect3DDevice9Ex* pDev9Ex = NULL; + cl_int errcode = CL_SUCCESS; + + // Check if this ressource has already been used for interop + IDirect3DSurface9 * pD3D9res = cl_surf_info->resource; + HANDLE shared_handle = cl_surf_info->shared_handle; + + if ((adapter_type == CL_ADAPTER_D3D9_KHR) || + (adapter_type == CL_ADAPTER_DXVA_KHR)) { + return CL_INVALID_DX9_MEDIA_ADAPTER_KHR; // Not supported yet + } + + std::vector>::iterator it; + for(it = resources_.begin(); it != resources_.end(); ++it) { + if((*it).first.surfInfo.resource == cl_surf_info->resource && (*it).first.surfPlane == plane) { + return CL_INVALID_D3D9_RESOURCE_KHR; + } + } + + HRESULT hr; + D3DQUERYTYPE desc = D3DQUERYTYPE_EVENT; + + D3DSURFACE_DESC resDesc; + if (D3D_OK != pD3D9res->GetDesc(&resDesc)) { + return CL_INVALID_D3D9_RESOURCE_KHR; + } + + hr = pD3D9res->GetContainer(IID_IDirect3DDevice9Ex, (void**)&pDev9Ex); + if (hr == D3D_OK) { + pDev9Ex->CreateQuery(desc, &(obj.pQuery_)); + } + else { + return CL_INVALID_D3D9_RESOURCE_KHR; // d3d9ex should be supported + } + + obj.handleShared_ = shared_handle; + obj.surfPlane_ = plane; + obj.surfInfo_ = *cl_surf_info; + obj.adapterType_ = adapter_type; + + // Init defaults + setObjDesc(obj.objDescOrig_, resDesc, plane); + obj.objDesc_ = obj.objDescOrig_; + + // shared handle cases if the shared_handle is NULL + // first check if the format is NV12 or YV12, which we need special handling + if (NULL == shared_handle) { + bool found = false; + std::vector>::iterator it; + for(it = resources_.begin(); it != resources_.end(); ++it) { + if((*it).first.surfInfo.resource == cl_surf_info->resource && + (*it).first.surfPlane != plane) { + obj.handleShared_ = (*it).second.surfInfo.shared_handle; + obj.pD3D9Res_ = (*it).second.surfInfo.resource; + obj.pD3D9Res_->AddRef(); + obj.objDesc_ = obj.objDescOrig_; + found = true; + break; + } + } + if (!found) { + obj.handleShared_ = 0; + hr = pDev9Ex->CreateOffscreenPlainSurface( + resDesc.Width, + resDesc.Height, + resDesc.Format, + resDesc.Pool, + &obj.pD3D9Res_, + &obj.handleShared_); + + if (D3D_OK != hr) { + errcode = CL_INVALID_D3D9_RESOURCE_KHR; + } + } + + // put the original info into the obj + obj.pD3D9ResOrig_ = pD3D9res; + obj.pD3D9ResOrig_->AddRef(); //addRef in case lost the resource + } + else { + // Share the original resource + obj.pD3D9ResOrig_ = NULL; + obj.pD3D9Res_ = pD3D9res; + obj.pD3D9Res_->AddRef(); + } + + // Release the Ex interface + if (pDev9Ex) pDev9Ex->Release(); + + // Check for CL format compatibilty + if(obj.objDesc_.resType_ == D3DRTYPE_SURFACE) { + cl_image_format clFmt = + obj.getCLFormatFromD3D9(obj.objDesc_.d3dFormat_, plane); + amd::Image::Format imageFormat(clFmt); + if(!imageFormat.isSupported(amdContext)) { + return CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + } + } + + TD3D9RESINFO d3d9ObjOri = {*cl_surf_info, plane}; + TD3D9RESINFO d3d9ObjShared = {{obj.pD3D9Res_, obj.handleShared_}, plane}; + + if (errcode == CL_SUCCESS) { + resources_.push_back(std::make_pair(d3d9ObjOri, d3d9ObjShared)); + } + + return errcode; +} +cl_uint +D3D9Object::getMiscFlag() +{ + switch (objDescOrig_.d3dFormat_) { + case D3DFMT_NV_12: + return 1; + break; + case D3DFMT_YV_12: + return 2; + break; + default: + return 0; + break; + } +} + +cl_image_format +D3D9Object::getCLFormatFromD3D9() +{ + return getCLFormatFromD3D9(objDesc_.d3dFormat_, surfPlane_); +} + +cl_image_format +D3D9Object::getCLFormatFromD3D9(D3DFORMAT d3d9Fmt, cl_uint plane) +{ + cl_image_format fmt; + + fmt.image_channel_order = 0;//CL_RGBA; + fmt.image_channel_data_type = 0;//CL_UNSIGNED_INT8; + + switch(d3d9Fmt) + { + case D3DFMT_R32F: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case D3DFMT_R16F: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case D3DFMT_L16: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case D3DFMT_A8: + fmt.image_channel_order = CL_A; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_L8: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_G32R32F: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case D3DFMT_G16R16F: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case D3DFMT_G16R16: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case D3DFMT_A8L8: + fmt.image_channel_order = CL_RG; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_A32B32G32R32F: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_FLOAT; + break; + + case D3DFMT_A16B16G16R16F: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_HALF_FLOAT; + break; + + case D3DFMT_A16B16G16R16: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT16; + break; + + case D3DFMT_A8B8G8R8: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_X8B8G8R8: + fmt.image_channel_order = CL_RGBA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_A8R8G8B8: + fmt.image_channel_order = CL_BGRA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_X8R8G8B8: + fmt.image_channel_order = CL_BGRA; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + case D3DFMT_NV_12: + fmt.image_channel_data_type = CL_UNORM_INT8; + if (plane == 0) { + fmt.image_channel_order = CL_R; + } + else if(plane == 1) { + fmt.image_channel_order = CL_RG; + } + break; + case D3DFMT_YV_12: + fmt.image_channel_order = CL_R; + fmt.image_channel_data_type = CL_UNORM_INT8; + break; + + case D3DFMT_UNKNOWN: + case D3DFMT_R8G8B8: + case D3DFMT_R5G6B5: + case D3DFMT_X1R5G5B5: + case D3DFMT_A1R5G5B5: + case D3DFMT_A4R4G4B4: + case D3DFMT_R3G3B2: + case D3DFMT_A8R3G3B2: + case D3DFMT_X4R4G4B4: + case D3DFMT_A2B10G10R10: + case D3DFMT_A2R10G10B10: + case D3DFMT_A8P8: + case D3DFMT_P8: + case D3DFMT_A4L4: + case D3DFMT_V8U8: + case D3DFMT_L6V5U5: + case D3DFMT_X8L8V8U8: + case D3DFMT_Q8W8V8U8: + case D3DFMT_V16U16: + case D3DFMT_A2W10V10U10: + case D3DFMT_UYVY: + case D3DFMT_R8G8_B8G8: + case D3DFMT_YUY2: + case D3DFMT_G8R8_G8B8: + case D3DFMT_DXT1: + case D3DFMT_DXT2: + case D3DFMT_DXT3: + case D3DFMT_DXT4: + case D3DFMT_DXT5: + case D3DFMT_D16_LOCKABLE: + case D3DFMT_D32: + case D3DFMT_D15S1: + case D3DFMT_D24S8: + case D3DFMT_D24X8: + case D3DFMT_D24X4S4: + case D3DFMT_D16: + case D3DFMT_D32F_LOCKABLE: + case D3DFMT_D24FS8: +//#if !defined(D3D_DISABLE_9EX) + case D3DFMT_D32_LOCKABLE: + case D3DFMT_S8_LOCKABLE: +//#endif // !D3D_DISABLE_9EX + case D3DFMT_VERTEXDATA: + case D3DFMT_INDEX16: + case D3DFMT_INDEX32: + case D3DFMT_Q16W16V16U16: + case D3DFMT_MULTI2_ARGB8: + case D3DFMT_CxV8U8: +//#if !defined(D3D_DISABLE_9EX) + case D3DFMT_A1: + case D3DFMT_A2B10G10R10_XR_BIAS: + case D3DFMT_BINARYBUFFER: + _ASSERT(FALSE); //NOT SURPPORTED + break; +//#endif // !D3D_DISABLE_9EX + default: + _ASSERT(FALSE); + break; + } + + return fmt; +} + +bool +D3D9Object::copyOrigToShared() +{ + // Don't copy if there is no orig + if (NULL == getD3D9ResOrig()) return true; + + IDirect3DDevice9Ex *d3dDev; + HRESULT hr; + ScopedLock sl(getResLock()); + + IDirect3DSurface9* srcSurf = getD3D9ResOrig(); + IDirect3DSurface9* dstSurf = getD3D9Resource(); + + hr = getD3D9Resource()->GetContainer(IID_IDirect3DDevice9Ex, (void**)&d3dDev); + if (hr != D3D_OK || !d3dDev) { + LogError("\nCannot get D3D9 device from D3D9 surface\n"); + return false; + } + + hr = d3dDev->StretchRect(srcSurf, NULL, dstSurf, NULL, D3DTEXF_NONE); + if (hr != D3D_OK ) { + LogError("\ncopy original surface to shared surface failed\n"); + return false; + } + // Flush D3D queues and make sure D3D stuff is finished + pQuery_->Issue(D3DISSUE_END); + BOOL data; + while ((D3D_OK != pQuery_->GetData(&data, sizeof(BOOL), D3DGETDATA_FLUSH)) && + (data != TRUE)) { + } + + if (d3dDev) d3dDev->Release(); + return true; +} + +bool +D3D9Object::copySharedToOrig() +{ + // Don't copy if there is no orig + if (NULL == getD3D9ResOrig()) return true; + + IDirect3DDevice9Ex *d3dDev; + HRESULT hr; + ScopedLock sl(getResLock()); + + hr = getD3D9Resource()->GetContainer(IID_IDirect3DDevice9Ex, (void**)&d3dDev); + if(hr != D3D_OK || !d3dDev) { + LogError("\nCannot get D3D9 device from D3D9 surface\n"); + return false; + } + + hr = d3dDev->StretchRect(getD3D9Resource(), NULL, getD3D9ResOrig(), NULL, D3DTEXF_NONE); + if(hr != D3D_OK ) { + LogError("\ncopy shared surface to original surface failed\n"); + return false; + } + + if (d3dDev) d3dDev->Release(); + return true; +} + +void +Image2DD3D9::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(Image2DD3D9)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +Image2DD3D9::mapExtObjectInCQThread() +{ + void* pCpuMem = NULL; + HRESULT hr; + DWORD lockFlags = 0; + + if (getMemFlags() & CL_MEM_READ_WRITE) { + lockFlags = 0; + } + else if (getMemFlags() & CL_MEM_READ_ONLY) { + lockFlags = D3DLOCK_READONLY; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + lockFlags = D3DLOCK_DISCARD; + } + else { + // Should not get here, the flags had been checked before + LogError("\nInvalid memrory flags"); + return false; + } + ScopedLock sl(getResLock()); + + D3DLOCKED_RECT lockedRect; + hr = getD3D9Resource()->LockRect( + &lockedRect, + NULL, + lockFlags); + if ((hr != D3D_OK) || !lockedRect.pBits) { + LogError("Cannot lock D3D9 surface for CPU access"); + return false; + } + + setHostMem(lockedRect.pBits); + return true; +} + +bool +Image2DD3D9::unmapExtObjectInCQThread() +{ + HRESULT hr; + ScopedLock sl(getResLock()); + hr = getD3D9Resource()->UnlockRect(); + if (hr != D3D_OK) { + LogError("Cannot unlock D3D9 surface"); + return false; + } + + setHostMem(NULL); + return true; +} + +} //namespace amd + +#endif //_WIN32 diff --git a/opencl/api/opencl/amdocl/cl_d3d9_amd.hpp b/opencl/api/opencl/amdocl/cl_d3d9_amd.hpp new file mode 100644 index 0000000000..1a83f29842 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_d3d9_amd.hpp @@ -0,0 +1,333 @@ +/* ============================================================ + +Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +/* $Revision$ on $Date$ */ + +#ifndef __OPENCL_CL_D3D9_AMD_H +#define __OPENCL_CL_D3D9_AMD_H + +#include "CL/cl_dx9_media_sharing.h" +#include +#include "platform/context.hpp" +#include "platform/memory.hpp" + +#include + +/* cl_amd_d3d9_sharing extension */ +#define cl_amd_d3d9_sharing 1 + +/* cl_amd_d3d9_sharing error codes */ +#define CL_INVALID_D3D9_DEVICE_KHR -1021 +#define CL_INVALID_D3D9_RESOURCE_KHR -1022 + +/* cl_amd_d3d9_sharing enumerations */ +#define CL_CONTEXT_D3D9_DEVICE_KHR 0x4039 + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromDX9MediaAdapterKHR( + cl_platform_id, + cl_uint, + cl_dx9_media_adapter_type_khr *, + void *, + cl_dx9_media_adapter_set_khr, + cl_uint, + cl_device_id *, + cl_uint *); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromDX9MediaSurfaceKHR( + cl_context, + cl_mem_flags, + cl_dx9_media_adapter_type_khr, + void *, + cl_uint, + cl_int *); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireDX9MediaSurfacesKHR( + cl_command_queue, + cl_uint, + const cl_mem *, + cl_uint, + const cl_event *, + cl_event *); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseDX9MediaSurfacesKHR( + cl_command_queue, + cl_uint, + cl_mem *, + cl_uint, + const cl_event *, + cl_event *); + +namespace amd +{ +typedef struct +{ + union + { + UINT ByteWidth; + UINT Width; + }; + UINT Height; + UINT Depth; +} D3D9ObjSize_t; + +typedef struct +{ + D3D9ObjSize_t objSize_; + D3DFORMAT d3dFormat_; + D3DRESOURCETYPE resType_; + UINT usage_; + D3DPOOL d3dPool_; + D3DMULTISAMPLE_TYPE msType_; + UINT msQuality_; + UINT mipLevels_; + UINT fvf_; + RECT surfRect_; +} D3D9ObjDesc_t; + +typedef struct d3d9ResInfo { + cl_dx9_surface_info_khr surfInfo; + cl_uint surfPlane; +} TD3D9RESINFO; + + +//typedef std::pair TD3D9OBJINFO; + +//! Class D3D9Object keeps all the info about the D3D9 object +//! from which the CL object is created +class D3D9Object : public InteropObject +{ +private: + IDirect3DSurface9* pD3D9Aux_; + cl_int cliChecksum_; + bool releaseResources_; + static bool createSharedResource(D3D9Object& obj); + static std::vector> resources_; + + //!Global lock + static Monitor resLock_; + cl_uint surfPlane_; + cl_dx9_surface_info_khr surfInfo_; + +protected: + IDirect3DSurface9* pD3D9Res_; + IDirect3DSurface9* pD3D9ResOrig_; + IDirect3DQuery9* pQuery_; + D3D9ObjDesc_t objDesc_; + D3D9ObjDesc_t objDescOrig_; + HANDLE handleOrig_; + HANDLE handleShared_; + RECT srcSurfRect; + RECT SharedSurfRect; + cl_dx9_media_adapter_type_khr adapterType_; + +public: +//! D3D9Object constructor initializes memeber variables + D3D9Object() + : releaseResources_(false), + pQuery_(NULL) + { + // @todo Incorrect initialization!!! + memset(this, 0, sizeof(D3D9Object)); + } + //copy constructor + D3D9Object(D3D9Object& d3d9obj) + :pQuery_(NULL) + { + *this = d3d9obj; + this->releaseResources_ = true; + } + + //virtual destructor + virtual ~D3D9Object() + { + ScopedLock sl(resLock_); + if(releaseResources_) { + if(pD3D9ResOrig_) pD3D9ResOrig_->Release(); + if(pD3D9Res_) pD3D9Res_->Release(); + if(pD3D9Aux_) pD3D9Aux_->Release(); + if(pQuery_) pQuery_->Release(); + //if the resouce is being used + std::vector>::iterator it; + if(resources_.size()) { + for(it = resources_.begin(); it != resources_.end(); ++it) { + if( surfInfo_.resource && + ((*it).first.surfInfo.resource == surfInfo_.resource) && + ((*it).first.surfPlane == surfPlane_)) { + resources_.erase(it); + break; + } + } + } + } + } + static int initD3D9Object(const Context& amdContext, cl_dx9_media_adapter_type_khr adapter_type, + cl_dx9_surface_info_khr* cl_surf_info, cl_uint plane, D3D9Object& obj); + cl_uint getMiscFlag(void); + + D3D9Object* asD3D9Object() {return this;} + IDirect3DSurface9* getD3D9Resource() const {return pD3D9Res_;} + HANDLE getD3D9SharedHandle() const {return handleShared_;} + IDirect3DSurface9* getD3D9ResOrig() const {return pD3D9ResOrig_;} + RECT* getSrcSurfRect() {return &objDesc_.surfRect_;} + RECT* getSharedSurfRect() {return &objDescOrig_.surfRect_;} + void setD3D9AuxRes(IDirect3DSurface9* pAux) {pD3D9Aux_ = pAux;} + IDirect3DSurface9* getD3D9AuxRes() {return pD3D9Aux_;} + IDirect3DQuery9* getQuery() const {return pQuery_;} + Monitor & getResLock() { return resLock_;} + UINT getWidth() const {return objDesc_.objSize_.Width;} + UINT getHeight() const {return objDesc_.objSize_.Height;} + cl_uint getPlane() const {return surfPlane_;} + cl_dx9_media_adapter_type_khr getAdapterType() const { return adapterType_;}; + const cl_dx9_surface_info_khr& getSurfInfo() const {return surfInfo_;}; + size_t getElementBytes(D3DFORMAT d3d9Format, cl_uint plane); + size_t getElementBytes() {return getElementBytes(objDesc_.d3dFormat_, surfPlane_);} + D3DFORMAT getD3D9Format() {return objDesc_.d3dFormat_;} + D3D9ObjDesc_t* getObjDesc() {return &objDesc_;} + cl_image_format getCLFormatFromD3D9(); + cl_image_format getCLFormatFromD3D9(D3DFORMAT d3d9Fmt, cl_uint plane); + // On acquire copy data from original resource to shared resource + virtual bool copyOrigToShared(); + // On release copy data from shared copy to the original resource + virtual bool copySharedToOrig(); +}; + +class Image2DD3D9 : public D3D9Object , public Image +{ +protected: + //! Initializes the device memory array which is nested + // after'Image2DD3D9' object in memory layout. + virtual void initDeviceMemory(); +public: +//! Image2DD3D9 constructor just calls constructors of base classes +//! to pass down the parameters + Image2DD3D9( + Context& amdContext, + cl_mem_flags clFlags, + D3D9Object& d3d9obj) + : // Call base classes constructors + D3D9Object(d3d9obj), + Image( + amdContext, + CL_MEM_OBJECT_IMAGE2D, + clFlags, + d3d9obj.getCLFormatFromD3D9(), + d3d9obj.getWidth(), + d3d9obj.getHeight(), + 1, + d3d9obj.getWidth() * d3d9obj.getElementBytes(), //rowPitch), + 0) + { + setInteropObj(this); + } + virtual ~Image2DD3D9() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + +cl_mem clCreateImage2DFromD3D9ResourceAMD( + Context& amdContext, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + cl_dx9_surface_info_khr* surface_info, + cl_uint plane, + int* errcode_ret); + +void SyncD3D9Objects(std::vector& memObjects); + +} //namespace amd + +#endif /* __OPENCL_CL_D3D9_AMD_H */ diff --git a/opencl/api/opencl/amdocl/cl_device.cpp b/opencl/api/opencl/amdocl/cl_device.cpp new file mode 100644 index 0000000000..ce29653566 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_device.cpp @@ -0,0 +1,797 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "device/device.hpp" +#include "platform/runtime.hpp" +#include "utils/versions.hpp" +#include "cl_semaphore_amd.h" + +#include "CL/cl_ext.h" + +#include // for alloca + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_PlatformInfo + * @{ + */ + +/*! \brief Get the list of available platforms. + * + * \param num_entries is the number of cl_platform_id entries that can be added + * to platforms. If \a platforms is not NULL, the \a num_entries must be greater + * than zero. + * + * \param platforms returns a list of OpenCL platforms found. The cl_platform_id + * values returned in \a platforms can be used to identify a specific OpenCL + * platform. If \a platforms argument is NULL, this argument is ignored. The + * number of OpenCL platforms returned is the mininum of the value specified by + * \a num_entries or the number of OpenCL platforms available. + * + * \param num_platforms returns the number of OpenCL platforms available. If + * \a num_platforms is NULL, this argument is ignored. + * + * \return CL_INVALID_VALUE if num_entries is equal to zero and platforms is not + * NULL or if both num_platforms and platforms are NULL, and returns CL_SUCCESS + * if the function is executed successfully. + * + * \version 1.0r33 + */ + +RUNTIME_ENTRY(cl_int, clGetPlatformIDs, ( + cl_uint num_entries, + cl_platform_id *platforms, + cl_uint *num_platforms)) +{ + if (!amd::Runtime::initialized()) { + amd::Runtime::init(); + } + + if (((num_entries > 0 || num_platforms == NULL) && platforms == NULL) + || (num_entries == 0 && platforms != NULL)) { + return CL_INVALID_VALUE; + } + if (num_platforms != NULL && platforms == NULL) { + *num_platforms = 1; + return CL_SUCCESS; + } + + assert(platforms != NULL && "check the code above"); + *platforms = AMD_PLATFORM; + + *not_null(num_platforms) = 1; + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Get specific information about the OpenCL platform. + * + * \param param_name is an enum that identifies the platform information being + * queried. + * + * \param param_value is a pointer to memory location where appropriate values + * for a given \a param_name will be returned. If \a param_value is NULL, + * it is ignored. + * + * \param param_value_size specifies the size in bytes of memory pointed to by + * \a param_value. This size in bytes must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_INVALID_VALUE if \a param_name is not one of the supported + * values or if size in bytes specified by \a param_value_size is < size of + * return type and \a param_value is not a NULL value. + * - CL_SUCCESS if the function is executed successfully. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetPlatformInfo, ( + cl_platform_id platform, + cl_platform_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret)) +{ + if (platform != NULL && platform != AMD_PLATFORM) { + return CL_INVALID_PLATFORM; + } + + const char* value = NULL; + switch (param_name) { + case CL_PLATFORM_PROFILE: + value = "FULL_PROFILE"; + break; + case CL_PLATFORM_VERSION: + value = "OpenCL " XSTR(OPENCL_MAJOR) "." XSTR(OPENCL_MINOR) \ + " " AMD_PLATFORM_INFO; + break; + case CL_PLATFORM_NAME: + value = AMD_PLATFORM_NAME; + break; + case CL_PLATFORM_VENDOR: + value = "Advanced Micro Devices, Inc."; + break; + case CL_PLATFORM_EXTENSIONS: + value = "cl_khr_icd " NOT_MAINLINE("cl_amd_object_metadata ") +#ifdef _WIN32 + "cl_khr_d3d10_sharing " + "cl_khr_d3d11_sharing " + "cl_khr_dx9_media_sharing " +#endif //_WIN32 + "cl_amd_event_callback cl_amd_offline_devices "; + break; + case CL_PLATFORM_ICD_SUFFIX_KHR: + value = "AMD"; + break; + case CL_PLATFORM_MAX_KEYS_AMD: { + size_t max_keys = OCL_MAX_KEYS; + return amd::clGetInfo( + max_keys, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + if (value != NULL) { + return amd::clGetInfo( + value, param_value_size, param_value, param_value_size_ret); + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_Devices + * @{ + */ + +/*! \brief Get the list of available devices. + * + * \param device_type is a bitfield that identifies the type of OpenCL device. + * The \a device_type can be used to query specific OpenCL devices or all + * OpenCL devices available. + * + * \param num_entries is the number of cl_device_id entries that can be added + * to devices. If devices is not NULL, the \a num_entries must be greater than + * zero. + * + * \param devices returns a list of OpenCL devices found. The cl_device_id + * values returned in devices can be used to identify a specific OpenCL device. + * If \a devices argument is NULL, this argument is ignored. The number of + * OpenCL devices returned is the mininum of value specified by \a num_entries + * or the number of OpenCL devices whose type matches device_type. + * + * \param num_devices returns the number of OpenCL devices available that match + * device_type. If \a num_devices is NULL, this argument is ignored. + * + * \return One of the following values: + * - CL_INVALID_DEVICE_TYPE if \a device_type is not a valid value. + * - CL_INVALID_VALUE if \a num_entries is equal to zero and devices is + * not NULL or if both \a num_devices and \a devices are NULL. + * - CL_DEVICE_ NOT_FOUND if no OpenCL devices that matched \a device_type + * were found. + * - CL_SUCCESS if the function is executed successfully. + * + * The application can query specific capabilities of the OpenCL device(s) + * returned by clGetDeviceIDs. This can be used by the application to + * determine which device(s) to use. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetDeviceIDs, ( + cl_platform_id platform, + cl_device_type device_type, + cl_uint num_entries, + cl_device_id *devices, + cl_uint *num_devices)) +{ + if (platform != NULL && platform != AMD_PLATFORM) { + return CL_INVALID_PLATFORM; + } + + if (((num_entries > 0 || num_devices == NULL) && devices == NULL) + || (num_entries == 0 && devices != NULL)) { + return CL_INVALID_VALUE; + } + + if (device_type & CL_HSA_ENABLED_AMD && + device_type & CL_HSA_DISABLED_AMD) { + device_type ^= (CL_HSA_ENABLED_AMD | CL_HSA_DISABLED_AMD); + } + + // Get all available devices + if (!amd::Device::getDeviceIDs(device_type, num_entries, + devices, num_devices, false)) { + return CL_DEVICE_NOT_FOUND; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \fn clGetDeviceInfo + * + * \brief Get specific information about an OpenCL device. + * + * \param device is a device returned by clGetDeviceIDs. + * + * \param param_name is an enum that identifies the device information being + * queried. + * + * \param param_value is a pointer to memory location where appropriate values + * for a given \a param_name will be returned. If \a param_value is NULL, + * it is ignored. + * + * \param param_value_size specifies the size in bytes of memory pointed to + * by \a param_value. This size in bytes must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_INVALID_DEVICE if device is not valid. + * - CL_INVALID_VALUE if param_name is not one of the supported values + * or if size in bytes specified by \a param_value_size is < size of return + * type and \a param_value is not a NULL value. + * - CL_SUCCESS if the function is executed successfully. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetDeviceInfo, ( + cl_device_id device, + cl_device_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + +#define CASE(param_name, field_name) \ + case param_name: \ + return amd::clGetInfo( \ + as_amd(device)->info().field_name, \ + param_value_size, \ + param_value, \ + param_value_size_ret); + + switch (param_name) { + case CL_DEVICE_TYPE: { + // For cl_device_type, we need to mask out the default bit. + cl_device_type device_type = as_amd(device)->type(); + return amd::clGetInfo( + device_type, param_value_size, param_value, param_value_size_ret); + } + CASE(CL_DEVICE_VENDOR_ID, vendorId_); + CASE(CL_DEVICE_MAX_COMPUTE_UNITS, maxComputeUnits_); + CASE(CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS, maxWorkItemDimensions_); + CASE(CL_DEVICE_MAX_WORK_GROUP_SIZE, maxWorkGroupSize_); + CASE(CL_DEVICE_MAX_WORK_ITEM_SIZES, maxWorkItemSizes_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR, preferredVectorWidthChar_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT, preferredVectorWidthShort_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT, preferredVectorWidthInt_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG, preferredVectorWidthLong_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT, preferredVectorWidthFloat_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE, preferredVectorWidthDouble_); + CASE(CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF, preferredVectorWidthDouble_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR, nativeVectorWidthChar_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT, nativeVectorWidthShort_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_INT, nativeVectorWidthInt_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG, nativeVectorWidthLong_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT, nativeVectorWidthFloat_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE, nativeVectorWidthDouble_); + CASE(CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF, nativeVectorWidthDouble_); + CASE(CL_DEVICE_MAX_CLOCK_FREQUENCY, maxClockFrequency_); + CASE(CL_DEVICE_ADDRESS_BITS, addressBits_); + CASE(CL_DEVICE_MAX_READ_IMAGE_ARGS, maxReadImageArgs_); + CASE(CL_DEVICE_MAX_WRITE_IMAGE_ARGS, maxWriteImageArgs_); + CASE(CL_DEVICE_MAX_READ_WRITE_IMAGE_ARGS, maxReadWriteImageArgs_); + CASE(CL_DEVICE_MAX_MEM_ALLOC_SIZE, maxMemAllocSize_); + CASE(CL_DEVICE_IMAGE2D_MAX_WIDTH, image2DMaxWidth_); + CASE(CL_DEVICE_IMAGE2D_MAX_HEIGHT, image2DMaxHeight_); + CASE(CL_DEVICE_IMAGE3D_MAX_WIDTH, image3DMaxWidth_); + CASE(CL_DEVICE_IMAGE3D_MAX_HEIGHT, image3DMaxHeight_); + CASE(CL_DEVICE_IMAGE3D_MAX_DEPTH, image3DMaxDepth_); + CASE(CL_DEVICE_IMAGE_SUPPORT, imageSupport_); + CASE(CL_DEVICE_MAX_PARAMETER_SIZE, maxParameterSize_); + CASE(CL_DEVICE_MAX_SAMPLERS, maxSamplers_); + CASE(CL_DEVICE_MEM_BASE_ADDR_ALIGN, memBaseAddrAlign_); + CASE(CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE, minDataTypeAlignSize_); + CASE(CL_DEVICE_HALF_FP_CONFIG, halfFPConfig_); + CASE(CL_DEVICE_SINGLE_FP_CONFIG, singleFPConfig_); + CASE(CL_DEVICE_DOUBLE_FP_CONFIG, doubleFPConfig_); + CASE(CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, globalMemCacheType_); + CASE(CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE, globalMemCacheLineSize_); + CASE(CL_DEVICE_GLOBAL_MEM_CACHE_SIZE, globalMemCacheSize_); + CASE(CL_DEVICE_GLOBAL_MEM_SIZE, globalMemSize_); + CASE(CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE, maxConstantBufferSize_); + CASE(CL_DEVICE_MAX_CONSTANT_ARGS, maxConstantArgs_); + CASE(CL_DEVICE_LOCAL_MEM_TYPE, localMemType_); + CASE(CL_DEVICE_LOCAL_MEM_SIZE, localMemSize_); + CASE(CL_DEVICE_ERROR_CORRECTION_SUPPORT, errorCorrectionSupport_); + CASE(CL_DEVICE_HOST_UNIFIED_MEMORY, hostUnifiedMemory_); + CASE(CL_DEVICE_PROFILING_TIMER_RESOLUTION, profilingTimerResolution_); + CASE(CL_DEVICE_PROFILING_TIMER_OFFSET_AMD, profilingTimerOffset_); + CASE(CL_DEVICE_ENDIAN_LITTLE, littleEndian_); + CASE(CL_DEVICE_AVAILABLE, available_); + CASE(CL_DEVICE_COMPILER_AVAILABLE, compilerAvailable_); + CASE(CL_DEVICE_EXECUTION_CAPABILITIES, executionCapabilities_); + CASE(CL_DEVICE_SVM_CAPABILITIES, svmCapabilities_); + CASE(CL_DEVICE_PREFERRED_PLATFORM_ATOMIC_ALIGNMENT, preferredPlatformAtomicAlignment_); + CASE(CL_DEVICE_PREFERRED_GLOBAL_ATOMIC_ALIGNMENT, preferredGlobalAtomicAlignment_); + CASE(CL_DEVICE_PREFERRED_LOCAL_ATOMIC_ALIGNMENT, preferredLocalAtomicAlignment_); + CASE(CL_DEVICE_QUEUE_ON_HOST_PROPERTIES, queueProperties_); + CASE(CL_DEVICE_PLATFORM, platform_); + CASE(CL_DEVICE_NAME, name_); + CASE(CL_DEVICE_VENDOR, vendor_); + CASE(CL_DRIVER_VERSION, driverVersion_); + CASE(CL_DEVICE_PROFILE, profile_); + CASE(CL_DEVICE_VERSION, version_); + CASE(CL_DEVICE_OPENCL_C_VERSION, oclcVersion_); + CASE(CL_DEVICE_EXTENSIONS, extensions_); + CASE(CL_DEVICE_MAX_ATOMIC_COUNTERS_EXT, maxAtomicCounters_); + CASE(CL_DEVICE_TOPOLOGY_AMD, deviceTopology_); + CASE(CL_DEVICE_MAX_SEMAPHORE_SIZE_AMD, maxSemaphoreSize_); + CASE(CL_DEVICE_BOARD_NAME_AMD, boardName_); + CASE(CL_DEVICE_SPIR_VERSIONS, spirVersions_); + CASE(CL_DEVICE_MAX_PIPE_ARGS, maxPipeArgs_); + CASE(CL_DEVICE_PIPE_MAX_ACTIVE_RESERVATIONS, maxPipeActiveReservations_); + CASE(CL_DEVICE_PIPE_MAX_PACKET_SIZE, maxPipePacketSize_); + CASE(CL_DEVICE_MAX_GLOBAL_VARIABLE_SIZE, maxGlobalVariableSize_); + CASE(CL_DEVICE_GLOBAL_VARIABLE_PREFERRED_TOTAL_SIZE, globalVariablePreferredTotalSize_); + CASE(CL_DEVICE_QUEUE_ON_DEVICE_PROPERTIES, queueOnDeviceProperties_); + CASE(CL_DEVICE_QUEUE_ON_DEVICE_PREFERRED_SIZE, queueOnDevicePreferredSize_); + CASE(CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE, queueOnDeviceMaxSize_); + CASE(CL_DEVICE_MAX_ON_DEVICE_QUEUES, maxOnDeviceQueues_); + CASE(CL_DEVICE_MAX_ON_DEVICE_EVENTS, maxOnDeviceEvents_); +#ifdef cl_ext_device_fission + case CL_DEVICE_AFFINITY_DOMAINS_EXT: { + const device::AffinityDomain& affinityDomain = + as_amd(device)->info().affinityDomain_; + + size_t valueSize = affinityDomain.getNumSet() * + sizeof(cl_device_partition_property_ext); + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + affinityDomain.toCLExt( + reinterpret_cast(param_value)); + if (param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_DEVICE_PARTITION_STYLE_EXT: { + const device::PartitionInfo& partitionInfo = + as_amd(device)->info().partitionCreateInfo_; + size_t valueSize = 0; + cl_device_partition_property_ext* properties = + reinterpret_cast(param_value); + + switch (partitionInfo.type_.value_) { + case device::PartitionType::EQUALLY: + valueSize = 3 * sizeof(cl_device_partition_property_ext); + if (param_value != NULL) { + if (param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + properties[0] = CL_DEVICE_PARTITION_EQUALLY_EXT; + properties[1] = (cl_device_partition_property_ext) + partitionInfo.equally_.numComputeUnits_; + properties[2] = CL_PROPERTIES_LIST_END_EXT; + } + break; + + case device::PartitionType::BY_COUNTS: + valueSize = (partitionInfo.byCounts_.listSize_ + 2) * + sizeof(cl_device_partition_property_ext); + if (param_value != NULL) { + if (param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *properties++ = CL_DEVICE_PARTITION_BY_COUNTS_EXT; + for (size_t i = 0; i < partitionInfo.byCounts_.listSize_; + ++i) { + *properties++ = partitionInfo.byCounts_.countsList_[i]; + } + *properties = CL_PROPERTIES_LIST_END_EXT; + } + break; + + case device::PartitionType::BY_AFFINITY_DOMAIN: + valueSize = 3 * sizeof(cl_device_partition_property_ext); + if (param_value != NULL) { + if (param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + properties[0] = CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN_EXT; + properties[1] = partitionInfo.byAffinityDomain_.toCLExt(); + properties[2] = CL_PROPERTIES_LIST_END_EXT; + } + break; + } + + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL && param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + return CL_SUCCESS; + } + case CL_DEVICE_PARTITION_TYPES_EXT: { + const device::PartitionType& partitionProperties = + as_amd(device)->info().partitionProperties_; + size_t valueSize = partitionProperties.getNumSet() * + sizeof(cl_device_partition_property_ext); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + partitionProperties.toCLExt( + reinterpret_cast(param_value)); + if (param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_DEVICE_PARENT_DEVICE_EXT: { + cl_device_id parent = !as_amd(device)->isRootDevice() + ? as_cl(as_amd(device)->parent()) : (cl_device_id)0; + return amd::clGetInfo( + parent, param_value_size, param_value, param_value_size_ret); + } + case CL_DEVICE_REFERENCE_COUNT_EXT: { + cl_uint count = as_amd(device)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } +#endif // cl_ext_device_fission + CASE(CL_DEVICE_LINKER_AVAILABLE, linkerAvailable_); + CASE(CL_DEVICE_BUILT_IN_KERNELS, builtInKernels_); + CASE(CL_DEVICE_IMAGE_MAX_BUFFER_SIZE, imageMaxBufferSize_); + CASE(CL_DEVICE_IMAGE_MAX_ARRAY_SIZE, imageMaxArraySize_); + case CL_DEVICE_PARENT_DEVICE: { + cl_device_id parent = !as_amd(device)->isRootDevice() + ? as_cl(as_amd(device)->parent()) : (cl_device_id)0; + return amd::clGetInfo( + parent, param_value_size, param_value, param_value_size_ret); + } + CASE(CL_DEVICE_PARTITION_MAX_SUB_DEVICES, maxComputeUnits_); + case CL_DEVICE_PARTITION_PROPERTIES: + { + const device::PartitionType& partitionProperties = + as_amd(device)->info().partitionProperties_; + size_t valueSize = partitionProperties.getNumSet() * + sizeof(cl_device_partition_property); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + partitionProperties.toCL( + reinterpret_cast(param_value)); + if (param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_DEVICE_PARTITION_AFFINITY_DOMAIN: { + cl_device_affinity_domain deviceAffinity = + as_amd(device)->info().affinityDomain_.toCL(); + return amd::clGetInfo( + deviceAffinity, param_value_size, param_value, param_value_size_ret); + } + case CL_DEVICE_PARTITION_TYPE: + { + const device::PartitionInfo& partitionInfo = + as_amd(device)->info().partitionCreateInfo_; + size_t valueSize = 0; + cl_device_partition_property* properties = + reinterpret_cast(param_value); + + switch (partitionInfo.type_.value_) { + case device::PartitionType::EQUALLY: + valueSize = 3 * sizeof(cl_device_partition_property); + if (param_value != NULL) { + if (param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + properties[0] = CL_DEVICE_PARTITION_EQUALLY; + properties[1] = (cl_device_partition_property) + partitionInfo.equally_.numComputeUnits_; + properties[2] = (cl_device_partition_property)0; + } + break; + + case device::PartitionType::BY_COUNTS: + valueSize = (partitionInfo.byCounts_.listSize_ + 2) * + sizeof(cl_device_partition_property); + if (param_value != NULL) { + if (param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *properties++ = CL_DEVICE_PARTITION_BY_COUNTS; + for (size_t i = 0; i < partitionInfo.byCounts_.listSize_; + ++i) { + *properties++ = partitionInfo.byCounts_.countsList_[i]; + } + *properties = (cl_device_partition_property)0; + } + break; + + case device::PartitionType::BY_AFFINITY_DOMAIN: + valueSize = 3 * sizeof(cl_device_partition_property); + if (param_value != NULL) { + if (param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + properties[0] = CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN; + properties[1] = (cl_device_partition_property) + partitionInfo.byAffinityDomain_.toCL(); + properties[2] = (cl_device_partition_property)0; + } + break; + } + + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL && param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + return CL_SUCCESS; + } + case CL_DEVICE_REFERENCE_COUNT: { + cl_uint count = as_amd(device)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + CASE(CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, preferredInteropUserSync_); + CASE(CL_DEVICE_PRINTF_BUFFER_SIZE, printfBufferSize_); + CASE(CL_DEVICE_IMAGE_PITCH_ALIGNMENT, imagePitchAlignment_); + CASE(CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT, imageBaseAddressAlignment_); + +#if cl_amd_open_video + CASE(CL_DEVICE_MAX_VIDEO_SESSIONS_AMD, maxVideoSessions_); + CASE(CL_DEVICE_NUM_VIDEO_ATTRIBS_AMD, numVideoAttribs_); + CASE(CL_DEVICE_NUM_VIDEO_ENC_ATTRIBS_AMD, numVideoEncAttribs_); + case CL_DEVICE_VIDEO_ATTRIBS_AMD: + { + size_t valueSize = sizeof(cl_video_attrib_amd) + * as_amd(device)->info().numVideoAttribs_; + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + ::memcpy(param_value, as_amd(device)->info().videoAttribs_, + valueSize); + if (param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_DEVICE_VIDEO_ATTRIBS_ENC_AMD: + { + size_t valueSize = sizeof(cl_video_attrib_encode_amd) // has profile & format + * as_amd(device)->info().numVideoEncAttribs_; + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + ::memcpy(param_value, as_amd(device)->info().videoEncAttribs_, valueSize); + // Zero out remaining bytes if encode CAP List + profile is < parameter value size. + if (param_value_size > valueSize) { + ::memset(static_cast(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } +#endif //cl_amd_open_video + default: + break; + } + if (as_amd(device)->type() == CL_DEVICE_TYPE_GPU) { + switch (param_name) { + case CL_DEVICE_GLOBAL_FREE_MEMORY_AMD: { + size_t freeMemory[2]; + if (as_amd(device)->globalFreeMemory(freeMemory)) { + return amd::clGetInfo( + freeMemory, param_value_size, param_value, param_value_size_ret); + } + else { + return CL_INVALID_DEVICE; + } + } + CASE(CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD, simdPerCU_); + CASE(CL_DEVICE_SIMD_WIDTH_AMD, simdWidth_); + CASE(CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD, simdInstructionWidth_); + CASE(CL_DEVICE_WAVEFRONT_WIDTH_AMD, wavefrontWidth_); + CASE(CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD, globalMemChannels_); + CASE(CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD, globalMemChannelBanks_); + CASE(CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD, globalMemChannelBankWidth_); + CASE(CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD, localMemSizePerCU_); + CASE(CL_DEVICE_LOCAL_MEM_BANKS_AMD, localMemBanks_); + CASE(CL_DEVICE_THREAD_TRACE_SUPPORTED_AMD, threadTraceEnable_); + case CL_DEVICE_GFXIP_MAJOR_AMD: { + cl_uint major= as_amd(device)->info().gfxipVersion_ / 100; + return amd::clGetInfo( + major, param_value_size, param_value, param_value_size_ret); + } + case CL_DEVICE_GFXIP_MINOR_AMD: { + cl_uint minor= as_amd(device)->info().gfxipVersion_ % 100; + return amd::clGetInfo( + minor, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + } +#undef CASE + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +#ifdef cl_ext_device_fission + +RUNTIME_ENTRY(cl_int, clCreateSubDevicesEXT, ( + cl_device_id in_device, + const cl_device_partition_property_ext * partition_properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices)) +{ + if (!is_valid(in_device)) { + return CL_INVALID_DEVICE; + } + if (partition_properties == NULL || *partition_properties == 0u) { + return CL_INVALID_VALUE; + } + if (((num_entries > 0 || num_devices == NULL) && out_devices == NULL) + || (num_entries == 0 && out_devices != NULL)) { + return CL_INVALID_VALUE; + } + + device::CreateSubDevicesInfoT create_info; + switch (*partition_properties) { + case CL_DEVICE_PARTITION_EQUALLY_EXT: + create_info.p_.type_.value_ = device::PartitionType::EQUALLY; + create_info.p_.equally_.numComputeUnits_ = + (size_t)partition_properties[1]; + break; + case CL_DEVICE_PARTITION_BY_COUNTS_EXT: + create_info.p_.type_.value_ = device::PartitionType::BY_COUNTS; + create_info.initCountsList(partition_properties + 1); + break; + case CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN_EXT: + create_info.p_.type_.value_ = device::PartitionType::BY_AFFINITY_DOMAIN; + create_info.p_.byAffinityDomain_.value_ = + (uint)partition_properties[1]; + break; + default: + return CL_INVALID_VALUE; + } + + cl_int ret = as_amd(in_device)->createSubDevices( + create_info, num_entries, out_devices, num_devices); + + if (ret == CL_DEVICE_PARTITION_FAILED) { + return CL_DEVICE_PARTITION_FAILED_EXT; + } + if (ret == CL_INVALID_DEVICE_PARTITION_COUNT) { + return CL_INVALID_PARTITION_COUNT_EXT; + } + return ret; +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clRetainDeviceEXT, (cl_device_id device)) +{ + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + as_amd(device)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clReleaseDeviceEXT, (cl_device_id device)) +{ + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + as_amd(device)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +#endif // cl_ext_device_fission + +RUNTIME_ENTRY(cl_int, clCreateSubDevices, ( + cl_device_id in_device, + const cl_device_partition_property * partition_properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices)) +{ + if (!is_valid(in_device)) { + return CL_INVALID_DEVICE; + } + if (partition_properties == NULL || *partition_properties == 0u) { + return CL_INVALID_VALUE; + } + if ((num_devices == NULL && out_devices == NULL) || + (num_entries == 0 && out_devices != NULL)) { + return CL_INVALID_VALUE; + } + + device::CreateSubDevicesInfoT create_info; + switch (*partition_properties) { + case CL_DEVICE_PARTITION_EQUALLY: + create_info.p_.type_.value_ = device::PartitionType::EQUALLY; + create_info.p_.equally_.numComputeUnits_ = + (size_t)partition_properties[1]; + break; + case CL_DEVICE_PARTITION_BY_COUNTS: + create_info.p_.type_.value_ = device::PartitionType::BY_COUNTS; + create_info.initCountsList(partition_properties + 1); + break; + case CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN: + create_info.p_.type_.value_ = device::PartitionType::BY_AFFINITY_DOMAIN; + create_info.p_.byAffinityDomain_.value_ = + (uint)partition_properties[1]; + break; + default: + return CL_INVALID_VALUE; + } + + return as_amd(in_device)->createSubDevices( + create_info, num_entries, out_devices, num_devices); +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clRetainDevice, (cl_device_id device)) +{ + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + as_amd(device)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clReleaseDevice, (cl_device_id device)) +{ + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + as_amd(device)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_event.cpp b/opencl/api/opencl/amdocl/cl_event.cpp new file mode 100644 index 0000000000..542ffc450e --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_event.cpp @@ -0,0 +1,393 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/object.hpp" +#include "platform/context.hpp" +#include "platform/command.hpp" + +/*! \addtogroup API + * @{ + * \addtogroup CL_Events + * + * Event objects can be used to refer to a kernel execution command: + * - clEnqueueNDRangeKernel + * - clEnqueueTask + * - clEnqueueNativeKernel + * + * or read, write, map and copy commands on memory objects: + * - clEnqueue{Read|Write|Map}{Buffer|Image} + * - clEnqueueCopy{Buffer|Image} + * - clEnqueueCopyBufferToImage + * - clEnqueueCopyImageToBuffer + * + * An event object can be used to track the execution status of a command. + * The execution status of a command at any given point in time can be + * CL_QUEUED (is currently in the command queue), + * CL_RUNNING (device is currently executing this command), + * CL_COMPLETE (command has successfully completed) or the appropriate error + * code if the command was abnormally terminated (this may be caused by a bad + * memory access etc.). The error code returned by a terminated command is + * a negative integer value. A command is considered to be complete if its + * execution status is CL_COMPLETE or is a negative integer value. + * + * If the execution of a command is terminated, the command-queue associated + * with this terminated command, and the associated context (and all other + * command-queues in this context) may no longer be available. The behavior of + * OpenCL API calls that use this context (and command-queues associated with + * this context) are now considered to be implementationdefined. The user + * registered callback function specified when context is created can be used + * to report appropriate error information. + * + * @{ + */ + + +/*! \brief Wait on the host thread for commands identified by event objects in + * event_list to complete. + * + * A command is considered complete if its execution status is CL_COMPLETE or + * a negative value. The events specified in event_list act as synchronization + * points. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully. + * - CL_INVALID_VALUE if \a num_events is zero + * - CL_INVALID_CONTEXT if events specified in \a event_list do not belong to + * the same context + * - CL_INVALID_EVENT if event objects specified in \a event_list are not valid + * event objects. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clWaitForEvents, ( + cl_uint num_events, + const cl_event *event_list)) +{ + if (num_events == 0 || event_list == NULL) { + return CL_INVALID_VALUE; + } + + const amd::Context* prevContext = NULL; + const amd::HostQueue* prevQueue = NULL; + + for (cl_uint i = 0; i < num_events; ++i) { + cl_event event = event_list[i]; + + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + + // Make sure all the events are associated with the same context + amd::Context* context = &as_amd(event)->context(); + if (prevContext != NULL && prevContext != context) { + return CL_INVALID_CONTEXT; + } + prevContext = context; + + // Flush the command queues associated with event1...eventN + amd::HostQueue* queue = as_amd(event)->command().queue(); + if (queue != NULL && prevQueue != queue) { + queue->flush(); + } + prevQueue = queue; + } + + bool allSucceeded = true; + while (num_events-- > 0) { + allSucceeded &= as_amd(*event_list++)->awaitCompletion(); + } + return allSucceeded ? CL_SUCCESS + : CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST; +} +RUNTIME_EXIT + +/*! \brief Return information about the event object. + * + * \param event specifies the event object being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * Using clGetEventInfo to determine if a command identified by event has + * finished execution (i.e. CL_EVENT_COMMAND_EXECUTION_STATUS returns + * CL_COMPLETE) is not a synchronization point i.e. there are no guarantees + * that the memory objects being modified by command associated with event will + * be visible to other enqueued commands. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_EVENT if \a event is a not a valid event object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetEventInfo, ( + cl_event event, + cl_event_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + + switch(param_name) { + case CL_EVENT_CONTEXT: { + cl_context context = as_cl(&as_amd(event)->context()); + return amd::clGetInfo( + context, param_value_size, param_value, param_value_size_ret); + } + case CL_EVENT_COMMAND_QUEUE: { + amd::Command& command = as_amd(event)->command(); + cl_command_queue queue = command.queue() == NULL + ? NULL : const_cast(as_cl(command.queue()->asCommandQueue())); + return amd::clGetInfo( + queue, param_value_size, param_value, param_value_size_ret); + } + case CL_EVENT_COMMAND_TYPE: { + cl_command_type type = as_amd(event)->command().type(); + return amd::clGetInfo( + type, param_value_size, param_value, param_value_size_ret); + } + case CL_EVENT_COMMAND_EXECUTION_STATUS: { + as_amd(event)->notifyCmdQueue(); + cl_int status = as_amd(event)->command().status(); + return amd::clGetInfo( + status, param_value_size, param_value, param_value_size_ret); + } + case CL_EVENT_REFERENCE_COUNT: { + cl_uint count = as_amd(event)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! \brief Increment the event reference count. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_EVENT if \a event is not a valid event object. + * + * The OpenCL commands that return an event perform an implicit retain. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainEvent, (cl_event event)) +{ + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + as_amd(event)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the event reference count. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_EVENT if \a event is not a valid event object. + * + * The event object is deleted once the reference count becomes zero, the + * specific command identified by this event has completed (or terminated) and + * there are no commands in the command-queues of a context that require a wait + * for this event to complete. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseEvent, (cl_event event)) +{ + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + as_amd(event)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Creates a user event object. + * + * User events allow applications to enqueue commands that wait on a user event + * to finish before the command is executed by the device. + * + * \return a valid non-zero event object and errcode_ret is set to CL_SUCCESS + * if the user event object is created successfully. Otherwise, it returns + * a NULL value with one of the following error values returned in errcode_ret: + * - CL_INVALID_CONTEXT if context is not a valid context. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * The execution status of the user event object created is set to CL_SUBMITTED. + * + * \version 1.1r15 + */ +RUNTIME_ENTRY_RET(cl_event, clCreateUserEvent, ( + cl_context context, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return (cl_event) 0; + } + + amd::Event* event = new amd::UserEvent(*as_amd(context)); + if (event == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_event) 0; + } + + event->retain(); + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(event); +} +RUNTIME_EXIT + +/*! \brief Sets the execution status of a user event object. + * + * \a event is a user event object created using clCreateUserEvent. + * \a execution_status specifies the new execution status to be set and can be + * CL_COMPLETE or a negative integer value to indicate an error. + * clSetUserEventStatus can only be called once to change the execution status + * of event. + * + * \return CL_SUCCESS if the function was executed successfully. Otherwise, + * it returns one of the following errors: + * - CL_INVALID_EVENT if event is not a valid user event object. + * - CL_INVALID_VALUE if the execution_status is not CL_COMPLETE or + * a negative integer value. + * - CL_INVALID_OPERATION if the execution_status for event has already been + * changed by a previous call to clSetUserEventStatus. + * + * \version 1.1r15 + */ +RUNTIME_ENTRY(cl_int, clSetUserEventStatus, ( + cl_event event, + cl_int execution_status)) +{ + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + if (execution_status > CL_COMPLETE) { + return CL_INVALID_VALUE; + } + + as_amd(event)->setStatus(execution_status); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Registers a user callback function for a specific command execution + * status. + * + * The registered callback function will be called when the execution status + * of command associated with event changes to the execution status specified + * by command_exec_status. + * + * Each call to clSetEventCallback registers the specified user callback + * function on a callback stack associated with event. The order in which the + * registered user callback functions are called is undefined. + * + * \a event is a valid event object. + * \a command_exec_callback_type specifies the command execution status for + * which the callback is registered. The command execution callback mask + * values for which a callback can be registered are: CL_COMPLETE. + * There is no guarantee that the callback functions registered for various + * execution status values for an event will be called in the exact order + * that the execution status of a command changes. + * \a pfn_event_notify is the event callback function that can be registered + * by the application. This callback function may be called asynchronously + * by the OpenCL implementation. It is the application’s responsibility to + * ensure that the callback function is thread-safe. The parameters to this + * callback function are: + * event is the event object for which the callback function is invoked. + * event_command_exec_status represents the execution status of command + * for which this callback function is invoked. If the callback is called + * as the result of the command associated with event being abnormally + * terminated, an appropriate error code for the error that caused the + * termination will be passed to event_command_exec_status instead. + * \a user_data is a pointer to user supplied data. user_data will be passed as + * the user_data argument when pfn_notify is called. user_data can be NULL. + * + * All callbacks registered for an event object must be called. All enqueued + * callbacks shall be called before the event object is destroyed. Callbacks + * must return promptly. The behavior of calling expensive system routines, + * OpenCL API calls to create contexts or command-queues, or blocking OpenCL + * operations from the following list below, in a callback is undefined. + * clFinish, clWaitForEvents, blocking calls to clEnqueueReadBuffer, + * clEnqueueReadBufferRect, clEnqueueWriteBuffer, clEnqueueWriteBufferRect, + * blocking calls to clEnqueueReadImage and clEnqueueWriteImage, blocking + * calls to clEnqueueMapBuffer and clEnqueueMapImage, blocking calls to + * clBuildProgram + * + * If an application needs to wait for completion of a routine from the above + * list in a callback, please use the non-blocking form of the function, and + * assign a completion callback to it to do the remainder of your work. + * Note that when a callback (or other code) enqueues commands to a + * command-queue, the commands are not required to begin execution until the + * queue is flushed. In standard usage, blocking enqueue calls serve this role + * by implicitly flushing the queue. Since blocking calls are not permitted in + * callbacks, those callbacks that enqueue commands on a command queue should + * either call clFlush on the queue before returning or arrange for clFlush + * to be called later on another thread. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise, + * it returns one of the following errors: + * - CL_INVALID_EVENT if event is not a valid event object or is a user event + * object created using clCreateUserEvent. + * - CL_INVALID_VALUE if pfn_event_notify is NULL or if + * command_exec_callback_type is not a valid command execution status. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.1r15 + */ +RUNTIME_ENTRY(cl_int, clSetEventCallback, ( + cl_event event, + cl_int command_exec_callback_type, + void (CL_CALLBACK * pfn_notify)( + cl_event event, cl_int command_exec_status, void *user_data), + void* user_data)) +{ + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + + if (pfn_notify == NULL + || command_exec_callback_type < CL_COMPLETE + || command_exec_callback_type > CL_QUEUED) { + return CL_INVALID_VALUE; + } + + if (!as_amd(event)->setCallback( + command_exec_callback_type, pfn_notify, user_data)) { + return CL_OUT_OF_HOST_MEMORY; + } + + as_amd(event)->notifyCmdQueue(); + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ + diff --git a/opencl/api/opencl/amdocl/cl_execute.cpp b/opencl/api/opencl/amdocl/cl_execute.cpp new file mode 100644 index 0000000000..95c5c7a85f --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_execute.cpp @@ -0,0 +1,1043 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/kernel.hpp" +#include "platform/ndrange.hpp" +#include "platform/command.hpp" +#include "platform/program.hpp" + +#include + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_Exec Executing Kernel Objects + * + * @{ + */ + +/*! \brief Enqueue a command to execute a kernel on a device. + * + * \param command_queue is a valid command-queue. The kernel will be queued + * for execution on the device associated with \a command_queue. + * + * \param kernel is a valid kernel object. The OpenCL context associated with + * \a kernel and \a command-queue must be the same. + * + * \param work_dim is the number of dimensions used to specify the global + * work-items and work-items in the work-group. \a work_dim must be greater + * than zero and less than or equal to three. + * + * \param global_work_offset must currently be a NULL value. In a future + * revision of OpenCL, \a global_work_offset can be used to specify an array + * of \a work_dim unsigned values that describe the offset used to calculate + * the global ID of a work-item instead of having the global IDs always start + * at offset (0, 0, 0). + * + * \param global_work_size points to an array of \a work_dim unsigned values + * that describe the number of global work-items in \a work_dim dimensions + * that will execute the kernel function. The total number of global + * work-items is computed as global_work_size[0] * ... + * * global_work_size[work_dim - 1]. + * + * \param local_work_size points to an array of \a work_dim unsigned values + * that describe the number of work-items that make up a work-group (also + * referred to as the size of the work-group) that will execue the kernel + * specified by kernel. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular kernel + * execution instance. Event objects are unique and can be used to identify a + * particular kernel execution instance later on. If \a event is NULL, no + * event will be created for this kernel execution instance and therefore it + * will not be possible for the application to query or queue a wait for this + * particular kernel execution instance. + * + * The total number of work-items in a work-group is computed as + * local_work_size[0] * ... * local_work_size[work_dim - 1]. + * The total number of work-items in the work-group must be less than or equal + * to the CL_DEVICE_MAX_WORK_GROUP_SIZE. The explicitly specified + * \a local_work_size will be used to determine how to break the global work- + * items specified by global_work_size into appropriate work-group instances. + * If \a local_work_size is specified, the values specified in + * \a global_work_size[0], ..., global_work_size[work_dim - 1] must be evenly + * divisable by the corresponding values specified in \a local_work_size[0], + * ..., local_work_size[work_dim - 1]. \a local_work_size can also be a NULL + * value in which case the OpenCL implementation will determine how to be + * break the global work-items into appropriate work-groups. + * + * If \a local_work_size is NULL and no work-group size is specified when the + * kernel is compiled, the OpenCL implementation will determine how to break + * the global work-items specified by \a global_work_size into appropriate + * work-group instances. The work-group size to be used for kernel can also be + * specified in the program source using the + * __attribute__((reqd_work_group_size(X, Y, Z))) qualifier. In this case the + * size of work group specified by \a local_work_size must match the value + * specified by the \a reqd_work_group_size attribute qualifier. + * + * These work-group instances are executed in parallel across multiple + * compute units or concurrently on the same compute unit. Each work-item + * is uniquely identified by a global identifier. The global ID, which can be + * read inside the kernel is computed using the value given by + * \a global_work_size and \a global_work_offset. + * + * \return One of the following values: + * + * - CL_SUCCESS if the kernel execution was successfully queued + * + * - CL_INVALID_PROGRAM_EXECUTABLE if there is no successfully built program + * executable available for device associated with \a command_queue. + * + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * + * - CL_INVALID_KERNEL if \a kernel is not a valid kernel object. + * + * - CL_INVALID_CONTEXT if context associated with command_queue and kernel are + * not the same or if the context associated with command_queue and events in + * event_wait_list are not the same. + * + * - CL_INVALID_KERNEL_ARGS if the kernel argument values have not been + * specified or are not valid for the device on which kernel will be + * executed. + * + * - CL_INVALID_WORK_DIMENSION if \a work_dim is not a valid value + * (i.e. a value between 1 and 3). + * + * - CL_INVALID_WORK_GROUP_SIZE if \a local_work_size is specified and number + * of workitems specified by \a global_work_size is not evenly divisable by + * size of work-given by \a local_work_size or does not match the work-group + * size specified for kernel using the + * __attribute__((reqd_work_group_size(X, Y, Z))) qualifier in program + * source. + * + * - CL_INVALID_GLOBAL_OFFSET if \a global_work_offset is not NULL. + * + * - CL_OUT_OF_RESOURCES if there is a failure to queue the execution instance + * of \a kernel on the command-queue because of insufficient resources + * needed to execute the kernel. For example, the explicitly specified + * \a local_work_dim in range causes a failure to execute the kernel because + * of insufficient resources such as registers or local memory. Another + * example would be the number of read-only image args used in kernel exceed + * the CL_DEVICE_MAX_READ_IMAGE_ARGS value for device or the number of + * write-only image args used in kernel exceed the + * CL_DEVICE_MAX_WRITE_IMAGE_ARGS value for device or the number of samplers + * used in kernel exceed CL_DEVICE_MAX_SAMPLERS for device. + * + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for image or buffer objects specified as arguments to kernel. + * + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in + * \a event_wait_list are not valid events. + * + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueNDRangeKernel, ( + cl_command_queue command_queue, + cl_kernel kernel, + cl_uint work_dim, + const size_t *global_work_offset, + const size_t *global_work_size, + const size_t *local_work_size, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + *not_null(event) = NULL; + + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + const amd::Kernel* amdKernel = as_amd(kernel); + if (&hostQueue.context() != &amdKernel->program().context()) { + return CL_INVALID_CONTEXT; + } + + const amd::Device& device = hostQueue.device(); + const device::Kernel* devKernel = amdKernel->getDeviceKernel(device); + if (devKernel == NULL) { + return CL_INVALID_PROGRAM_EXECUTABLE; + } + + if (amdKernel->parameters().getSvmSystemPointersSupport() == FGS_YES && + !(device.info().svmCapabilities_ & CL_DEVICE_SVM_FINE_GRAIN_SYSTEM)) { + // The user indicated that this kernel will access SVM system pointers, + // but the device does not support them. + return CL_INVALID_OPERATION; + } + + if (work_dim < 1 || work_dim > 3) { + return CL_INVALID_WORK_DIMENSION; + } +#if !defined(CL_VERSION_1_1) + if (global_work_offset != NULL) { + return CL_INVALID_GLOBAL_OFFSET; + } +#endif // CL_VERSION + if (global_work_size == NULL) { + return CL_INVALID_VALUE; + } + + if (local_work_size == NULL) { + static size_t zeroes[3] = { 0, 0, 0 }; + local_work_size = zeroes; + } + else { + size_t numWorkItems = 1; + for (cl_uint dim = 0; dim < work_dim; ++dim) { + if (local_work_size[dim] == 0 || local_work_size[dim] + > device.info().maxWorkItemSizes_[dim]) { + return CL_INVALID_WORK_ITEM_SIZE; + } + if ((local_work_size[dim] != 0) && + (devKernel->workGroupInfo()->compileSize_[0] != 0) && (local_work_size[dim] != + devKernel->workGroupInfo()->compileSize_[dim])) { + return CL_INVALID_WORK_GROUP_SIZE; + } + if ((global_work_size[dim] == 0) || + (((global_work_size[dim] % local_work_size[dim]) != 0) && + (!device.settings().partialDispatch_ || + devKernel->workGroupInfo()->uniformWorkGroupSize_))) { + return CL_INVALID_WORK_GROUP_SIZE; + } + numWorkItems *= local_work_size[dim]; + } + if (numWorkItems > devKernel->workGroupInfo()->size_) { + return CL_INVALID_WORK_GROUP_SIZE; + } + } + + // Check that all parameters have been defined. + if (!amdKernel->parameters().check()) { + return CL_INVALID_KERNEL_ARGS; + } + + // Check that we do not exceed the amount of available local memory. + const size_t align = device.info().minDataTypeAlignSize_; + cl_ulong requiredLocalMemSize = + static_cast(amdKernel->parameters().localMemSize(align)) + + amd::alignUp(devKernel->workGroupInfo()->localMemSize_, align); + + if (requiredLocalMemSize > device.info().localMemSize_) { + return CL_OUT_OF_RESOURCES; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::NDRangeContainer ndrange((size_t) work_dim, + global_work_offset, global_work_size, local_work_size); + amd::NDRangeKernelCommand* command = new amd::NDRangeKernelCommand( + hostQueue, eventWaitList, *as_amd(kernel), ndrange); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + // ndrange is now owned by command. Do not delete it! + + // Make sure we have memory for the command execution + cl_int result = command->validateMemory(); + if (result != CL_SUCCESS) { + delete command; + return result; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to execute a kernel on a device. + * The kernel is executed using a single work-item. + * + * \param command_queue is a valid command-queue. The kernel will be queued + * for execution on the device associated with \a command_queue. + * + * \param kernel is a valid kernel object. The OpenCL context associated with + * \a kernel and \a command-queue must be the same. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event objects that identifies this particular kernel + * execution instance. Event objects are unique and can be used to identify a + * particular kernel execution instance later on. If \a event is NULL, no event + * will be created for this kernel execution instance and therefore it will not + * be possible for the application to query or queue a wait for this particular + * kernel execution instance. + * + * \return One of the following values: + * - CL_SUCCESS if the kernel execution was successfully queued. + * - CL_INVALID_PROGRAM_EXECUTABLE if there is no successfully built program + * executable available for device associated with \a command_queue. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_KERNEL if \a kernel is not a valid kernel object. + * - CL_INVALID_KERNEL_ARGS if the kernel argument values have not been + * specified or are not valid for the device on which kernel will be + * executed. + * - CL_INVALID_WORK_GROUP_SIZE if a work-group size is specified for + * kernel using the __attribute__((reqd_work_group_size(X, Y, Z))) + * qualifier in program source and is not (1, 1, 1). + * - CL_OUT_OF_RESOURCES if there is a failure to queue the execution instance + * of kernel on the command-queue because of insufficient resources needed + * to execute the kernel. For example, the explicitly specified + * \a local_work_dim in range causes a failure to execute the kernel because + * of insufficient resources such as registers or local memory. Another + * example would be the number of read-only image args used in kernel exceed + * the CL_DEVICE_MAX_READ_IMAGE_ARGS value for device or the number of + * write-only image args used in kernel exceed the + * CL_DEVICE_MAX_WRITE_IMAGE_ARGS value for device or the number of samplers + * used in kernel exceed CL_DEVICE_MAX_SAMPLERS for device. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for image or buffer objects specified as arguments to kernel. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueTask, ( + cl_command_queue command_queue, + cl_kernel kernel, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + static size_t const globalWorkSize[3] = {1, 0, 0}; + static size_t const localWorkSize[3] = {1, 0, 0}; + + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + return hostQueue->dispatch_->clEnqueueNDRangeKernel( + command_queue, kernel, 1, NULL, globalWorkSize, localWorkSize, + num_events_in_wait_list, event_wait_list, event); +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to execute a native C/C++ function not compiled + * using the OpenCL compiler. + * + * \param command_queue is a valid command-queue. A native user function can + * only be executed on a command-queue created on a device that has + * CL_EXEC_NATIVE_KERNEL capability set in CL_DEVICE_EXECUTION_CAPABILITIES. + * + * \param user_func is a pointer to a host-callable user function. + * + * \param args is a pointer to the args list that \a user_func should be called + * with. + * + * \param cb_args is the size in bytes of the args list that args points to. + * The data pointed to by \a args and \a cb_args bytes in size will be copied + * and a pointer to this copied region will be passed to \a user_func. The copy + * needs to be done because the memory objects (cl_mem values) that args may + * contain need to be modified and replaced by appropriate pointers to global + * memory. When clEnqueueNativeKernel returns, the memory region pointed to by + * args can be reused by the application. + * + * \param num_mem_objects is the number of buffer objects that are passed in + * args. + * + * \param mem_list is a list of valid buffer objects, if \a num_mem_objects > 0 + * + * \param args_mem_loc is a pointer to appropriate locations that args points + * to where memory object handles (cl_mem values) are stored. Before the user + * function is executed, the memory object handles are replaced by pointers to + * global memory. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list + * + * \param event_wait_list as described in clEnqueueNDRangeKernel. + * + * \param event returns an event objects that identifies this particular kernel + * execution instance. Event objects are unique and can be used to identify a + * particular kernel execution instance later on. If \a event is NULL, no event + * will be created for this kernel execution instance and therefore it will not + * be possible for the application to query or queue a wait for this particular + * kernel execution instance. + * + * \return One of the following values: + * - CL_SUCCESS if the user function execution instance was successfully queued + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_VALUE if \a user_func is NULL, or if \a args is a NULL value + * and \a num_mem_objects > 0 or if \a num_mem_objects > 0 and \a mem_list + * is NULL. + * - CL_INVALID_OPERATION if device cannot execute the native kernel. + * - CL_INVALID_MEM_OBJECT if one or more memory objects specified in + * \a mem_list are not valid or are not buffer objects. + * - CL_OUT_OF_RESOURCES if there is a failure to queue the execution instance + * of kernel on the command-queue because of insufficient resources needed + * to execute the kernel. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for buffer objects specified as arguments to \a kernel. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueNativeKernel, ( + cl_command_queue command_queue, + void (CL_CALLBACK * user_func)(void *), + void *args, + size_t cb_args, + cl_uint num_mem_objects, + const cl_mem *mem_list, + const void **args_mem_loc, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + *not_null(event) = NULL; + + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + const amd::Device& device = hostQueue.device(); + + if (!(device.info().executionCapabilities_ & CL_EXEC_NATIVE_KERNEL)) { + return CL_INVALID_OPERATION; + } + + if (user_func == NULL + || (num_mem_objects > 0 && (mem_list == NULL || args_mem_loc == NULL)) + || (num_mem_objects == 0 && (mem_list != NULL || args_mem_loc != NULL)) + || (args == NULL && (cb_args > 0 || num_mem_objects > 0)) + || (args != NULL && cb_args == 0)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + for (size_t i = 0; i < num_mem_objects; ++i) { + cl_mem obj = mem_list[i]; + if (!is_valid(obj)) { + return CL_INVALID_MEM_OBJECT; + } + } + + amd::NativeFnCommand* command = new amd::NativeFnCommand( + hostQueue, eventWaitList, + user_func, args, cb_args, num_mem_objects, mem_list, args_mem_loc); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * + * \addtogroup CL_Order Out of order Execution of Kernels and Memory Commands + * + * The OpenCL functions that are submitted to a command-queue are queued in + * the order the calls are made but can be configured to execute in-order or + * out-of-order. The properties argument in clCreateCommandQueue can be used + * to specify the execution order. + * + * If the CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE property of a command-queue + * is not set, the commands queued to a command-queue execute in order. + * For example, if an application calls clEnqueueNDRangeKernel to execute + * kernel A followed by a clEnqueueNDRangeKernel to execute kernel B, + * the application can assume that kernel A finishes first and then kernel B + * is executed. If the memory objects output by kernel A are inputs to kernel B + * then kernel B will see the correct data in memory objects produced + * by execution of kernel A. If the CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE + * property of a commandqueue is set, then there is no guarantee that kernel A + * will finish before kernel B starts execution. + * + * Applications can configure the commands queued to a command-queue to + * execute out-of-order by setting the CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE + * property of the commandqueue. This can be specified when the command-queue + * is created or can be changed dynamically using clSetCommandQueueProperty. + * In out-of-order execution mode there is no guarantee that the queued + * commands will finish execution in the order they were queued. As there is + * no guarantee that kernels will be executed in order i.e. based on when + * the clEnqueueNDRangeKernel calls are made within a command-queue, it is + * therefore possible that an earlier clEnqueueNDRangeKernel call to execute + * kernel A identified by event A may execute and/or finish later than a + * clEnqueueNDRangeKernel call to execute kernel B which was called by the + * application at a later point in time. To guarantee a specific order of + * execution of kernels, a wait on a particular event (in this case event A) + * can be used. The wait for event A can be specified in the event_wait_list + * argument to clEnqueueNDRangeKernel for kernel B. + * + * In addition, a wait for events or a barrier function can be queued to the + * command-queue. The wait for events command ensures that previously queued + * commands identified by the list of events to wait for have finished before + * the next batch of commands is executed. The barrier ensures that all + * previously queued commands in a command-queue have finished execution + * before the next batch of commands is executed. + * + * Similarly, commands to read, write, copy or map memory objects that are + * queued after clEnqueueNDRangeKernel, clEnqueueTask or clEnqueueNativeKernel + * commands are not guaranteed to wait for kernels scheduled for execution + * to have completed (if the CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE property + * is set). To ensure correct ordering of commands, the event object returned + * by clEnqueueNDRangeKernel, clEnqueueTask or clEnqueueNativeKernel can be + * used to queue a wait for event or a barrier command can be queued that must + * complete before reads or writes to the memory object(s) occur. + * + * @{ + */ + +/*! \brief Enqueue a marker command to \a command_queue. + * + * The marker command returns an event which can be used by to queue a wait on + * this marker event i.e. wait for all commands queued before the marker + * command to complete. + * + * \return One of the following values: + * - CL_SUCCESS if the function is successfully executed + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_INVALID_VALUE if \a event is a NULL value + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueMarker, ( + cl_command_queue command_queue, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::Command* command = new amd::Marker(*hostQueue, true); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief enqueues a marker command which waits for either a list of events + * to complete, or if the list is empty it waits for all commands previously + * enqueued in \a command_queue to complete before it completes. This command + * returns an event which can be waited on, i.e. this event can be waited on + * to insure that all events either in the \a event_wait_list or all + * previously enqueued commands, queued before this command to + * \a command_queue, have completed. + * + * \param command_queue is a valid command-queue. + * + * \param num_events_in_wait_list specifies the number of events given + * by \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must + * be greater than 0. The events specified in event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The + * memory associated with \a event_wait_list can be reused or freed after + * the function returns. + * If \a event_wait_list is NULL, then this particular command waits until + * all previous enqueued commands to \a command_queue have completed. + * + * \param event returns an event object that identifies this particular + * kernel execution instance. Event objects are unique and can be used to + * identify this marker command later on. + * + * \return CL_SUCCESS if the function is successfully executed. + * Otherwise, it returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid \a command-queue. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by + * the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueMarkerWithWaitList, ( + cl_command_queue command_queue, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue->context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::Command* command = new amd::Marker(*hostQueue, true, eventWaitList); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a wait for a specific event or a list of events to complete + * before any future commands queued in the command-queue are executed. + * + * \param command_queue is a valid command-queue. + * + * \param num_events specifies the number of events given by \a event_list. + * + * \param event_list is the list of events. Each event in \a event_list must + * be a valid event object returned by a previous call to: + * - clEnqueueNDRangeKernel + * - clEnqueueTask + * - clEnqueueNativeKernel + * - clEnqueue{Read|Write|Map}{Buffer|Image} + * - clEnqueueCopy{Buffer|Image} + * - clEnqueueCopyBufferToImage + * - clEnqueueCopyImageToBuffer + * - clEnqueueMarker. + * The events specified in \a event_list act as synchronization points. + * + * \return One of the following values: + * - CL_SUCCESS if the function was successfully executed. + * - CL_INVALID_COMMAND_QUEUE if c\a ommand_queue is not a valid command-queue + * - CL_INVALID_VALUE if \a num_events is zero or \a event_list is NULL + * - CL_INVALID_EVENT if event objects specified in \a event_list are not valid + * events + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueWaitForEvents, ( + cl_command_queue command_queue, + cl_uint num_events, + const cl_event *event_list)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events, event_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::Command* command = new amd::Marker(hostQueue, false, eventWaitList); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + command->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a barrier operation. + * + * The clEnqueueBarrier command ensures that all queued commands in + * \a command_queue have finished execution before the next batch of commands + * can begin execution. clEnqueueBarrier is a synchronization point. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueBarrier, (cl_command_queue command_queue)) +{ + //! @todo: Unimplemented(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief enqueues a barrier command which waits for either a list of events + * to complete, or if the list is empty it waits for all commands previously + * enqueued in \a command_queue to complete before it completes. This command + * blocks command execution, that is, any following commands enqueued after it + * do not execute until it completes. This command returns an event which can + * be waited on, i.e. this event can be waited on to insure that all events + * either in the \a event_wait_list or all previously enqueued commands, + * queued before this command to command_queue, have completed + * + * \param command_queue is a valid command-queue. + * + * \param num_events_in_wait_list specifies the number of events given + * by \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must + * be greater than 0. The events specified in event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The + * memory associated with \a event_wait_list can be reused or freed after + * the function returns. + * If \a event_wait_list is NULL, then this particular command waits until + * all previous enqueued commands to \a command_queue have completed. + * + * \param event returns an event object that identifies this particular + * kernel execution instance. Event objects are unique and can be used to + * identify this marker command later on. + * + * \return CL_SUCCESS if the function is successfully executed. + * Otherwise, it returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid \a command-queue. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by + * the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueBarrierWithWaitList, ( + cl_command_queue command_queue, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue->context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + //!@note: with the current runtime architecture and in-order execution + //! barrier and marker should be the same operation + amd::Command* command = new amd::Marker(*hostQueue, true, eventWaitList); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * + * \addtogroup CL_Profiling Profiling Operations on Memory Objects and Kernels + * + * Profiling of OpenCL functions that are enqueued as commands to a + * command-queue. The specific functions being referred to are: + * - clEnqueue{Read|Write|Map}Buffer, + * - clEnqueue{Read|Write|Map}Image, + * - clEnqueueCopy{Buffer|Image}, + * - clEnqueueCopyImageToBuffer, + * - clEnqueueCopyBufferToImage, + * - clEnqueueNDRangeKernel , + * - clEnqueueTask and + * - clEnqueueNativeKernel. + * These enqueued commands are identified by unique event objects. + * + * Event objects can be used to capture profiling information that measure + * execution time of a command. Profiling of OpenCL commands can be enabled + * either by using a command-queue created with CL_QUEUE_PROFILING_ENABLE + * flag set in properties arguments to clCreateCommandQueue or by setting the + * CL_QUEUE_PROFILING_ENABLE flag in properties arguments to + * clSetCommandQueueProperty. + * + * @{ + */ + +/*! \brief Return profiling information for the command associated with event. + * + * \param event specifies the event object. + * + * \param param_name specifies the profiling data to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * The unsigned 64-bit values returned can be used to measure the time in + * nano-seconds consumed by OpenCL commands. OpenCL devices are required to + * correctly track time across changes in frequency and p-states. The + * CL_DEVICE_PROFILING_TIMER_RESOLUTION specifies the resolution of the timer + * i.e. the number of nanoseconds elapsed before the timer is incremented. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully and the profiling + * information has been recorded + * - CL_PROFILING_INFO_NOT_AVAILABLE if the profiling information is currently + * not available (because the command identified by event has not completed) + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by param_value_size is < size of return type and \a param_value + * is not NULL + * - CL_INVALID_EVENT if \a event is a not a valid event object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetEventProfilingInfo, ( + cl_event event, + cl_profiling_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(event)) { + return CL_INVALID_EVENT; + } + + if (!as_amd(event)->profilingInfo().enabled_) { + return CL_PROFILING_INFO_NOT_AVAILABLE; + } + + if (param_value != NULL && param_value_size < sizeof(cl_ulong)) { + return CL_INVALID_VALUE; + } + + *not_null(param_value_size_ret) = sizeof(cl_ulong); + if (param_value != NULL) { + cl_ulong value = 0; + switch (param_name) { + case CL_PROFILING_COMMAND_END: + value = as_amd(event)->profilingInfo().end_; + break; + + case CL_PROFILING_COMMAND_START: + value = as_amd(event)->profilingInfo().start_; + break; + + case CL_PROFILING_COMMAND_SUBMIT: + value = as_amd(event)->profilingInfo().submitted_; + break; + + case CL_PROFILING_COMMAND_QUEUED: + value = as_amd(event)->profilingInfo().queued_; + break; + + default: + return CL_INVALID_VALUE; + } + if (value == 0) { + return CL_PROFILING_INFO_NOT_AVAILABLE; + } + *(cl_ulong*)param_value = value; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_FlushFinish Flush and Finish + * @{ + */ + +/*! \brief Issue all previously queued OpenCL commands in \a command_queue to + * the device associated with command_queue. + * + * clFlush only guarantees that all queued commands to \a command_queue get + * issued to the appropriate device. There is no guarantee that they will be + * complete after clFlush returns. + * + * \return One of the following values: + * - CL_SUCCESS if the function call was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * Any blocking commands queued in a command-queue such as + * clEnqueueRead{Image|Buffer} with \a blocking_read set to CL_TRUE, + * clEnqueueWrite{Image|Buffer} with \a blocking_write set to CL_TRUE, + * clEnqueueMap{Buffer|Image} with \a blocking_map set to CL_TRUE or + * clWaitForEvents perform an implicit flush of the command-queue. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clFlush, (cl_command_queue command_queue)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::Command* command = new amd::Marker(*hostQueue, false); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + command->release(); + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Block until all previously queued OpenCL runtime commands in + * \a command_queue are issued to the associated device and have completed. + * + * clFinish does not return until all queued commands in \a command_queue have + * been processed and completed. clFinish is also a synchronization point. + * + * \return One of the following values: + * - CL_SUCCESS if the function call was executed successfully. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clFinish, (cl_command_queue command_queue)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + hostQueue->finish(); + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_gl.cpp b/opencl/api/opencl/amdocl/cl_gl.cpp new file mode 100644 index 0000000000..3847712907 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_gl.cpp @@ -0,0 +1,2601 @@ +// +// Copyright 2010 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifdef _WIN32 +#include +#include +#include +// This is necessary since there are common GL/D3D10 functions +#include "cl_d3d9_amd.hpp" +#include "cl_d3d10_amd.hpp" +#include "cl_d3d11_amd.hpp" +#endif //_WIN32 + +#include +#include + +#include "cl_common.hpp" +#include "cl_gl_amd.hpp" + +#include "device/device.hpp" + +#include +#include + + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_GL_Interops + * + * This section discusses OpenCL functions that allow applications to + * use OpenGL buffer/texture/render-buffer objects as OpenCL memory + * objects. This allows efficient sharing of data between these OpenCL + * and OpenGL. The OpenCL API can be used to execute kernels that read + * and/or write memory objects that are also an OpenGL buffer object + * or a texture. An OpenCL image object can be created from an OpenGL + * texture or renderbuffer object. An OpenCL buffer object can be + * created from an OpenGL buffer object. An OpenCL memory object can + * be created from an OpenGL texture/buffer/render-buffer object or + * the default system provided framebuffer if any only if the OpenCL + * clContext has been created from a GL clContext. OpenGL contexts are + * created using platform specific APIs (EGL, CGL, WGL, GLX are some + * of the platform specific APIs that allow applications to create GL + * contexts). The appropriate platform API (such as EGL, CGL, WGL, + * GLX) will be extended to allow a CL clContext to be created from a + * GL clContext. Creating an OpenCL memory object from the default + * system provided framebuffer will also require an appropriate + * extension to the platform API. Refer to the appropriate platform + * API documentation to understand how to create a CL clContext from a + * GL clContext and creating a CL memory object from the default + * system provided framebuffer. + * + * @{ + * + * \addtogroup clCreateFromGLBuffer + * + * @{ + */ + +/*! \brief Creates an OpenCL buffer object from an OpenGL buffer object. + * + * \param clContext is a valid OpenCL clContext created from an OpenGL clContext. + * + * \param clFlags is a bit-field that is used to specify usage information. Only + * CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE can be used. + * + * \param glBufferName is a GL buffer object name. The GL buffer + * object must have a data store created though it does not need to + * be initialized. The size of the data store will be used to + * determine the size of the CL buffer object. + * + * \param pCpuMem is a pointer to the buffer data that may already be + * allocated by the application. The size of the buffer that pCpuMem points + * to must be >= \a size bytes. Passing in a pointer to an already allocated + * buffer on the host and using it as a buffer object allows applications to + * share data efficiently with kernels and the host. + * + * \param errcode_ret will return an appropriate error code. If errcode_ret + * is NULL, no error code is returned. + * + * \return valid non-zero OpenCL buffer object and errcode_ret is set + * to CL_SUCCESS if the buffer object is created successfully. It + * returns a NULL value with one of the following error values + * returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a clContext is not a valid clContext. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_GL_OBJECT if glBufferName is not a GL buffer object or is a + * GL buffer object but does not have a data store created. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromGLBuffer, ( + cl_context context, + cl_mem_flags flags, + GLuint bufobj, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + + if (!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + + return(amd::clCreateFromGLBufferAMD(*as_amd(context), flags, bufobj, errcode_ret)); +} +RUNTIME_EXIT + +/*! \brief creates the following: + * - an OpenCL 2D image object from an OpenGL 2D texture object + * or a single face of an OpenGL cubemap texture object, + * - an OpenCL 2D image array object from an OpenGL 2D texture array object, + * - an OpenCL 1D image object from an OpenGL 1D texture object, + * - an OpenCL 1D image buffer object from an OpenGL texture buffer object, + * - an OpenCL 1D image array object from an OpenGL 1D texture array object, + * - an OpenCL 3D image object from an OpenGL 3D texture object. + * + * \param clContext is a valid OpenCL clContext created from an OpenGL clContext. + * + * \param clFlags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param texture_target must be GL_TEXTURE_1D, GL_TEXTURE_1D_ARRAY, + * GL_TEXTURE_BUFFER, GL_TEXTURE_2D_ARRAY, GL_TEXTURE_3D, + * GL_TEXTURE_2D, GL_TEXTURE_CUBE_MAP_POSITIVE_X, + * GL_TEXTURE_CUBE_MAP_POSITIVE_Y, GL_TEXTURE_CUBE_MAP_POSITIVE_Z, + * GL_TEXTURE_CUBE_MAP_NEGATIVE_X, GL_TEXTURE_CUBE_MAP_NEGATIVE_Y, + * GL_TEXTURE_CUBE_MAP_NEGATIVE_Z or GL_TEXTURE_RECTANGLE_ARB. + * + * \param miplevel is the mipmap level to be used. If \a texture_target + * is GL_TEXTURE_BUFFER, \a miplevel must be 0. + * + * \param texture is a GL 1D, 2D, 3D, 1D array, 2D array, cubemap, + * rectangle or buffer texture object. + * The texture object must be a complete texture as per + * OpenGL rules on texture completeness. The texture format and dimensions + * defined by OpenGL for the specified miplevel of the texture will be + * used to create the OpenCL image memory object. Only GL texture formats + * that map to appropriate image channel order and data type can be used + * to create the the OpenCL image memory object. + * + * \param errcode_ret will return an appropriate error code. If \a + * errcode_ret is NULL, no error code is returned. + * + * \return A valid non-zero OpenCL image object and \a errcode_ret is set to + * CL_SUCCESS if the image object is created successfully. It returns a NULL value + * with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a clContext is not a valid clContext or was not + * created from a GL clContext. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_MIP_LEVEL if \a miplevel is not a valid mip-level for \a texture. + * - CL_INVALID_GL_OBJECT if \a texture is not an appropriate GL 2D texture, + * cubemap or texture rectangle. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if the OpenGL texture format does not + * map to an appropriate OpenCL image format. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromGLTexture, ( + cl_context context, + cl_mem_flags flags, + GLenum texture_target, + GLint miplevel, + GLuint texture, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + + if (!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().imageSupport_) { + supportPass = true; + } + } + if (!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return static_cast(0); + } + + return amd::clCreateFromGLTextureAMD(*as_amd(context), flags, + texture_target, miplevel, texture, errcode_ret); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromGLTexture2D + * @{ + */ + +/*! \brief Create an OpenCL 2D image object from an OpenGL 2D texture object. + * + * \param clContext is a valid OpenCL clContext created from an OpenGL clContext. + * + * \param clFlags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param target must be GL_TEXTURE_2D, GL_TEXTURE_CUBE_MAP_POSITIVE_X, + * GL_TEXTURE_CUBE_MAP_POSITIVE_Y, GL_TEXTURE_CUBE_MAP_POSITIVE_Z, + * GL_TEXTURE_CUBE_MAP_NEGATIVE_X, GL_TEXTURE_CUBE_MAP_NEGATIVE_Y, + * GL_TEXTURE_CUBE_MAP_NEGATIVE_Z or GL_TEXTURE_RECTANGLE_ARB. + * + * \param miplevel is the mipmap level to be used. + * + * \param texture is a GL 2D texture, cubemap or texture rectangle + * object name. The texture object must be a complete texture as per + * OpenGL rules on texture completeness. The \a texture format and + * dimensions specified using appropriate glTexImage2D call for \a + * miplevel will be used to create the 2D image object. Only GL + * texture formats that map to appropriate image channel order and + * data type can be used to create the 2D image object. + * + * \param errcode_ret will return an appropriate error code. If \a + * errcode_ret is NULL, no error code is returned. + * + * \return A valid non-zero OpenCL image object and \a errcode_ret is set to + * CL_SUCCESS if the image object is created successfully. It returns a NULL value + * with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a clContext is not a valid clContext or was not + * created from a GL clContext. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_MIP_LEVEL if \a miplevel is not a valid mip-level for \a texture. + * - CL_INVALID_GL_OBJECT if \a texture is not an appropriate GL 2D texture, + * cubemap or texture rectangle. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if the OpenGL texture format does not + * map to an appropriate OpenCL image format. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromGLTexture2D, ( + cl_context context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texture, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + + if (!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().imageSupport_) { + supportPass = true; + } + } + if (!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return static_cast(0); + } + + return amd::clCreateFromGLTextureAMD(*as_amd(context), flags, target, + miplevel, texture, errcode_ret); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromGLTexture3D + * @{ + */ + +/*! \brief Create an OpenCL 3D image object from an OpenGL 3D texture object. + * + * \param clContext is a valid OpenCL clContext created from an OpenGL clContext. + * + * \param clFlags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param target must be GL_TEXTURE_3D. + * + * \param miplevel is the mipmap level to be used. + * + * \param texture is a GL 3D texture object [name]. + * The texture object must be a complete texture as per OpenGL rules on texture + * completeness. The \a texture format and dimensions specified using appropriate + * glTexImage3D call for \a miplevel will be used to create the 3D image object. + * Only GL texture formats that map to appropriate image channel order and + * data type can be used to create the 3D image object. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero OpenCL image object and \a errcode_ret is set to + * CL_SUCCESS if the image object is created successfully. It returns a NULL value + * with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a clContext is not a valid clContext or was not + * created from a GL clContext. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_MIP_LEVEL if \a miplevel is not a valid mip-level for \a texture. + * - CL_INVALID_GL_OBJECT if \a texture is not an GL 3D texture. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if the OpenGL texture format does not + * map to an appropriate OpenCL image format. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromGLTexture3D, ( + cl_context context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texture, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + + if (!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().imageSupport_) { + supportPass = true; + } + } + if (!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return static_cast(0); + } + + return amd::clCreateFromGLTextureAMD(*as_amd(context), flags, target, + miplevel, texture, errcode_ret); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clCreateFromGLRenderbuffer + * @{ + */ + +/*! \brief Create an OpenCL 2D image object from an OpenGL renderbuffer object. + * + * \param clContext is a valid OpenCL clContext created from an OpenGL clContext. + * + * \param clFlags is a bit-field that is used to specify usage information. + * Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY and CL_MEM_READ_WRITE values + * can be used. + * + * \param renderbuffer is a GL renderbuffer object name. The renderbuffer + * storage must be specified before the image object can be created. Only + * GL renderbuffer formats that map to appropriate image channel order and + * data type can be used to create the 2D image object. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero OpenCL image object and \a errcode_ret is set + * to CL_SUCCESS if the image object is created successfully. It returns a + * NULL value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a clContext is not a valid clContext or was not + * created from a GL clContext. + * - CL_INVALID_VALUE if values specified in \a clFlags are not valid. + * - CL_INVALID_GL_OBJECT if \a renderbuffer is not an GL renderbuffer object. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if the OpenGL renderbuffer format + * does not map to an appropriate OpenCL image format. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateFromGLRenderbuffer, ( + cl_context context, + cl_mem_flags flags, + GLuint renderbuffer, + cl_int* errcode_ret)) +{ + cl_mem clMemObj = NULL; + + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return clMemObj; + } + + if (!(((flags & CL_MEM_READ_ONLY) == CL_MEM_READ_ONLY) + || ((flags & CL_MEM_WRITE_ONLY) == CL_MEM_WRITE_ONLY) + || ((flags & CL_MEM_READ_WRITE) == CL_MEM_READ_WRITE))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return clMemObj; + } + + return(amd::clCreateFromGLRenderbufferAMD(*as_amd(context), flags, + renderbuffer, errcode_ret)); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clGetGLObjectInfo + * @{ + */ + +/*! \brief Query GL object type from a CL memory object. + * + * \param memobj [is a valid cl_mem object created from a GL object]. + * + * \param gl_object_type returns the type of GL object attached to memobj + * and can be CL_GL_OBJECT_BUFFER, CL_GL_OBJECT_TEXTURE2D, + * CL_GL_OBJECT_TEXTURE_RECTANGLE, CL_GL_OBJECT_TEXTURE3D, or + * CL_GL_OBJECT_RENDERBUFFER. If \a gl_object_type is NULL, it is ignored. + * + * \param gl_object_name returns the GL object name used to create memobj. + * If \a gl_object_name is NULL, it is ignored. + * + * \return One of the following values is returned: + * - CL_SUCCESS if the call was executed successfully. + * - CL_INVALID_MEM_OBJECT if \a memobj is not a valid OpenCL memory object. + * - CL_INVALID_GL_OBJECT if there is no GL object associated with \a memobj. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY(cl_int, clGetGLObjectInfo, ( + cl_mem memobj, + cl_gl_object_type* gl_object_type, + GLuint* gl_object_name)) +{ + if (!is_valid(memobj)) { + LogWarning("\"memobj\" is not a valid cl_mem object"); + return CL_INVALID_MEM_OBJECT; + } + + amd::InteropObject* interop = as_amd(memobj)->getInteropObj(); + if (NULL == interop) { + LogWarning("CL object \"memobj\" is not created from GL object"); + return CL_INVALID_GL_OBJECT; + } + + amd::GLObject* glObject = interop->asGLObject(); + if (NULL == glObject) { + LogWarning("CL object \"memobj\" is not created from GL object"); + return CL_INVALID_GL_OBJECT; + } + + cl_int result; + + cl_gl_object_type clGLType = glObject->getCLGLObjectType(); + result = amd::clGetInfo(clGLType, + sizeof(cl_gl_object_type), gl_object_type, NULL); + + GLuint glName = glObject->getGLName(); + result |= amd::clGetInfo(glName, sizeof(GLuint), gl_object_name, NULL); + + return result; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clGetGLTextureInfo + * @{ + */ + +/*! \brief Query additional information about the GL texture object associated + * with \a memobj. + * + * \param memobj [is a valid cl_mem object created from a GL object]. + * + * \param param_name specifies what additional information about the GL + * texture object associated with \a memobj to query: + * - CL_GL_TEXTURE_TARGET (GLenum) to query the \a target argument specified + * in clCreateGLTexture2D or clCreateGLTexture3D calls. + * - CL_GL_MIPMAP_LEVEL (GLint) to query the \a miplevel argument specified + * in clCreateGLTexture2D or clCreateGLTexture3D calls. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type as + * described for \a param_name argumnet (GLenum or GLint). + * \a param_value_size_ret returns the actual size in bytes of data copied to + * \a param_value. If \a param_value_size_ret is NULL, it is ignored + * + * \return One of the following values is returned: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_MEM_OBJECT if \a memobj is not a valid OpenCL memory object. + * - CL_INVALID_GL_OBJECT if there is no GL texture object (2D or 3D texture) + * associated with \a memobj. + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type required by + * \a param_name and \a param_value is not NULL, or if \a param_value and + * \a param_value_size_ret are NULL. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY(cl_int, clGetGLTextureInfo, ( + cl_mem memobj, + cl_gl_texture_info param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret)) +{ + if (!is_valid(memobj)) { + LogWarning("\"memobj\" is not a valid cl_mem object"); + return CL_INVALID_MEM_OBJECT; + } + amd::InteropObject* interop = as_amd(memobj)->getInteropObj(); + if (NULL == interop) { + LogWarning("CL object \"memobj\" is not created from GL object"); + return CL_INVALID_GL_OBJECT; + } + amd::GLObject* glObject = interop->asGLObject(); + if ((NULL == glObject) || (NULL != glObject->asBufferGL())) { + LogWarning("CL object \"memobj\" is not created from GL texture"); + return CL_INVALID_GL_OBJECT; + } + + switch (param_name) { + case CL_GL_TEXTURE_TARGET: { + GLenum glTarget = glObject->getGLTarget(); + if (glTarget == GL_TEXTURE_CUBE_MAP) { + glTarget = glObject->getCubemapFace(); + } + return amd::clGetInfo( + glTarget, param_value_size, param_value, param_value_size_ret); + } + case CL_GL_MIPMAP_LEVEL: { + GLint mipLevel = glObject->getGLMipLevel(); + return amd::clGetInfo( + mipLevel, param_value_size, param_value, param_value_size_ret); + } + case CL_GL_NUM_SAMPLES: { + GLsizei numSamples = glObject->getNumSamples(); + return amd::clGetInfo( + numSamples,param_value_size, param_value, param_value_size_ret); + } + default: + LogWarning("Unknown param_name in clGetGLTextureInfoAMD"); + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clEnqueueAcquireExtObjects + * @{ + */ + +/*! \brief Acquire OpenCL memory objects that have been created from external + * objects (OpenGL, D3D). + * + * \param command_queue is a valid command-queue. + * + * \param num_objects is the number of memory objects to be acquired + * in \a mem_objects. + * + * \param mem_objects is a pointer to a list of CL memory objects that refer + * to a GL object (buffer/texture/renderbuffer objects or the framebuffer). + * + * \param event_wait_list specify [is a pointer to] events that need to + * complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait + * on any event to complete. If \a event_wait_list is NULL, + * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, + * the list of events pointed to by \a event_wait_list must be valid and + * \a num_events_in_wait_list must be greater than 0. The events specified in + * \a event_wait_list act as synchronization points. + * + * \param num_events_in_wait_list specify the number of events in + * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be + * greater than 0 if \a event_wait_list is not NULL. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. + * + * \return One of the following values is returned: + * - CL_SUCCESS if the function is executed successfully. + * - CL_SUCCESS if \a num_objects is 0 and \a mem_objects is NULL; the + * function does nothing. + * - CL_INVALID_VALUE if \a num_objects is zero and \a mem_objects is not a + * NULL value or if \a num_objects > 0 and \a mem_objects is NULL. + * - CL_INVALID_MEM_OBJECT if memory objects in \a mem_objects are not valid + * OpenCL memory objects. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if clContext associated with \a command_queue was not + * created from an OpenGL clContext. + * - CL_INVALID_GL_OBJECT if memory objects in \a mem_objects have not been + * created from a GL object(s). + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY(cl_int, clEnqueueAcquireGLObjects, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + return amd::clEnqueueAcquireExtObjectsAMD( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event, + CL_COMMAND_ACQUIRE_GL_OBJECTS); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clEnqueueReleaseGLObjects + * @{ + */ + +/*! \brief Release OpenCL memory objects that have been created from OpenGL + * objects. + * + * \param command_queue is a valid command-queue [which is associated with the + * OpenCL clContext releasing the OpenGL objects]. + * + * \param num_objects is the number of memory objects to be released + * in \a mem_objects. + * + * \param mem_objects is a pointer to a list of CL memory objects that refer + * to a GL object (buffer/texture/renderbuffer objects or the framebuffer). + * + * \param event_wait_list specify [is a pointer to] events that need to + * complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait + * on any event to complete. If \a event_wait_list is NULL, + * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, + * the list of events pointed to by \a event_wait_list must be valid and + * \a num_events_in_wait_list must be greater than 0. The events specified in + * \a event_wait_list act as synchronization points. + * + * \param num_events_in_wait_list specify the number of events in + * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be + * greater than 0 if \a event_wait_list is not NULL. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. + * + * \return One of the following values is returned: + * - CL_SUCCESS if the function is executed successfully. + * - CL_SUCCESS if \a num_objects is 0 and \a mem_objects is NULL; the + * function does nothing. + * - CL_INVALID_VALUE if \a num_objects is zero and \a mem_objects is not a + * NULL value or if \a num_objects > 0 and \a mem_objects is NULL. + * - CL_INVALID_MEM_OBJECT if memory objects in \a mem_objects are not valid + * OpenCL memory objects. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if clContext associated with \a command_queue was not + * created from an OpenGL clContext. + * - CL_INVALID_GL_OBJECT if memory objects in \a mem_objects have not been + * created from a GL object(s). + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.0r29 + */ +RUNTIME_ENTRY(cl_int, clEnqueueReleaseGLObjects, ( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + return amd::clEnqueueReleaseExtObjectsAMD( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event, + CL_COMMAND_RELEASE_GL_OBJECTS); +} +RUNTIME_EXIT + +/*! @} +* \addtogroup clCreateEventFromGLsyncKHR +* @{ +*/ + +/*! \brief Creates an event object linked to an OpenGL sync object. +* Completion of such an event object is equivalent to waiting for completion +* of the fence command associated with the linked GL sync object. +* +* \param context is valid OpenCL context created from an OpenGL context +* or share group, using the cl_khr_gl_sharing extension. +* +* \param sync is the 'name' of a sync object in the GL share group associated +* with context. +* +* \param errcode_ret Returns an appropriate error code as described below. +* If errcode_ret is NULL, no error code is returned. +* +* \return a valid OpenCL event object and errcode_ret is set to CL_SUCCESS +* if the event object is created successfully.Otherwise, it returns a NULL +* value with one of the following error values returned in errcode_ret: +* - CL_INVALID_CONTEXT if context is not a valid context or was not created +* from a GL context. +* - CL_INVALID_GL_OBJECT if sync is not the name of a sync object in the +* GL share group associated with context. +* +* \version 1.1 +*/ + +RUNTIME_ENTRY_RET(cl_event, clCreateEventFromGLsyncKHR, ( +cl_context context, +cl_GLsync clGLsync, +cl_int* errcode_ret)) +{ + // create event of fence sync type + amd::ClGlEvent* clglEvent = new amd::ClGlEvent(*as_amd(context)); + clglEvent->context().glenv()->glFlush_(); + // initially set the status of fence as queued + clglEvent->setStatus(CL_SUBMITTED); + // store GLsync id of the fence in event in order to associate them together + clglEvent->setData(clGLsync); + amd::Event* evt = dynamic_cast(clglEvent); + evt->retain(); + return as_cl(evt); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup clGetGLContextInfoKHR + * @{ + */ + +/*! \brief This f-n is defined in CL extension cl_khr_gl_sharing and serves + * the purpose of quering current device and all devices that support + * CL-GL interoperability. + * + * \param properties points to an , which is a array of + * ordered pairs terminated with zero. If an + * attribute is not specified in , then its default value + * (listed in table 4.attr) is used (it is said to be specified + * implicitly). If is NULL or empty (points to a list + * whose first value is zero), all attributes take on their default + * values. + * + * \param param_name may accept one of the following enumerated values: + * - CL_CURRENT_DEVICE_FOR_GL_CONTEXT_KHR 0x2006 + * - CL_DEVICES_FOR_GL_CONTEXT_KHR 0x2007. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type as + * described for \a param_name argumnet (GLenum or GLint). + * \a param_value_size_ret returns the actual size in bytes of data copied to + * \a param_value. If \a param_value_size_ret is NULL, it is ignored + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type as + * described for \a param_name argumnet (GLenum or GLint). + * \a param_value_size_ret returns the actual size in bytes of data copied to + * \a param_value. If \a param_value_size_ret is NULL, it is ignored + * + * \return one of the following values is returned: + * - CL_SUCCESS if the function is executed successfully. + * - CL_SUCCESS if \a num_objects is 0 and \a mem_objects is NULL; the + * function does nothing. + * - CL_INVALID_VALUE if \a num_objects is zero and \a mem_objects is not a + * NULL value or if \a num_objects > 0 and \a mem_objects is NULL. + * - CL_INVALID_MEM_OBJECT if memory objects in \a mem_objects are not valid + * OpenCL memory objects. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if clContext associated with \a command_queue was not + * created from an OpenGL clContext. + * - CL_INVALID_GL_OBJECT if memory objects in \a mem_objects have not been + * created from a GL object(s). + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * - CL_INVALID_GL_SHAREGROUP_REFERENCE_KHR if + * + * \version 1.0r47 + */ +RUNTIME_ENTRY(cl_int, clGetGLContextInfoKHR, ( + const cl_context_properties *properties, + cl_gl_context_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + cl_int errcode; + cl_device_id* gpu_devices; + cl_device_id* cpu_devices; + cl_uint num_gpu_devices = 0; + cl_uint num_cpu_devices = 0; + amd::Context::Info info; + static const bool VALIDATE_ONLY = true; + + errcode = amd::Context::checkProperties(properties, &info); + if (CL_SUCCESS != errcode) { + return errcode; + } + + if (!(info.flags_ & amd::Context::GLDeviceKhr)) { + // No GL context is specified + return CL_INVALID_GL_SHAREGROUP_REFERENCE_KHR; + } + + // Get devices + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, 0, NULL, &num_gpu_devices); + if (errcode != CL_SUCCESS && errcode != CL_DEVICE_NOT_FOUND) { + return CL_INVALID_VALUE; + } + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_CPU, 0, NULL, &num_cpu_devices); + if (errcode != CL_SUCCESS && errcode != CL_DEVICE_NOT_FOUND) { + return CL_INVALID_VALUE; + } + + if (!num_gpu_devices && !num_cpu_devices) { + return CL_INVALID_GL_SHAREGROUP_REFERENCE_KHR; + } + + switch(param_name) { + + case CL_CURRENT_DEVICE_FOR_GL_CONTEXT_KHR: + // Return the CL device currently associated with the specified OpenGL context. + if (num_gpu_devices) { + gpu_devices = (cl_device_id *) alloca(num_gpu_devices * sizeof(cl_device_id)); + + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU, + num_gpu_devices, gpu_devices, NULL); + if (errcode != CL_SUCCESS) { + return errcode; + } + + for (cl_uint i = 0; i < num_gpu_devices; ++i) { + cl_device_id device = gpu_devices[i]; + if (is_valid(device) && + as_amd(device)->bindExternalDevice(info.type_, info.hDev_, info.hCtx_, VALIDATE_ONLY)) { + return amd::clGetInfo( + device, param_value_size, param_value, param_value_size_ret); + } + } + + *not_null(param_value_size_ret) = 0; + } + else { + cpu_devices = (cl_device_id *) alloca(num_cpu_devices * sizeof(cl_device_id)); + + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_CPU, + num_cpu_devices, cpu_devices, NULL); + if (errcode != CL_SUCCESS) { + return errcode; + } + return amd::clGetInfo( + cpu_devices[0], param_value_size, param_value, param_value_size_ret); + } + break; + + case CL_DEVICES_FOR_GL_CONTEXT_KHR: + { + //List of all CL devices that can be associated with the specified OpenGL context. + cl_uint total_devices = num_gpu_devices + num_cpu_devices; + size_t size = total_devices * sizeof(cl_device_id); + + cl_device_id* devices = (cl_device_id *) alloca(size); + + errcode = clGetDeviceIDs(NULL, CL_DEVICE_TYPE_GPU | CL_DEVICE_TYPE_CPU, + total_devices, devices, NULL); + if (errcode != CL_SUCCESS) { + return errcode; + } + + std::vector compatible_devices; + + for (cl_uint i = 0; i < total_devices; ++i) { + cl_device_id device = devices[i]; + if (is_valid(device) && + as_amd(device)->bindExternalDevice(info.type_, info.hDev_, info.hCtx_, VALIDATE_ONLY)) { + compatible_devices.push_back(as_amd(device)); + } + } + + size_t deviceCount = compatible_devices.size(); + size_t deviceCountSize = deviceCount * sizeof(cl_device_id); + + if (param_value != NULL && param_value_size < deviceCountSize) { + return CL_INVALID_VALUE; + } + + *not_null(param_value_size_ret) = deviceCountSize; + + if (param_value != NULL) { + cl_device_id* deviceList = (cl_device_id*) param_value; + std::vector::const_iterator it; + for (it = compatible_devices.begin(); it != compatible_devices.end(); ++it) { + *deviceList++ = as_cl(*it); + } + } + + return CL_SUCCESS; + } + break; + + default: + LogWarning("\"param_name\" is not valid"); + return CL_INVALID_VALUE; + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +// +// +// namespace amd +// +// +namespace amd +{ + +typedef struct +{ + GLenum glBinding; + GLenum glTarget; +} TargetBindings_t; + +/*! @} + * \addtogroup CL-GL interop helper functions + * @{ + */ + +//! Function clearGLErrors() to clear all GL error bits, if any +void +clearGLErrors(const Context &amdContext) +{ + GLenum glErr, glLastErr = GL_NO_ERROR; + while(1) { + glErr = amdContext.glenv()->glGetError_(); + if (glErr == GL_NO_ERROR || glErr == glLastErr) { + break; + } + glLastErr = glErr; + LogWarning("GL error"); + } +} + +GLenum +checkForGLError(const Context &amdContext) +{ + GLenum glRetErr = GL_NO_ERROR; + GLenum glErr; + while(GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) + { + glRetErr = glErr; // Just return the last GL error + LogWarning("Check GL error"); + } + return glRetErr; +} + +//! Function getCLFormatFromGL returns "true" if GL format +//! is compatible with CL format, "false" otherwise. +bool +getCLFormatFromGL(const Context& amdContext, GLint gliInternalFormat, + cl_image_format* pclImageFormat, + int* piBytesPerPixel) +{ + bool bRetVal = false; + +/* +Available values for "image_channel_order" +========================================== +CL_R +CL_A +CL_INTENSITY +CL_LUMINANCE +CL_RG +CL_RA +CL_RGB +CL_RGBA +CL_ARGB +CL_BGRA + +Available values for "image_channel_data_type" +============================================== +CL_SNORM_INT8 +CL_SNORM_INT16 +CL_UNORM_INT8 +CL_UNORM_INT16 +CL_UNORM_SHORT_565 +CL_UNORM_SHORT_555 +CL_UNORM_INT_101010 +CL_SIGNED_INT8 +CL_SIGNED_INT16 +CL_SIGNED_INT32 +CL_UNSIGNED_INT8 +CL_UNSIGNED_INT16 +CL_UNSIGNED_INT32 +CL_HALF_FLOAT +CL_FLOAT +*/ + + switch(gliInternalFormat) + { + case GL_ALPHA8: + pclImageFormat->image_channel_order = CL_A; + pclImageFormat->image_channel_data_type = CL_UNORM_INT8;//CL_UNSIGNED_INT8; + *piBytesPerPixel = 1; + bRetVal = true; + break; + + case GL_R8: + case GL_R8UI: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_R8)? CL_UNORM_INT8:CL_UNSIGNED_INT8; + *piBytesPerPixel = 1; + bRetVal = true; + break; + + case GL_R8I: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT8; + *piBytesPerPixel = 1; + bRetVal = true; + break; + + case GL_RG8: + case GL_RG8UI: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_RG8)? CL_UNORM_INT8:CL_UNSIGNED_INT8; + *piBytesPerPixel = 2; + bRetVal = true; + break; + + case GL_RG8I: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT8; + *piBytesPerPixel = 2; + bRetVal = true; + break; + + case GL_RGB8: + case GL_RGB8UI: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_RGB8)? CL_UNORM_INT8:CL_UNSIGNED_INT8; + *piBytesPerPixel = 3; + bRetVal = true; + break; + + case GL_RGB8I: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT8; + *piBytesPerPixel = 3; + bRetVal = true; + break; + + case GL_RGBA: + case GL_RGBA8: + case GL_RGBA8UI: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_RGBA8UI)? CL_UNSIGNED_INT8:CL_UNORM_INT8; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_RGBA8I: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT8; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_R16: + case GL_R16UI: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_R16)? CL_UNORM_INT16:CL_UNSIGNED_INT16; + bRetVal = true; + *piBytesPerPixel = 2; + break; + + case GL_R16I: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT16; + *piBytesPerPixel = 2; + bRetVal = true; + break; + + case GL_R16F: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = CL_HALF_FLOAT; + *piBytesPerPixel = 2; + bRetVal = true; + break; + + case GL_RG16: + case GL_RG16UI: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_RG16)? CL_UNORM_INT16:CL_UNSIGNED_INT16; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_RG16I: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT16; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_RG16F: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = CL_HALF_FLOAT; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_RGB16: + case GL_RGB16UI: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_RGB16)? CL_UNORM_INT16:CL_UNSIGNED_INT16; + *piBytesPerPixel = 6; + bRetVal = true; + break; + + case GL_RGB16I: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT16; + *piBytesPerPixel = 6; + bRetVal = true; + break; + + case GL_RGB16F: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = CL_HALF_FLOAT; + *piBytesPerPixel = 6; + bRetVal = true; + break; + + case GL_RGBA16: + case GL_RGBA16UI: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = (gliInternalFormat == GL_RGBA16)? CL_UNORM_INT16:CL_UNSIGNED_INT16; + *piBytesPerPixel = 8; + bRetVal = true; + break; + + case GL_RGBA16I: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT16; + *piBytesPerPixel = 8; + bRetVal = true; + break; + + case GL_RGBA16F: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = CL_HALF_FLOAT; + *piBytesPerPixel = 8; + bRetVal = true; + break; + + case GL_R32I: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT32; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_R32UI: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = CL_UNSIGNED_INT32; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_R32F: + pclImageFormat->image_channel_order = CL_R; + pclImageFormat->image_channel_data_type = CL_FLOAT; + *piBytesPerPixel = 4; + bRetVal = true; + break; + + case GL_RG32I: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT32; + *piBytesPerPixel = 8; + bRetVal = true; + break; + + case GL_RG32UI: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = CL_UNSIGNED_INT32; + *piBytesPerPixel = 8; + bRetVal = true; + break; + + case GL_RG32F: + pclImageFormat->image_channel_order = CL_RG; + pclImageFormat->image_channel_data_type = CL_FLOAT; + *piBytesPerPixel = 8; + bRetVal = true; + break; + + case GL_RGB32I: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT32; + *piBytesPerPixel = 12; + bRetVal = true; + break; + + case GL_RGB32UI: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = CL_UNSIGNED_INT32; + *piBytesPerPixel = 12; + bRetVal = true; + break; + + case GL_RGB32F: + pclImageFormat->image_channel_order = CL_RGB; + pclImageFormat->image_channel_data_type = CL_FLOAT; + *piBytesPerPixel = 12; + bRetVal = true; + break; + + case GL_RGBA32I: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = CL_SIGNED_INT32; + *piBytesPerPixel = 16; + bRetVal = true; + break; + + case GL_RGBA32UI: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = CL_UNSIGNED_INT32; + *piBytesPerPixel = 16; + bRetVal = true; + break; + + case GL_RGBA32F: + pclImageFormat->image_channel_order = CL_RGBA; + pclImageFormat->image_channel_data_type = CL_FLOAT; + *piBytesPerPixel = 16; + bRetVal = true; + break; + case GL_DEPTH_COMPONENT32F: + pclImageFormat->image_channel_order = CL_DEPTH; + pclImageFormat->image_channel_data_type = CL_FLOAT; + *piBytesPerPixel = 4; + bRetVal = true; + break; + case GL_DEPTH_COMPONENT16: + pclImageFormat->image_channel_order = CL_DEPTH; + pclImageFormat->image_channel_data_type = CL_UNORM_INT16; + *piBytesPerPixel = 2; + bRetVal = true; + break; + case GL_DEPTH24_STENCIL8: + pclImageFormat->image_channel_order = CL_DEPTH_STENCIL; + pclImageFormat->image_channel_data_type = CL_UNORM_INT24; + *piBytesPerPixel = 4; + bRetVal = true; + break; + case GL_DEPTH32F_STENCIL8: + pclImageFormat->image_channel_order = CL_DEPTH_STENCIL; + pclImageFormat->image_channel_data_type = CL_FLOAT; + *piBytesPerPixel = 5; + bRetVal = true; + break; + default: + LogWarning("unsupported GL internal format"); + break; + } + amd::Image::Format imageFormat(*pclImageFormat); + if (bRetVal && !imageFormat.isSupported(amdContext)) { + bRetVal = false; + } + return bRetVal; +} + +void +BufferGL::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(BufferGL)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +BufferGL::mapExtObjectInCQThread() +{ + GLFunctions::SetIntEnv ie(context_().glenv()); + if (!ie.isValid()) { + return false; + } + + GLenum glAccess = GL_READ_WRITE; // Default + if (getMemFlags() & CL_MEM_READ_ONLY) { + glAccess = GL_READ_ONLY; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + glAccess = GL_WRITE_ONLY; + } + clearGLErrors(context_()); + context_().glenv()->glBindBuffer_(GL_ARRAY_BUFFER, gluiName_); + + void* pCpuMem = context_().glenv()->glMapBuffer_(GL_ARRAY_BUFFER, glAccess); + + if (checkForGLError(context_()) != GL_NO_ERROR || !pCpuMem) { + LogError("cannot map GL buffer"); + return false; + } + + setHostMem(pCpuMem); + + return true; +} + +bool +BufferGL::unmapExtObjectInCQThread() +{ + GLFunctions::SetIntEnv ie(context_().glenv()); + if (!ie.isValid()) { + return false; + } + + clearGLErrors(context_()); + context_().glenv()->glBindBuffer_(GL_ARRAY_BUFFER, gluiName_); + + if (GL_FALSE == context_().glenv()->glUnmapBuffer_(GL_ARRAY_BUFFER)) { + LogError("context_().glenv()->glUnmapBuffer_ returned GL_FALSE - buffer may be corrupted"); + return false; + } + if (checkForGLError(context_()) != GL_NO_ERROR) { + LogWarning("Error unmapping GL buffer"); + return false; + } + + setHostMem(NULL); + + return true; +} + +static GLenum +clChannelDataTypeToGlType(cl_channel_type channel_type) +{ + // Pick + // GL_BYTE, GL_UNSIGNED_BYTE, GL_SHORT, GL_UNSIGNED_SHORT, GL_INT, + // GL_UNSIGNED_INT, GL_FLOAT, GL_2_BYTES, GL_3_BYTES, GL_4_BYTES + // or GL_DOUBLE + switch (channel_type) { + case CL_SNORM_INT8: return GL_BYTE; + case CL_SNORM_INT16: return GL_SHORT; + case CL_UNORM_INT8: return GL_UNSIGNED_BYTE; + case CL_UNORM_INT16: return GL_UNSIGNED_SHORT; + case CL_SIGNED_INT8: return GL_BYTE; + case CL_SIGNED_INT16: return GL_SHORT; + case CL_SIGNED_INT32: return GL_INT; + case CL_UNSIGNED_INT8: return GL_UNSIGNED_BYTE; + case CL_UNSIGNED_INT16: return GL_UNSIGNED_SHORT; + case CL_UNSIGNED_INT32: return GL_UNSIGNED_INT; + case CL_FLOAT: return GL_FLOAT; + case CL_HALF_FLOAT: + case CL_UNORM_SHORT_565: + case CL_UNORM_SHORT_555: + case CL_UNORM_INT_101010: + default: + guarantee(false && "Unexpected CL type."); + return 0; + } +} + +static GLenum +glInternalFormatToGlFormat(GLenum internalFormat) +{ + switch (internalFormat) { + // Base internal formats + case GL_RGBA: + case GL_BGRA: + return internalFormat; + // Sized internal formats + case GL_RGBA8: + case GL_RGBA16: + case GL_RGBA16F: + case GL_RGBA32F: + return GL_RGBA; + case GL_RGBA8I: + case GL_RGBA8UI: + case GL_RGBA16I: + case GL_RGBA16UI: + case GL_RGBA32I: + case GL_RGBA32UI: + return GL_RGBA_INTEGER; + + default: + guarantee(false && "Unexpected GL internal format."); + return 0; + } +} + +void +ImageGL::initDeviceMemory() +{ + deviceMemories_ = reinterpret_cast( + reinterpret_cast(this) + sizeof(ImageGL)); + memset(deviceMemories_, 0, + context_().devices().size() * sizeof(DeviceMemory)); +} + +bool +ImageGL::mapExtObjectInCQThread() +{ + GLFunctions::SetIntEnv ie(context_().glenv()); + if (!ie.isValid()) { + return false; + } + + GLenum glAccess = GL_READ_WRITE; // Default + + if (getMemFlags() & CL_MEM_READ_ONLY) { + glAccess = GL_READ_ONLY; + } + else if (getMemFlags() & CL_MEM_WRITE_ONLY) { + glAccess = GL_WRITE_ONLY; + } + clearGLErrors(context_()); + context_().glenv()->glBindTexture_(getGLTarget(), gluiName_); + + size_t mem_size = getSize(); + + char* pCpuMem = new char[mem_size]; + if (pCpuMem == NULL) { + LogError("Cannot alloc host memory for ImageGL"); + return false; + } + + context_().glenv()->glGetTexImage_( + getGLTarget(), + gliMipLevel_, + glInternalFormatToGlFormat(glInternalFormat_), + clChannelDataTypeToGlType(getImageFormat().image_channel_data_type), + pCpuMem); + + if (checkForGLError(context_()) != GL_NO_ERROR) { + LogError("cannot map GL texture"); + free(pCpuMem); + return false; + } + + setHostMem(pCpuMem); + + return true; +} + +bool +ImageGL::unmapExtObjectInCQThread() +{ + GLFunctions::SetIntEnv ie(context_().glenv()); + if (!ie.isValid()) { + return false; + } + + bool status = true; + + clearGLErrors(context_()); + context_().glenv()->glBindTexture_(getGLTarget(), gluiName_); + + char* pCpuMem = (char *)getHostMem(); + + if (checkForGLError(context_()) != GL_NO_ERROR) { + LogError("Cannot map GL texture"); + status = false; + goto cleanup; + } + + context_().glenv()->glTexImage2D_( + getGLTarget(), // target + gliMipLevel_, // miplevel + glInternalFormat_, // internalFormat or bytes per pixel + gliWidth_, // width + gliHeight_, // height + 0, // border + // format + glInternalFormatToGlFormat(glInternalFormat_), + // type + clChannelDataTypeToGlType(getImageFormat().image_channel_data_type), + pCpuMem); // data + + if (checkForGLError(context_()) != GL_NO_ERROR) { + LogError("Cannot update GL texture"); + status = false; + goto cleanup; + } + + cleanup: + delete [] pCpuMem; + setHostMem(NULL); + + return status; +} + +//******************************************************************* +// +// Internal implementation of CL API functions +// +//******************************************************************* + +// +// clCreateFromGLBufferAMD +// +cl_mem +clCreateFromGLBufferAMD( + Context& amdContext, + cl_mem_flags flags, + GLuint bufobj, + cl_int* errcode_ret) +{ + BufferGL* pBufferGL = NULL; + GLenum glErr; + GLenum glTarget = GL_ARRAY_BUFFER; + GLint gliSize = 0; + GLint gliMapped = 0; + + // Verify context init'ed for interop + if (!amdContext.glenv() || !amdContext.glenv()->isAssociated()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from GL context or share list"); + return (cl_mem) 0; + } + + // Add this scope to bound the scoped lock + { + GLFunctions::SetIntEnv ie(amdContext.glenv()); + if (!ie.isValid()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from GL context or share list"); + return as_cl(0); + } + + // Verify GL buffer object + clearGLErrors(amdContext); + if ((GL_FALSE == amdContext.glenv()->glIsBuffer_(bufobj)) + || (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_()))) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("\"bufobj\" is not a GL buffer object"); + return (cl_mem) 0; + } + + // It seems that CL spec is not concerned with GL_BUFFER_USAGE, so skip it + + // Check if size is available - data store is created + + amdContext.glenv()->glBindBuffer_(glTarget, bufobj); + clearGLErrors(amdContext); + amdContext.glenv()->glGetBufferParameteriv_(glTarget, GL_BUFFER_SIZE, &gliSize); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("cannot get the GL buffer size"); + return (cl_mem) 0; + } + if (gliSize == 0) { + //@todo - check why sometime the size is zero + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("the GL buffer's data store is not created"); + return (cl_mem) 0; + } + + // Mapping will be done at acquire time (sync point) + + } // Release scoped lock + + // Now create BufferGL object + pBufferGL = new(amdContext) BufferGL(amdContext, flags, gliSize, 0, bufobj); + + if (!pBufferGL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("cannot create object of class BufferGL"); + return (cl_mem) 0; + } + + if (!pBufferGL->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pBufferGL->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + + // Create interop object + if (pBufferGL->getInteropObj() == NULL) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("cannot create object of class BufferGL"); + return (cl_mem)0; + } + + // Fixme: If more than one device is present in the context, we choose the first device. + // We should come up with a more elegant solution to handle this. + assert(amdContext.devices().size() == 1); + + std::vector::const_iterator itr = amdContext.devices().begin(); + amd::Device& dev = *(*itr); + + if (dev.type() != CL_DEVICE_TYPE_CPU){ + device::Memory* mem = pBufferGL->getDeviceMemory(dev); + if (NULL == mem) { + LogPrintfError("Can't allocate memory size - 0x%08X bytes!", + pBufferGL->getSize()); + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + } + mem->processGLResource(device::Memory::GLDecompressResource); + } + + return as_cl(pBufferGL); +} + +cl_mem +clCreateFromGLTextureAMD( + Context& amdContext, + cl_mem_flags clFlags, + GLenum target, + GLint miplevel, + GLuint texture, + int* errcode_ret) +{ + ImageGL* pImageGL = NULL; + GLenum glErr; + GLenum glTarget = 0; + GLenum glInternalFormat; + cl_image_format clImageFormat; + uint dim = 1; + cl_mem_object_type clType; + cl_gl_object_type clGLType; + GLsizei numSamples = 1; + + // Verify context init'ed for interop + if (!amdContext.glenv() || !amdContext.glenv()->isAssociated()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from GL context or share list"); + return static_cast(0); + } + + GLint gliTexWidth = 1; + GLint gliTexHeight = 1; + GLint gliTexDepth = 1; + + // Add this scope to bound the scoped lock + { + GLFunctions::SetIntEnv ie(amdContext.glenv()); + if (!ie.isValid()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from GL context or share list"); + return as_cl(0); + } + + // Verify GL texture object + clearGLErrors(amdContext); + if ((GL_FALSE == amdContext.glenv()->glIsTexture_(texture)) + || (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_()))) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("\"texture\" is not a GL texture object"); + return static_cast(0); + } + + bool image = true; + + // Check target value validity + switch(target) + { + case GL_TEXTURE_BUFFER: + glTarget = GL_TEXTURE_BUFFER; + target = 0; + dim = 1; + clType = CL_MEM_OBJECT_IMAGE1D_BUFFER; + clGLType = CL_GL_OBJECT_TEXTURE_BUFFER; + image = false; + break; + + case GL_TEXTURE_1D: + glTarget = GL_TEXTURE_1D; + target = 0; + dim = 1; + clType = CL_MEM_OBJECT_IMAGE1D; + clGLType = CL_GL_OBJECT_TEXTURE1D; + break; + + case GL_TEXTURE_CUBE_MAP_POSITIVE_X: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y: + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z: + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z: + glTarget = GL_TEXTURE_CUBE_MAP; + dim = 2; + clType = CL_MEM_OBJECT_IMAGE2D; + clGLType = CL_GL_OBJECT_TEXTURE2D; + break; + + case GL_TEXTURE_1D_ARRAY: + glTarget = GL_TEXTURE_1D_ARRAY; + target = 0; + dim = 2; + clType = CL_MEM_OBJECT_IMAGE1D_ARRAY; + clGLType = CL_GL_OBJECT_TEXTURE1D_ARRAY; + break; + + case GL_TEXTURE_2D: + glTarget = GL_TEXTURE_2D; + target = 0; + dim = 2; + clType = CL_MEM_OBJECT_IMAGE2D; + clGLType = CL_GL_OBJECT_TEXTURE2D; + break; + + case GL_TEXTURE_2D_MULTISAMPLE: + glTarget = GL_TEXTURE_2D_MULTISAMPLE; + target = 0; + dim = 2; + clType = CL_MEM_OBJECT_IMAGE2D; + clGLType = CL_GL_OBJECT_TEXTURE2D; + break; + + case GL_TEXTURE_RECTANGLE_ARB: + glTarget = GL_TEXTURE_RECTANGLE_ARB; + target = 0; + dim = 2; + clType = CL_MEM_OBJECT_IMAGE2D; + clGLType = CL_GL_OBJECT_TEXTURE2D; + break; + + case GL_TEXTURE_2D_ARRAY: + glTarget = GL_TEXTURE_2D_ARRAY; + target = 0; + dim = 3; + clType = CL_MEM_OBJECT_IMAGE2D_ARRAY; + clGLType = CL_GL_OBJECT_TEXTURE2D_ARRAY; + break; + + case GL_TEXTURE_3D: + glTarget = GL_TEXTURE_3D; + target = 0; + dim = 3; + clType = CL_MEM_OBJECT_IMAGE3D; + clGLType = CL_GL_OBJECT_TEXTURE3D; + break; + + default: + // wrong value + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid \"target\" value"); + return static_cast(0); + break; + } + + amdContext.glenv()->glBindTexture_(glTarget, texture); + + // Check if size is available - data store is created + if (image) { + // Check mipmap level for "texture" name + GLint gliTexBaseLevel; + GLint gliTexMaxLevel; + + clearGLErrors(amdContext); + amdContext.glenv()->glGetTexParameteriv_(glTarget, GL_TEXTURE_BASE_LEVEL, &gliTexBaseLevel); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_MIP_LEVEL; + LogWarning("Cannot get base mipmap level of a GL \"texture\" object"); + return static_cast(0); + } + clearGLErrors(amdContext); + amdContext.glenv()->glGetTexParameteriv_(glTarget, GL_TEXTURE_MAX_LEVEL, &gliTexMaxLevel); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_MIP_LEVEL; + LogWarning("Cannot get max mipmap level of a GL \"texture\" object"); + return static_cast(0); + } + if ((gliTexBaseLevel > miplevel) || (miplevel > gliTexMaxLevel)) { + *not_null(errcode_ret) = CL_INVALID_MIP_LEVEL; + LogWarning("\"miplevel\" is not a valid mipmap level of the GL \"texture\" object"); + return static_cast(0); + } + + // Get GL texture format and check if it's compatible with CL format + clearGLErrors(amdContext); + amdContext.glenv()->glGetTexLevelParameteriv_(glTarget, miplevel, GL_TEXTURE_INTERNAL_FORMAT, + (GLint*) &glInternalFormat); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("Cannot get internal format of \"miplevel\" of GL \"texture\" object"); + return static_cast(0); + } + + amdContext.glenv()->glGetTexLevelParameteriv_(glTarget, miplevel, GL_TEXTURE_SAMPLES, + (GLint*) &numSamples); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("Cannot get numbers of samples of GL \"texture\" object"); + return static_cast(0); + } + if (numSamples > 1 && ! (amdContext.devices()[0]->settings().depthMSAAInterop_) ) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("MSAA \"texture\" object is not suppoerted for the device"); + return static_cast(0); + } + + // Now get CL format from GL format and bytes per pixel + int iBytesPerPixel = 0; + if (!getCLFormatFromGL(amdContext, glInternalFormat, &clImageFormat, &iBytesPerPixel)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("\"texture\" format does not map to an appropriate CL image format"); + return static_cast(0); + } + + switch (dim) { + case 3: + clearGLErrors(amdContext); + amdContext.glenv()->glGetTexLevelParameteriv_(glTarget, miplevel, GL_TEXTURE_DEPTH, &gliTexDepth); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("Cannot get the depth of \"miplevel\" of GL \"texure\""); + return static_cast(0); + } + // Fall trough to process other dimensions... + case 2: + clearGLErrors(amdContext); + amdContext.glenv()->glGetTexLevelParameteriv_(glTarget, miplevel, GL_TEXTURE_HEIGHT, &gliTexHeight); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("Cannot get the height of \"miplevel\" of GL \"texure\""); + return static_cast(0); + } + // Fall trough to process other dimensions... + case 1: + clearGLErrors(amdContext); + amdContext.glenv()->glGetTexLevelParameteriv_(glTarget, miplevel, GL_TEXTURE_WIDTH, &gliTexWidth); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("Cannot get the width of \"miplevel\" of GL \"texure\""); + return static_cast(0); + } + break; + default: + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid \"target\" value"); + return static_cast(0); + } + } + else { + GLint size; + + // In case target is GL_TEXTURE_BUFFER + amdContext.glenv()->glBindBuffer_(glTarget, texture); + + // Get GL texture format and check if it's compatible with CL format + clearGLErrors(amdContext); + amdContext.glenv()->glGetIntegerv_(GL_TEXTURE_BUFFER_FORMAT, + reinterpret_cast(&glInternalFormat)); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("Cannot get internal format of \"miplevel\" of GL \"texture\" object"); + return static_cast(0); + } + + // Now get CL format from GL format and bytes per pixel + int iBytesPerPixel = 0; + if (!getCLFormatFromGL(amdContext, glInternalFormat, &clImageFormat, &iBytesPerPixel)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("\"texture\" format does not map to an appropriate CL image format"); + return static_cast(0); + } + + clearGLErrors(amdContext); + amdContext.glenv()->glGetBufferParameteriv_(glTarget, GL_BUFFER_SIZE, &size); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("Cannot get internal format of \"miplevel\" of GL \"texture\" object"); + return static_cast(0); + } + + gliTexWidth = size / iBytesPerPixel; + } + size_t imageSize = (clType == CL_MEM_OBJECT_IMAGE1D_ARRAY) ? + static_cast(gliTexHeight) : static_cast(gliTexDepth); + + if (!amd::Image::validateDimensions( + amdContext.devices(), clType, + static_cast(gliTexWidth), static_cast(gliTexHeight), + static_cast(gliTexDepth), imageSize)) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("The GL \"texture\" data store is not created or out of supported dimensions"); + return static_cast(0); + } + + // PBO and mapping will be done at "acquire" time (sync point) + + } // Release scoped lock + + pImageGL = new(amdContext) + ImageGL(amdContext, clType, clFlags, clImageFormat, + static_cast(gliTexWidth), static_cast(gliTexHeight), + static_cast(gliTexDepth), + glTarget, texture, miplevel, glInternalFormat, clGLType,numSamples, + target); + + if (!pImageGL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("Cannot create class ImageGL - out of memory?"); + return static_cast(0); + } + + if (!pImageGL->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImageGL->release(); + return static_cast(0); + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImageGL); +} + +// +// clCreateFromGLRenderbufferDAMD +// +cl_mem +clCreateFromGLRenderbufferAMD( + Context& amdContext, + cl_mem_flags clFlags, + GLuint renderbuffer, + int* errcode_ret) +{ + ImageGL* pImageGL = NULL; + GLenum glErr; + + GLenum glTarget = GL_RENDERBUFFER; + GLenum glInternalFormat; + cl_image_format clImageFormat; + + // Verify context init'ed for interop + if (!amdContext.glenv() || !amdContext.glenv()->isAssociated()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from GL context or share list"); + return (cl_mem) 0; + } + + GLint gliRbWidth; + GLint gliRbHeight; + + // Add this scope to bound the scoped lock + { + GLFunctions::SetIntEnv ie(amdContext.glenv()); + if (!ie.isValid()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("\"amdContext\" is not created from GL context or share list"); + return as_cl(0); + } + + // Verify GL renderbuffer object + clearGLErrors(amdContext); + if ((GL_FALSE == amdContext.glenv()->glIsRenderbufferEXT_(renderbuffer)) + || (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_()))) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("\"renderbuffer\" is not a GL texture object"); + return (cl_mem) 0; + } + + amdContext.glenv()->glBindRenderbuffer_(glTarget, renderbuffer); + + // Get GL RB format and check if it's compatible with CL format + clearGLErrors(amdContext); + amdContext.glenv()->glGetRenderbufferParameterivEXT_(glTarget, GL_RENDERBUFFER_INTERNAL_FORMAT, + (GLint*) &glInternalFormat); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("Cannot get internal format of GL \"renderbuffer\" object"); + return (cl_mem) 0; + } + + // Now get CL format from GL format and bytes per pixel + int iBytesPerPixel = 0; + if (!getCLFormatFromGL(amdContext, glInternalFormat, &clImageFormat, &iBytesPerPixel)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("\"renderbuffer\" format does not map to an appropriate CL image format"); + return (cl_mem) 0; + } + + // Check if size is available - data store is created + clearGLErrors(amdContext); + amdContext.glenv()->glGetRenderbufferParameterivEXT_(glTarget, GL_RENDERBUFFER_WIDTH, + &gliRbWidth); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("Cannot get the width of GL \"renderbuffer\""); + return (cl_mem) 0; + } + if (gliRbWidth == 0) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("The GL \"renderbuffer\" data store is not created"); + return (cl_mem) 0; + } + clearGLErrors(amdContext); + amdContext.glenv()->glGetRenderbufferParameterivEXT_(glTarget, GL_RENDERBUFFER_HEIGHT, + &gliRbHeight); + if (GL_NO_ERROR != (glErr = amdContext.glenv()->glGetError_())) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("Cannot get the height of GL \"renderbuffer\""); + return (cl_mem) 0; + } + if (gliRbHeight == 0) { + *not_null(errcode_ret) = CL_INVALID_GL_OBJECT; + LogWarning("The GL \"renderbuffer\" data store is not created"); + return (cl_mem) 0; + } + + // PBO and mapping will be done at "acquire" time (sync point) + + } // Release scoped lock + + pImageGL = new(amdContext) + ImageGL(amdContext, CL_MEM_OBJECT_IMAGE2D, clFlags, clImageFormat, + (size_t) gliRbWidth, (size_t) gliRbHeight, 1, + glTarget, renderbuffer, 0, glInternalFormat, CL_GL_OBJECT_RENDERBUFFER, 0); + + if (!pImageGL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("Cannot create class ImageGL from renderbuffer - out of memory?"); + return (cl_mem) 0; + } + + if (!pImageGL->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + pImageGL->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(pImageGL); +} + +// +// clEnqueueAcquireExtObjectsAMD +// + +static cl_int +clSetInteropObjects(cl_uint num_objects, + const cl_mem* mem_objects, + std::vector& interopObjects) +{ + if ((num_objects == 0 && mem_objects != NULL) + || (num_objects != 0 && mem_objects == NULL)) { + return CL_INVALID_VALUE; + } + + while (num_objects-- > 0) { + + cl_mem obj = *mem_objects++; + if (!is_valid(obj)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Memory* mem = as_amd(obj); + if (mem->getInteropObj() == NULL) { + return CL_INVALID_GL_OBJECT; + } + + interopObjects.push_back(mem); + } + return CL_SUCCESS; +} + +cl_int +clEnqueueAcquireExtObjectsAMD(cl_command_queue command_queue, + cl_uint num_objects, const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, const cl_event* event_wait_list, + cl_event* event, cl_command_type cmd_type) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (cmd_type == CL_COMMAND_ACQUIRE_GL_OBJECTS) { + // Verify context init'ed for interop + if (!hostQueue.context().glenv() || !hostQueue.context().glenv()->isAssociated()) { + LogWarning("\"amdContext\" is not created from GL context or share list"); + return CL_INVALID_CONTEXT; + } + hostQueue.context().glenv()->glFlush_(); + } + + std::vector memObjects; + cl_int err = clSetInteropObjects(num_objects, mem_objects, memObjects); + if (err != CL_SUCCESS){ + return err; + } + + amd::Command::EventWaitList eventWaitList; + err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + +#ifdef _WIN32 + if ((hostQueue.context().info().flags_ & amd::Context::InteropUserSync) == 0) + { + //! Make sure D3D10 queues are flushed and all commands are finished + //! before CL side would access interop objects + if (cmd_type == CL_COMMAND_ACQUIRE_D3D10_OBJECTS_KHR) { + SyncD3D10Objects(memObjects); + } + //! Make sure D3D11 queues are flushed and all commands are finished + //! before CL side would access interop objects + if (cmd_type == CL_COMMAND_ACQUIRE_D3D11_OBJECTS_KHR) { + SyncD3D11Objects(memObjects); + } + //! Make sure D3D9 queues are flushed and all commands are finished + //! before CL side would access interop objects + if (cmd_type == CL_COMMAND_ACQUIRE_DX9_MEDIA_SURFACES_KHR) { + SyncD3D9Objects(memObjects); + } + } +#endif //_WIN32 + + //! Now create command and enqueue + amd::AcquireExtObjectsCommand* command = new amd::AcquireExtObjectsCommand( + hostQueue, eventWaitList, num_objects, memObjects, cmd_type); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} + + +// +// clEnqueueReleaseExtObjectsAMD +// +cl_int +clEnqueueReleaseExtObjectsAMD(cl_command_queue command_queue, + cl_uint num_objects, const cl_mem* mem_objects, + cl_uint num_events_in_wait_list, const cl_event* event_wait_list, + cl_event* event, cl_command_type cmd_type) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + std::vector memObjects; + cl_int err = clSetInteropObjects(num_objects, mem_objects, memObjects); + if (err != CL_SUCCESS){ + return err; + } + + amd::Command::EventWaitList eventWaitList; + err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + //! Now create command and enqueue + amd::ReleaseExtObjectsCommand* command = new amd::ReleaseExtObjectsCommand( + hostQueue, eventWaitList, num_objects, memObjects, cmd_type); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + +#ifdef _WIN32 + if ((hostQueue.context().info().flags_ & amd::Context::InteropUserSync) == 0) + { + //! Make sure CL command queue is flushed and all commands are finished + //! before D3D10 side would access interop resources + if (cmd_type == CL_COMMAND_RELEASE_DX9_MEDIA_SURFACES_KHR || + cmd_type == CL_COMMAND_RELEASE_D3D10_OBJECTS_KHR || + cmd_type == CL_COMMAND_RELEASE_D3D11_OBJECTS_KHR) { + command->awaitCompletion(); + } + } +#endif //_WIN32 + + *not_null(event) = as_cl(&command->event()); + + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} + +// Placed here as opposed to command.cpp, as glext.h and cl_gl_amd.hpp will have +// to be included because of the GL calls +bool ClGlEvent::waitForFence() +{ + GLenum ret; + // get fence id associated with fence event + GLsync gs = reinterpret_cast (command().data()); + if (!gs) return false; + + // Try to use DC and GLRC of current thread, if it doesn't exist + // create a new GL context on this thread, which is shared with the original context + +#ifdef _WIN32 + HDC tempDC_ = wglGetCurrentDC(); + HGLRC tempGLRC_ = wglGetCurrentContext(); + // Set DC and GLRC + if (tempDC_ && tempGLRC_) { + ret = context().glenv()->glClientWaitSync_(gs, GL_SYNC_FLUSH_COMMANDS_BIT, static_cast(-1)); + if (!(ret == GL_ALREADY_SIGNALED || ret == GL_CONDITION_SATISFIED)) return false; + } + else + { + tempDC_ = context().glenv()->getDC(); + tempGLRC_ = context().glenv()->getIntGLRC(); + if (!context().glenv()->init(reinterpret_cast(tempDC_), reinterpret_cast(tempGLRC_))) return false; + + // Make the newly created GL context current to this thread + context().glenv()->setIntEnv(); + // If fence has not yet executed, wait till it finishes + ret = context().glenv()->glClientWaitSync_(gs, GL_SYNC_FLUSH_COMMANDS_BIT, static_cast(-1)); + if (!(ret == GL_ALREADY_SIGNALED || ret == GL_CONDITION_SATISFIED)) return false; + // Since we're done making GL calls, restore whatever context was previously current to this thread + context().glenv()->restoreEnv(); + } +#else // Lnx + Display* tempDpy_ = context().glenv()->glXGetCurrentDisplay_(); + GLXDrawable tempDrawable_ = context().glenv()->glXGetCurrentDrawable_(); + GLXContext tempCtx_ = context().glenv()->glXGetCurrentContext_(); + // Set internal Display and GLXContext + if (tempDpy_ && tempCtx_) { + ret = context().glenv()->glClientWaitSync_(gs, GL_SYNC_FLUSH_COMMANDS_BIT, static_cast(-1)); + if (!(ret == GL_ALREADY_SIGNALED || ret == GL_CONDITION_SATISFIED)) return false; + } + else { + if (!context().glenv()->init(reinterpret_cast(context().glenv()->getIntDpy()), + reinterpret_cast(context().glenv()->getIntCtx()))) return false; + + // Make the newly created GL context current to this thread + context().glenv()->setIntEnv(); + // If fence has not yet executed, wait till it finishes + ret = context().glenv()->glClientWaitSync_(gs, GL_SYNC_FLUSH_COMMANDS_BIT, static_cast(-1)); + if (!(ret == GL_ALREADY_SIGNALED || ret == GL_CONDITION_SATISFIED)) return false; + // Since we're done making GL calls, restore whatever context was previously current to this thread + context().glenv()->restoreEnv(); + } +#endif + // If we reach this point, fence should have completed + setStatus(CL_COMPLETE); + return true; +} + +// +// GLFunctions implementation +// + +#ifdef _WIN32 +#define CONVERT_CHAR_GLUBYTE +#else //!_WIN32 +#define CONVERT_CHAR_GLUBYTE (GLubyte*) +#endif //!_WIN32 + +#define GLPREFIX(rtype, fcn, dclargs) \ + if (!(fcn##_ = (PFN_##fcn) GETPROCADDRESS( \ + libHandle_, #fcn))) { \ + if (!(fcn##_ = (PFN_##fcn) GetProcAddress_( \ + reinterpret_cast(#fcn)))) ++missed_; \ + } + +GLFunctions::SetIntEnv::SetIntEnv(GLFunctions* env) + : env_(env) +{ + env_->getLock().lock(); + + // Set environment (DC and GLRC) + isValid_ = env_->setIntEnv(); + +} + +GLFunctions::SetIntEnv::~SetIntEnv() +{ + // Restore environment (CL DC and CL GLRC) + env_->restoreEnv(); + + env_->getLock().unlock(); +} + +GLFunctions::GLFunctions(HMODULE h) : + libHandle_(h), + missed_(0), +#ifdef _WIN32 + hOrigGLRC_(0), + hDC_(0), + hIntGLRC_(0) +#else //!_WIN32 + Dpy_(0), + Drawable_(0), + origCtx_(0), + intDpy_(0), + intDrawable_(0), + intCtx_(0), + XOpenDisplay_(NULL), + XCloseDisplay_(NULL), + glXGetCurrentDrawable_(NULL), + glXGetCurrentDisplay_(NULL), + glXGetCurrentContext_(NULL), + glXChooseVisual_(NULL), + glXCreateContext_(NULL), + glXDestroyContext_(NULL), + glXMakeCurrent_(NULL) +#endif //!_WIN32 +{ + GetProcAddress_ = (PFN_xxxGetProcAddress) GETPROCADDRESS(h, API_GETPROCADDR); + +#ifndef _WIN32 + // Initialize pointers to X11/GLX functions + // We can not link with these functions on compile time since we need to support + // console mode. In console mode X server and X server components may be absent. + // Hence linking with X11 or libGL will fail module image loading in console mode.-tzachi cohen +#define VERIFY_POINTER(p) if (NULL == p) {missed_++;} + + glXGetCurrentDrawable_ = (PFNglXGetCurrentDrawable)GETPROCADDRESS(h,"glXGetCurrentDrawable"); + VERIFY_POINTER(glXGetCurrentDrawable_) + glXGetCurrentDisplay_ = (PFNglXGetCurrentDisplay)GETPROCADDRESS(h,"glXGetCurrentDisplay"); + VERIFY_POINTER(glXGetCurrentDisplay_) + glXGetCurrentContext_ = (PFNglXGetCurrentContext) GETPROCADDRESS(h,"glXGetCurrentContext"); + VERIFY_POINTER(glXGetCurrentContext_) + glXChooseVisual_ = (PFNglXChooseVisual)GETPROCADDRESS(h,"glXChooseVisual"); + VERIFY_POINTER(glXChooseVisual_) + glXCreateContext_ = (PFNglXCreateContext)GETPROCADDRESS(h,"glXCreateContext"); + VERIFY_POINTER(glXCreateContext_) + glXDestroyContext_ = (PFNglXDestroyContext) GETPROCADDRESS(h,"glXDestroyContext"); + VERIFY_POINTER(glXDestroyContext_) + glXMakeCurrent_ = (PFNglXMakeCurrent) GETPROCADDRESS(h,"glXMakeCurrent"); + VERIFY_POINTER(glXMakeCurrent_) + + HMODULE hXModule = (HMODULE) Os::loadLibrary("libX11.so.6"); + if (NULL != hXModule) { + XOpenDisplay_ = (PFNXOpenDisplay)GETPROCADDRESS(hXModule,"XOpenDisplay"); + VERIFY_POINTER(XOpenDisplay_) + XCloseDisplay_= (PFNXCloseDisplay)GETPROCADDRESS(hXModule,"XCloseDisplay"); + VERIFY_POINTER(XCloseDisplay_) + } + else{ + missed_ += 2; + } +#endif + + // Initialize pointers to GL functions + #include "gl_functions.hpp" +} + +GLFunctions::~GLFunctions() +{ +#ifdef _WIN32 + if (hIntGLRC_) { + if (!wglDeleteContext_(hIntGLRC_)) { + DWORD dwErr = GetLastError(); + LogWarning("Cannot delete GLRC"); + } + } +#else //!_WIN32 + if (intDpy_) { + if (intCtx_) { + glXDestroyContext_(intDpy_, intCtx_); + intCtx_ = NULL; + } + XCloseDisplay_(intDpy_); + intDpy_ = NULL; + } +#endif //!_WIN32 +} + +bool +GLFunctions::init(intptr_t hdc, intptr_t hglrc) +{ + +#ifdef _WIN32 + DWORD err; +#endif //_WIN32 + + if (!missed_) { +#ifdef _WIN32 + if (!hdc) { + hDC_ = wglGetCurrentDC_(); + } + else + { + hDC_ = (HDC) hdc; + } + hOrigGLRC_ = (HGLRC) hglrc; + if (!(hIntGLRC_ = wglCreateContext_(hDC_))) { + err = GetLastError(); + return false; + } + if (!wglShareLists_(hOrigGLRC_, hIntGLRC_)) { + err = GetLastError(); + return false; + } +#else //!_WIN32 + if (!hdc) { + Dpy_ = glXGetCurrentDisplay_(); + } + else { + Dpy_ = (Display*) hdc; + } + Drawable_ = glXGetCurrentDrawable_(); + origCtx_ = (GLXContext) hglrc; + + int attribList[] = { + GLX_RGBA, + None}; + if (!(intDpy_ = XOpenDisplay_(DisplayString(Dpy_)))) { +#if defined(ATI_ARCH_X86) + asm("int $3"); +#endif + } + intDrawable_ = DefaultRootWindow(intDpy_); + + XVisualInfo* vis; + int defaultScreen = DefaultScreen(intDpy_); + if (!(vis = glXChooseVisual_(intDpy_, defaultScreen , attribList))) { + return false; + } + if (!(intCtx_ = glXCreateContext_(intDpy_, vis, origCtx_, true))) { + return false; + } +#endif //!_WIN32 + return true; + } + return false; +} + +bool +GLFunctions::setIntEnv() +{ +#ifdef _WIN32 + // Save current DC and GLRC + tempDC_ = wglGetCurrentDC_(); + tempGLRC_ = wglGetCurrentContext_(); + // Set internal DC and GLRC + if (tempDC_ != getDC() || tempGLRC_ != getIntGLRC()) { + if (!wglMakeCurrent_(getDC(), getIntGLRC())) { + DWORD err = GetLastError(); + LogWarning("cannot set internal GL environment"); + return false; + } + } +#else //!_WIN32 + tempDpy_ = glXGetCurrentDisplay_(); + tempDrawable_ = glXGetCurrentDrawable_(); + tempCtx_ = glXGetCurrentContext_(); + // Set internal Display and GLXContext + if (tempDpy_ != getDpy() || tempCtx_ != getIntCtx()) { + if (!glXMakeCurrent_( + getIntDpy(), getIntDrawable(), getIntCtx())) { + LogWarning("cannot set internal GL environment"); + return false; + } + } +#endif //!_WIN32 + + return true; +} + +bool +GLFunctions::restoreEnv() +{ +#ifdef _WIN32 + // Restore original DC and GLRC + if (!wglMakeCurrent_(tempDC_, tempGLRC_)) { + DWORD err = GetLastError(); + LogWarning("cannot restore original GL environment"); + return false; + } +#else //!_WIN32 + // Restore Display and GLXContext + if (tempDpy_) { + if (!glXMakeCurrent_(tempDpy_, tempDrawable_, tempCtx_)) { + LogWarning("cannot restore original GL environment"); + return false; + } + } + else { + // Just release internal context + if (!glXMakeCurrent_(getIntDpy(), None, NULL)) { + LogWarning("cannot reelase internal GL environment"); + return false; + } + } +#endif //!_WIN32 + + return true; +} + +} //namespace amd diff --git a/opencl/api/opencl/amdocl/cl_gl_amd.hpp b/opencl/api/opencl/amdocl/cl_gl_amd.hpp new file mode 100644 index 0000000000..6f05917e56 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_gl_amd.hpp @@ -0,0 +1,348 @@ +// +// Copyright 2010 Advanced Micro Devices, Inc. All rights reserved. +// + +#ifndef CL_GL_AMD_HPP_ +#define CL_GL_AMD_HPP_ + +#ifdef _WIN32 +#include +#else //!_WIN32 +#include +#endif //!_WIN32 + +#include +#include +#include "CL/cl_gl.h" +#ifndef _WIN32 +#include +#endif //!_WIN32 +#include "platform/context.hpp" +#include "platform/command.hpp" + +namespace amd +{ + +//! Class GLObject keeps all the info about the GL object +//! from which the CL object is created +class GLObject : public InteropObject +{ +protected: + cl_gl_object_type clGLType_; //!< CL GL object type + GLenum glTarget_; + GLuint gluiName_; + GLint gliMipLevel_; + GLenum glInternalFormat_; + GLint gliWidth_; + GLint gliHeight_; + GLint gliDepth_; + GLenum glCubemapFace_; + GLsizei glNumSamples_; + //! Need to pass PBO name from acquire to release + GLuint gluiPBO_; + + // @todo: TBD: Do we need to sync data after access + // or it'll be done by the GL driver? + cl_int cliChecksum_; + +public: +//! GLObject constructor initializes member variables + GLObject( + GLenum glTarget, + GLuint gluiName, + GLint gliMipLevel, + GLenum glInternalFormat, + GLint gliWidth, + GLint gliHeight, + GLint gliDepth, + cl_gl_object_type clGLType, + GLenum glCubemapFace, + GLsizei glNumSamples + ): // Initialization of member variables + clGLType_(clGLType), + glTarget_(glTarget), + gluiName_(gluiName), + gliMipLevel_(gliMipLevel), + glInternalFormat_(glInternalFormat), + gliWidth_(gliWidth), + gliHeight_(gliHeight), + gliDepth_(gliDepth), + glCubemapFace_(glCubemapFace), + glNumSamples_(glNumSamples), + cliChecksum_(0) + { + } + + virtual ~GLObject() {} + virtual GLObject* asGLObject() {return this;} + +//! GLObject query functions to get GL info from member variables + GLenum getGLTarget() const {return glTarget_;} + GLuint getGLName() const {return gluiName_;} + GLint getGLMipLevel() const {return gliMipLevel_;} + GLenum getGLInternalFormat() const {return glInternalFormat_;} + GLint getGLSize() const {return gliWidth_;} + GLint getGLWidth() const {return gliWidth_;} + GLint getGLHeight() const {return gliHeight_;} + GLint getGLDepth() const {return gliDepth_;} + cl_gl_object_type getCLGLObjectType() const { return clGLType_; } + GLenum getCubemapFace() const {return glCubemapFace_;} + GLsizei getNumSamples() const { return glNumSamples_;} + void setPBOName(GLuint gluiName) {gluiPBO_ = gluiName;} + GLuint getPBOName() const {return gluiPBO_;} + +}; + + +//! Class BufferGL is drived from classes Buffer and GLObject +//! where the former keeps all data for CL object and +//! the latter keeps all data for GL object +class BufferGL : public Buffer, public GLObject +{ +protected: + //! Initializes the device memory array which is nested + // after'BufferGL' object in memory layout. + virtual void initDeviceMemory(); +public: +//! BufferGL constructor just calls constructors of base classes +//! to pass down the parameters + BufferGL( + Context& amdContext, + cl_mem_flags clFlags, + size_t uiSizeInBytes, + GLenum glTarget, + GLuint gluiName) + : // Call base classes constructors + Buffer( + amdContext, + clFlags, + uiSizeInBytes + ), + GLObject( + glTarget, + gluiName, + 0, // Mipmap level default + GL_ARRAY_BUFFER, // Just init to some value + (GLint) uiSizeInBytes, + 1, + 1, + CL_GL_OBJECT_BUFFER, + 0, + 0 + ) + { + setInteropObj(this); + } + virtual ~BufferGL() {} + + virtual BufferGL* asBufferGL() { return this; } + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); +}; + + +//! Class ImageGL is derived from classes Image and GLObject +//! where the former keeps all data for CL object and +//! the latter keeps all data for GL object +class ImageGL : public Image, public GLObject +{ +public: + //! ImageGL constructor just calls constructors of base classes + //! to pass down the parameters + ImageGL( + Context& amdContext, + cl_mem_object_type clType, + cl_mem_flags clFlags, + const Format& format, + size_t width, + size_t height, + size_t depth, + GLenum glTarget, + GLuint gluiName, + GLint gliMipLevel, + GLenum glInternalFormat, + cl_gl_object_type clGLType, + GLsizei numSamples, + GLenum glCubemapFace = 0) + : Image(amdContext, clType, clFlags, format, width, height, depth, + Format(format).getElementSize() * width, + Format(format).getElementSize() * width * depth) + , GLObject(glTarget, gluiName, gliMipLevel, glInternalFormat, + static_cast(width), static_cast(height), + static_cast(depth), clGLType, glCubemapFace,numSamples) + { + setInteropObj(this); + } + + virtual ~ImageGL() {} + + //! For CPU device only! + virtual bool mapExtObjectInCQThread(void); + virtual bool unmapExtObjectInCQThread(void); + +protected: + //! Initializes the device memory array which is nested + // after'BufferGL' object in memory layout. + virtual void initDeviceMemory(); +}; + +#ifdef _WIN32 +#define APICALL WINAPI +#define GETPROCADDRESS GetProcAddress +#define API_GETPROCADDR "wglGetProcAddress" +#define FCN_STR_TYPE LPCSTR + typedef PROC (WINAPI* PFN_xxxGetProcAddress) (LPCSTR fcnName); +#else //!_WIN32 +#define APICALL // __stdcall //??? todo odintsov +#define API_GETPROCADDR "glXGetProcAddress" +#define GETPROCADDRESS dlsym +#define FCN_STR_TYPE const GLubyte* +#define WINAPI +#define PROC void* + typedef void* (*PFN_xxxGetProcAddress) (const GLubyte* procName); + // X11 typedef + typedef Display* (*PFNXOpenDisplay)(_Xconst char* display_name ); + typedef int (*PFNXCloseDisplay)(Display* display ); + + //glx typedefs + typedef GLXDrawable (*PFNglXGetCurrentDrawable)(); + typedef Display* (*PFNglXGetCurrentDisplay)(); + typedef GLXContext (*PFNglXGetCurrentContext)( void ); + typedef XVisualInfo* (*PFNglXChooseVisual)(Display *dpy, int screen, int *attribList); + typedef GLXContext(*PFNglXCreateContext)(Display* dpy,XVisualInfo* vis,GLXContext shareList,Bool direct); + typedef void(*PFNglXDestroyContext)(Display* dpy, GLXContext ctx); + typedef Bool(*PFNglXMakeCurrent)( Display* dpy, GLXDrawable drawable, GLXContext ctx); + typedef void* HMODULE; + +#endif //!_WIN32 + +#define GLPREFIX(rtype, fcn, dclargs) \ + typedef rtype (APICALL* PFN_##fcn) dclargs; + +// Declare prototypes for GL functions +#include "gl_functions.hpp" + +class GLFunctions +{ +public: + //! Locks any access to the virtual GPUs + class SetIntEnv : public amd::StackObject { + public: + //! Default constructor + SetIntEnv(GLFunctions* env); + + //! Destructor + ~SetIntEnv(); + + //! Checks if the environment setup was successful + bool isValid() const { return isValid_; } + + private: + GLFunctions* env_; //!< GL environment + bool isValid_; //!< If TRUE, then it's a valid setup + }; + +private: + HMODULE libHandle_; + int missed_; // Indicates how many GL functions not init'ed, if any + + amd::Monitor lock_; + +#ifdef _WIN32 + HGLRC hOrigGLRC_; + HDC hDC_; + HGLRC hIntGLRC_; // handle for internal GLRC to access shared context + HDC tempDC_; + HGLRC tempGLRC_; +#else +public: + Display* Dpy_; + GLXDrawable Drawable_; + GLXContext origCtx_; + Display* intDpy_; + Window intDrawable_; + GLXContext intCtx_; + Display* tempDpy_; + GLXDrawable tempDrawable_; + GLXContext tempCtx_; + + //pointers to X11 functions + PFNXOpenDisplay XOpenDisplay_; + PFNXCloseDisplay XCloseDisplay_; + + //pointers to GLX functions + PFNglXGetCurrentDrawable glXGetCurrentDrawable_; + PFNglXGetCurrentDisplay glXGetCurrentDisplay_; + PFNglXGetCurrentContext glXGetCurrentContext_; + PFNglXChooseVisual glXChooseVisual_; + PFNglXCreateContext glXCreateContext_; + PFNglXDestroyContext glXDestroyContext_; + PFNglXMakeCurrent glXMakeCurrent_; +#endif +public: + + GLFunctions(HMODULE h); + ~GLFunctions(); + + // Query CL-GL context association + bool isAssociated() const + { +#ifdef _WIN32 + if(hDC_ && hOrigGLRC_) return true; +#else //!_WIN32 + if(Dpy_ && origCtx_) return true; +#endif //!_WIN32 + return false; + } + // Accessor methods +#ifdef _WIN32 + HGLRC getOrigGLRC() const {return hOrigGLRC_;} + HDC getDC() const {return hDC_;} + HGLRC getIntGLRC() const {return hIntGLRC_;} +#else //!_WIN32 + Display* getDpy() const {return Dpy_;} + GLXDrawable getDrawable() const {return Drawable_;} + GLXContext getOrigCtx() const {return origCtx_;} + + Display* getIntDpy() const {return intDpy_;} + GLXDrawable getIntDrawable() const {return intDrawable_;} + GLXContext getIntCtx() const {return intCtx_;} +#endif //!_WIN32 + + // Initialize GL dynamic library and function pointers + bool init(intptr_t hdc, intptr_t hglrc); + + // Return true if successful, false - if error occurred + bool setIntEnv(); + bool restoreEnv(); + + amd::Monitor& getLock() { return lock_; } + + PFN_xxxGetProcAddress GetProcAddress_; + +#define GLPREFIX(rtype, fcn, dclargs) \ + PFN_##fcn fcn##_; +// Declare pointers to GL functions +#include "gl_functions.hpp" +}; + +//! Functions for executing the GL related stuff +cl_mem clCreateFromGLBufferAMD(Context& amdContext, cl_mem_flags flags, + GLuint bufobj, cl_int* errcode_ret); +cl_mem clCreateFromGLTextureAMD(Context& amdContext, cl_mem_flags flags, + GLenum target, GLint miplevel, GLuint texture, int* errcode_ret); +cl_mem clCreateFromGLRenderbufferAMD(Context& amdContext, cl_mem_flags flags, + GLuint renderbuffer, int* errcode_ret); + +bool +getCLFormatFromGL( + const Context& amdContext, + GLint gliInternalFormat, + cl_image_format* pclImageFormat, + int* piBytesPerPixel); + +} //namespace amd + +#endif //CL_GL_AMD_HPP_ diff --git a/opencl/api/opencl/amdocl/cl_icd.cpp b/opencl/api/opencl/amdocl/cl_icd.cpp new file mode 100644 index 0000000000..4dd85c2721 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_icd.cpp @@ -0,0 +1,235 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" +#ifdef _WIN32 +#include +#include "cl_d3d9_amd.hpp" +#include "cl_d3d10_amd.hpp" +#include "cl_d3d11_amd.hpp" +#endif //_WIN32 + +#include + +amd::PlatformIDS amd::PlatformID::Platform = //{ NULL }; + { amd::ICDDispatchedObject::icdVendorDispatch_ }; + +static cl_int CL_API_CALL +icdGetPlatformInfo( + cl_platform_id platform, + cl_platform_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) +{ + return clGetPlatformInfo( + NULL, param_name, param_value_size, param_value, param_value_size_ret); +} + +static cl_int CL_API_CALL +icdGetDeviceIDs( + cl_platform_id platform, + cl_device_type device_type, + cl_uint num_entries, + cl_device_id *devices, + cl_uint *num_devices) +{ + return clGetDeviceIDs( + NULL, device_type, num_entries, devices, num_devices); +} + +static cl_int CL_API_CALL +icdGetDeviceInfo( + cl_device_id device, + cl_device_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) +{ + if (param_name == CL_DEVICE_PLATFORM) { + // Return the ICD platform instead of the default NULL platform. + cl_platform_id platform = reinterpret_cast(&amd::PlatformID::Platform); + return amd::clGetInfo( + platform, param_value_size, param_value, param_value_size_ret); + } + + return clGetDeviceInfo( + device, param_name, param_value_size, param_value, param_value_size_ret); +} + +KHRicdVendorDispatch +amd::ICDDispatchedObject::icdVendorDispatch_[] = {{ + NULL /* should not get called */, + icdGetPlatformInfo, + icdGetDeviceIDs, + icdGetDeviceInfo, + clCreateContext, + clCreateContextFromType, + clRetainContext, + clReleaseContext, + clGetContextInfo, + clCreateCommandQueue, + clRetainCommandQueue, + clReleaseCommandQueue, + clGetCommandQueueInfo, + clSetCommandQueueProperty, + clCreateBuffer, + clCreateImage2D, + clCreateImage3D, + clRetainMemObject, + clReleaseMemObject, + clGetSupportedImageFormats, + clGetMemObjectInfo, + clGetImageInfo, + clCreateSampler, + clRetainSampler, + clReleaseSampler, + clGetSamplerInfo, + clCreateProgramWithSource, + clCreateProgramWithBinary, + clRetainProgram, + clReleaseProgram, + clBuildProgram, + clUnloadCompiler, + clGetProgramInfo, + clGetProgramBuildInfo, + clCreateKernel, + clCreateKernelsInProgram, + clRetainKernel, + clReleaseKernel, + clSetKernelArg, + clGetKernelInfo, + clGetKernelWorkGroupInfo, + clWaitForEvents, + clGetEventInfo, + clRetainEvent, + clReleaseEvent, + clGetEventProfilingInfo, + clFlush, + clFinish, + clEnqueueReadBuffer, + clEnqueueWriteBuffer, + clEnqueueCopyBuffer, + clEnqueueReadImage, + clEnqueueWriteImage, + clEnqueueCopyImage, + clEnqueueCopyImageToBuffer, + clEnqueueCopyBufferToImage, + clEnqueueMapBuffer, + clEnqueueMapImage, + clEnqueueUnmapMemObject, + clEnqueueNDRangeKernel, + clEnqueueTask, + clEnqueueNativeKernel, + clEnqueueMarker, + clEnqueueWaitForEvents, + clEnqueueBarrier, + clGetExtensionFunctionAddress, + clCreateFromGLBuffer, + clCreateFromGLTexture2D, + clCreateFromGLTexture3D, + clCreateFromGLRenderbuffer, + clGetGLObjectInfo, + clGetGLTextureInfo, + clEnqueueAcquireGLObjects, + clEnqueueReleaseGLObjects, + clGetGLContextInfoKHR, + WINDOWS_SWITCH(clGetDeviceIDsFromD3D10KHR,NULL), + WINDOWS_SWITCH(clCreateFromD3D10BufferKHR,NULL), + WINDOWS_SWITCH(clCreateFromD3D10Texture2DKHR,NULL), + WINDOWS_SWITCH(clCreateFromD3D10Texture3DKHR,NULL), + WINDOWS_SWITCH(clEnqueueAcquireD3D10ObjectsKHR,NULL), + WINDOWS_SWITCH(clEnqueueReleaseD3D10ObjectsKHR,NULL), + clSetEventCallback, + clCreateSubBuffer, + clSetMemObjectDestructorCallback, + clCreateUserEvent, + clSetUserEventStatus, + clEnqueueReadBufferRect, + clEnqueueWriteBufferRect, + clEnqueueCopyBufferRect, + clCreateSubDevicesEXT, + clRetainDeviceEXT, + clReleaseDeviceEXT, + clCreateEventFromGLsyncKHR, + + /* OpenCL 1.2*/ + clCreateSubDevices, + clRetainDevice, + clReleaseDevice, + clCreateImage, + clCreateProgramWithBuiltInKernels, + clCompileProgram, + clLinkProgram, + clUnloadPlatformCompiler, + clGetKernelArgInfo, + clEnqueueFillBuffer, + clEnqueueFillImage, + clEnqueueMigrateMemObjects, + clEnqueueMarkerWithWaitList, + clEnqueueBarrierWithWaitList, + clGetExtensionFunctionAddressForPlatform, + clCreateFromGLTexture, + + WINDOWS_SWITCH(clGetDeviceIDsFromD3D11KHR,NULL), + WINDOWS_SWITCH(clCreateFromD3D11BufferKHR,NULL), + WINDOWS_SWITCH(clCreateFromD3D11Texture2DKHR,NULL), + WINDOWS_SWITCH(clCreateFromD3D11Texture3DKHR,NULL), + WINDOWS_SWITCH(clCreateFromDX9MediaSurfaceKHR, NULL), + WINDOWS_SWITCH(clEnqueueAcquireD3D11ObjectsKHR,NULL), + WINDOWS_SWITCH(clEnqueueReleaseD3D11ObjectsKHR,NULL), + + WINDOWS_SWITCH(clGetDeviceIDsFromDX9MediaAdapterKHR,NULL),//KHRpfn_clGetDeviceIDsFromDX9MediaAdapterKHR clGetDeviceIDsFromDX9MediaAdapterKHR; + WINDOWS_SWITCH(clEnqueueAcquireDX9MediaSurfacesKHR, NULL), //KHRpfn_clEnqueueAcquireDX9MediaSurfacesKHR clEnqueueAcquireDX9MediaSurfacesKHR; + WINDOWS_SWITCH(clEnqueueReleaseDX9MediaSurfacesKHR, NULL), //KHRpfn_clEnqueueReleaseDX9MediaSurfacesKHR clEnqueueReleaseDX9MediaSurfacesKHR; + + NULL, + NULL, + NULL, + NULL, + + clCreateCommandQueueWithProperties, + clCreatePipe, + clGetPipeInfo, + clSVMAlloc, + clSVMFree, + clEnqueueSVMFree, + clEnqueueSVMMemcpy, + clEnqueueSVMMemFill, + clEnqueueSVMMap, + clEnqueueSVMUnmap, + clCreateSamplerWithProperties, + clSetKernelArgSVMPointer, + clSetKernelExecInfo, + + clGetKernelSubGroupInfoKHR, + clTerminateContextKHR +}}; + +CL_API_ENTRY cl_int CL_API_CALL +clIcdGetPlatformIDsKHR( + cl_uint num_entries, + cl_platform_id * platforms, + cl_uint * num_platforms) +{ + if (!amd::Runtime::initialized()) { + amd::Runtime::init(); + } + + if (((num_entries > 0 || num_platforms == NULL) && platforms == NULL) + || (num_entries == 0 && platforms != NULL)) { + return CL_INVALID_VALUE; + } + if (num_platforms != NULL && platforms == NULL) { + *num_platforms = 1; + return CL_SUCCESS; + } + + assert(platforms != NULL && "check the code above"); + *platforms = reinterpret_cast(&amd::PlatformID::Platform); + + *not_null(num_platforms) = 1; + return CL_SUCCESS; + +} diff --git a/opencl/api/opencl/amdocl/cl_icd_amd.h b/opencl/api/opencl/amdocl/cl_icd_amd.h new file mode 100644 index 0000000000..999fd180f7 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_icd_amd.h @@ -0,0 +1,981 @@ +/******************************************************************************* + * Copyright (c) 2008-2010 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +#ifndef __OPENCL_CL_ICD_H +#define __OPENCL_CL_ICD_H + +#include +#include + +#define cl_khr_icd 1 + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +typedef cl_int (CL_API_CALL * clGetPlatformIDs_fn)( + cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetPlatformInfo_fn)( + cl_platform_id /* platform */, + cl_platform_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetDeviceIDs_fn)( + cl_platform_id /* platform */, + cl_device_type /* device_type */, + cl_uint /* num_entries */, + cl_device_id * /* devices */, + cl_uint * /* num_devices */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetDeviceInfo_fn)( + cl_device_id /* device */, + cl_device_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_context (CL_API_CALL * clCreateContext_fn)( + const cl_context_properties * /* properties */, + cl_uint /* num_devices */, + const cl_device_id * /* devices */, + void (CL_CALLBACK * /* pfn_notify */)(const char *, const void *, size_t, void *), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_context (CL_API_CALL * clCreateContextFromType_fn)( + const cl_context_properties * /* properties */, + cl_device_type /* device_type */, + void (CL_CALLBACK * /* pfn_notify*/ )(const char *, const void *, size_t, void *), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clRetainContext_fn)( + cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseContext_fn)( + cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetContextInfo_fn)( + cl_context /* context */, + cl_context_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_command_queue (CL_API_CALL * clCreateCommandQueue_fn)( + cl_context /* context */, + cl_device_id /* device */, + cl_command_queue_properties /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clRetainCommandQueue_fn)( + cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseCommandQueue_fn)( + cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetCommandQueueInfo_fn)( + cl_command_queue /* command_queue */, + cl_command_queue_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clSetCommandQueueProperty_fn)( + cl_command_queue /* command_queue */, + cl_command_queue_properties /* properties */, + cl_bool /* enable */, + cl_command_queue_properties * /* old_properties */) /*CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED*/; + +typedef cl_mem (CL_API_CALL * clCreateBuffer_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + size_t /* size */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_mem (CL_API_CALL * clCreateSubBuffer_fn)( + cl_mem /* buffer */, + cl_mem_flags /* flags */, + cl_buffer_create_type /* buffer_create_type */, + const void * /* buffer_create_info */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_mem (CL_API_CALL * clCreateImage2D_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + size_t /* image_width */, + size_t /* image_height */, + size_t /* image_row_pitch */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_mem (CL_API_CALL * clCreateImage3D_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + size_t /* image_width */, + size_t /* image_height */, + size_t /* image_depth */, + size_t /* image_row_pitch */, + size_t /* image_slice_pitch */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clRetainMemObject_fn)( + cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseMemObject_fn)( + cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetSupportedImageFormats_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_mem_object_type /* image_type */, + cl_uint /* num_entries */, + cl_image_format * /* image_formats */, + cl_uint * /* num_image_formats */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetMemObjectInfo_fn)( + cl_mem /* memobj */, + cl_mem_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetImageInfo_fn)( + cl_mem /* image */, + cl_image_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clSetMemObjectDestructorCallback_fn)( + cl_mem /* memobj */, + void (CL_CALLBACK * /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), + void * /*user_data */ ) CL_API_SUFFIX__VERSION_1_1; + +/* Sampler APIs */ +typedef cl_sampler (CL_API_CALL * clCreateSampler_fn)( + cl_context /* context */, + cl_bool /* normalized_coords */, + cl_addressing_mode /* addressing_mode */, + cl_filter_mode /* filter_mode */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clRetainSampler_fn)( + cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseSampler_fn)( + cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetSamplerInfo_fn)( + cl_sampler /* sampler */, + cl_sampler_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Program Object APIs */ +typedef cl_program (CL_API_CALL * clCreateProgramWithSource_fn)( + cl_context /* context */, + cl_uint /* count */, + const char ** /* strings */, + const size_t * /* lengths */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_program (CL_API_CALL * clCreateProgramWithBinary_fn)( + cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const size_t * /* lengths */, + const unsigned char ** /* binaries */, + cl_int * /* binary_status */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clRetainProgram_fn)( + cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseProgram_fn)( + cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clBuildProgram_fn)( + cl_program /* program */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clUnloadCompiler_fn)(void) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetProgramInfo_fn)( + cl_program /* program */, + cl_program_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetProgramBuildInfo_fn)( + cl_program /* program */, + cl_device_id /* device */, + cl_program_build_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Kernel Object APIs */ +typedef cl_kernel (CL_API_CALL * clCreateKernel_fn)( + cl_program /* program */, + const char * /* kernel_name */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clCreateKernelsInProgram_fn)( + cl_program /* program */, + cl_uint /* num_kernels */, + cl_kernel * /* kernels */, + cl_uint * /* num_kernels_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clRetainKernel_fn)( + cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseKernel_fn)( + cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clSetKernelArg_fn)( + cl_kernel /* kernel */, + cl_uint /* arg_index */, + size_t /* arg_size */, + const void * /* arg_value */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetKernelInfo_fn)( + cl_kernel /* kernel */, + cl_kernel_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetKernelWorkGroupInfo_fn)( + cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_work_group_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Event Object APIs */ +typedef cl_int (CL_API_CALL * clWaitForEvents_fn)( + cl_uint /* num_events */, + const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetEventInfo_fn)( + cl_event /* event */, + cl_event_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_event (CL_API_CALL * clCreateUserEvent_fn)( + cl_context /* context */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_int (CL_API_CALL * clRetainEvent_fn)( + cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clReleaseEvent_fn)( + cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clSetUserEventStatus_fn)( + cl_event /* event */, + cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_int (CL_API_CALL * clSetEventCallback_fn)( + cl_event /* event */, + cl_int /* command_exec_callback_type */, + void (CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_1; + +/* Profiling APIs */ +typedef cl_int (CL_API_CALL * clGetEventProfilingInfo_fn)( + cl_event /* event */, + cl_profiling_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Flush and Finish APIs */ +typedef cl_int (CL_API_CALL * clFlush_fn)( + cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clFinish_fn)( + cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +/* Enqueued Commands APIs */ +typedef cl_int (CL_API_CALL * clEnqueueReadBuffer_fn)( + cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + size_t /* offset */, + size_t /* cb */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueReadBufferRect_fn)( + cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + const size_t * /* buffer_offset */, + const size_t * /* host_offset */, + const size_t * /* region */, + size_t /* buffer_row_pitch */, + size_t /* buffer_slice_pitch */, + size_t /* host_row_pitch */, + size_t /* host_slice_pitch */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_int (CL_API_CALL * clEnqueueWriteBuffer_fn)( + cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_write */, + size_t /* offset */, + size_t /* cb */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueWriteBufferRect_fn)( + cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + const size_t * /* buffer_offset */, + const size_t * /* host_offset */, + const size_t * /* region */, + size_t /* buffer_row_pitch */, + size_t /* buffer_slice_pitch */, + size_t /* host_row_pitch */, + size_t /* host_slice_pitch */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_int (CL_API_CALL * clEnqueueCopyBuffer_fn)( + cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_buffer */, + size_t /* src_offset */, + size_t /* dst_offset */, + size_t /* cb */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueCopyBufferRect_fn)( + cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_buffer */, + const size_t * /* src_origin */, + const size_t * /* dst_origin */, + const size_t * /* region */, + size_t /* src_row_pitch */, + size_t /* src_slice_pitch */, + size_t /* dst_row_pitch */, + size_t /* dst_slice_pitch */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_int (CL_API_CALL * clEnqueueReadImage_fn)( + cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_read */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t /* row_pitch */, + size_t /* slice_pitch */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueWriteImage_fn)( + cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_write */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t /* input_row_pitch */, + size_t /* input_slice_pitch */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueCopyImage_fn)( + cl_command_queue /* command_queue */, + cl_mem /* src_image */, + cl_mem /* dst_image */, + const size_t * /* src_origin[3] */, + const size_t * /* dst_origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueCopyImageToBuffer_fn)( + cl_command_queue /* command_queue */, + cl_mem /* src_image */, + cl_mem /* dst_buffer */, + const size_t * /* src_origin[3] */, + const size_t * /* region[3] */, + size_t /* dst_offset */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueCopyBufferToImage_fn)( + cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_image */, + size_t /* src_offset */, + const size_t * /* dst_origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef void * (CL_API_CALL * clEnqueueMapBuffer_fn)( + cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + size_t /* offset */, + size_t /* cb */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef void * (CL_API_CALL * clEnqueueMapImage_fn)( + cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t * /* image_row_pitch */, + size_t * /* image_slice_pitch */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueUnmapMemObject_fn)( + cl_command_queue /* command_queue */, + cl_mem /* memobj */, + void * /* mapped_ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueNDRangeKernel_fn)( + cl_command_queue /* command_queue */, + cl_kernel /* kernel */, + cl_uint /* work_dim */, + const size_t * /* global_work_offset */, + const size_t * /* global_work_size */, + const size_t * /* local_work_size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueTask_fn)( + cl_command_queue /* command_queue */, + cl_kernel /* kernel */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueNativeKernel_fn)( + cl_command_queue /* command_queue */, + void (CL_CALLBACK *user_func)(void *), + void * /* args */, + size_t /* cb_args */, + cl_uint /* num_mem_objects */, + const cl_mem * /* mem_list */, + const void ** /* args_mem_loc */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueMarker_fn)( + cl_command_queue /* command_queue */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueWaitForEvents_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_events */, + const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueBarrier_fn)( + cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +typedef void * (CL_API_CALL * clGetExtensionFunctionAddress_fn)( + const char * /* func_name */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_mem (CL_API_CALL * clCreateFromGLBuffer_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLuint /* bufobj */, + int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_mem (CL_API_CALL * clCreateFromGLTexture2D_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_mem (CL_API_CALL * clCreateFromGLTexture3D_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_mem (CL_API_CALL * clCreateFromGLRenderbuffer_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLuint /* renderbuffer */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetGLObjectInfo_fn)( + cl_mem /* memobj */, + cl_gl_object_type * /* gl_object_type */, + cl_GLuint * /* gl_object_name */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clGetGLTextureInfo_fn)( + cl_mem /* memobj */, + cl_gl_texture_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_event (CL_API_CALL * clCreateEventFromGLsyncKHR_fn) ( + cl_context /* context */, + cl_GLsync /* cl_GLsync */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +typedef cl_int (CL_API_CALL * clEnqueueAcquireGLObjects_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clEnqueueReleaseGLObjects_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef cl_int (CL_API_CALL * clCreateSubDevices_fn)( + cl_device_id /* in_device */, + const cl_device_partition_property * /* properties */, + cl_uint /* num_entries */, + cl_device_id * /* out_devices */, + cl_uint * /* num_devices */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clRetainDevice_fn)( + cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clReleaseDevice_fn)( + cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_mem (CL_API_CALL * clCreateImage_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format*/, + const cl_image_desc * /* image_desc*/, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_program (CL_API_CALL * clCreateProgramWithBuiltInKernels_fn)( + cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* kernel_names */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clCompileProgram_fn)( + cl_program /* program */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + cl_uint /* num_input_headers */, + const cl_program * /* input_headers */, + const char ** /* header_include_names */, + void (CL_CALLBACK *pfn_notify)(cl_program program, void *user_data), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_program (CL_API_CALL * clLinkProgram_fn)( + cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + cl_uint /* num_input_programs */, + const cl_program * /* input_programs */, + void (CL_CALLBACK *pfn_notify)(cl_program program, void *user_data), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clUnloadPlatformCompiler_fn)( + cl_platform_id /* platform */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clGetKernelArgInfo_fn)( + cl_kernel /* kernel */, + cl_uint /* arg_indx */, + cl_kernel_arg_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clEnqueueFillBuffer_fn)( + cl_command_queue /* command_queue */, + cl_mem /* buffer */, + const void * /* pattern */, + size_t /* pattern_size */, + size_t /* offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clEnqueueFillImage_fn)( + cl_command_queue /* command_queue */, + cl_mem /* image */, + const void * /* fill_color */, + const size_t * /* origin */, + const size_t * /* region */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clEnqueueMigrateMemObjects_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_mem_objects */, + const cl_mem * /* mem_objects */, + cl_mem_migration_flags /* flags */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clEnqueueMarkerWithWaitList_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_int (CL_API_CALL * clEnqueueBarrierWithWaitList_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +typedef void* (CL_API_CALL * clGetExtensionFunctionAddressForPlatform_fn)( + cl_platform_id /* platform */, + const char * /* funcname */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_mem (CL_API_CALL * clCreateFromGLTexture_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* texture_target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef cl_command_queue (CL_API_CALL * clCreateCommandQueueWithProperties_fn)( + cl_context /* context */, + cl_device_id /* device */, + const cl_queue_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_sampler (CL_API_CALL * clCreateSamplerWithProperties_fn)( + cl_context /* context */, + const cl_sampler_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef void * (CL_API_CALL * clSVMAlloc_fn)( + cl_context /* context */, + cl_svm_mem_flags /* flags */, + size_t /* size */, + cl_uint /* alignment */) CL_API_SUFFIX__VERSION_2_0; + +typedef void (CL_API_CALL * clSVMFree_fn)( + cl_context /* context */, + void * /* svm_pointer */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clSetKernelArgSVMPointer_fn)( + cl_kernel /* kernel */, + cl_uint /* arg_index */, + const void * /* arg_value */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clSetKernelExecInfo_fn)( + cl_kernel /* kernel */, + cl_kernel_exec_info /* param_name */, + size_t /* param_value_size */, + const void * /* param_value */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clEnqueueSVMFree_fn)( + cl_command_queue /* command_queue */, + cl_uint /* num_svm_pointers */, + void *[] /* svm_pointers */, + void (CL_CALLBACK * /* pfn_free_func */)(cl_command_queue /* queue */, + cl_uint /* num_svm_pointers */, + void *[] /* svm_pointers */, + void * /* user_data */), + void * /* user_data */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clEnqueueSVMMemcpy_fn)( + cl_command_queue /* command_queue */, + cl_bool /* blocking_copy */, + void * /* dst_ptr */, + const void * /* src_ptr */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clEnqueueSVMMemFill_fn)( + cl_command_queue /* command_queue */, + void * /* svm_ptr */, + const void * /* pattern */, + size_t /* pattern_size */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clEnqueueSVMMap_fn)( + cl_command_queue /* command_queue */, + cl_bool /* blocking_map */, + cl_map_flags /* flags */, + void * /* svm_ptr */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clEnqueueSVMUnmap_fn)( + cl_command_queue /* command_queue */, + void * /* svm_ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_mem (CL_API_CALL * clCreatePipe_fn)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_uint /* pipe_packet_size */, + cl_uint /* pipe_max_packets */, + const cl_pipe_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clGetPipeInfo_fn)( + cl_mem /* pipe */, + cl_pipe_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef cl_int (CL_API_CALL * clGetKernelSubGroupInfoKHR_fn)( + cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_sub_group_info /* param_name */, + size_t /* input_value_size */, + const void * /* input_value */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_2_0; + + +typedef struct _cl_icd_dispatch_table +{ + /* OpenCL 1.0 */ + clGetPlatformIDs_fn GetPlatformIDs; + clGetPlatformInfo_fn GetPlatformInfo; + clGetDeviceIDs_fn GetDeviceIDs; + clGetDeviceInfo_fn GetDeviceInfo; + clCreateContext_fn CreateContext; + clCreateContextFromType_fn CreateContextFromType; + clRetainContext_fn RetainContext; + clReleaseContext_fn ReleaseContext; + clGetContextInfo_fn GetContextInfo; + clCreateCommandQueue_fn CreateCommandQueue; + clRetainCommandQueue_fn RetainCommandQueue; + clReleaseCommandQueue_fn ReleaseCommandQueue; + clGetCommandQueueInfo_fn GetCommandQueueInfo; + clSetCommandQueueProperty_fn SetCommandQueueProperty; + clCreateBuffer_fn CreateBuffer; + clCreateImage2D_fn CreateImage2D; + clCreateImage3D_fn CreateImage3D; + clRetainMemObject_fn RetainMemObject; + clReleaseMemObject_fn ReleaseMemObject; + clGetSupportedImageFormats_fn GetSupportedImageFormats; + clGetMemObjectInfo_fn GetMemObjectInfo; + clGetImageInfo_fn GetImageInfo; + clCreateSampler_fn CreateSampler; + clRetainSampler_fn RetainSampler; + clReleaseSampler_fn ReleaseSampler; + clGetSamplerInfo_fn GetSamplerInfo; + clCreateProgramWithSource_fn CreateProgramWithSource; + clCreateProgramWithBinary_fn CreateProgramWithBinary; + clRetainProgram_fn RetainProgram; + clReleaseProgram_fn ReleaseProgram; + clBuildProgram_fn BuildProgram; + clUnloadCompiler_fn UnloadCompiler; + clGetProgramInfo_fn GetProgramInfo; + clGetProgramBuildInfo_fn GetProgramBuildInfo; + clCreateKernel_fn CreateKernel; + clCreateKernelsInProgram_fn CreateKernelsInProgram; + clRetainKernel_fn RetainKernel; + clReleaseKernel_fn ReleaseKernel; + clSetKernelArg_fn SetKernelArg; + clGetKernelInfo_fn GetKernelInfo; + clGetKernelWorkGroupInfo_fn GetKernelWorkGroupInfo; + clWaitForEvents_fn WaitForEvents; + clGetEventInfo_fn GetEventInfo; + clRetainEvent_fn RetainEvent; + clReleaseEvent_fn ReleaseEvent; + clGetEventProfilingInfo_fn GetEventProfilingInfo; + clFlush_fn Flush; + clFinish_fn Finish; + clEnqueueReadBuffer_fn EnqueueReadBuffer; + clEnqueueWriteBuffer_fn EnqueueWriteBuffer; + clEnqueueCopyBuffer_fn EnqueueCopyBuffer; + clEnqueueReadImage_fn EnqueueReadImage; + clEnqueueWriteImage_fn EnqueueWriteImage; + clEnqueueCopyImage_fn EnqueueCopyImage; + clEnqueueCopyImageToBuffer_fn EnqueueCopyImageToBuffer; + clEnqueueCopyBufferToImage_fn EnqueueCopyBufferToImage; + clEnqueueMapBuffer_fn EnqueueMapBuffer; + clEnqueueMapImage_fn EnqueueMapImage; + clEnqueueUnmapMemObject_fn EnqueueUnmapMemObject; + clEnqueueNDRangeKernel_fn EnqueueNDRangeKernel; + clEnqueueTask_fn EnqueueTask; + clEnqueueNativeKernel_fn EnqueueNativeKernel; + clEnqueueMarker_fn EnqueueMarker; + clEnqueueWaitForEvents_fn EnqueueWaitForEvents; + clEnqueueBarrier_fn EnqueueBarrier; + clGetExtensionFunctionAddress_fn GetExtensionFunctionAddress; + clCreateFromGLBuffer_fn CreateFromGLBuffer; + clCreateFromGLTexture2D_fn CreateFromGLTexture2D; + clCreateFromGLTexture3D_fn CreateFromGLTexture3D; + clCreateFromGLRenderbuffer_fn CreateFromGLRenderbuffer; + clGetGLObjectInfo_fn GetGLObjectInfo; + clGetGLTextureInfo_fn GetGLTextureInfo; + clEnqueueAcquireGLObjects_fn EnqueueAcquireGLObjects; + clEnqueueReleaseGLObjects_fn EnqueueReleaseGLObjects; + clGetGLContextInfoKHR_fn GetGLContextInfoKHR; + void* _reservedForD3D10KHR[6]; + /* OpenCL 1.1 */ + clSetEventCallback_fn SetEventCallback; + clCreateSubBuffer_fn CreateSubBuffer; + clSetMemObjectDestructorCallback_fn SetMemObjectDestructorCallback; + clCreateUserEvent_fn CreateUserEvent; + clSetUserEventStatus_fn SetUserEventStatus; + clEnqueueReadBufferRect_fn EnqueueReadBufferRect; + clEnqueueWriteBufferRect_fn EnqueueWriteBufferRect; + clEnqueueCopyBufferRect_fn EnqueueCopyBufferRect; + + void* _reservedForDeviceFissionEXT[3]; + clCreateEventFromGLsyncKHR_fn CreateEventFromGLsyncKHR; + + /* OpenCL 1.2 */ + clCreateSubDevices_fn CreateSubDevices; + clRetainDevice_fn RetainDevice; + clReleaseDevice_fn ReleaseDevice; + clCreateImage_fn CreateImage; + clCreateProgramWithBuiltInKernels_fn CreateProgramWithBuiltInKernels; + clCompileProgram_fn CompileProgram; + clLinkProgram_fn LinkProgram; + clUnloadPlatformCompiler_fn UnloadPlatformCompiler; + clGetKernelArgInfo_fn GetKernelArgInfo; + clEnqueueFillBuffer_fn EnqueueFillBuffer; + clEnqueueFillImage_fn EnqueueFillImage; + clEnqueueMigrateMemObjects_fn EnqueueMigrateMemObjects; + clEnqueueMarkerWithWaitList_fn EnqueueMarkerWithWaitList; + clEnqueueBarrierWithWaitList_fn EnqueueBarrierWithWaitList; + clGetExtensionFunctionAddressForPlatform_fn GetExtensionFunctionAddressForPlatform; + clCreateFromGLTexture_fn CreateFromGLTexture; + + /* cl_khr_d3d11_sharing, cl_khr_dx9_media_sharing */ + void* _reservedD3DExtensions[10]; + + /* cl_khr_egl_image, cl_khr_egl_event */ + void* _reservedEGLExtensions[4]; + + /* OpenCL 2.0 */ + clCreateCommandQueueWithProperties_fn CreateCommandQueueWithProperties; + clCreatePipe_fn CreatePipe; + clGetPipeInfo_fn GetPipeInfo; + clSVMAlloc_fn SVMAlloc; + clSVMFree_fn SVMFree; + clEnqueueSVMFree_fn EnqueueSVMFree; + clEnqueueSVMMemcpy_fn EnqueueSVMMemcpy; + clEnqueueSVMMemFill_fn EnqueueSVMMemFill; + clEnqueueSVMMap_fn EnqueueSVMMap; + clEnqueueSVMUnmap_fn EnqueueSVMUnmap; + clCreateSamplerWithProperties_fn CreateSamplerWithProperties; + clSetKernelArgSVMPointer_fn SetKernelArgSVMPointer; + clSetKernelExecInfo_fn SetKernelExecInfo; + + clGetKernelSubGroupInfoKHR_fn GetKernelSubGroupInfoKHR; + clTerminateContextKHR_fn TerminateContextKHR; +} cl_icd_dispatch_table; + +#ifdef __cplusplus +} +#endif /* __cplusplus */ + +#endif /* __OPENCL_CL_ICD_H */ diff --git a/opencl/api/opencl/amdocl/cl_kernel.h b/opencl/api/opencl/amdocl/cl_kernel.h new file mode 100644 index 0000000000..39a6bc6612 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_kernel.h @@ -0,0 +1,121 @@ +/* + * Copyright (c) 2012 Advanced Micro Devices, Inc. All rights reserved. + */ + +#ifndef CL_KERNEL_H_ +#define CL_KERNEL_H_ + +struct clk_builtins_t; + +// This must be a multiple of sizeof(cl_ulong16) +#define __CPU_SCRATCH_SIZE 128 + +#define CLK_PRIVATE_MEMORY_SIZE (8*1024) + +struct clk_thread_info_block_t +{ + // Warning! The size of this struct needs to be a multiple + // of 16 when compiling 64 bit + + struct clk_builtins_t const * builtins; + void* local_mem_base; + const void * table_base; + + uint work_dim; + size_t global_offset[4]; /*dim0,dim1,dim2,invalid(dim<0||dim>2)*/ + size_t global_size[4]; /*dim0,dim1,dim2,invalid(dim<0||dim>2)*/ + + size_t enqueued_local_size[4]; + size_t local_size[4]; /*dim0,dim1,dim2,invalid(dim<0||dim>2)*/ + size_t local_id[4]; /*dim0,dim1,dim2,invalid(dim<0||dim>2)*/ + size_t group_id[4]; /*dim0,dim1,dim2,invalid(dim<0||dim>2)*/ +}; + +typedef enum clk_value_type_t +{ + T_VOID, T_CHAR, T_SHORT, T_INT, + T_LONG, T_FLOAT, T_DOUBLE, T_POINTER, + T_CHAR2, T_CHAR3, T_CHAR4, T_CHAR8, T_CHAR16, + T_SHORT2, T_SHORT3, T_SHORT4, T_SHORT8, T_SHORT16, + T_INT2, T_INT3, T_INT4, T_INT8, T_INT16, + T_LONG2, T_LONG3, T_LONG4, T_LONG8, T_LONG16, + T_FLOAT2, T_FLOAT3, T_FLOAT4, T_FLOAT8, T_FLOAT16, + T_DOUBLE2, T_DOUBLE3, T_DOUBLE4, T_DOUBLE8, T_DOUBLE16, + T_SAMPLER, T_SEMA, T_STRUCT, T_QUEUE +} clk_value_type_t; + +typedef enum clk_address_space_t +{ + A_PRIVATE, A_LOCAL, A_CONSTANT, A_GLOBAL, A_REGION +} clk_address_space_t; + +//kernel arg access qualifier and type qualifier +typedef enum clk_arg_qualifier_t +{ + Q_NONE = 0, + + //for image type only, access qualifier + Q_READ = 1, + Q_WRITE = 2, + + //for pointer type only + Q_CONST = 4, // pointee + Q_RESTRICT = 8, + Q_VOLATILE = 16, // pointee + Q_PIPE = 32 // pipe + +} clk_arg_qualifier_t; + +#pragma pack(push, 4) +struct clk_parameter_descriptor_t +{ + clk_value_type_t type; + clk_address_space_t space; + uint qualifier; + const char* name; +}; +#pragma pack(pop) + +//#define CLK_LOCAL_MEM_FENCE (1 << 0) +//#define CLK_GLOBAL_MEM_FENCE (1 << 1) + +struct clk_builtins_t +{ + /* Synchronization functions */ + void (*barrier_ptr)(cl_mem_fence_flags flags); + + /* AMD Only builtins: FIXME_lmoriche (extension) */ + void* reserved; + int (*printf_ptr)(const char *format, ...); +}; + +enum clk_natures_t +{ + KN_HAS_BARRIER = 1 << 0, + KN_WG_LEVEL = 1 << 1 +}; + +#if defined(_MSC_VER) +#pragma warning( push ) +#pragma warning( disable : 4200 ) +#endif + +#if !defined(__OPENCL_VERSION__) || __OPENCL_VERSION__ >= 200 + +typedef struct clk_pipe_t +{ + size_t read_idx; + size_t write_idx; + size_t end_idx; + char padding[128 - 3*sizeof(size_t)]; + char packets[]; +} clk_pipe_t; + +#endif + +#if defined(_MSC_VER) +#pragma warning( pop ) +#endif + +#endif /*CL_KERNEL_H_*/ + diff --git a/opencl/api/opencl/amdocl/cl_kernel_info_amd.cpp b/opencl/api/opencl/amdocl/cl_kernel_info_amd.cpp new file mode 100644 index 0000000000..37ed252d12 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_kernel_info_amd.cpp @@ -0,0 +1,136 @@ +// +// Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. +// +#include "cl_common.hpp" +#include "cl_kernel_info_amd.h" +#include "platform/kernel.hpp" +#include "platform/ndrange.hpp" +#include "platform/command.hpp" + +/*! \addtogroup API + * @{ + * + * \addtogroup AMD_Extensions + * @{ + * + */ + +/*! \brief Retrieves the kernel information. + * + * \param kernel specifies the kernel object being queried. + * + * \param device identifies a specific device in the list of devices associated + * with \a kernel. The list of devices is the list of devices in the OpenCL + * context that is associated with \a kernel. If the list of devices associated + * with kernel is a single device, \a device can be a NULL value. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_KERNEL if \a kernel is a not a valid program object + */ +RUNTIME_ENTRY(cl_int, clGetKernelInfoAMD, ( + cl_kernel kernel, + cl_device_id device, + cl_kernel_info_amd param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret)) +{ + // Check if we have a valid device + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + + // Check if we have a valid performance counter + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + // Find the kernel, associated with the specified device + const device::Kernel* devKernel = + as_amd(kernel)->getDeviceKernel(*as_amd(device)); + + // Make sure we found a valid kernel + if (devKernel == NULL) { + return CL_INVALID_KERNEL; + } + + // Get the corresponded parameters + switch (param_name) { + case CL_KERNELINFO_SCRATCH_REGS: + return amd::clGetInfo( + devKernel->workGroupInfo()->scratchRegs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_WAVEFRONT_PER_SIMD: + return amd::clGetInfo( + devKernel->workGroupInfo()->wavefrontPerSIMD_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_WAVEFRONT_SIZE: + return amd::clGetInfo( + devKernel->workGroupInfo()->wavefrontSize_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_AVAILABLE_GPRS: + return amd::clGetInfo( + devKernel->workGroupInfo()->availableGPRs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_USED_GPRS: + return amd::clGetInfo( + devKernel->workGroupInfo()->usedGPRs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_AVAILABLE_SGPRS: + return amd::clGetInfo( + devKernel->workGroupInfo()->availableSGPRs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_USED_SGPRS: + return amd::clGetInfo( + devKernel->workGroupInfo()->usedSGPRs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_AVAILABLE_VGPRS: + return amd::clGetInfo( + devKernel->workGroupInfo()->availableVGPRs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_USED_VGPRS: + return amd::clGetInfo( + devKernel->workGroupInfo()->usedVGPRs_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_AVAILABLE_LDS_SIZE: + return amd::clGetInfo( + devKernel->workGroupInfo()->availableLDSSize_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_USED_LDS_SIZE: + return amd::clGetInfo( + devKernel->workGroupInfo()->usedLDSSize_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_AVAILABLE_STACK_SIZE: + return amd::clGetInfo( + devKernel->workGroupInfo()->availableStackSize_, + param_value_size, param_value, param_value_size_ret); + case CL_KERNELINFO_USED_STACK_SIZE: + return amd::clGetInfo( + devKernel->workGroupInfo()->usedStackSize_, + param_value_size, param_value, param_value_size_ret); + default: + return CL_INVALID_VALUE; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_kernel_info_amd.h b/opencl/api/opencl/amdocl/cl_kernel_info_amd.h new file mode 100644 index 0000000000..f24222044c --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_kernel_info_amd.h @@ -0,0 +1,165 @@ +/* ============================================================ + +Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance + +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +#ifndef __CL_KERNEL_INFO_AMD_H +#define __CL_KERNEL_INFO_AMD_H + +#include "CL/cl_platform.h" + +#ifdef __cplusplus +extern "C" { +#endif /*__cplusplus*/ + +typedef cl_uint cl_kernel_info_amd; + +/* cl_kernel_info */ +enum KernelInfoAMD +{ + CL_KERNELINFO_NONE = 0x0, + CL_KERNELINFO_SCRATCH_REGS, + CL_KERNELINFO_WAVEFRONT_PER_SIMD, + CL_KERNELINFO_WAVEFRONT_SIZE, + CL_KERNELINFO_AVAILABLE_GPRS, + CL_KERNELINFO_USED_GPRS, + CL_KERNELINFO_AVAILABLE_LDS_SIZE, + CL_KERNELINFO_USED_LDS_SIZE, + CL_KERNELINFO_AVAILABLE_STACK_SIZE, + CL_KERNELINFO_USED_STACK_SIZE, + CL_KERNELINFO_AVAILABLE_SGPRS, + CL_KERNELINFO_USED_SGPRS, + CL_KERNELINFO_AVAILABLE_VGPRS, + CL_KERNELINFO_USED_VGPRS, + CL_KERNELINFO_LAST +}; + +/*! \brief Retrieves the kernel information. + * + * \param kernel specifies the kernel object being queried. + * + * \param device identifies a specific device in the list of devices associated + * with \a kernel. The list of devices is the list of devices in the OpenCL + * context that is associated with \a kernel. If the list of devices associated + * with kernel is a single device, \a device can be a NULL value. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_KERNEL if \a kernel is a not a valid program object + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelInfoAMD( + cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_info_amd /* param_name */, + size_t /* param_value_size */, + void* /* param_value */, + size_t* /* param_value_size_ret */ + ) CL_API_SUFFIX__VERSION_1_0; + +#ifdef __cplusplus +} /*extern "C"*/ +#endif /*__cplusplus*/ + +#endif /*__CL_KERNEL_INFO_AMD_H*/ diff --git a/opencl/api/opencl/amdocl/cl_memobj.cpp b/opencl/api/opencl/amdocl/cl_memobj.cpp new file mode 100644 index 0000000000..454238e816 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_memobj.cpp @@ -0,0 +1,4936 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/context.hpp" +#include "platform/command.hpp" +#include "platform/memory.hpp" +#include + +#ifdef _WIN32 +#include +#include "cl_d3d9_amd.hpp" +#include "cl_d3d10_amd.hpp" +#include "cl_d3d11_amd.hpp" +#endif //_WIN32 + +#include + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_MemObjs + * + * Memory objects are categorized into two types: buffer objects, and image + * objects. A buffer object stores a one-dimensional collection of elements + * whereas an image object is used to store a two- or three- dimensional + * texture, frame-buffer or image. + * + * Elements of a buffer object can be a scalar data type (such as an int, + * float), vector data type, or a user-defined structure. An image object is + * used to represent a buffer that can be used as a texture or a frame-buffer. + * The elements of an image object are selected from a list of predefined + * image formats. The minimum number of elements in a memory object is one. + * + * @{ + * + * \addtogroup CL_CreatingBuffer + * + * @{ + */ + +/*! \brief Helper function to validate cl_mem_flags + * + * \return true of flags are valid, otherwise - false +*/ +static bool +validateFlags( cl_mem_flags flags) +{ + // check flags for validity + cl_bitfield temp = flags + & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY | CL_MEM_READ_ONLY); + + if(temp + && !(CL_MEM_READ_WRITE == temp + || CL_MEM_WRITE_ONLY == temp + || CL_MEM_READ_ONLY == temp)) { + return false; + } + + if((flags & (CL_MEM_USE_HOST_PTR | CL_MEM_ALLOC_HOST_PTR)) + == (CL_MEM_USE_HOST_PTR | CL_MEM_ALLOC_HOST_PTR)) { + return false; + } + if((flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) + == (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) { + return false; + } + + if ((flags & CL_MEM_EXTERNAL_PHYSICAL_AMD) && (flags & (CL_MEM_USE_HOST_PTR + | CL_MEM_COPY_HOST_PTR | CL_MEM_ALLOC_HOST_PTR | CL_MEM_READ_WRITE + | CL_MEM_READ_ONLY))) { + return false; + } + + if ((flags & CL_MEM_BUS_ADDRESSABLE_AMD) && (flags & (CL_MEM_USE_HOST_PTR + | CL_MEM_ALLOC_HOST_PTR))) { + return false; + } + + return true; +} + +/*! \brief Helper function to validate cl_image_desc + * + * \return true of cl_image_desc parameters are valid, otherwise - false + * + * image_type describes the image type and must be either CL_MEM_OBJECT_IMAGE1D, + * CL_MEM_OBJECT_IMAGE1D_BUFFER, CL_MEM_OBJECT_IMAGE1D_ARRAY, + * CL_MEM_OBJECT_IMAGE2D, CL_MEM_OBJECT_IMAGE2D_ARRAY or CL_MEM_OBJECT_IMAGE3D. + * + * image_width is the width of the image in pixels. For a 2D image and + * image array, the image width must be <= CL_DEVICE_IMAGE2D_MAX_WIDTH. + * For a 3D image, the image width must be <= CL_DEVICE_IMAGE3D_MAX_WIDTH. + * For a 1D image buffer, the image width must be <= CL_DEVICE_IMAGE_MAX_BUFFER_SIZE. + * For a 1D image and 1D image array, the image width must be + * <= CL_DEVICE_IMAGE2D_MAX_WIDTH. + * + * image_height is height of the image in pixels. This is only used if + * the image is a 2D, 3D or 2D image array. For a 2D image or image array, + * the image height must be <= CL_DEVICE_IMAGE2D_MAX_HEIGHT. For a 3D image, + * the image height must be <= CL_DEVICE_IMAGE3D_MAX_HEIGHT. + * + * image_depth is the depth of the image in pixels. This is only used if + * the image is a 3D image and must be a value > 1 and + * <= CL_DEVICE_IMAGE3D_MAX_DEPTH. + * + * image_array_size is the number of images in the image array. This is only + * used if the image is a 1D or 2D image array. The values for + * image_array_size, if specified, must be between 1 and + * CL_DEVICE_IMAGE_MAX_ARRAY_SIZE. + * + * image_row_pitch is the scan-line pitch in bytes. This must be 0 if + * host_ptr is NULL and can be either 0 or >= image_width * size of element in + * bytes if host_ptr is not NULL. If host_ptr is not NULL and image_row_pitch = 0, + * image_row_pitch is calculated as image_width * size of element in bytes. + * If image_row_pitch is not 0, it must be a multiple of the image element + * size in bytes. + * + * image_slice_pitch is the size in bytes of each 2D slice in the 3D image or + * the size in bytes of each image in a 1D or 2D image array. This must be 0 + * if host_ptr is NULL. If host_ptr is not NULL, image_slice_pitch can be either + * 0 or >= image_row_pitch * image_height for a 2D image array or 3D image and + * can be either 0 or >= image_row_pitch for a 1D image array. If host_ptr is + * not NULL and image_slice_pitch = 0, image_slice_pitch is calculated as + * image_row_pitch * image_height for a 2D image array or 3D image and + * image_row_pitch for a 1D image array. If image_slice_pitch is not 0, it must + * be a multiple of the image_row_pitch. + * + * num_mip_levels and num_samples must be 0. + * + * buffer refers to a valid buffer memory object if image_type is + * CL_MEM_OBJECT_IMAGE1D_BUFFER. Otherwise it must be NULL. For a 1D image + * buffer object, the image pixels are taken from the buffer object’s + * data store. When the contents of a buffer object’s data store are modified, + * those changes are reflected in the contents of the 1D image buffer object + * and vice-versa at corresponding sychronization points. The image_width + * size of element in bytes must be <= size of buffer object data store. + */ +static bool +validateImageDescriptor( + const std::vector& devices, + const amd::Image::Format imageFormat, + const cl_image_desc* desc, + void* hostPtr, + size_t& imageRowPitch, + size_t& imageSlicePitch) +{ + if (desc == NULL) { + return false; + } + + if ((desc->num_mip_levels != 0) || (desc->num_samples != 0)) { + return false; + } + + amd::Buffer* buffer = NULL; + size_t elemSize = imageFormat.getElementSize(); + bool imageBuffer = false; + + if (desc->image_type == CL_MEM_OBJECT_IMAGE1D_BUFFER || + (desc->mem_object != NULL && desc->image_type == CL_MEM_OBJECT_IMAGE2D)) { + if (desc->mem_object == NULL) { + return false; + } + buffer = as_amd(desc->mem_object)->asBuffer(); + if (buffer == NULL) { + return false; + } + if ((desc->image_width * desc->image_height * elemSize) > buffer->getSize()) { + return false; + } + imageBuffer = true; + } + else if (desc->mem_object != NULL) { + return false; + } + + imageRowPitch = desc->image_row_pitch; + imageSlicePitch = desc->image_slice_pitch; + + switch (desc->image_type) { + case CL_MEM_OBJECT_IMAGE3D: + case CL_MEM_OBJECT_IMAGE2D_ARRAY: + case CL_MEM_OBJECT_IMAGE1D_ARRAY: + // check slice pitch + if (hostPtr == NULL) { + if (imageSlicePitch != 0) { + return false; + } + } + // Fall through to process pitch... + case CL_MEM_OBJECT_IMAGE2D: + case CL_MEM_OBJECT_IMAGE1D: + // check row pitch rules + if (hostPtr == NULL && !imageBuffer) { + if (imageRowPitch != 0) { + return false; + } + } + else if (imageRowPitch != 0) { + if ((imageRowPitch < desc->image_width * elemSize) || + ((imageRowPitch % elemSize) != 0)) { + return false; + } + } + if (imageRowPitch == 0) { + imageRowPitch = desc->image_width * elemSize; + } + break; + case CL_MEM_OBJECT_IMAGE1D_BUFFER: + break; + default: + return false; + break; + } + + // Extra slice validation for three dimensional images + if ((desc->image_type == CL_MEM_OBJECT_IMAGE3D) || + (desc->image_type == CL_MEM_OBJECT_IMAGE2D_ARRAY)) { + if (imageSlicePitch != 0) { + if ((imageSlicePitch < (imageRowPitch * desc->image_height)) || + ((imageSlicePitch % imageRowPitch) != 0)) { + return false; + } + } + if (imageSlicePitch == 0) { + imageSlicePitch = imageRowPitch * desc->image_height; + } + } + else if (desc->image_type == CL_MEM_OBJECT_IMAGE1D_ARRAY) { + if (imageSlicePitch != 0) { + if ((imageSlicePitch % imageRowPitch) != 0) { + return false; + } + } + if (imageSlicePitch == 0) { + imageSlicePitch = imageRowPitch; + } + } + + return true; +} + +/*! \brief Create a buffer object. + * + * \param context is a valid OpenCL context used to create the buffer object. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information such as the memory arena that should be used to allocate the + * buffer object and how it will be used. + * + * \param size is the size in bytes of the buffer memory object to be + * allocated. + * + * \param host_ptr is a pointer to the buffer data that may already be + * allocated by the application. The size of the buffer that host_ptr points + * to must be >= \a size bytes. Passing in a pointer to an already allocated + * buffer on the host and using it as a buffer object allows applications to + * share data efficiently with kernels and the host. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero buffer object and \a errcode_ret is set to + * CL_SUCCESS if the buffer object is created successfully or a NULL value + * with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if values specified in \a flags are not valid. + * - CL_INVALID_BUFFER_SIZE if \a size is 0 or is greater than + * CL_DEVICE_MAX_MEM_ALLOC_SIZE value. + * - CL_INVALID_HOST_PTR if host_ptr is NULL and CL_MEM_USE_HOST_PTR or + * CL_MEM_COPY_HOST_PTR are set in \a flags or if \a host_ptr is not NULL but + * CL_MEM_COPY_HOST_PTR or CL_MEM_USE_HOST_PTR are not set in \a flags. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for buffer object. + * - CL_INVALID_OPERATION if the buffer object cannot be created for all + * devices in \a context. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateBuffer, ( + cl_context context, + cl_mem_flags flags, + size_t size, + void *host_ptr, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return NULL; + } + // check flags for validity + if (!validateFlags(flags)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return (cl_mem) 0; + } + // check size + if (size == 0) { + *not_null(errcode_ret) = CL_INVALID_BUFFER_SIZE; + LogWarning("invalid parameter \"size = 0\""); + return (cl_mem)0; + } + const std::vector& devices = as_amd(context)->devices(); + bool sizePass = false; + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + if (((*it)->info().maxMemAllocSize_ >= size) || + ((flags & (CL_MEM_USE_HOST_PTR | CL_MEM_ALLOC_HOST_PTR)) && + (*it)->settings().largeHostMemAlloc_)) { + sizePass = true; + break; + } + } + if (!sizePass) { + *not_null(errcode_ret) = CL_INVALID_BUFFER_SIZE; + LogWarning("invalid parameter \"size\""); + return (cl_mem) 0; + } + + // check host_ptr consistency + if (host_ptr == NULL) { + if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR + | CL_MEM_EXTERNAL_PHYSICAL_AMD)) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"host_ptr\""); + return (cl_mem) 0; + } + } + else { + if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR + | CL_MEM_EXTERNAL_PHYSICAL_AMD))) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"host_ptr\""); + return (cl_mem) 0; + } + + if (flags & CL_MEM_EXTERNAL_PHYSICAL_AMD) { + + flags |= CL_MEM_WRITE_ONLY; + + cl_bus_address_amd * bus_address = + reinterpret_cast(host_ptr); + + if (bus_address->surface_bus_address == 0) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"surface bus address\""); + return static_cast(NULL); + } + + if (bus_address->surface_bus_address & (amd::Os::pageSize()-1)) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"surface bus address\""); + return static_cast(NULL); + } + + if (bus_address->marker_bus_address == 0) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"marker bus address\""); + return static_cast(NULL); + } + + if (bus_address->marker_bus_address & (amd::Os::pageSize()-1)) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"marker bus address\""); + return static_cast(NULL); + } + + } + } + + // check extensions flag consistency + if ((flags & CL_MEM_USE_PERSISTENT_MEM_AMD) && + (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_ALLOC_HOST_PTR | + CL_MEM_EXTERNAL_PHYSICAL_AMD | CL_MEM_BUS_ADDRESSABLE_AMD))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("conflicting flags CL_MEM_USE_PERSISTENT_MEM_AMD and host memory specific flags"); + return (cl_mem) 0; + } + + if ((flags & CL_MEM_EXTERNAL_PHYSICAL_AMD) || + (flags & CL_MEM_BUS_ADDRESSABLE_AMD)) { + size = (size+(amd::Os::pageSize()-1))&(~(amd::Os::pageSize()-1)); + } + + amd::Context& amdContext = *as_amd(context); + amd::Memory* mem = NULL; + //check if the ptr is in the svm space, if yes, we need return SVM buffer + amd::Memory * svmMem = amd::SvmManager::FindSvmBuffer(host_ptr); + if ((NULL != svmMem) && (flags & CL_MEM_USE_HOST_PTR)) { + size_t svmSize = svmMem->getSize(); + size_t offset = static_cast
(host_ptr) - static_cast
(svmMem->getSvmPtr()); + if (size + offset > svmSize) { + LogWarning("invalid parameter \"size\""); + return (cl_mem) 0; + } + mem = new(amdContext) amd::Buffer(*svmMem, flags, offset, size); + host_ptr = NULL; + } + else { + mem = new(amdContext) amd::Buffer(amdContext, flags, size); + } + + if (mem == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem)0; + } + + if (!mem->create(host_ptr)) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + mem->release(); + return NULL; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(mem); +} +RUNTIME_EXIT + +RUNTIME_ENTRY_RET(cl_mem, clCreateSubBuffer, ( + cl_mem mem, + cl_mem_flags flags, + cl_buffer_create_type buffer_create_type, + const void *buffer_create_info, + cl_int *errcode_ret)) +{ + if (!is_valid(mem) || as_amd(mem)->asBuffer() == NULL) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return NULL; + } + amd::Buffer& buffer = *as_amd(mem)->asBuffer(); + + // check flags for validity + if (!validateFlags(flags) || + (buffer_create_type != CL_BUFFER_CREATE_TYPE_REGION)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return NULL; + } + + if (buffer.getMemFlags() & + (CL_MEM_EXTERNAL_PHYSICAL_AMD|CL_MEM_BUS_ADDRESSABLE_AMD)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return NULL; + } + + const cl_buffer_region* region = + (const cl_buffer_region*) buffer_create_info; + + // Check sub buffer offset alignment + bool alignmentPass = false; + const std::vector& devices = + buffer.getContext().devices(); + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + cl_uint deviceAlignmentBytes = (*it)->info().memBaseAddrAlign_>>3; + if (region->origin == + amd::alignDown(region->origin, deviceAlignmentBytes)) { + alignmentPass = true; + } + } + + // Return an error if the offset is misaligned on all devices + if (!alignmentPass) { + *not_null(errcode_ret) = CL_MISALIGNED_SUB_BUFFER_OFFSET; + return NULL; + } + + // check size + if ((region->size == 0) || + (region->origin + region->size) > buffer.getSize()) { + *not_null(errcode_ret) = CL_INVALID_BUFFER_SIZE; + return NULL; + } + + amd::Memory* mem = new(buffer.getContext()) amd::Buffer( + buffer, (flags) ? flags : buffer.getMemFlags(), + region->origin, region->size); + if (mem == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return NULL; + } + + if (!mem->create(NULL)) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + mem->release(); + return NULL; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(mem); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_ReadWriteBuffer + * @{ + */ + +/*! \brief Enqueue a command to read from a buffer object to host memory. + * + * \param command_queue refers to the command-queue in which the read / write + * command will be queued. \a command_queue and \a buffer must be created with + * the same OpenCL context. + * + * \param buffer refers to a valid buffer object. + * + * \param blocking_read indicates if the read operation is blocking or + * nonblocking. If \a blocking_read is CL_TRUE i.e. the read command is + * blocking, clEnqueueReadBuffer does not return until the buffer data has been + * read and copied into memory pointed to by ptr. + * If \a blocking_read is CL_FALSE i.e. the read command is non-blocking, + * clEnqueueReadBuffer queues a non-blocking read command and returns. The + * contents of the buffer that ptr points to cannot be used until the read + * command has completed. The \a event argument returns an event object which + * can be used to query the execution status of the read command. When the read + * command has completed, the contents of the buffer that ptr points to can be + * used by the application. + * + * \param offset is the offset in bytes in the buffer object to read from or + * write to. + * + * \param cb is the size in bytes of data being read or written. + * + * \param ptr is the pointer to buffer in host memory where data is to be read + * into or to be written from. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, + * then this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular read + * command and can be used to query or queue a wait for this particular command + * to complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue a wait for this + * command to complete. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue and + * \a buffer are not the same. + * - CL_INVALID_MEM_OBJECT if \a buffer is not a valid buffer object. + * - CL_INVALID_VALUE if the region being read or written specified by (offset, + * cb) is out of bounds or if \a ptr is a NULL value. + * - CL_INVALID_OPERATION if \a clEnqueueReadBuffer is called on buffer which + * has been created with CL_MEM_HOST_WRITE_ONLY or CL_MEM_HOST_NO_ACCESS. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and \a + * num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueReadBuffer, ( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + size_t offset, + size_t cb, + void *ptr, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(buffer)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* srcBuffer = as_amd(buffer)->asBuffer(); + if (srcBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (srcBuffer->getMemFlags() & + (CL_MEM_HOST_WRITE_ONLY | CL_MEM_HOST_NO_ACCESS)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + if (ptr == NULL) { + return CL_INVALID_VALUE; + } + + amd::Coord3D srcOffset(offset, 0, 0); + amd::Coord3D srcSize(cb, 1, 1); + + if(!srcBuffer->validateRegion(srcOffset, srcSize)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::ReadMemoryCommand *command = new amd::ReadMemoryCommand( + hostQueue, + CL_COMMAND_READ_BUFFER, + eventWaitList, + *srcBuffer, + srcOffset, srcSize, ptr); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + if (blocking_read) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to write to a buffer object from host memory. + * + * \param command_queue refers to the command-queue in which the read / write + * command will be queued. \a command_queue and \a buffer must be created with + * the same OpenCL context. + * + * \param buffer refers to a valid buffer object. + * + * \param blocking_write indicates if the write operation is blocking or + * non-blocking. If \a blocking_write is CL_TRUE, the OpenCL implementation + * copies the data referred to by \a ptr and enqueues the write operation in + * the command-queue. The memory pointed to by \a ptr can be reused by the + * application after the clEnqueueWriteBuffer call returns. If + * \a blocking_write is CL_FALSE, the OpenCL implementation will use \a ptr to + * perform a nonblocking write. As the write is non-blocking the implementation + * can return immediately. The memory pointed to by \a ptr cannot be reused by + * the application after the call returns. The \a event argument returns an + * event object which can be used to query the execution status of the write + * command. When the write command has completed, the memory pointed to by + * \a ptr can then be reused by the application + * + * \param offset is the offset in bytes in the buffer object to read from or + * write to. + * + * \param cb is the size in bytes of data being read or written. + * + * \param ptr is the pointer to buffer in host memory where data is to be read + * into or to be written from. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, + * then this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular write + * command and can be used to query or queue a wait for this particular command + * to complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue a wait for this + * command to complete. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue and + * \a buffer are not the same. + * - CL_INVALID_MEM_OBJECT if \a buffer is not a valid buffer object. + * - CL_INVALID_VALUE if the region being read or written specified by (offset, + * cb) is out of bounds or if \a ptr is a NULL value. + * - CL_INVALID_OPERATION if \a clEnqueueWriteBuffer is called on buffer which + * has been created with CL_MEM_HOST_READ_ONLY or CL_MEM_HOST_NO_ACCESS. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and \a + * num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueWriteBuffer, ( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_write, + size_t offset, + size_t cb, + const void *ptr, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(buffer)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* dstBuffer = as_amd(buffer)->asBuffer(); + if (dstBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (dstBuffer->getMemFlags() & + (CL_MEM_HOST_READ_ONLY | CL_MEM_HOST_NO_ACCESS)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if(hostQueue.context() != dstBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + if (ptr == NULL) { + return CL_INVALID_VALUE; + } + + amd::Coord3D dstOffset(offset, 0, 0); + amd::Coord3D dstSize(cb, 1, 1); + + if(!dstBuffer->validateRegion(dstOffset, dstSize)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::WriteMemoryCommand *command = new amd::WriteMemoryCommand( + hostQueue, + CL_COMMAND_WRITE_BUFFER, + eventWaitList, + *dstBuffer, + dstOffset, dstSize, ptr); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + if (blocking_write) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueues a command to copy a buffer object to another + * + * \param command_queue refers to the command-queue in which the copy command + * will be queued. The OpenCL context associated with \a command_queue, + * \a src_buffer and \a dst_buffer must be the same. + * + * \param src_buffer is the source buffer object. + * + * \param dst_buffer is the destination buffer object. + * + * \param src_offset refers to the offset where to begin reading data in + * \a src_buffer. + * + * \param dst_offset refers to the offset where to begin copying data in + * \a dst_buffer. + * + * \param cb refers to the size in bytes to copy. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, + * then this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular copy + * command and can be used to query or queue a wait for this particular command + * to complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue and wait for + * this command to complete. clEnqueueBarrier can be used instead. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue, + * \a src_buffer and \a dst_buffer are not the same. + * - CL_INVALID_MEM_OBJECT if \a src_buffer and \a dst_buffer are not valid + * buffer objects. + * - CL_INVALID_VALUE if \a src_offset, \a dst_offset, \a cb, \a src_offset + + * \a cb or \a dst_offset + \a cb require accessing elements outside the + * buffer memory objects. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueCopyBuffer, ( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_buffer, + size_t src_offset, + size_t dst_offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(src_buffer) || !is_valid(dst_buffer)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* srcBuffer = as_amd(src_buffer)->asBuffer(); + amd::Buffer* dstBuffer = as_amd(dst_buffer)->asBuffer(); + if (srcBuffer == NULL || dstBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcBuffer->getContext() + || hostQueue.context() != dstBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Coord3D srcOffset(src_offset, 0, 0); + amd::Coord3D dstOffset(dst_offset, 0, 0); + amd::Coord3D size(cb, 1, 1); + + if(!srcBuffer->validateRegion(srcOffset, size) || + !dstBuffer->validateRegion(dstOffset, size)) { + return CL_INVALID_VALUE; + } + + if(srcBuffer == dstBuffer + && ((src_offset <= dst_offset && dst_offset < src_offset + cb) + || (dst_offset <= src_offset && src_offset < dst_offset + cb))) { + return CL_MEM_COPY_OVERLAP; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::CopyMemoryCommand *command = new amd::CopyMemoryCommand( + hostQueue, + CL_COMMAND_COPY_BUFFER, + eventWaitList, + *srcBuffer, *dstBuffer, + srcOffset, + dstOffset, + size); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief clEnqueueReadBufferRect enqueues commands to read a 2D or 3D rectangular + * region from a buffer object to host memory. + * + * \param command_queue refers to the command-queue in which the read / write + * command will be queued. command_queue and buffer must be created with the same + * OpenCL context. buffer refers to a valid buffer object. + * + * \param blocking_read indicates if the read operations are blocking or + * nonblocking. + * If \a blocking_read is CL_TRUE i.e. the read command is blocking, + * clEnqueueReadBufferRect does not return until the buffer data has been read + * and copied into memory pointed to by ptr. + * If blocking_read is CL_FALSE i.e. the read command is non-blocking, + * clEnqueueReadBufferRect queues a non-blocking read command and returns. + * The contents of the buffer that ptr points to cannot be used until + * the read command has completed. The event argument returns an event object + * which can be used to query the execution status of the read command. + * When the read command has completed, the contents of the buffer that + * ptr points to can be used by the application. + * + * \buffer_origin defines the (x, y, z) offset in the memory region associated + * with buffer. For a 2D rectangle region, the z value given by buffer_origin[2] + * should be 0. The offset in bytes is computed as + * buffer_origin[2] * buffer_slice_pitch + buffer_origin[1] * buffer_row_pitch + + * buffer_origin[0]. + * + * \host_origin defines the (x, y, z) offset in the memory region pointed to + * by ptr. For a 2D rectangle region, the z value given by host_origin[2] + * should be 0. The offset in bytes is computed as + * host_origin[2] * host_slice_pitch + host_origin[1] * host_row_pitch + + * host_origin[0]. + * + * \param region defines the (width, height, depth) in bytes of the 2D or 3D + * rectangle being read or written. + * For a 2D rectangle copy, the depth value given by region[2] should be 1. + * + * \param buffer_row_pitch is the length of each row in bytes to be used for + * the memory region associated with buffer. If \a buffer_row_pitch is 0, + * \a buffer_row_pitch is computed as region[0]. + * + * \param buffer_slice_pitch is the length of each 2D slice in bytes to be used + * for the memory region associated with buffer. If \a buffer_slice_pitch is 0, + * \a buffer_slice_pitch is computed as region[1] * \a buffer_row_pitch. + * + * \param host_row_pitch is the length of each row in bytes to be used for + * the memory region pointed to by ptr. If \a host_row_pitch is 0, \a host_row_pitch + * is computed as region[0]. + * + * \param host_slice_pitch is the length of each 2D slice in bytes to be used + * for the memory region pointed to by ptr. If \a host_slice_pitch is 0, + * \a host_slice_pitch is computed as region[1] * \a host_row_pitch. + * ptr is the pointer to buffer in host memory where data is to be read into + * or to be written from. + * + * \param event_wait_list and \a num_events_in_wait_list specify events that + * need to complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait on any + * event to complete. If \a event_wait_list is NULL, \a num_events_in_wait_list + * must be 0. If \a event_wait_list is not NULL, the list of events pointed to + * by \a event_wait_list must be valid and \a num_events_in_wait_list + * must be greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. + * + * \param event returns an event object that identifies this particular + * read / write command and can be used to query or queue a wait for this + * particular command to complete. event can be NULL in which case it will not + * be possible for the application to query the status of this command or queue a + * wait for this command to complete. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise, + * it returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with command_queue and + * buffer are not the same or if the context associated with \a command_queue + * and events in event_wait_list are not the same. + * - CL_INVALID_MEM_OBJECT if buffer is not a valid buffer object. + * - CL_INVALID_VALUE if the region being read or written specified by + * (buffer_origin, region) is out of bounds. + * - CL_INVALID_VALUE if ptr is a NULL value. + * - CL_INVALID_OPERATION if \a clEnqueueReadBufferRect is called on buffer which + * has been created with CL_MEM_HOST_WRITE_ONLY or CL_MEM_HOST_NO_ACCESS. + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_MISALIGNED_SUB_BUFFER_OFFSET if buffer is a sub-buffer object and offset + * specified when the sub-buffer object is created is not aligned to + * - CL_DEVICE_MEM_BASE_ADDR_ALIGN value for device associated with queue. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for data store associated with buffer. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueReadBufferRect, ( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + const size_t buffer_origin[3], + const size_t host_origin[3], + const size_t region[3], + size_t buffer_row_pitch, + size_t buffer_slice_pitch, + size_t host_row_pitch, + size_t host_slice_pitch, + void* ptr, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + // Validate command queue + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + // Validate opencl buffer + if (!is_valid(buffer)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* srcBuffer = as_amd(buffer)->asBuffer(); + if (srcBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (srcBuffer->getMemFlags() & + (CL_MEM_HOST_WRITE_ONLY | CL_MEM_HOST_NO_ACCESS)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + // Make sure we have a valid system memory pointer + if (ptr == NULL) { + return CL_INVALID_VALUE; + } + + // Create buffer rectangle info structure + amd::BufferRect bufRect; + amd::BufferRect hostRect; + + if (!bufRect.create(buffer_origin, region, buffer_row_pitch, buffer_slice_pitch) || + !hostRect.create(host_origin, region, host_row_pitch, host_slice_pitch)) { + return CL_INVALID_VALUE; + } + + amd::Coord3D srcStart(bufRect.start_, 0, 0); + amd::Coord3D srcEnd(bufRect.end_, 1, 1); + + if (!srcBuffer->validateRegion(srcStart, srcEnd)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::Coord3D size(region[0], region[1], region[2]); + amd::ReadMemoryCommand *command = new amd::ReadMemoryCommand( + hostQueue, + CL_COMMAND_READ_BUFFER_RECT, + eventWaitList, + *srcBuffer, + srcStart, size, ptr, + bufRect, + hostRect); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + if (blocking_read) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief clEnqueueWriteBufferRect enqueues commands to write a 2D or 3D + * rectangular region to a buffer object from host memory. + * + * \param command_queue refers to the command-queue in which the read / write + * command will be queued. command_queue and buffer must be created with the same + * OpenCL context. buffer refers to a valid buffer object. + * + * \param blocking_write indicates if the write operations are blocking or + * nonblocking. + * If \a blocking_write is CL_TRUE, the OpenCL implementation copies the data + * referred to by ptr and enqueues the write operation in the command-queue. + * The memory pointed to by ptr can be reused by the application after + * the clEnqueueWriteBufferRect call returns. + * If \a blocking_write is CL_FALSE, the OpenCL implementation will use ptr to + * perform a nonblocking write. As the write is non-blocking the implementation + * can return immediately. The memory pointed to by ptr cannot be reused by + * the application after the call returns. The event argument returns + * an event object which can be used to query the execution status of the write + * command. When the write command has completed, the memory pointed to by ptr + * can then be reused by the application. + * + * \buffer_origin defines the (x, y, z) offset in the memory region associated + * with buffer. For a 2D rectangle region, the z value given by buffer_origin[2] + * should be 0. The offset in bytes is computed as + * buffer_origin[2] * buffer_slice_pitch + buffer_origin[1] * buffer_row_pitch + + * buffer_origin[0]. + * + * \host_origin defines the (x, y, z) offset in the memory region pointed to + * by ptr. For a 2D rectangle region, the z value given by host_origin[2] + * should be 0. The offset in bytes is computed as + * host_origin[2] * host_slice_pitch + host_origin[1] * host_row_pitch + + * host_origin[0]. + * + * \param region defines the (width, height, depth) in bytes of the 2D or 3D + * rectangle being read or written. + * For a 2D rectangle copy, the depth value given by region[2] should be 1. + * + * \param buffer_row_pitch is the length of each row in bytes to be used for + * the memory region associated with buffer. If \a buffer_row_pitch is 0, + * \a buffer_row_pitch is computed as region[0]. + * + * \param buffer_slice_pitch is the length of each 2D slice in bytes to be used + * for the memory region associated with buffer. If \a buffer_slice_pitch is 0, + * \a buffer_slice_pitch is computed as region[1] * \a buffer_row_pitch. + * + * \param host_row_pitch is the length of each row in bytes to be used for + * the memory region pointed to by ptr. If \a host_row_pitch is 0, \a host_row_pitch + * is computed as region[0]. + * + * \param host_slice_pitch is the length of each 2D slice in bytes to be used + * for the memory region pointed to by ptr. If \a host_slice_pitch is 0, + * \a host_slice_pitch is computed as region[1] * \a host_row_pitch. + * ptr is the pointer to buffer in host memory where data is to be read into + * or to be written from. + * + * \param event_wait_list and \a num_events_in_wait_list specify events that + * need to complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait on any + * event to complete. If \a event_wait_list is NULL, \a num_events_in_wait_list + * must be 0. If \a event_wait_list is not NULL, the list of events pointed to + * by \a event_wait_list must be valid and \a num_events_in_wait_list + * must be greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. + * + * \param event returns an event object that identifies this particular + * read / write command and can be used to query or queue a wait for this + * particular command to complete. event can be NULL in which case it will not + * be possible for the application to query the status of this command or queue a + * wait for this command to complete. + * + * clEnqueueReadBufferRect and clEnqueueWriteBufferRect + * \return CL_SUCCESS if the function is executed successfully. Otherwise, + * it returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with command_queue and + * buffer are not the same or if the context associated with \a command_queue + * and events in event_wait_list are not the same. + * - CL_INVALID_MEM_OBJECT if buffer is not a valid buffer object. + * - CL_INVALID_VALUE if the region being read or written specified by + * (buffer_origin, region) is out of bounds. + * - CL_INVALID_VALUE if ptr is a NULL value. + * - CL_INVALID_OPERATION if \a clEnqueueWriteBufferRect is called on buffer + * which has been created with CL_MEM_HOST_READ_ONLY or CL_MEM_HOST_NO_ACCESS. + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_MISALIGNED_SUB_BUFFER_OFFSET if buffer is a sub-buffer object and offset + * specified when the sub-buffer object is created is not aligned to + * - CL_DEVICE_MEM_BASE_ADDR_ALIGN value for device associated with queue. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for data store associated with buffer. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + */ +RUNTIME_ENTRY(cl_int, clEnqueueWriteBufferRect, ( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_write, + const size_t buffer_origin[3], + const size_t host_origin[3], + const size_t region[3], + size_t buffer_row_pitch, + size_t buffer_slice_pitch, + size_t host_row_pitch, + size_t host_slice_pitch, + const void *ptr, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(buffer)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* dstBuffer = as_amd(buffer)->asBuffer(); + if (dstBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (dstBuffer->getMemFlags() & + (CL_MEM_HOST_READ_ONLY | CL_MEM_HOST_NO_ACCESS)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if(hostQueue.context() != dstBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + if (ptr == NULL) { + return CL_INVALID_VALUE; + } + + // Create buffer rectangle info structure + amd::BufferRect bufRect; + amd::BufferRect hostRect; + + if (!bufRect.create(buffer_origin, region, buffer_row_pitch, buffer_slice_pitch) || + !hostRect.create(host_origin, region, host_row_pitch, host_slice_pitch)) { + return CL_INVALID_VALUE; + } + + amd::Coord3D dstStart(bufRect.start_, 0, 0); + amd::Coord3D dstEnd(bufRect.end_, 1, 1); + + if(!dstBuffer->validateRegion(dstStart, dstEnd)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::Coord3D size(region[0], region[1], region[2]); + amd::WriteMemoryCommand *command = new amd::WriteMemoryCommand( + hostQueue, + CL_COMMAND_WRITE_BUFFER_RECT, + eventWaitList, + *dstBuffer, + dstStart, size, ptr, + bufRect, + hostRect); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + if (blocking_write) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueues a command to copy a 2D or 3D rectangular region from + * the buffer object identified by \a src_buffer to a 2D or 3D region + * in the buffer object identified by \a dst_buffer. + * + * \param command_queue refers to the command-queue in which the copy command + * will be queued. The OpenCL context associated with command_queue, + * \a src_buffer and \a dst_buffer must be the same. + * + * \param src_origin defines the (x, y, z) offset in the memory region + * associated with \a src_buffer. For a 2D rectangle region, the z value given + * by src_origin[2] should be 0. The offset in bytes is computed as + * src_origin[2] * src_slice_pitch + src_origin[1] * src_row_pitch + src_origin[0]. + * + * \param dst_origin defines the (x, y, z) offset in the memory region + * associated with \a dst_buffer. For a 2D rectangle region, the z value given + * by dst_origin[2] should be 0. The offset in bytes is computed as + * dst_origin[2] * dst_slice_pitch + dst_origin[1] * dst_row_pitch + dst_origin[0]. + * + * \param region defines the (width, height, depth) in bytes of the 2D or 3D + * rectangle being copied. For a 2D rectangle, the depth value given by + * region[2] should be 1. + * + * \param pasrc_row_pitch is the length of each row in bytes to be used for + * the memory region associated with src_buffer. If src_row_pitch is 0, + * src_row_pitch is computed as region[0]. + * + * \param src_slice_pitch is the length of each 2D slice in bytes to be used + * for the memory region associated with src_buffer. If src_slice_pitch is 0, + * src_slice_pitch is computed as region[1] * src_row_pitch. + * + * \param dst_row_pitch is the length of each row in bytes to be used for + * the memory region associated with dst_buffer. If dst_row_pitch is 0, + * dst_row_pitch is computed as region[0]. + * + * \param dst_slice_pitch is the length of each 2D slice in bytes to be used + * for the memory region associated with dst_buffer. If dst_slice_pitch is 0, + * dst_slice_pitch is computed as region[1] * dst_row_pitch. + * + * \param event_wait_list and num_events_in_wait_list specify events that + * need to complete before this particular command can be executed. + * If event_wait_list is NULL, then this particular command does not wait on + * any event to complete. If event_wait_list is NULL, num_events_in_wait_list + * must be 0. If event_wait_list is not NULL, the list of events pointed to by + * event_wait_list must be valid and num_events_in_wait_list must be greater + * than 0. The events specified in event_wait_list act as synchronization + * points. The context associated with events in event_wait_list and + * command_queue must be the same. + * + * \param event returns an event object that identifies this particular copy + * command and can be used to query or queue a wait for this particular + * command to complete. event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue + * a wait for this command to complete. clEnqueueBarrier can be used instead. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise, + * it returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with command_queue, + * \a src_buffer and \a dst_buffer are not the same or if the context + * associated with \a command_queue and in \a event_wait_list are not the same. + * - CL_INVALID_MEM_OBJECT if \a src_buffer and \a dst_buffer are not valid + * buffer objects. + * - CL_INVALID_VALUE if (\a src_offset, \a region) or (\a dst_offset, + * \a region) require accessing elements outside the \a src_buffer and + * \a dst_buffer buffer objects respectively. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in + * \a event_wait_list are not valid events. + * - CL_MEM_COPY_OVERLAP if \a src_buffer and \a dst_buffer are the same + * buffer object and the source and destination regions overlap. + * - CL_MISALIGNED_SUB_BUFFER_OFFSET if \a src_buffer is a sub-buffer object + * and offset specified when the sub-buffer object is created is + * not aligned to CL_DEVICE_MEM_BASE_ADDR_ALIGN value for device + * associated with queue. + * - CL_MISALIGNED_SUB_BUFFER_OFFSET if dst_buffer is a sub-buffer object + * and offset specified when the sub-buffer object is created is not + * aligned to CL_DEVICE_MEM_BASE_ADDR_ALIGN value for device associated + * with queue. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate + * memory for data store associated with src_buffer or dst_buffer. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources + * required by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host + * + */ +RUNTIME_ENTRY(cl_int, clEnqueueCopyBufferRect, ( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_buffer, + const size_t src_origin[3], + const size_t dst_origin[3], + const size_t region[3], + size_t src_row_pitch, + size_t src_slice_pitch, + size_t dst_row_pitch, + size_t dst_slice_pitch, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(src_buffer) || !is_valid(dst_buffer)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* srcBuffer = as_amd(src_buffer)->asBuffer(); + amd::Buffer* dstBuffer = as_amd(dst_buffer)->asBuffer(); + if (srcBuffer == NULL || dstBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcBuffer->getContext() + || hostQueue.context() != dstBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + // Create buffer rectangle info structure + amd::BufferRect srcRect; + amd::BufferRect dstRect; + + if (!srcRect.create(src_origin, region, src_row_pitch, src_slice_pitch) || + !dstRect.create(dst_origin, region, dst_row_pitch, dst_slice_pitch)) { + return CL_INVALID_VALUE; + } + + amd::Coord3D srcStart(srcRect.start_, 0, 0); + amd::Coord3D dstStart(dstRect.start_, 0, 0); + amd::Coord3D srcEnd(srcRect.end_, 1, 1); + amd::Coord3D dstEnd(dstRect.end_, 1, 1); + + if (!srcBuffer->validateRegion(srcStart, srcEnd) || + !dstBuffer->validateRegion(dstStart, dstEnd)) { + return CL_INVALID_VALUE; + } + + // Check if regions overlap each other + if ((srcBuffer == dstBuffer) && + (std::abs(static_cast(src_origin[0]) - + static_cast(dst_origin[0])) < static_cast(region[0])) && + (std::abs(static_cast(src_origin[1]) - + static_cast(dst_origin[1])) < static_cast(region[1])) && + (std::abs(static_cast(src_origin[2]) - + static_cast(dst_origin[2])) < static_cast(region[2]))){ + return CL_MEM_COPY_OVERLAP; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::Coord3D size(region[0], region[1], region[2]); + amd::CopyMemoryCommand *command = new amd::CopyMemoryCommand( + hostQueue, + CL_COMMAND_COPY_BUFFER_RECT, + eventWaitList, + *srcBuffer, *dstBuffer, + srcStart, + dstStart, + size, + srcRect, + dstRect); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_MemoryCallback + * @{ + */ + +/*! \brief Registers a user callback function that will be called when the + * memory object is deleted and its resources freed. + * + * Each call to clSetMemObjectDestructorCallback registers the specified user + * callback function on a callback stack associated with memobj. The registered + * user callback functions are called in the reverse order in which they were + * registered. The user callback functions are called and then the memory + * object’s resources are freed and the memory object is deleted. + * This provides a mechanism for the application (and libraries) using memobj + * to be notified when the memory referenced by host_ptr, specified when + * the memory object is created and used as the storage bits for the memory + * object, can be reused or freed. + * + * \a memobj is a valid memory object. + * \a pfn_notify is the callback function that can be registered by the + * application. This callback function may be called asynchronously by the + * OpenCL implementation. It is the application’s responsibility to ensure + * that the callback function is thread-safe. The parameters to this callback + * function are: + * - memobj is the memory object being deleted. + * - user_data is a pointer to user supplied data. + * If pfn_notify is NULL, no callback function is registered for memobj. + * \a user_data will be passed as the user_data argument when pfn_notify is + * called. user_data can be NULL. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_MEM_OBJECT if memobj is not a valid memory object. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * NOTE: When the user callback function is called by the implementation, the + * contents of the memory region pointed to by host_ptr (if the memory object is + * created with CL_MEM_USE_HOST_PTR) are undefined. The callback function is + * typically used by the application to either free or reuse the memory region + * pointed to by host_ptr. The behavior of calling expensive system routines, + * OpenCL API calls to create contexts or command-queues, or blocking OpenCL + * operations from the following list below, in a callback is undefined. + * + * \version 1.1r17 + */ +RUNTIME_ENTRY(cl_int, clSetMemObjectDestructorCallback, ( + cl_mem memobj, + void (CL_CALLBACK * pfn_notify)(cl_mem memobj, void *user_data), + void *user_data)) +{ + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + + if (pfn_notify == NULL) { + return CL_INVALID_VALUE; + } + + if (!as_amd(memobj)->setDestructorCallback(pfn_notify, user_data)) { + return CL_OUT_OF_HOST_MEMORY; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_RetRelMemory + * @{ + */ + +/*! \brief Increment the \a memobj reference count. + * + * \return CL_SUCCESS if the function is executed successfully or + * CL_INVALID_MEM_OBJECT if \a memobj is not a valid memory object. + * + * clCreateBuffer and clCreateImage{2D|3D} perform an implicit retain. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainMemObject, (cl_mem memobj)) +{ + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + as_amd(memobj)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the \a memobj reference count. + * + * After the \a memobj reference count becomes zero and commands queued for + * execution on a command-queue(s) that use \a memobj have finished, the + * memory object is deleted. + * + * \return CL_SUCCESS if the function is executed successfully or + * CL_INVALID_MEM_OBJECT if \a memobj is not a valid memory object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseMemObject, (cl_mem memobj)) +{ + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + as_amd(memobj)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_CreatingImage + * @{ + */ + +/*! \brief Create a (1D, or 2D) image object. + * + * \param context is a valid OpenCL context on which the image object is to be + * created. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information about the image memory object being created. + * + * \param image_format is a pointer to a structure that describes format + * properties of the image to be allocated. + * + * \param image_width is the width of the image in pixels. Must be greater + * than or equal to 1. + * + * \param image_height is the height of the image in pixels. Must be greater + * than or equal to 1. + * + * \param image_row_pitch is the scan-line pitch in bytes. This must be 0 if + * \a host_ptr is NULL and can be either 0 or >= \a image_width * size of + * element in bytes if \a host_ptr is not NULL. If \a host_ptr is not NULL and + * \a image_row_pitch = 0, \a image_row_pitch is calculated as + * \a image_width * size of element in bytes. + * + * \param host_ptr is a pointer to the image data that may already be allocated + * by the application. The size of the buffer that \a host_ptr points to must + * be >= \a image_row_pitch * \a image_height. The size of each element in + * bytes must be a power of 2. Passing in a pointer to an already allocated + * buffer on the host and using it as a memory object allows applications to + * share data efficiently with kernels and the host. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero image object and errcode_ret is set to CL_SUCCESS + * if the image object is created successfully. It returns a NULL value with + * one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if values specified in \a flags are not valid. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if values specified in \a image_format + * are not valid or if \a image_format is NULL. + * - CL_INVALID_IMAGE_SIZE if \a image_width or \a image_height are 0 or if + * they exceed values specified in CL_DEVICE_IMAGE2D_MAX_WIDTH or + * CL_DEVICE_IMAGE2D_MAX_HEIGHT respectively or if values specified by + * \a image_row_pitch do not follow rules described in the argument + * description above. + * - CL_INVALID_HOST_PTR if \a host_ptr is NULL and CL_MEM_USE_HOST_PTR or + * CL_MEM_COPY_HOST_PTR are set in \a flags or if \a host_ptr is not NULL + * but CL_MEM_COPY_HOST_PTR or CL_MEM_USE_HOST_PTR are not set in \a flags. + * - CL_IMAGE_FORMAT_NOT_SUPPORTED if the \a image_format is not supported. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for image object. + * - CL_INVALID_OPERATION if the image object as specified by the + * \a image_format, \a flags and dimensions cannot be created for all devices + * in context that support images or if there are no devices in context that + * support images. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateImage2D, ( + cl_context context, + cl_mem_flags flags, + const cl_image_format *image_format, + size_t image_width, + size_t image_height, + size_t image_row_pitch, + void *host_ptr, + cl_int *errcode_ret)) +{ + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return (cl_mem) 0; + } + // check flags for validity + if(!validateFlags(flags)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return (cl_mem) 0; + } + // check format + if(image_format == NULL) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter \"image_format\""); + return (cl_mem) 0; + } + + const amd::Image::Format imageFormat(*image_format); + if(!imageFormat.isValid()) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter \"image_format\""); + return (cl_mem) 0; + } + + amd::Context& amdContext = *as_amd(context); + if(!imageFormat.isSupported(amdContext)) { + *not_null(errcode_ret) = CL_IMAGE_FORMAT_NOT_SUPPORTED; + LogWarning("invalid parameter \"image_format\""); + return (cl_mem) 0; + } + // check size parameters + if(image_width == 0 || image_height == 0) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_width\" or \"image_height\""); + return (cl_mem) 0; + } + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if((*it)->info().imageSupport_) { + supportPass = true; + if((*it)->info().image2DMaxWidth_ >= image_width + && (*it)->info().image2DMaxHeight_ >= image_height) { + sizePass = true; + break; + } + } + } + if(!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return (cl_mem) 0; + } + if(!sizePass) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_width\" or \"image_height\""); + return (cl_mem) 0; + } + // check row pitch rules + if(host_ptr == NULL) { + if(image_row_pitch) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_row_pitch\""); + return (cl_mem) 0; + } + } + else if(image_row_pitch) { + size_t elemSize = imageFormat.getElementSize(); + if((image_row_pitch < image_width * elemSize) + || (image_row_pitch % elemSize)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_row_pitch\""); + return (cl_mem) 0; + } + } + // check host_ptr consistency + if(host_ptr == NULL) { + if(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"host_ptr\""); + return (cl_mem) 0; + } + } + else { + if(!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"host_ptr\""); + return (cl_mem) 0; + } + } + + // CL_IMAGE_FORMAT_NOT_SUPPORTED ??? + + if(image_row_pitch == 0) { + image_row_pitch = image_width * imageFormat.getElementSize(); + } + + amd::Image* image = new(amdContext) + amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE2D, + flags, + imageFormat, + image_width, + image_height, + 1, + image_row_pitch, + 0); + if(image == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("cannot allocate resources"); + return (cl_mem) 0; + } + + // CL_MEM_OBJECT_ALLOCATION_FAILURE + if(!image->create(host_ptr)) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + image->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return (cl_mem) as_cl(image); +} +RUNTIME_EXIT + +/*! \brief Create a 3D image object. + * + * \param context is a valid OpenCL context on which the image object is to be + * created. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information about the image memory object being created. + * + * \param image_format is a pointer to a structure that describes format + * properties of the image to be allocated. + * + * \param image_width is the width of the image in pixels. Must be greater + * than or equal to 1. + * + * \param image_height is the height of the image in pixels. Must be greater + * than or equal to 1. + * + * \param image_depth is the depth of the image in pixels. This must be a + * value > 1. + * + * \param image_row_pitch is the scan-line pitch in bytes. This must be 0 if + * \a host_ptr is NULL and can be either 0 or >= \a image_width * size of + * element in bytes if \a host_ptr is not NULL. If \a host_ptr is not NULL and + * \a image_row_pitch = 0, \a image_row_pitch is calculated as + * \a image_width * size of element in bytes. + * + * \param image_slice_pitch is the size in bytes of each 2D slice in the 3D + * image. This must be 0 if \a host_ptr is NULL and can be either 0 or >= + * \a image_row_pitch * \a image_height if \a host_ptr is not NULL. + * If \a host_ptr is not NULL and \a image_slice_pitch = 0, + * \a image_slice_pitch is calculated as \a image_row_pitch * \a image_height. + * + * \param host_ptr is a pointer to the image data that may already be allocated + * by the application. The size of the buffer that \a host_ptr points to must + * be >= \a image_row_pitch * \a image_height * \a image_depth. The size of + * each element in bytes must be a power of 2. Passing in a pointer to an + * already allocated buffer on the host and using it as a memory object allows + * applications to share data efficiently with kernels and the host. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return valid non-zero image object created and the \a errcode_ret is set to + * CL_SUCCESS if the image object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if values specified in \a flags are not valid. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if values specified in \a image_format + * are not valid or if \a image_format is NULL. + * - CL_INVALID_IMAGE_SIZE if \a image_width, \a image_height or \a image_depth + * are 0 or if they exceed values specified in CL_DEVICE_IMAGE3D_MAX_WIDTH, + * CL_DEVICE_IMAGE3D_MAX_HEIGHT or CL_DEVICE_IMAGE3D_MAX_DEPTH respectively + * or if values specified by \a image_row_pitch and \a image_slice_pitch do + * not follow rules described in the argument description above. + * - CL_INVALID_HOST_PTR if \a host_ptr is NULL and CL_MEM_USE_HOST_PTR or + * CL_MEM_COPY_HOST_PTR are set in \a flags or if \a host_ptr is not NULL but + * CL_MEM_COPY_HOST_PTR or CL_MEM_USE_HOST_PTR are not set in \a flags. + * - CL_IMAGE_FORMAT_NOT_SUPPORTED if the \a image_format is not supported. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for image object. + * - CL_INVALID_OPERATION if the image object as specified by the + * \a image_format, \a flags and dimensions cannot be created for all devices + * in context that support images, or if there are no devices in context that + * support images. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateImage3D, ( + cl_context context, + cl_mem_flags flags, + const cl_image_format *image_format, + size_t image_width, + size_t image_height, + size_t image_depth, + size_t image_row_pitch, + size_t image_slice_pitch, + void *host_ptr, + cl_int *errcode_ret)) +{ + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return (cl_mem) 0; + } + // check flags for validity + if(!validateFlags(flags)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return (cl_mem) 0; + } + // check format + if(image_format == NULL) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter \"image_format\""); + return (cl_mem) 0; + } + amd::Image::Format imageFormat(*image_format); + + if(!imageFormat.isValid()) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter \"image_format\""); + return (cl_mem) 0; + } + + amd::Context& amdContext = *as_amd(context); + if(!imageFormat.isSupported(amdContext)) { + *not_null(errcode_ret) = CL_IMAGE_FORMAT_NOT_SUPPORTED; + LogWarning("invalid parameter \"image_format\""); + return (cl_mem) 0; + } + // check size parameters + if(image_width == 0 || image_height == 0 || image_depth <= 1) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid size parameter(s)"); + return (cl_mem) 0; + } + const std::vector& devices = as_amd(context)->devices(); + bool supportPass = false; + bool sizePass = false; + std::vector::const_iterator it; + for(it = devices.begin(); it != devices.end(); ++it) { + if((*it)->info().imageSupport_) { + supportPass = true; + if((*it)->info().image3DMaxWidth_ >= image_width + && (*it)->info().image3DMaxHeight_ >= image_height + && (*it)->info().image3DMaxDepth_ >= image_depth) { + sizePass = true; + break; + } + } + } + if(!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return (cl_mem) 0; + } + if(!sizePass) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid size parameter(s)"); + return (cl_mem) 0; + } + // check row pitch rules + if(host_ptr == NULL) { + if(image_row_pitch) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_row_pitch\""); + return (cl_mem) 0; + } + } + else if(image_row_pitch) { + size_t elemSize = imageFormat.getElementSize(); + if((image_row_pitch < image_width * elemSize) + || (image_row_pitch % elemSize)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_row_pitch\""); + return (cl_mem) 0; + } + } + // check slice pitch + if(host_ptr == NULL) { + if(image_slice_pitch) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_row_pitch\""); + return (cl_mem) 0; + } + } + else if(image_slice_pitch) { + size_t elemSize = imageFormat.getElementSize(); + if((image_slice_pitch < image_row_pitch * image_height) + || (image_slice_pitch % image_row_pitch)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter \"image_row_pitch\""); + return (cl_mem) 0; + } + } + // check host_ptr consistency + if(host_ptr == NULL) { + if(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"host_ptr\""); + return (cl_mem) 0; + } + } + else { + if(!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter \"host_ptr\""); + return (cl_mem) 0; + } + } + + // CL_IMAGE_FORMAT_NOT_SUPPORTED ??? + + if(image_row_pitch == 0) { + image_row_pitch = image_width * imageFormat.getElementSize(); + } + if(image_slice_pitch == 0) { + image_slice_pitch = image_row_pitch * image_height; + } + + amd::Image* image = new(amdContext) + amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE3D, + flags, + imageFormat, + image_width, + image_height, + image_depth, + image_row_pitch, + image_slice_pitch); + if(image == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("cannot allocate resources"); + return (cl_mem) 0; + } + + // CL_MEM_OBJECT_ALLOCATION_FAILURE + if(!image->create(host_ptr)) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + image->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return (cl_mem) as_cl(image); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_QueryImageFormat + * @{ + */ + +/*! \brief Get the list of supported image formats. + * + * \param context is a valid OpenCL context on which the image object(s) will + * be created. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information about the image memory object being created. + * + * \param image_type describes the image type and must be either + * CL_MEM_OBJECT_IMAGE1D, CL_MEM_OBJECT_IMAGE1D_BUFFER, CL_MEM_OBJECT_IMAGE2D, + * CL_MEM_OBJECT_IMAGE3D, CL_MEM_OBJECT_IMAGE1D_ARRAY or + * CL_MEM_OBJECT_IMAGE2D_ARRAY. + * + * \param num_entries specifies the number of entries that can be returned in + * the memory location given by \a image_formats. + * + * \param image_formats is a pointer to a memory location where the list of + * supported image formats are returned. Each entry describes a cl_image_format + * structure supported by the runtime. If \a image_formats is NULL, it is + * ignored. + * + * \param num_image_formats is the actual number of supported image formats for + * a specific context and values specified by \a flags. If \a num_image_formats + * is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_CONTEXT if \a context is not a valid context + * - CL_INVALID_VALUE if \a flags or \a image_type are not valid, or if + * \a num_entries is 0 and \a image_formats is not NULL + * + * \version 1.2r08 + */ +RUNTIME_ENTRY(cl_int, clGetSupportedImageFormats, ( + cl_context context, + cl_mem_flags flags, + cl_mem_object_type image_type, + cl_uint num_entries, + cl_image_format *image_formats, + cl_uint *num_image_formats)) +{ + if(!is_valid(context)) { + LogWarning("invalid parameter \"context\""); + return CL_INVALID_CONTEXT; + } + // check flags for validity + if(!validateFlags(flags)) { + LogWarning("invalid parameter \"flags\""); + return CL_INVALID_VALUE; + } + // chack image_type + switch(image_type) + { + case CL_MEM_OBJECT_IMAGE1D_BUFFER: + case CL_MEM_OBJECT_IMAGE1D: + case CL_MEM_OBJECT_IMAGE1D_ARRAY: + case CL_MEM_OBJECT_IMAGE2D: + case CL_MEM_OBJECT_IMAGE2D_ARRAY: + case CL_MEM_OBJECT_IMAGE3D: + break; + + default: + LogWarning("invalid parameter \"image_type\""); + return CL_INVALID_VALUE; + } + if(num_entries == 0 && image_formats != NULL) { + LogWarning("invalid parameter \"num_entries\""); + return CL_INVALID_VALUE; + } + + const amd::Context& amdContext = *as_amd(context); + + if(image_formats != NULL) { + amd::Image::getSupportedFormats(amdContext, image_type, num_entries, image_formats, flags); + } + if(num_image_formats != NULL) { + *num_image_formats = amd::Image::numSupportedFormats(amdContext, image_type, flags); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + + +/*! @} + * \addtogroup CL_ReadWriteImage + * @{ + */ + +/*! \brief Enqueue a command to read from a 2D or 3D image object to host memory + * + * \param command_queue refers to the command-queue in which the read + * command will be queued. \a command_queue and \a image must be created with + * the same OpenCL context. + * + * \param image refers to a valid 2D or 3D image object. + * + * \param blocking_read indicates if the read is blocking or nonblocking. If + * \a blocking_read is CL_TRUE i.e. the read command is blocking, + * clEnqueueReadImage does not return until the buffer data has been read and + * copied into memory pointed to by \a ptr. If \a blocking_read is CL_FALSE + * i.e. the read command is non-blocking, clEnqueueReadImage queues a + * non-blocking read command and returns. The contents of the buffer that + * \a ptr points to cannot be used until the read command has completed. + * The \a event argument returns an event object which can be used to query the + * execution status of the read command. When the read command has completed, + * the contents of the buffer that ptr points to can be used by the application + * + * \param origin defines the (x, y, z) offset in the image from where to read + * or write. If image is a 2D image object, the z value given by origin[2] must + * be 0. + * + * \param region defines the (width, height, depth) of the 2D or 3D rectangle + * being read or written. If image is a 2D image object, the depth value given + * by region[2] must be 1. + * + * \param row_pitch in clEnqueueReadImage is the length of each row in bytes. + * This value must be greater than or equal to the element size in bytes + * width. If \a row_pitch is set to 0, the appropriate row pitch is calculated + * based on the size of each element in bytes multiplied by width. + * + * \param slice_pitch in clEnqueueReadImage clEnqueueWriteImage is the size + * in bytes of the 2D slice of the 3D region of a 3D image being read or + * written respectively. This must be 0 if image is a 2D image. This value + * must be greater than or equal to row_pitch * height. If \a slice_pitch is + * set to 0, the appropriate slice pitch is calculated based on the + * \a row_pitch * \a height. + * + * \param ptr is the pointer to a buffer in host memory where image data is + * to be read from. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then this + * particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular read + * command and can be used to query or queue a wait for this particular command + * to complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue a wait for this + * command to complete. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue and + * \a image are not the same. + * - CL_INVALID_MEM_OBJECT if \a image is not a valid image object. + * - CL_INVALID_VALUE if the region being read specified by \a origin and + * \a region is out of bounds or if \a ptr is a NULL value. + * - CL_INVALID_VALUE if \a image is a 2D image object and \a origin[2] is not + * equal to 0 or \a region[2] is not equal to 1 or \a slice_pitch is not + * equal to 0. + * - CL_INVALID_OPERATION if \a clEnqueueReadImage is called on image which + * has been created with CL_MEM_HOST_WRITE_ONLY or CL_MEM_HOST_NO_ACCESS. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_INVALID_VALUE if blocking_read is CL_FALSE and \a event is NULL. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueReadImage, ( + cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_read, + const size_t origin[3], + const size_t region[3], + size_t row_pitch, + size_t slice_pitch, + void *ptr, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(image)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Image* srcImage = as_amd(image)->asImage(); + if (srcImage == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (srcImage->getMemFlags() & + (CL_MEM_HOST_WRITE_ONLY | CL_MEM_HOST_NO_ACCESS)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcImage->getContext()) { + return CL_INVALID_CONTEXT; + } + + if (ptr == NULL) { + return CL_INVALID_VALUE; + } + + amd::Coord3D srcOrigin(origin[0], origin[1], origin[2]); + amd::Coord3D srcRegion(region[0], region[1], region[2]); + + if (!srcImage->validateRegion(srcOrigin, srcRegion) || + !srcImage->isSliceValid(row_pitch, slice_pitch, region[1])) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::ReadMemoryCommand *command = new amd::ReadMemoryCommand( + hostQueue, + CL_COMMAND_READ_IMAGE, + eventWaitList, + *srcImage, + srcOrigin, + srcRegion, + ptr, + row_pitch, + slice_pitch); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + if (blocking_read) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to write to a 2D or 3D image object from host + * memory + * + * \param command_queue refers to the command-queue in which the write + * command will be queued. \a command_queue and \a image must be created with + * the same OpenCL context. + * + * \param image refers to a valid 2D or 3D image object. + * + * \param blocking_write indicates if the write operation is blocking or + * nonblocking. If blocking_write is CL_TRUE, the OpenCL implementation copies + * the data referred to by \a ptr and enqueues the write command in the + * command-queue. The memory pointed to by ptr can be reused by the application + * after the clEnqueueWriteImage call returns. If blocking_write is CL_FALSE, + * the OpenCL implementation will use ptr to perform a nonblocking write. As + * the write is non-blocking the implementation can return immediately. The + * memory pointed to by ptr cannot be reused by the application after the call + * returns. The event argument returns an event object which can be used to + * query the execution status of the write command. When the write command has + * completed, the memory pointed to by ptr can then be reused by the + * application. + * + * \param origin defines the (x, y, z) offset in the image from where to read + * or write. If image is a 2D image object, the z value given by origin[2] must + * be 0. + * + * \param region defines the (width, height, depth) of the 2D or 3D rectangle + * being read or written. If image is a 2D image object, the depth value given + * by region[2] must be 1. + * + * \param input_row_pitch in is the length of each row in bytes. + * This value must be greater than or equal to the element size in bytes + * width. If \a input_row_pitch is set to 0, the appropriate row pitch is + * calculated based on the size of each element in bytes multiplied by width. + * + * \param input_slice_pitch is the size + * in bytes of the 2D slice of the 3D region of a 3D image being read or + * written respectively. This must be 0 if image is a 2D image. This value + * must be greater than or equal to input_row_pitch * height. If + * \a input_slice_pitch is set to 0, the appropriate slice pitch is calculated + * based on the \a input_row_pitch * \a height. + * + * \param ptr is the pointer to a buffer in host memory where image data is + * to be written to. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then this + * particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular write + * command and can be used to query or queue a wait for this particular command + * to complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue a wait for this + * command to complete. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue and + * \a image are not the same. + * - CL_INVALID_MEM_OBJECT if \a image is not a valid image object. + * - CL_INVALID_VALUE if the region being written specified by \a origin and + * \a region is out of bounds or if \a ptr is a NULL value. + * - CL_INVALID_VALUE if \a image is a 2D image object and \a origin[2] is not + * equal to 0 or \a region[2] is not equal to 1 or \a slice_pitch is not + * equal to 0. + * - CL_INVALID_OPERATION if \a clEnqueueWriteImage is called on image which + * has been created with CL_MEM_HOST_READ_ONLY or CL_MEM_HOST_NO_ACCESS. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_INVALID_VALUE if blocking_write is CL_FALSE and \a event is NULL. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueWriteImage, ( + cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_write, + const size_t origin[3], + const size_t region[3], + size_t input_row_pitch, + size_t input_slice_pitch, + const void *ptr, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(image)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Image* dstImage = as_amd(image)->asImage(); + if (dstImage == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (dstImage->getMemFlags() & + (CL_MEM_HOST_READ_ONLY | CL_MEM_HOST_NO_ACCESS)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if(hostQueue.context() != dstImage->getContext()) { + return CL_INVALID_CONTEXT; + } + + if (ptr == NULL) { + return CL_INVALID_VALUE; + } + + amd::Coord3D dstOrigin(origin[0], origin[1], origin[2]); + amd::Coord3D dstRegion(region[0], region[1], region[2]); + + if (!dstImage->validateRegion(dstOrigin, dstRegion) || + !dstImage->isSliceValid(input_row_pitch, input_slice_pitch, region[1])) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list , event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::WriteMemoryCommand *command = new amd::WriteMemoryCommand( + hostQueue, + CL_COMMAND_WRITE_IMAGE, + eventWaitList, + *dstImage, + dstOrigin, + dstRegion, + ptr, + input_row_pitch, + input_slice_pitch); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + if (blocking_write) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to copy image objects. + * + * \param command_queue refers to the command-queue in which the copy command + * will be queued. The OpenCL context associated with \a command_queue, + * \a src_image and \a dst_image must be the same. + * + * \param src_image is the source image object. + * + * \param dst_image is the destination image object. + * + * \param src_origin defines the starting (x, y, z) location in \a src_image + * from where to start the data copy. If \a src_image is a 2D image object, + * the z value given by \a src_origin[2] must be 0. + * + * \param dst_origin defines the starting (x, y, z) location in \a dst_image + * from where to start the data copy. If \a dst_image is a 2D image object, + * the z value given by \a dst_origin[2] must be 0. + * + * \param region defines the (width, height, depth) of the 2D or 3D rectangle + * to copy. If \a src_image or \a dst_image is a 2D image object, the depth + * value given by \a region[2] must be 1. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular copy + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue + * a wait for this command to complete. clEnqueueBarrier can be used instead. + * It is currently a requirement that the \a src_image and \a dst_image image + * memory objects for clEnqueueCopyImage must have the exact image format + * (i.e. channel order and channel data type must match). + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue, + * \a src_image and \a dst_image are not the same. + * - CL_INVALID_MEM_OBJECT if \a src_image and \a dst_image are not valid image + * objects. + * - CL_IMAGE_FORMAT_MISMATCH if src_image and dst_image do not use the same + * image format. + * - CL_INVALID_VALUE if the 2D or 3D rectangular region specified by + * \a src_origin and \a src_origin + \a region refers to a region outside + * \a src_image, or if the 2D or 3D rectangular region specified by + * \a dst_origin and \a dst_origin + \a region refers to a region outside + * \a dst_image. + * - CL_INVALID_VALUE if \a src_image is a 2D image object and \a origin[2] is + * not equal to 0 or \a region[2] is not equal to 1. + * - CL_INVALID_VALUE if \a dst_image is a 2D image object and \a dst_origin[2] + * is not equal to 0 or \a region[2] is not equal to 1. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueCopyImage, ( + cl_command_queue command_queue, + cl_mem src_image, + cl_mem dst_image, + const size_t src_origin[3], + const size_t dst_origin[3], + const size_t region[3], + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(src_image) || !is_valid(dst_image)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Image* srcImage = as_amd(src_image)->asImage(); + amd::Image* dstImage = as_amd(dst_image)->asImage(); + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcImage->getContext() + || hostQueue.context() != dstImage->getContext()) { + return CL_INVALID_CONTEXT; + } + + if(srcImage->getImageFormat() != dstImage->getImageFormat()) { + return CL_IMAGE_FORMAT_MISMATCH; + } + + amd::Coord3D srcOrigin(src_origin[0], src_origin[1], src_origin[2]); + amd::Coord3D dstOrigin(dst_origin[0], dst_origin[1], dst_origin[2]); + amd::Coord3D copyRegion(region[0], region[1], region[2]); + + if (!srcImage->validateRegion(srcOrigin, copyRegion)) { + return CL_INVALID_VALUE; + } + + if (!dstImage->validateRegion(dstOrigin, copyRegion)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + if(src_image == dst_image) { + if((src_origin[0] <= dst_origin[0] + && dst_origin[0] < src_origin[0] + region[0]) + || (dst_origin[0] <= src_origin[0] + && src_origin[0] < dst_origin[0] + region[0]) + || (src_origin[1] <= dst_origin[1] + && dst_origin[1] < src_origin[1] + region[1]) + || (dst_origin[1] <= src_origin[1] + && src_origin[1] < dst_origin[1] + region[1])) { + return CL_MEM_COPY_OVERLAP; + } + if(srcImage->getDims() > 2) { + if((src_origin[2] <= dst_origin[2] + && dst_origin[2] < src_origin[2] + region[2]) + || (dst_origin[2] <= src_origin[2] + && src_origin[2] < dst_origin[2] + region[2])) { + return CL_MEM_COPY_OVERLAP; + } + } + } + + amd::CopyMemoryCommand *command = new amd::CopyMemoryCommand( + hostQueue, + CL_COMMAND_COPY_IMAGE, + eventWaitList, + *srcImage, *dstImage, + srcOrigin, + dstOrigin, + copyRegion); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_CopyingImageBuffer + * @{ + */ + +/*! \brief Enqueue a command to copy an image object to a buffer object. + * + * \param command_queue must be a valid command-queue. The OpenCL context + * associated with \a command_queue, \a src_image and \a dst_buffer must be + * the same. + * + * \param src_image is a valid image object. + * + * \param dst_buffer is a valid buffer object. + * + * \param src_origin defines the (x, y, z) offset in the image from where to + * copy. If \a src_image is a 2D image object, the z value given by + * \a src_origin[2] must be 0. + * + * \param region defines the (width, height, depth) of the 2D or 3D rectangle + * to copy. If \a src_image is a 2D image object, the depth value given by + * \a region[2] must be 1. + * + * \param dst_offset refers to the offset where to begin copying data in + * \a dst_buffer. The size in bytes of the region to be copied referred to as + * \a dst_cb is computed as width * height * depth * bytes/image element if + * \a src_image is a 3D image object and is computed as + * width * height * bytes/image element if \a src_image is a 2D image object. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then this + * particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular copy + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. clEnqueueBarrier can be used instead. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue, + * \a src_image and \a dst_buffer are not the same. + * - CL_INVALID_MEM_OBJECT if \a src_image is not a valid image object or + * \a dst_buffer is not a valid buffer object. + * - CL_INVALID_VALUE if the 2D or 3D rectangular region specified by + * \a src_origin and \a src_origin + \a region refers to a region outside + * \a src_image, or if the region specified by \a dst_offset and + * \a dst_offset + \a dst_cb to a region outside \a dst_buffer. + * - CL_INVALID_VALUE if \a src_image is a 2D image object and \a src_origin[2] + * is not equal to 0 or \a region[2] is not equal to 1. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueCopyImageToBuffer, ( + cl_command_queue command_queue, + cl_mem src_image, + cl_mem dst_buffer, + const size_t* src_origin, + const size_t* region, + size_t dst_offset, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(src_image) || !is_valid(dst_buffer)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Image* srcImage = as_amd(src_image)->asImage(); + amd::Buffer* dstBuffer = as_amd(dst_buffer)->asBuffer(); + if (srcImage == NULL || dstBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcImage->getContext() + || hostQueue.context() != dstBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Coord3D srcOrigin(src_origin[0], src_origin[1], src_origin[2]); + amd::Coord3D dstOffset(dst_offset, 0, 0); + amd::Coord3D srcRegion(region[0], region[1], region[2]); + amd::Coord3D copySize(region[0] * region[1] * region[2] * + srcImage->getImageFormat().getElementSize(), 0, 0); + + if (!srcImage->validateRegion(srcOrigin, srcRegion) || + !dstBuffer->validateRegion(dstOffset, copySize)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::CopyMemoryCommand *command = new amd::CopyMemoryCommand( + hostQueue, + CL_COMMAND_COPY_IMAGE_TO_BUFFER, + eventWaitList, + *srcImage, *dstBuffer, + srcOrigin, + dstOffset, + srcRegion); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to copy a buffer object to an image object. + * + * \param command_queue must be a valid command-queue. The OpenCL context + * associated with \a command_queue, \a src_buffer and \a dst_image must be + * the same. + * + * \param src_buffer is a valid buffer object. + * + * \param dst_image is a valid image object. + * + * \param src_offset refers to the offset where to begin copying data in + * \a src_buffer. + * + * \param dst_origin defines the (x, y, z) offset in the image from where to + * copy. If \a dst_image is a 2D image object, the z value given by + * \a dst_origin[2] must be 0. + * + * \param region defines the (width, height, depth) of the 2D or 3D rectangle + * to copy. If dst_image is a 2D image object, the depth value given by + * \a region[2] must be 1. The size in bytes of the region to be copied from + * \a src_buffer referred to as \a src_cb is computed as + * width * height * depth * bytes/image element if \a dst_image is a 3D image + * object and is computed as width * height * bytes/image element if + * \a dst_image is a 2D image object. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular copy + * command and can be used to query or queue a wait for this particular command + * to complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue a wait for + * this command to complete. clEnqueueBarrier can be used instead. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise it + * returns one of the following errors: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with \a command_queue, + * \a src_buffer and \a dst_image are not the same. + * - CL_INVALID_MEM_OBJECT if \a src_buffer is not a valid buffer object or + * \a dst_image is not a valid image object. + * - CL_INVALID_VALUE if the 2D or 3D rectangular region specified by + * \a dst_origin and \a dst_origin + \a region refers to a region outside + * \a dst_image, or if the region specified by \a src_offset and + * \a src_offset + \a src_cb to a region outside \a src_buffer. + * - CL_INVALID_VALUE if \a dst_image is a 2D image object and \a dst_origin[2] + * is not equal to 0 or \a region[2] is not equal to 1. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in + * \a event_wait_list are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueCopyBufferToImage, ( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_image, + size_t src_offset, + const size_t* dst_origin, + const size_t* region, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(src_buffer) || !is_valid(dst_image)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Buffer* srcBuffer = as_amd(src_buffer)->asBuffer(); + amd::Image* dstImage = as_amd(dst_image)->asImage(); + if (srcBuffer == NULL || dstImage == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcBuffer->getContext() + || hostQueue.context() != dstImage->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Coord3D dstOrigin(dst_origin[0], dst_origin[1], dst_origin[2]); + amd::Coord3D srcOffset(src_offset, 0, 0); + amd::Coord3D dstRegion(region[0], region[1], region[2]); + amd::Coord3D copySize(region[0] * region[1] * region[2] * + dstImage->getImageFormat().getElementSize(), 0, 0); + + if (!srcBuffer->validateRegion(srcOffset, copySize) || + !dstImage->validateRegion(dstOrigin, dstRegion)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::CopyMemoryCommand *command = new amd::CopyMemoryCommand( + hostQueue, + CL_COMMAND_COPY_BUFFER_TO_IMAGE, + eventWaitList, + *srcBuffer, *dstImage, + srcOffset, + dstOrigin, + dstRegion); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_MapUnmap + * @{ + */ + +/*! \brief Enqueue a command to map a region of a buffer object into the + * host address. + * + * \param command_queue must be a valid command-queue. + * + * \param blocking_map indicates if the map operation is blocking or + * non-blocking. If \a blocking_map is CL_TRUE, clEnqueueMapBuffer does not + * return until the specified region in \a buffer can be mapped. If + * \a blocking_map is CL_FALSE i.e. map operation is non-blocking, the pointer + * to the mapped region returned by clEnqueueMapBuffer cannot be used until the + * map command has completed. The event argument returns an event object which + * can be used to query the execution status of the map command. When the map + * command is completed, the application can access the contents of the mapped + * region using the pointer returned by clEnqueueMapBuffer. + * + * \param map_flags is a bit-field and can be set to CL_MAP_READ to indicate + * that the region specified by (\a offset, \a cb) in the buffer object is + * being mapped for reading, and/or CL_MAP_WRITE to indicate that the region + * specified by (\a offset, \a cb) in the buffer object is being mapped for + * writing. + * + * \param buffer is a valid buffer object. The OpenCL context associated with + * \a command_queue and \a buffer must be the same. + * + * \param offset is the offset in bytes of the region in the buffer object + * that is being mapped + * + * \param cb is the size in bytes of the region in the buffer object that + * is being mapped. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue + * a wait for this command to complete. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A pointer to the mapped region if buffer is a memory object + * created with clCreateBuffer and the region specified by (offset , cb) + * is a valid region in the buffer object and is successfully mapped into the + * host address space . The \a errcode_ret is set to CL_SUCCESS. + * A NULL pointer is returned otherwise with one of the following error values + * returned in \a errcode_ret: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * \a buffer are not the same. + * - CL_INVALID_MEM_OBJECT if \a buffer is not a valid buffer object. + * - CL_INVALID_OPERATION if buffer has been created with + * CL_MEM_HOST_WRITE_ONLY or CL_MEM_HOST_NO_ACCESS and CL_MAP_READ + * is set in map_flags or if buffer has been created with + * CL_MEM_HOST_READ_ONLY or CL_MEM_HOST_NO_ACCESS and CL_MAP_WRITE or + * CL_MAP_WRITE_INVALIDATE_REGION is set in map_flags. + * - CL_INVALID_VALUE if region being mapped given by (\a offset, \a cb) is out + * of bounds or if values specified in \a map_flags are not valid. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in + * \a event_wait_list are not valid events. + * - CL_MEM_O BJECT_MAP_FAILURE if there is a failure to map the specified + * region in the host address space. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * The pointer returned maps a region starting at \a offset and is atleast + * \a cb bytes in size. The result of a memory access outside this region is + * undefined. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY_RET(void *, clEnqueueMapBuffer, ( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_map, + cl_map_flags map_flags, + size_t offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event, + cl_int *errcode_ret)) +{ + if (!is_valid(command_queue)) { + *not_null(errcode_ret) = CL_INVALID_COMMAND_QUEUE; + return NULL; + } + + if (!is_valid(buffer)) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return NULL; + } + amd::Buffer* srcBuffer = as_amd(buffer)->asBuffer(); + if (srcBuffer == NULL) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return NULL; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + *not_null(errcode_ret) = CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcBuffer->getContext()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return NULL; + } + + if ((srcBuffer->getMemFlags() & + (CL_MEM_HOST_WRITE_ONLY | CL_MEM_HOST_NO_ACCESS)) && + (map_flags & CL_MAP_READ)) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + if ((srcBuffer->getMemFlags() & + (CL_MEM_HOST_READ_ONLY | CL_MEM_HOST_NO_ACCESS)) && + (map_flags & (CL_MAP_WRITE | CL_MAP_WRITE_INVALIDATE_REGION))) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + if (srcBuffer->getMemFlags() & CL_MEM_EXTERNAL_PHYSICAL_AMD) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + amd::Coord3D srcOffset(offset); + amd::Coord3D srcSize(cb); + + if (!srcBuffer->validateRegion(srcOffset, srcSize)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return NULL; + } + + // Wait for possible pending operations + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + *not_null(errcode_ret) = err; + return (void*) 0; + } + + // Allocate a map command for the queue thread + amd::MapMemoryCommand *command = new amd::MapMemoryCommand( + hostQueue, + CL_COMMAND_MAP_BUFFER, + eventWaitList, + *srcBuffer, + map_flags, + blocking_map ? true : false, + srcOffset, + srcSize); + if (command == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return NULL; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + return NULL; + } + // Make sure the svm host memory is commited if the buffer is SVM buffer + void* svmPtr = srcBuffer->getSvmPtr(); + if (NULL != svmPtr) { + srcBuffer->commitSvmMemory(); + } + + if (srcBuffer->getMemFlags() & CL_MEM_USE_PERSISTENT_MEM_AMD) { + // [Windows VidMM restriction] + // Runtime can't map persistent memory if it's still busy or + // even wasn't submitted to HW from the worker thread yet + hostQueue.finish(); + } + + // Attempt to allocate the map target now (whether blocking or non-blocking) + void* mapPtr = hostQueue.device().allocMapTarget( + *srcBuffer, srcOffset, srcSize); + if (NULL == mapPtr) { + delete command; + *not_null(errcode_ret) = CL_MAP_FAILURE; + return NULL; + } + + // Send the map command for processing + command->enqueue(); + + // A blocking map has to wait for completion + if (blocking_map) { + command->awaitCompletion(); + } + + // Save the command event if applicaiton has requested it + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + *not_null(errcode_ret) = CL_SUCCESS; + srcBuffer->incMapCount(); + return mapPtr; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to map a region in an image object given into + * the host address. + * + * \param command_queue must be a valid command-queue. + * + * \param image is a valid image object. The OpenCL context associated with + * \a command_queue and \a image must be the same. + * + * \param blocking_map indicates if the map operation is blocking or + * non-blocking. If \a blocking_map is CL_TRUE, clEnqueueMapImage does not + * return until the specified region in image is mapped. If \a blocking_map is + * CL_FALSE i.e. map operation is non-blocking, the pointer to the mapped + * region returned by clEnqueueMapImage cannot be used until the map command + * has completed. The event argument returns an event object which can be used + * to query the execution status of the map command. When the map command is + * completed, the application can access the contents of the mapped region + * using the pointer returned by clEnqueueMapImage. + * + * \param map_flags is a bit-field and can be set to CL_MAP_READ to indicate + * that the region specified by (\a origin, \a region) in the image object is + * being mapped for reading, and/or CL_MAP_WRITE to indicate that the region + * specified by (\a origin, \a region) in the image object is being mapped for + * writing. + * + * \param origin defines the (x, y, z) offset in pixels in the image or (x, y) + * offset and the image index in the image array. If image is a 2D image + * object, origin[2] must be 0. If image is a 1D image or 1D image buffer + * object, origin[1] and origin[2] must be 0. If image is a 1D image array + * object, origin[2] must be 0. If image is a 1D image array object, origin[1] + * describes the image index in the 1D image array. If image is a 2D image + * array object, origin[2] describes the image index in the 2D image array. + * + * \param region defines the (width, height, depth) in pixels of the 1D, 2D or + * 3D rectangle or the (width, height) in pixels in pixels of the 1D or 2D + * rectangle and the image index of an image array. If image is a 2D image + * object, region[2] must be 1. If image is a 1D image or 1D image buffer + * object, region[1] and region[2] must be 1. If image is a 1D image array + * object, region[1] and region[2] must be 1. If image is a 2D image array + * object, region[2] must be 1. + * + * \param origin define the (x, y, z) offset of the 2D or 3D rectangle region + * that is to be mapped. If image is a 2D image object, the z value given by + * \a origin[2] must be 0. + * + * \param region define the (width, height, depth) of the 2D or 3D rectangle + * region that is to be mapped. If image is a 2D image object, the depth value + * given by \a region[2] must be 1. + * + * \param image_row_pitch returns the scan-line pitch in bytes for the mapped + * region. This must be a non- NULL value. + * + * \param image_slice_pitch returns the size in bytes of each 2D slice for the + * mapped region. For a 2D image this argument is ignored. For a 3D image this + * must be a non-NULL value. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before + * clEnqueueMapImage can be executed. If \a event_wait_list is NULL, then + * clEnqueueMapImage does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A pointer to the mapped region if image is a memory object + * created with clCreateImage {2D|3D}, and the 2D or 3D rectangle specified + * by origin and region is a valid region in the image object and can be + * mapped into the host address space. + * The \a errcode_ret is set to CL_SUCCESS. A NULL pointer is returned + * otherwise with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * \a image are not the same. + * - CL_INVALID_MEM_OBJECT if \a image is not a valid image object. + * - CL_INVALID_VALUE if region being mapped given by + * (\a origin, \a origin + \a region) is out of bounds or if values + * specified in \a map_flags are not valid. + * - CL_INVALID_VALUE if values in origin and region do not follow rules + * described in the argument description for origin and region. + * - CL_INVALID_VALUE if \a image is a 2D image object and \a origin[2] is not + * equal to 0 or \a region[2] is not equal to 1. + * - CL_INVALID_VALUE if \a image_row_pitch is NULL. + * - CL_INVALID_VALUE if \a image is a 3D image object and \a image_slice_pitch + * is NULL. + * - CL_INVALID_IMAGE_FORMAT if image format (image channel order and data + * type) for image are not supported by device associated with queue. + * - CL_INVALID_OPERATION if buffer has been created with + * CL_MEM_HOST_WRITE_ONLY or CL_MEM_HOST_NO_ACCESS and CL_MAP_READ + * is set in map_flags or if buffer has been created with + * CL_MEM_HOST_READ_ONLY or CL_MEM_HOST_NO_ACCESS and CL_MAP_WRITE or + * CL_MAP_WRITE_INVALIDATE_REGION is set in map_flags. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_MEM_OBJECT_MAP_FAILURE if there is a failure to map the specified + * region in the host address space. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * The pointer returned maps a 2D or 3D region starting at origin and is + * at least (\a image_row_pitch * \a region[1] + \a region[0]) pixels in size + * for a 2D image, and is at least (\a image_slice_pitch * \a region[2] + + * \a image_row_pitch * \a region[1] + \a region[0]) pixels in size for a 3D + * image. The result of a memory access outside this region is undefined. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY_RET(void *, clEnqueueMapImage, ( + cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_map, + cl_map_flags map_flags, + const size_t* origin, + const size_t* region, + size_t *image_row_pitch, + size_t *image_slice_pitch, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event, + cl_int *errcode_ret)) +{ + if (!is_valid(command_queue)) { + *not_null(errcode_ret) = CL_INVALID_COMMAND_QUEUE; + return NULL; + } + + if (!is_valid(image)) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return NULL; + } + amd::Image* srcImage = as_amd(image)->asImage(); + if (srcImage == NULL) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return NULL; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + *not_null(errcode_ret) = CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != srcImage->getContext()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return NULL; + } + + if ((srcImage->getMemFlags() & + (CL_MEM_HOST_WRITE_ONLY | CL_MEM_HOST_NO_ACCESS)) && + (map_flags & CL_MAP_READ)) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + if ((srcImage->getMemFlags() & + (CL_MEM_HOST_READ_ONLY | CL_MEM_HOST_NO_ACCESS)) && + (map_flags & (CL_MAP_WRITE | CL_MAP_WRITE_INVALIDATE_REGION))) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + if ((srcImage->getDims() == 1) && + ((region[1] != 1) || (region[2] != 1))) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return NULL; + } + + if ((srcImage->getDims() == 2) && (region[2] != 1)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return NULL; + } + + amd::Coord3D srcOrigin(origin[0], origin[1], origin[2]); + amd::Coord3D srcRegion(region[0], region[1], region[2]); + if (!srcImage->validateRegion(srcOrigin, srcRegion)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return NULL; + } + + // Wait for possible pending operations + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + *not_null(errcode_ret) = err; + return (void*) 0; + } + + // Allocate a map command for the queue thread + amd::MapMemoryCommand *command = new amd::MapMemoryCommand( + hostQueue, + CL_COMMAND_MAP_IMAGE, + eventWaitList, + *srcImage, + map_flags, + blocking_map ? true : false, + srcOrigin, + srcRegion); + if (command == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return NULL; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + return NULL; + } + + if (srcImage->getMemFlags() & CL_MEM_USE_PERSISTENT_MEM_AMD) { + // [Windows VidMM restriction] + // Runtime can't map persistent memory if it's still busy or + // even wasn't submitted to HW from the worker thread yet + hostQueue.finish(); + } + + // Attempt to allocate the map target now (whether blocking or non-blocking) + void *mapPtr = hostQueue.device().allocMapTarget( + *srcImage, srcOrigin, srcRegion, image_row_pitch, image_slice_pitch); + if (NULL == mapPtr) { + delete command; + *not_null(errcode_ret) = CL_MAP_FAILURE; + return NULL; + } + + // Send the map command for processing + command->enqueue(); + + // A blocking map has to wait for completion + if (blocking_map) { + command->awaitCompletion(); + } + + // Save the command event if applicaiton has requested it + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + *not_null(errcode_ret) = CL_SUCCESS; + srcImage->incMapCount(); + return mapPtr; +} +RUNTIME_EXIT + +/*! \brief Enqueue a command to unmap a previously mapped region of a memory i + * object. + * + * Reads or writes from the host using the pointer returned by + * clEnqueueMapBuffer or clEnqueueMapImage are considered to be complete. + * + * \param command_queue must be a valid command-queue. + * + * \param memobj is a valid memory object. The OpenCL context associated with + * \a command_queue and \a memobj must be the same. + * + * \param mapped_ptr is the host address returned by a previous call to + * clEnqueueMapBuffer or clEnqueueMapImage for \a memobj. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifies events that need to complete before + * clEnqueueUnmapMemObject can be executed. If \a event_wait_list is NULL, + * then clEnqueueUnmapMemObject does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. clEnqueueBarrier can be used instead. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_INVALID_MEM_OBJECT if \a memobj is not a valid memory object. + * - CL_INVALID_VALUE if \a mapped_ptr is not a valid pointer returned by + * clEnqueueMapBuffer or clEnqueueMapImage for \a memobj. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or if \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * \a memobj are not the same. + * + * clEnqueueMapBuffer and clEnqueueMapImage increments the mapped count of the + * memory object. Multiple calls to clEnqueueMapBuffer or clEnqueueMapImage on + * the same memory object will increment this mapped count by appropriate number + * of calls. clEnqueueUnmapMemObject decrements the mapped count of the memory + * object. clEnqueueMapBuffer and clEnqueueMapImage act as synchronization + * points for a region of the memory object being mapped. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueUnmapMemObject, ( + cl_command_queue command_queue, + cl_mem memobj, + void *mapped_ptr, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Memory* amdMemory = as_amd(memobj); + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != amdMemory->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::UnmapMemoryCommand *command = new amd::UnmapMemoryCommand( + hostQueue, + CL_COMMAND_UNMAP_MEM_OBJECT, + eventWaitList, + *amdMemory, mapped_ptr); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + amdMemory->decMapCount(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_MemObjQuery + * @{ + */ + +/*! \brief Get information that is common to all memory objects (buffer and + * image objects) + * + * \param memobj specifies the memory object being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by \a param_value. If \a param_value_size_ret is NULL, it is + * ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type. + * - CL_INVALID_MEM_OBJECT if \a memobj is a not a valid memory object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetMemObjectInfo, ( + cl_mem memobj, + cl_mem_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + + switch (param_name) { + case CL_MEM_TYPE: { + cl_mem_object_type type = as_amd(memobj)->getType(); + return amd::clGetInfo( + type, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_FLAGS: { + cl_mem_flags flags = as_amd(memobj)->getMemFlags(); + return amd::clGetInfo( + flags, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_SIZE: { + size_t size = as_amd(memobj)->getSize(); + return amd::clGetInfo( + size, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_HOST_PTR: { + amd::Memory* memory = as_amd(memobj); + const void* hostPtr = (memory->getMemFlags() & CL_MEM_USE_HOST_PTR) ? + memory->getHostMem() : NULL; + return amd::clGetInfo( + hostPtr, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_MAP_COUNT: { + cl_uint count = as_amd(memobj)->mapCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_REFERENCE_COUNT: { + cl_uint count = as_amd(memobj)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_CONTEXT: { + cl_context context = as_cl(&as_amd(memobj)->getContext()); + return amd::clGetInfo( + context, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_ASSOCIATED_MEMOBJECT: { + amd::Memory* amdParent = as_amd(memobj)->parent(); + if ((NULL != amdParent) && (NULL != amdParent->getSvmPtr()) && + (NULL == amdParent->parent())) { + amdParent = NULL; + } + cl_mem parent = as_cl(amdParent); + return amd::clGetInfo( + parent, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_OFFSET: { + size_t mem_offset = as_amd(memobj)->getOrigin(); + return amd::clGetInfo( + mem_offset, param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_USES_SVM_POINTER: { + cl_bool usesSvmPointer = as_amd(memobj)->usesSvmPointer(); + return amd::clGetInfo( + usesSvmPointer, param_value_size, param_value, param_value_size_ret); + } +#ifdef _WIN32 + case CL_MEM_D3D10_RESOURCE_KHR: { + ID3D10Resource *pRes; + + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(interop) { + amd::D3D10Object *d3d10obj = interop->asD3D10Object(); + if(d3d10obj) { + pRes = d3d10obj->getD3D10ResOrig(); + if(!pRes) { + pRes = d3d10obj->getD3D10Resource(); + } + } + return amd::clGetInfo( + pRes, param_value_size, param_value, param_value_size_ret); + } + break; + } + case CL_MEM_D3D11_RESOURCE_KHR: { + ID3D11Resource *pRes; + + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(interop) { + amd::D3D11Object *d3d11obj = interop->asD3D11Object(); + if(d3d11obj) { + pRes = d3d11obj->getD3D11ResOrig(); + if(!pRes) { + pRes = d3d11obj->getD3D11Resource(); + } + } + return amd::clGetInfo( + pRes, param_value_size, param_value, param_value_size_ret); + } + break; + } + case CL_MEM_DX9_MEDIA_SURFACE_INFO_KHR: { + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(interop) { + amd::D3D9Object *d3d9obj = interop->asD3D9Object(); + if(d3d9obj) + return amd::clGetInfo(d3d9obj->getSurfInfo(), param_value_size, + param_value, param_value_size_ret); + else + return CL_INVALID_MEM_OBJECT; + } + else + return CL_INVALID_MEM_OBJECT; + break; + } + case CL_MEM_DX9_MEDIA_ADAPTER_TYPE_KHR: { + cl_dx9_media_adapter_type_khr adapterType; + + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(interop) { + amd::D3D9Object *d3d9obj = interop->asD3D9Object(); + if(d3d9obj) { + adapterType = d3d9obj->getAdapterType(); + } + return amd::clGetInfo( + adapterType, param_value_size, param_value, param_value_size_ret); + } + break; + } +#endif //_WIN32 + default: + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! \brief Get information specific to an image object. + * + * \param obj specifies the image object being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by \a param_value. If \a param_value_size_ret is NULL, it is + * ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL. + * - CL_INVALID_MEM_OBJECT if \a image is a not a valid image object. + * + * \version 1.2r09 + */ +RUNTIME_ENTRY(cl_int, clGetImageInfo, ( + cl_mem memobj, + cl_image_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + amd::Image* image = as_amd(memobj)->asImage(); + if (image == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + switch(param_name) + { + case CL_IMAGE_FORMAT: { + cl_image_format format = image->getImageFormat(); + return amd::clGetInfo( + format, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_ELEMENT_SIZE: { + size_t elementSize = image->getImageFormat().getElementSize(); + return amd::clGetInfo( + elementSize, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_ROW_PITCH: { + size_t rowPitch = image->getRowPitch(); + return amd::clGetInfo( + rowPitch, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_SLICE_PITCH: { + size_t slicePitch = image->getSlicePitch(); + return amd::clGetInfo( + slicePitch, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_WIDTH: { + size_t width = image->getWidth(); + return amd::clGetInfo( + width, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_HEIGHT: { + size_t height = image->getHeight(); + if ((image->getType() == CL_MEM_OBJECT_IMAGE1D) || + (image->getType() == CL_MEM_OBJECT_IMAGE1D_ARRAY) || + (image->getType() == CL_MEM_OBJECT_IMAGE1D_BUFFER)) { + height = 0; + } + return amd::clGetInfo( + height, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_DEPTH: { + size_t depth = image->getDepth(); + if ((image->getType() == CL_MEM_OBJECT_IMAGE1D_BUFFER) || + (image->getType() == CL_MEM_OBJECT_IMAGE1D_ARRAY) || + (image->getType() == CL_MEM_OBJECT_IMAGE2D_ARRAY) || + (image->getType() == CL_MEM_OBJECT_IMAGE1D) || + (image->getType() == CL_MEM_OBJECT_IMAGE2D)) { + depth = 0; + } + return amd::clGetInfo( + depth, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_ARRAY_SIZE: { + size_t arraySize = 0; + if (image->getType() == CL_MEM_OBJECT_IMAGE1D_ARRAY) { + arraySize = image->getHeight(); + } + else if (image->getType() == CL_MEM_OBJECT_IMAGE2D_ARRAY) { + arraySize = image->getDepth(); + } + return amd::clGetInfo( + arraySize, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_BUFFER: { + cl_mem buffer = 0; + amd::Memory* parent = image->parent(); + while (parent && (parent->asBuffer() == NULL)) { + parent = parent->parent(); + } + buffer = as_cl(parent); + return amd::clGetInfo( + buffer, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_NUM_MIP_LEVELS: { + cl_uint numMipLevels = 0; + return amd::clGetInfo( + numMipLevels, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_NUM_SAMPLES: { + cl_uint numSamples = 0; + return amd::clGetInfo( + numSamples, param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_BYTE_PITCH_AMD: { + size_t bytePitch = image->getBytePitch(); + return amd::clGetInfo( + bytePitch, param_value_size, param_value, param_value_size_ret); + } +#ifdef _WIN32 + case CL_IMAGE_D3D10_SUBRESOURCE_KHR: { + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(!interop) { + return CL_INVALID_MEM_OBJECT; + } + amd::D3D10Object *d3d10obj = interop->asD3D10Object(); + if(!d3d10obj) { + return CL_INVALID_MEM_OBJECT; + } + UINT subresource = d3d10obj->getSubresource(); + return amd::clGetInfo( subresource, + param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_D3D11_SUBRESOURCE_KHR: { + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(!interop) { + return CL_INVALID_MEM_OBJECT; + } + amd::D3D11Object *d3d11obj = interop->asD3D11Object(); + if(!d3d11obj) { + return CL_INVALID_MEM_OBJECT; + } + UINT subresource = d3d11obj->getSubresource(); + return amd::clGetInfo( subresource, + param_value_size, param_value, param_value_size_ret); + } + case CL_MEM_DX9_MEDIA_SURFACE_INFO_KHR: { + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(!interop) { + return CL_INVALID_MEM_OBJECT; + } + amd::D3D9Object *d3d9obj = interop->asD3D9Object(); + if(!d3d9obj) { + return CL_INVALID_MEM_OBJECT; + } + return amd::clGetInfo( d3d9obj->getSurfInfo(), + param_value_size, param_value, param_value_size_ret); + } + case CL_IMAGE_DX9_MEDIA_PLANE_KHR: { + amd::InteropObject* interop = ((amd::Memory*) as_amd(memobj))->getInteropObj(); + if(!interop) { + return CL_INVALID_MEM_OBJECT; + } + amd::D3D9Object *d3d9obj = interop->asD3D9Object(); + if(!d3d9obj) { + return CL_INVALID_MEM_OBJECT; + } + cl_uint plane = d3d9obj->getPlane(); + return amd::clGetInfo( plane, + param_value_size, param_value, param_value_size_ret); + } +#endif //_WIN32 + default: + break; + } + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! \brief creates a 1D image, 1D image buffer, 1D image array, 2D image, + * 2D image array and 3D image object + * + * \param context is a valid OpenCL context on which the image object is + * to be created. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information about the image memory object being created and is described + * in table 5.3. If value specified for flags is 0, the default is used which + * is CL_MEM_READ_WRITE. + * + * \param image_format is a pointer to a structure that describes format + * properties of the image to be allocated. Refer to section 5.3.1.1 for + * a detailed description of the image format descriptor. + * + * \param image_desc is a pointer to a structure that describes type and + * dimensions of the image to be allocated. Refer to section 5.3.1.2 for + * a detailed description of the image descriptor. + * + * \param host_ptr is a pointer to the image data that may already be + * allocated by the application. Refer to table below for a description of + * how large the buffer that host_ptr points to must be. + * CL_MEM_OBJECT_IMAGE1D >= image_row_pitch + * CL_MEM_OBJECT_IMAGE1D_BUFFER >= image_row_pitch + * CL_MEM_OBJECT_IMAGE2D >= image_row_pitch * image_height + * CL_MEM_OBJECT_IMAGE3D >= image_slice_pitch * image_depth + * CL_MEM_OBJECT_IMAGE1D_ARRAY >= image_slice_pitch * image_array_size + * CL_MEM_OBJECT_IMAGE2D_ARRAY >= image_slice_pitch * image_array_size + * For a 3D image or 2D image array, the image data specified by \a host_ptr + * is stored as a linear sequence of adjacent 2D image slices or 2D images + * respectively. Each 2D image is a linear sequence of adjacent scanlines. + * Each scanline is a linear sequence of image elements. + * For a 2D image array, the image data specified by \a host_ptr is stored + * as a linear sequence of adjacent scanlines. Each scanline is a linear + * sequence of image elements. + * For a 1D image array, the image data specified by \a host_ptr is stored + * as a linear sequence of adjacent 1D images respectively. Each 1D image + * or 1D image buffer is a single scanline which is a linear sequence of + * adjacent elements. + * + * \param errcode_ret will return an appropriate error code. + * If \a errcode_ret is NULL, no error code is returned. + * + * \return a valid non-zero image object created and the \a errcode_ret is + * set to CL_SUCCESS if the image object is created successfully. Otherwise, + * it returns a NULL value with one of the following error values + * returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if values specified in \a flags are not valid. + * - CL_INVALID_IMAGE_FORMAT_DESCRIPTOR if values specified in \a image_format + * are not valid or if \a image_format is NULL. + * - CL_INVALID_IMAGE_DESCRIPTOR if values specified in \a image_desc are + * not valid or if \a image_desc is NULL. + * - CL_INVALID_HOST_PTR if \a host_ptr in \a image_desc is NULL and + * CL_MEM_USE_HOST_PTR or CL_MEM_COPY_HOST_PTR are set in \a flags or + * if \a host_ptr is not NULL, but CL_MEM_COPY_HOST_PTR or + * CL_MEM_USE_HOST_PTR are not set in \a flags. + * - CL_INVALID_VALUE if a 1D image buffer is being created and + * the buffer object was created with CL_MEM_WRITE_ONLY and \a flags + * specifies CL_MEM_READ_WRITE or CL_MEM_READ_ONLY, or if the buffer object + * was created with CL_MEM_READ_ONLY and \a flags specifies + * CL_MEM_READ_WRITE or CL_MEM_WRITE_ONLY, or if \a flags specifies + * CL_MEM_USE_HOST_PTR or CL_MEM_ALLOC_HOST_PTR or CL_MEM_COPY_HOST_PTR. + * - CL_IMAGE_FORMAT_NOT_SUPPORTED if the image_format is not supported. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for image object. + * - CL_INVALID_OPERATION if there are no devices in \a context that support + * images + * - CL_DEVICE_IMAGE_SUPPORT specified in table 4.3 is CL_FALSE). + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreateImage, ( + cl_context context, + cl_mem_flags flags, + const cl_image_format *image_format, + const cl_image_desc *image_desc, + void* host_ptr, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter: context"); + return (cl_mem) 0; + } + // check flags for validity + if (!validateFlags(flags)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter: flags"); + return (cl_mem) 0; + } + // check format + if (image_format == NULL) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter: image_format"); + return (cl_mem) 0; + } + + const amd::Image::Format imageFormat(*image_format); + if (!imageFormat.isValid()) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter: image_format"); + return (cl_mem) 0; + } + + amd::Context& amdContext = *as_amd(context); + + if (!imageFormat.isSupported(amdContext, image_desc->image_type)) { + *not_null(errcode_ret) = CL_IMAGE_FORMAT_NOT_SUPPORTED; + LogWarning("invalid parameter: image_format"); + return (cl_mem) 0; + } + + // check host_ptr consistency + if (host_ptr == NULL) { + if(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter: host_ptr"); + return (cl_mem) 0; + } + } + else { + if(!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) { + *not_null(errcode_ret) = CL_INVALID_HOST_PTR; + LogWarning("invalid parameter: host_ptr"); + return (cl_mem) 0; + } + } + + const std::vector& devices = as_amd(context)->devices(); + std::vector::const_iterator it; + + bool supportPass = false; + for (it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().imageSupport_) { + supportPass = true; + break; + } + } + + if (!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support images"); + return (cl_mem) 0; + } + + if (!amd::Image::validateDimensions( + devices, image_desc->image_type, image_desc->image_width, + image_desc->image_height, image_desc->image_depth, image_desc->image_array_size)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_SIZE; + LogWarning("invalid parameter: image dimensions exceeding max"); + return (cl_mem) 0; + } + + size_t imageRowPitch = 0; + size_t imageSlicePitch = 0; + if (!validateImageDescriptor(devices, imageFormat, + image_desc, host_ptr, imageRowPitch, imageSlicePitch)) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_DESCRIPTOR; + LogWarning("invalid parameter: image_desc"); + return (cl_mem) 0; + } + + amd::Image* image = NULL; + + switch (image_desc->image_type) { + case CL_MEM_OBJECT_IMAGE1D: + image = new (amdContext) amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE1D, + flags, + imageFormat, + image_desc->image_width, + 1, + 1, + imageRowPitch, + 0); + break; + case CL_MEM_OBJECT_IMAGE2D: + if (image_desc->mem_object != NULL) { + amd::Buffer& buffer = *(as_amd(image_desc->mem_object)->asBuffer()); + if (&amdContext != &buffer.getContext()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter: context"); + return (cl_mem) 0; + } + + // host_ptr is not supported, the buffer object is used instead. + if ((flags & (CL_MEM_USE_HOST_PTR | + CL_MEM_ALLOC_HOST_PTR | + CL_MEM_COPY_HOST_PTR)) != 0) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter: flags"); + return (cl_mem) 0; + } + + cl_uint pitchAlignment = 0; + for (unsigned int i=0; iinfo().imagePitchAlignment_) { + pitchAlignment = devices[i]->info().imagePitchAlignment_; + } + } + if ((imageRowPitch % pitchAlignment) != 0) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter: flags"); + return (cl_mem) 0; + } + + image = new (amdContext) amd::Image( + buffer, + CL_MEM_OBJECT_IMAGE2D, + (flags != 0) ? flags : buffer.getMemFlags(), + imageFormat, + image_desc->image_width, + image_desc->image_height, + 1, + imageRowPitch, + imageSlicePitch); + } + else { + image = new (amdContext) amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE2D, + flags, + imageFormat, + image_desc->image_width, + image_desc->image_height, + 1, + imageRowPitch, + 0); + } + break; + case CL_MEM_OBJECT_IMAGE3D: + image = new (amdContext) amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE3D, + flags, + imageFormat, + image_desc->image_width, + image_desc->image_height, + image_desc->image_depth, + imageRowPitch, + imageSlicePitch); + break; + case CL_MEM_OBJECT_IMAGE1D_BUFFER: + { + amd::Buffer& buffer = *(as_amd(image_desc->mem_object)->asBuffer()); + if (&amdContext != &buffer.getContext()) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter: context"); + return (cl_mem) 0; + } + + // host_ptr is not supported, the buffer object is used instead. + if ((flags & (CL_MEM_USE_HOST_PTR | + CL_MEM_ALLOC_HOST_PTR | + CL_MEM_COPY_HOST_PTR)) != 0) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter: flags"); + return (cl_mem) 0; + } + + image = new (amdContext) amd::Image( + buffer, + CL_MEM_OBJECT_IMAGE1D_BUFFER, + (flags != 0) ? flags : buffer.getMemFlags(), + imageFormat, + image_desc->image_width, + 1, + 1, + imageRowPitch, + imageSlicePitch); + } + break; + case CL_MEM_OBJECT_IMAGE1D_ARRAY: + image = new (amdContext) amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE1D_ARRAY, + flags, + imageFormat, + image_desc->image_width, + image_desc->image_array_size, + 1, + imageRowPitch, + imageSlicePitch); + break; + case CL_MEM_OBJECT_IMAGE2D_ARRAY: + image = new (amdContext) amd::Image( + amdContext, + CL_MEM_OBJECT_IMAGE2D_ARRAY, + flags, + imageFormat, + image_desc->image_width, + image_desc->image_height, + image_desc->image_array_size, + imageRowPitch, + imageSlicePitch); + break; + default: { + *not_null(errcode_ret) = CL_INVALID_IMAGE_DESCRIPTOR; + LogWarning("invalid parameter: image_desc"); + return reinterpret_cast(image); + } + break; + } + + if (image == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("cannot allocate resources"); + return (cl_mem) 0; + } + + if(!image->create(host_ptr)) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + image->release(); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return (cl_mem) as_cl(image); +} +RUNTIME_EXIT + +/*! \brief Enqueues a command to fill a buffer object with + * a pattern of a given pattern size. + * + * \param command_queue refers to the command-queue in which + * the fill command will be queued. The OpenCL context associated with + * command_queue and buffer must be the same. + * + * \param buffer is a valid buffer object. + * + * \param pattern is a pointer to the data pattern of size pattern_size + * in bytes. pattern will be used to fill a region in buffer starting + * at offset and is cb bytes in size. The data pattern must be a scalar or + * vector integer or floating-point data type supported by OpenCL + * as described in sections 6.1.1 and 6.1.2. For example, if buffer is + * to be filled with a pattern of float4 values, then pattern will be + * a pointer to a cl_float4 value and pattern_size will be sizeof(cl_float4). + * The maximum value of pattern_size is the size of the largest integer or + * floating-point vector data type supported by the OpenCL device. + * + * \param offset is the location in bytes of the region being filled + * in buffer and must be a multiple of pattern_size. size is the size + * in bytes of region being filled in buffer and must be a multiple + * of pattern_size. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifes events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, + * then this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and a\ num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. + * The memory associated with \a event_wait_list can be reused or + * freed after the function returns. + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. clEnqueueBarrierWithWaitList can be used instead. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * \a buffer are not the same or if the \a context associated with + * \a command_queue and \a events in \a event_wait_list are not the same. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_INVALID_MEM_OBJECT if \a memobj is not a valid memory object. + * - CL_INVALID_VALUE if pattern is NULL or if pattern_size is 0 or if + * \a pattern_size is one of {1, 2, 4, 8, 16, 32, 64, 128}. + * - CL_INVALID_VALUE if \a offset or \a offset + \a size require accessing + * elements outside the \a buffer object respectively. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or if \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueFillBuffer, ( + cl_command_queue command_queue, + cl_mem buffer, + const void* pattern, + size_t pattern_size, + size_t offset, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + amd::Buffer* fillBuffer; + + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(buffer)) { + return CL_INVALID_MEM_OBJECT; + } + + fillBuffer = as_amd(buffer)->asBuffer(); + if (fillBuffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if ((pattern == NULL) || (pattern_size == 0) || + (pattern_size > amd::FillMemoryCommand::MaxFillPatterSize) || + ((pattern_size & (pattern_size - 1)) != 0)) { + return CL_INVALID_VALUE; + } + + // Offset must be a multiple of pattern_size + if ((offset % pattern_size) != 0) { + return CL_INVALID_VALUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != fillBuffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Coord3D fillOffset(offset, 0, 0); + amd::Coord3D fillSize(size, 1, 1); + if(!fillBuffer->validateRegion(fillOffset, fillSize)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::FillMemoryCommand *command = new amd::FillMemoryCommand( + hostQueue, + CL_COMMAND_FILL_BUFFER, + eventWaitList, + *fillBuffer, + pattern, + pattern_size, + fillOffset, + fillSize); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +static cl_uint +sRGBmap(float fc) +{ + double c = (double)fc; + +#ifdef ATI_OS_LINUX + if (isnan(c)) + c = 0.0; +#else + if (_isnan(c)) + c = 0.0; +#endif + + if (c > 1.0) + c = 1.0; + else if (c < 0.0) + c = 0.0; + else if (c < 0.0031308) + c = 12.92 * c; + else + c = (1055.0/1000.0) * pow(c, 5.0/12.0) - (55.0/1000.0); + + return (cl_uint)(c * 255.0 + 0.5); +} + +/*! \brief enqueues a command to fill an image object with + * a specified color. + * + * \param command_queue refers to the command-queue in which + * the fill command will be queued. The OpenCL context associated with + * command_queue and buffer must be the same. + * + * \param buffer is a valid buffer object. + * + * \param fill_color is the fill color. The fill color is a four + * component RGBA floating-point color value if the image channel data type + * is not an unnormalized signed and unsigned integer type, is a four + * component signed integer value if the image channel data type is + * an unnormalized signed integer type and is a four component unsigned + * integer value if the image channel data type is an unormalized + * unsigned integer type. The fill color will be converted to + * the appropriate image channel format and order associated with image + * as described in sections 6.11.13 and 8.3. + * + * \param origin defines the (x, y, z) offset in pixels in the image + * or (x, y) offset and the image index in the image array. If image is + * a 2D image object, origin[2] must be 0. If image is a 1D image or 1D + * image buffer object, origin[1] and origin[2] must be 0. If image is + * a 1D image array object, origin[2] must be 0. If image is a 1D image array + * object, origin[1] describes the image index in the 1D image array. + * If image is a 2D image array object, origin[2] describes the image index + * in the 2D image array. + * + * \param region defines the (width, height, depth) in pixels of + * the 1D, 2D or 3D rectangle or the (width, height) in pixels in pixels of + * the 1D or 2D rectangle and the image index of an image array. If image is + * a 2D image object, region[2] must be 1. If image is a 1D image or + * 1D image buffer object, region[1] and region[2] must be 1. If image is + * a 1D image array object, region[1] and region[2] must be 1. + * If image is a 2D image array object, region[2] must be 1. + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifes events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, + * then this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and a\ num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. + * The memory associated with \a event_wait_list can be reused or + * freed after the function returns. + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for + * the application to query the status of this command or queue a wait for this + * command to complete. clEnqueueBarrierWithWaitList can be used instead. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * \a buffer are not the same or if the \a context associated with + * \a command_queue and \a events in \a event_wait_list are not the same. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_INVALID_MEM_OBJECT if \a memobj is not a valid memory object. + * - CL_INVALID_VALUE if fill_color is NULL. + * - CL_INVALID_VALUE if the region being filled as specified by origin and + * region is out of bounds. + * - CL_INVALID_VALUE if values in origin and region do not follow rules + * described in the argument description for origin and region. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or if \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_INVALID_IMAGE_SIZE if image dimensions (image width, height, specified + * or compute row + * - CL_INVALID_IMAGE_FORMAT if image format (image channel order and data type) + * for image are not supported by device associated with queue. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clEnqueueFillImage, ( + cl_command_queue command_queue, + cl_mem image, + const void* fill_color, + const size_t* origin, + const size_t* region, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event)) +{ + amd::Image* fillImage; + + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(image)) { + return CL_INVALID_MEM_OBJECT; + } + + if (fill_color == NULL) { + return CL_INVALID_VALUE; + } + + fillImage = as_amd(image)->asImage(); + if (fillImage == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != fillImage->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Coord3D fillOrigin(origin[0], origin[1], origin[2]); + amd::Coord3D fillRegion(region[0], region[1], region[2]); + if(!fillImage->validateRegion(fillOrigin, fillRegion)) { + return CL_INVALID_VALUE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + void *new_fill_color = const_cast(fill_color); + cl_uint4 i_fill_color; + cl_float4 f_fill_color; + + // Convert int to float for sRGBA because hw is not supporting this conversion yet. + if (fillImage->getImageFormat().image_channel_order == CL_sRGBA) { + float *f_color = (float *)(new_fill_color); + if (hostQueue.device().info().type_ == CL_DEVICE_TYPE_CPU) { + f_fill_color.s[0] = sRGBmap(f_color[0]) / 255.0; + f_fill_color.s[1] = sRGBmap(f_color[1]) / 255.0; + f_fill_color.s[2] = sRGBmap(f_color[2]) / 255.0; + f_fill_color.s[3] = f_color[3]; + new_fill_color = (void *)&f_fill_color; + } + else { + i_fill_color.s[0] = sRGBmap(f_color[0]); + i_fill_color.s[1] = sRGBmap(f_color[1]); + i_fill_color.s[2] = sRGBmap(f_color[2]); + i_fill_color.s[3] = (cl_uint)(f_color[3]*255.0); + new_fill_color = (void *)&i_fill_color; + } + } + + amd::FillMemoryCommand *command = new amd::FillMemoryCommand( + hostQueue, + CL_COMMAND_FILL_IMAGE, + eventWaitList, + *fillImage, + const_cast(new_fill_color), + sizeof(cl_float4), // @note color size is always 16 bytes value + fillOrigin, + fillRegion); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueues a command to indicate which device a set of memory objects + * should be associated with. Typically, memory objects are implicitly + * migrated to a device for which enqueued commands, using the memory object, + * are targeted. \a clEnqueueMigrateMemObjects allows this migration to be + * explicitly performed ahead of the dependent commands. This allows a user to + * preemptively change the association of a memory object, through regular + * command queue scheduling, in order to prepare for another upcoming + * command. This also permits an application to overlap the placement of + * memory objects with other unrelated operations before these memory objects + * are needed potentially hiding transfer latencies. Once the event, returned + * from \a clEnqueueMigrateMemObjects, has been marked \a CL_COMPLETE + * the memory objects specified in \a mem_objects have been successfully + * migrated to the device associated with \a command_queue. The migrated memory + * object shall remain resident on the device until another command is enqueued + * that either implicitly or explicitly migrates it away. + * \a clEnqueueMigrateMemObjects can also be used to direct the initial + * placement of a memory object, after creation, possibly avoiding the initial + * overhead of instantiating the object on the first enqueued command to use it. + * The user is responsible for managing the event dependencies, associated with + * this command, in order to avoid overlapping access to memory objects. + * Improperly specified event dependencies passed to + * \a clEnqueueMigrateMemObjects could result in undefined results. + * + * \param command_queue is a valid command-queue. The specified set of memory + * objects in \a mem_objects will be migrated to the OpenCL device associated + * with \a command_queue or to the host if the \a CL_MIGRATE_MEM_OBJECT_HOST + * has been specified. + * + * \param num_mem_objects is the number of memory objects specified in + * \a mem_objects. \a mem_objects is a pointer to a list of memory objects. + * + * \param flags is a bit-field that is used to specify migration options. + * The following table describes the possible values for flags. + * cl_mem_migration flags Description + * CL_MIGRATE_MEM_OBJECT_HOST This flag indicates that the specified set + * of memory objects are to be migrated to the + * host, regardless of the target command-queue. + * CL_MIGRATE_MEM_OBJECT_ This flag indicates that the contents of the set + * CONTENT_UNDEFINED of memory objects are undefined after migration. + * The specified set of memory objects are migrated + * to the device associated with \a command_queue + * without incurring + * + * \param num_events_in_wait_list specifies the number of event objects in + * \a event_wait_list. + * + * \param event_wait_list specifes events that need to complete before this + * particular command can be executed. If \a event_wait_list is NULL, + * then this particular command does not wait on any event to complete. + * If \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. + * If \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and a\ num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. + * The memory associated with \a event_wait_list can be reused or + * freed after the function returns. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid command-queue + * - CL_INVALID_CONTEXT if the context associated with \a command_queue + * and memory objects in \a mem_objects are not the same or if the context + * associated with \a command_queue and events in \a event_wait_list + * are not the same. + * - CL_INVALID_MEM_OBJECT if any of the memory objects in \a mem_objects + * is not a valid memory object. + * - CL_INVALID_VALUE if \a num_mem_objects is zero or + * if \a mem_objects is NULL. + * - CL_INVALID_VALUE if flags is not 0 or any of the values described + * in the table above + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or if \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate + * memory for the specified set of memory objects in \a mem_objects. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the OpenCL implementation on the host. + * + * \version 1.2r15 + */ +RUNTIME_ENTRY(cl_int, clEnqueueMigrateMemObjects, ( + cl_command_queue command_queue, + cl_uint num_mem_objects, + const cl_mem* mem_objects, + cl_mem_migration_flags flags, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if ((num_mem_objects == 0) || (mem_objects == NULL)) { + return CL_INVALID_VALUE; + } + + if (flags & ~(CL_MIGRATE_MEM_OBJECT_HOST | + CL_MIGRATE_MEM_OBJECT_CONTENT_UNDEFINED)) { + return CL_INVALID_VALUE; + } + + std::vector memObjects; + for (uint i = 0; i < num_mem_objects; ++i) { + if (!is_valid(mem_objects[i])) { + return CL_INVALID_MEM_OBJECT; + } + amd::Memory* memory = as_amd(mem_objects[i]); + if (hostQueue.context() != memory->getContext()) { + return CL_INVALID_CONTEXT; + } + memObjects.push_back(memory); + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::MigrateMemObjectsCommand* command = new amd::MigrateMemObjectsCommand( + hostQueue, + CL_COMMAND_MIGRATE_MEM_OBJECTS, + eventWaitList, + memObjects, + flags); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +RUNTIME_ENTRY_RET(cl_mem, clConvertImageAMD, ( + cl_context context, + cl_mem image, + const cl_image_format * image_format, + cl_int * errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter: context"); + return (cl_mem) 0; + } + // check format + if (image_format == NULL) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter: image_format"); + return (cl_mem) 0; + } + const amd::Image::Format imageFormat(*image_format); + if (!imageFormat.isValid()) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("invalid parameter: image_format"); + return (cl_mem) 0; + } + + amd::Context& amdContext = *as_amd(context); + if (!imageFormat.isSupported(amdContext)) { + *not_null(errcode_ret) = CL_IMAGE_FORMAT_NOT_SUPPORTED; + LogWarning("invalid parameter: image_format"); + return (cl_mem) 0; + } + amd::Image* amdImage = as_amd(image)->asImage(); + amd::Image* converted_image = amdImage->createView(amdContext, imageFormat, NULL); + + if (converted_image == NULL) { + *not_null(errcode_ret) = CL_INVALID_IMAGE_FORMAT_DESCRIPTOR; + LogWarning("cannot allocate resources"); + return (cl_mem) 0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return (cl_mem) as_cl(converted_image); +} +RUNTIME_EXIT + +RUNTIME_ENTRY_RET(cl_mem, clCreateBufferFromImageAMD, ( + cl_context context, + cl_mem image, + cl_int * errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter: context"); + return (cl_mem) 0; + } + + amd::Context& amdContext = *as_amd(context); + if (amdContext.devices().size() > 1) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter: context"); + return (cl_mem) 0; + } + + const std::vector& devices = amdContext.devices(); + std::vector::const_iterator it; + + bool supportPass = false; + for (it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().bufferFromImageSupport_) { + supportPass = true; + break; + } + } + + if (!supportPass) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("there are no devices in context to support buffer from image"); + return (cl_mem) 0; + } + + amd::Image* amdImage = as_amd(image)->asImage(); + if (!is_valid(image) || amdImage == NULL) { + *not_null(errcode_ret) = CL_INVALID_MEM_OBJECT; + return NULL; + } + + amd::Memory* mem = new(amdContext) amd::Buffer(*amdImage, 0, 0, amdImage->getSize()); + if (mem == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem)0; + } + + if (!mem->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + mem->release(); + return NULL; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return (cl_mem) as_cl(mem); +} +RUNTIME_EXIT + +/*! @} + * @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_object.cpp b/opencl/api/opencl/amdocl/cl_object.cpp new file mode 100644 index 0000000000..dd505fa144 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_object.cpp @@ -0,0 +1,72 @@ +// +// Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +RUNTIME_ENTRY_RET(cl_key_amd, clCreateKeyAMD, ( + cl_platform_id platform, + void (CL_CALLBACK * destructor)( void* ), + cl_int *errcode_ret)) +{ + cl_key_amd key = amd::ObjectMetadata::createKey(destructor); + + *not_null(errcode_ret) = amd::ObjectMetadata::check(key) + ? CL_SUCCESS : CL_OUT_OF_RESOURCES; + + return key; +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clObjectGetValueForKeyAMD, ( + void * object, + cl_key_amd key, + void ** ret_val)) +{ + if (ret_val == NULL) { + return CL_INVALID_VALUE; + } + *ret_val = NULL; + + if (!amd::RuntimeObject::isValidHandle(object)) { + return CL_INVALID_OBJECT_AMD; + } + if (!amd::ObjectMetadata::check(key)) { + return CL_INVALID_KEY_AMD; + } + + amd::ObjectMetadata& metadata = + amd::RuntimeObject::fromHandle(object)->metadata(); + + void* value = metadata.getValueForKey(key); + if (value == NULL) { + return CL_INVALID_KEY_AMD; + } + + *ret_val = value; + return CL_SUCCESS; +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clObjectSetValueForKeyAMD, ( + void * object, + cl_key_amd key, + void * value)) +{ + if (!amd::RuntimeObject::isValidHandle(object)) { + return CL_INVALID_OBJECT_AMD; + } + if (!amd::ObjectMetadata::check(key)) { + return CL_INVALID_KEY_AMD; + } + if (value == NULL) { + return CL_INVALID_VALUE; + } + + amd::ObjectMetadata& metadata = + amd::RuntimeObject::fromHandle(object)->metadata(); + + metadata.setValueForKey(key, value); + return CL_SUCCESS; +} +RUNTIME_EXIT diff --git a/opencl/api/opencl/amdocl/cl_open_video.cpp b/opencl/api/opencl/amdocl/cl_open_video.cpp new file mode 100644 index 0000000000..f27e2c3535 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_open_video.cpp @@ -0,0 +1,829 @@ +// +// Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" +#include "platform/command.hpp" +#include + +#if cl_amd_open_video + +#include "platform/video_session.hpp" + +#include "cal.h" +#include "calcl.h" + +amd::HostQueue* staticQueue; //@todo fix the interface and remove hardcoded queue +/* Helper function GetCalDecodeProfile() +*/ +static bool +GetCalVideoProfile(cl_video_decode_profile_amd clVideoProfile, + CALdecodeProfile& calVideoProfile) +{ + switch(clVideoProfile) { + case CL_VIDEO_DECODE_PROFILE_H264_BASELINE_AMD: + calVideoProfile = CAL_VID_H264_BASELINE; + break; + case CL_VIDEO_DECODE_PROFILE_H264_MAIN_AMD: + calVideoProfile = CAL_VID_H264_MAIN; + break; + case CL_VIDEO_DECODE_PROFILE_H264_HIGH_AMD: + calVideoProfile = CAL_VID_H264_HIGH; + break; + case CL_VIDEO_DECODE_PROFILE_VC1_SIMPLE_AMD: + calVideoProfile = CAL_VID_VC1_SIMPLE; + break; + case CL_VIDEO_DECODE_PROFILE_VC1_MAIN_AMD: + calVideoProfile = CAL_VID_VC1_MAIN; + break; + case CL_VIDEO_DECODE_PROFILE_VC1_ADVANCED_AMD: + calVideoProfile = CAL_VID_VC1_ADVANCED; + break; + case CL_VIDEO_DECODE_PROFILE_MPEG2_VLD_AMD: + calVideoProfile = CAL_VID_MPEG2_VLD; + break; + //case CL_VIDEO_DECODE_PROFILE_MPEG2_SIMPLE_AMD: + //case CL_VIDEO_DECODE_PROFILE_MPEG2_MAIN_AMD: + default: + return false; + break; + } + return true; +} +/* Helper function GetCalDecodeProfile() +*/ +static bool +GetCalVideoFormat(cl_video_format_amd clVideoFormat, + CALdecodeFormat& calVideoFormat) +{ + switch(clVideoFormat) { + case CL_VIDEO_NV12_INTERLEAVED_AMD: + calVideoFormat = CAL_VID_NV12_INTERLEAVED; + break; + case CL_VIDEO_YV12_INTERLEAVED_AMD: + calVideoFormat = CAL_VID_YV12_INTERLEAVED; + break; + default: + return false; + } + return true; +} +/*! \addtogroup API + * @{ + * + * \addtogroup cl_amd_open_video + * + * This section provides OpenCL extension functions that allow applications + * to use kernel decoding of video streams using corresponding HW capabilities. + * + * @} + * \addtogroup clQueryVidDecoderCapsAMD + * @{ + */ + + +/*! \brief Creates a video session object. + * + * \param context is a valid OpenCL context on which the image object + * is to be created. + * + * \param device must be a device associated with context. It can either be in + * the list of devices specified when context is created using + * clCreateContext or have the same device type as the device type + * specified when the context is created using clCreateContextFromType + * + * \param flags is a bit-field that is used to specify the video session + * operation mode, available flags are as follows: + * - CL_VIDEO_DECODE_ACCELERATION_AMD - the video session supports HW + * acceleration of video decoding operation. + * + * \param errcode_ret will return CL_SUCCESS if no error occurs; + * or an appropriate error code: + * - CL_INVALID_CONTEXT if \a context is not a valid context; + * - CL_INVALID_DEVICE if \a device is not a valid device; + * - CL_INVALID_VIDEO_SESSION_FLAGS_AMD if flags is NULL or contains invalid + * configuration; + * - CL_INVALID_OPERATION if video decode extension is not supported by + * any device associated with context; + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.1r25 + */ +RUNTIME_ENTRY_RET(cl_video_session_amd, clCreateVideoSessionAMD, ( + cl_context context, + cl_device_id device, + cl_video_session_flags_amd flags, + cl_video_config_type_amd config_buffer_type, + cl_uint config_buffer_size, + void* config_buffer, + cl_int* errcode_ret)) +{ + *not_null(errcode_ret) = CL_SUCCESS; + + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return (cl_video_session_amd) 0; + } + if (!is_valid(device)) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + LogWarning("invalid parameter \"device\""); + return (cl_video_session_amd) 0; + } + amd::Device* amdDevice = as_amd(device); + if (!amdDevice->info().openVideo_) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("Device or CAL does not support Open Video extension"); + return (cl_video_session_amd) 0; + } + + // Create video command queue + cl_video_encode_desc_amd ovSessionProperties; + CALvideoProperties calVideoProperties; + calVideoProperties.size = sizeof(calVideoProperties); + calVideoProperties.flags = static_cast(flags); + switch(config_buffer_type) { + case CL_VIDEO_DECODE_CONFIGURATION_AMD: + { + cl_video_decode_desc_amd* videoDecodeDesc = + static_cast(config_buffer); + if (!GetCalVideoProfile(videoDecodeDesc->attrib.profile, + calVideoProperties.profile)) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("Profile is not supported or invalid"); + return (cl_video_session_amd) 0; + } + if (!GetCalVideoFormat(videoDecodeDesc->attrib.format, + calVideoProperties.format)) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("Format is not supported or invalid"); + return (cl_video_session_amd) 0; + } + calVideoProperties.width = videoDecodeDesc->image_width; + calVideoProperties.height = videoDecodeDesc->image_height; + calVideoProperties.VideoEngine_name = CAL_CONTEXT_VIDEO; + } + break; + default: + *not_null(errcode_ret) = CL_INVALID_VIDEO_CONFIG_TYPE_AMD; + LogWarning("invalid parameter \"config_buffer_type\""); + return (cl_video_session_amd) 0; + break; + } + + ovSessionProperties.calVideoProperties = &calVideoProperties; + + amd::HostQueue* queue = new amd::HostQueue( + *as_amd(context), *as_amd(device), 0, &ovSessionProperties); + + if (queue == NULL || !queue->create()) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("not enough host memory"); + return (cl_video_session_amd) 0; + } + + staticQueue = queue; + amd::VideoSession* video_session = new amd::VideoSession( + *as_amd(context), *as_amd(device), queue, flags, config_buffer_type, + config_buffer_size, config_buffer); + if (!video_session) { + queue->release(); + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("not enough host memory"); + return (cl_video_session_amd) 0; + } + + return as_cl(video_session); +} +RUNTIME_EXIT + + +RUNTIME_ENTRY_RET(cl_video_session_amd, clCreateVideoEncSessionAMD, ( + cl_context context, + cl_device_id device, + cl_video_session_flags_amd flags, + cl_video_config_type_amd config_buffer_type, + cl_uint config_buffer_size, + void* config_buffer, + cl_int* errcode_ret)) +{ + *not_null(errcode_ret) = CL_SUCCESS; + + // Make sure the context is valid + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return (cl_video_session_amd) 0; + } + + // Make sure the device is valid + if (!is_valid(device)) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + LogWarning("invalid parameter \"device\""); + return (cl_video_session_amd) 0; + } + + // Make sure the device supports Open Video extensions + amd::Device* amdDevice = as_amd(device); + if (!amdDevice->info().openVideo_) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + LogWarning("Device or CAL does not support Open Video extension"); + return (cl_video_session_amd) 0; + } + + cl_video_encode_desc_amd *ovSessionProperties = (cl_video_encode_desc_amd *)config_buffer; + + // Create video command queue + CALvideoProperties calVideoProperties; + + switch(config_buffer_type) + { + case CL_VIDEO_ENCODE_CONFIGURATION_AMD: + calVideoProperties.size = sizeof(calVideoProperties); + calVideoProperties.flags = static_cast(flags); + calVideoProperties.profile = (CALdecodeProfile)ovSessionProperties->attrib.codec_profile; + calVideoProperties.format = (CALdecodeFormat)ovSessionProperties->attrib.format; + calVideoProperties.width = ovSessionProperties->image_width; + calVideoProperties.height = ovSessionProperties->image_height; + calVideoProperties.VideoEngine_name = CAL_CONTEXT_VIDEO_VCE; + ovSessionProperties->calVideoProperties = &calVideoProperties; + break; + default: + *not_null(errcode_ret) = CL_INVALID_VIDEO_CONFIG_TYPE_AMD; + LogWarning("invalid parameter \"config_buffer_type\""); + return (cl_video_session_amd) 0; + break; + } + + amd::HostQueue* queue = new amd::HostQueue( + *as_amd(context), *as_amd(device), 0, ovSessionProperties); + + if (queue == NULL || !queue->create()) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("not enough host memory"); + return (cl_video_session_amd) 0; + } + staticQueue = queue; +#if 0 // this doesn't work here due to context value... + CAL_VID_PROFILE_LEVEL encode_profile_level; + encode_profile_level.profile = buffer->attrib.profile; + encode_profile_level.level = buffer->attrib.level; + CALuint frameRate = 0; + CALuint YUVwidth = buffer->image_width; + CALuint YUVhigh = buffer->image_height; + + cal::details::extensions_.extEncodeCreateSession_((CALcontext)as_amd(context), 0, CAL_VID_encode_AVC_FULL, encode_profile_level, + CAL_VID_PICTURE_NV12, YUVwidth, YUVhigh, frameRate, CAL_VID_ENCODE_JOB_PRIORITY_LEVEL1); +#endif + + amd::VideoSession* video_session = new amd::VideoSession( + *as_amd(context), *as_amd(device), queue, flags, config_buffer_type, + config_buffer_size, config_buffer); + if (!video_session) + { + queue->release(); + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("not enough host memory"); + return (cl_video_session_amd) 0; + } + + return as_cl(video_session); +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clDestroyVideoEncSessionAMD, + (cl_video_session_amd video_session)) +{ + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + + cl_int errcode_ret = CL_SUCCESS; + amd::VideoSession* session = as_amd(video_session); + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + session->context(), 0, NULL); + amd::Command* command = new amd::SetVideoSessionCommand( + session->queue(), eventWaitList, + amd::SetVideoSessionCommand::CloseSession, NULL); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + command->release(); + + session->queue().finish(); + session->release(); + + return errcode_ret; +} +RUNTIME_EXIT + +/*! \brief Increments the video session reference count. + * + * \param video_session is a valid OpenCL Video Session. + * + * \return CL_SUCCESS if the function executes successfully; + * Otherwise an error code is returned: + * - CL_INVALID_VIDEO_SESSION_AMD if video_session is not a valid + * video session; + * + * \version 1.1r25 + */ +RUNTIME_ENTRY(cl_int, clRetainVideoSessionAMD, + (cl_video_session_amd video_session)) +{ + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + as_amd(video_session)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrements the video session reference count. + * + * \param video_session is a valid OpenCL Video Session. + * + * \return CL_SUCCESS if the function executes successfully; + * Otherwise an error code is returned: + * - CL_INVALID_VIDEO_SESSION_AMD if video_session is not a valid + * video session; + * + * \version 1.1r25 + */ +RUNTIME_ENTRY(cl_int, clReleaseVideoSessionAMD, + (cl_video_session_amd video_session)) +{ + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + as_amd(video_session)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Queries the configuration of a specific video session. + * + * \param video_session is a valid OpenCL Video Session. + * + * \param config_buffer_type is an enumeration constant that identifies the type + * of the configuration buffer. It can be one of the following values: + * - CL_VIDEO_DECODE_CONFIGURATION_AMD - The buffer contains decoder configuration, + * according to the cl_video_decode_desc_amd structure definition. + * + * \param config_buffer_size is a pointer to size of config_buffer in bytes. + * The implementation returns the size of the config_buffer in this argument. + * + * \param config_buffer is a pointer to a configuration buffer. The + * implementation writes the configuration query data into this + * buffer. If this pointer is NULL, it is ignored (which enables the + * application to query the required buffer size for this configuration type). + * + * \returns CL_SUCCESS if the video session configuration was set properly; + * Otherwise an appropriate error code is returned: + * - CL_INVALID_VIDEO_SESSION_AMD if video_session is not a valid video session; + * - CL_INVALID_VIDEO_CONFIG_TYPE_AMD if config_buffer_type is NULL or contains a + * configuration type which is not supported by the specific implementation; + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.1r25 + */ +RUNTIME_ENTRY(cl_int, clGetVideoSessionInfoAMD, ( + cl_video_session_amd video_session, + cl_video_session_info_amd param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret)) +{ + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + //switch(config_buffer_type) { + //case CL_VIDEO_DECODE_CONFIGURATION_AMD: + // break; + //default: + // return CL_INVALID_VIDEO_CONFIG_TYPE_AMD; + //} +/* + return amd::clGetVideoSessionInfoAMD(*as_amd(video_session), + config_buffer_type, config_buffer_size, config_buffer); +*/ + return CL_INVALID_VALUE; +// return CL_SUCCESS; +} +RUNTIME_EXIT + + +/*! \brief Queries the configuration of a specific video session. + * + * \param video_session is a valid OpenCL Video Session. + * + * \param config_buffer_type is an enumeration constant that identifies the type + * of the configuration buffer. It can be one of the following values: + * - CL_VIDEO_ENCODE_CONFIGURATION_AMD - The buffer contains encoder configuration, + * according to the cl_video_decode_desc_amd structure definition. + * + * \param config_buffer_size is a pointer to size of config_buffer in bytes. + * The implementation returns the size of the config_buffer in this argument. + * + * \param config_buffer is a pointer to a configuration buffer. The + * implementation writes the configuration query data into this + * buffer. If this pointer is NULL, it is ignored (which enables the + * application to query the required buffer size for this configuration type). + * + * \returns CL_SUCCESS if the video session configuration was set properly; + * Otherwise an appropriate error code is returned: + * - CL_INVALID_VIDEO_SESSION_AMD if video_session is not a valid video session; + * - CL_INVALID_VIDEO_CONFIG_TYPE_AMD if config_buffer_type is NULL or contains a + * configuration type which is not supported by the specific implementation; + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.1r25 + */ +RUNTIME_ENTRY(cl_int, clGetVideoSessionEncInfoAMD, ( + cl_video_session_amd video_session, + cl_video_session_enc_info_amd param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret)) +{ + cl_uint errcode_ret = CL_SUCCESS; + + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + + amd::VideoSession& session = *as_amd(video_session); + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, session.context(), 0, NULL); + amd::Command* command = NULL; + + switch(param_name) { + case CL_CONFIG_TYPE_PICTURE_CONTROL : + command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::ConfigTypePictureControl, param_value); + break; + case CL_CONFIG_TYPE_RATE_CONTROL : + command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::ConfigTypeRateControl, param_value); + break; + case CL_CONFIG_TYPE_MOTION_ESTIMATION : + command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::ConfigTypeMotionEstimation, param_value); + break; + case CL_CONFIG_TYPE_RDO : + command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::ConfigTypeRDO, param_value); + break; + default: + errcode_ret = CL_INVALID_VIDEO_CONFIG_TYPE_AMD; + break; + } + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + command->release(); + + session.queue().finish(); + * param_value_size_ret = sizeof(param_value); + + return errcode_ret; +} +RUNTIME_EXIT + + +RUNTIME_ENTRY(cl_int, clSendEncodeConfigInfoAMD, ( + cl_video_session_amd video_session, + size_t numBuffers, + void* pConfigBuffers)) +{ + cl_uint errcode_ret = CL_SUCCESS; + + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + + amd::VideoSession& session = *as_amd(video_session); + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, session.context(), 0, NULL); + amd::Command* command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::SendEncodeConfig, pConfigBuffers, numBuffers); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + command->release(); + + session.queue().finish(); + + return errcode_ret; +} +RUNTIME_EXIT + +/*! \brief Enqueues execution of a decode/encode command on UVD unit of the device + * specified at creation of video session. + * + * \param video_session is a valid OpenCL video session. + * + * \param video_data_structure is a valid pointer to a structure + * containing control and video stream data for decoding/encoding. + * + * \param config_buffer_size is a pointer to size of config_buffer in bytes. + * The implementation returns the size of the config_buffer in this argument. + * + * \param event_wait_list and \param num_events_in_wait_list specify events + * that need to complete before this particular command can be executed. + * + * \param event returns an event object that identifies this particular + * video program execution instance. + * + * \returns CL_SUCCESS if the video program execution was + * successfully passed to the CAL for execution and CAL returned CAL_RESULT_OK. + * Otherwise, it returnes one of the following errors: + * + * - CL_INVALID_VIDEO_SESSION_AMD if the parameter video_session is not + * a valid video session object returned by a call to function + * clCreateVideoSessionAMD; + * - CL_IVALID_VIDEO_DATA_AMD if pointer video_data_struct is NULL, points + * to invalid memory, or any of the pointers/objects provided in the + * video_data_structure is NULL or invalid; + * - CL_INVALID_CONTEXT if context associated with video_session is not + * the same as context associated with event_wait_list; + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and + * num_events_in_wait_list > 0, or event_wait_list is not NULL and + * num_events_in_wait_list is 0, or if event objects in event_wait_list + * are not valid events; + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources + * required by the Open CL implementation on the host. + * + * \version 1.1r33 + */ +RUNTIME_ENTRY(cl_int, clEnqueueRunVideoProgramAMD, ( + cl_video_session_amd video_session, + void* video_data, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + cl_int errcode = CL_SUCCESS; + + amd::VideoSession& session = *as_amd(video_session); + + switch(session.type()) { + case CL_VIDEO_DECODE_CONFIGURATION_AMD: + errcode = amd::clEnqueueVideoDecodeAMD(session, + (cl_video_decode_data_amd*) video_data, + num_events_in_wait_list, event_wait_list, event); + break; + case CL_VIDEO_ENCODE_CONFIGURATION_AMD: + errcode = amd::clEnqueueVideoEncodeAMD(session, + (cl_video_encode_data_amd*) video_data, + num_events_in_wait_list, event_wait_list, event); + break; + + default: + errcode = CL_INVALID_VIDEO_CONFIG_TYPE_AMD; + } + return errcode; +} +RUNTIME_EXIT + +RUNTIME_ENTRY(cl_int, clEncodeGetDeviceCapAMD, ( + cl_device_id device_id, + cl_uint encode_mode, + cl_uint encode_cap_total_size, + cl_uint *num_encode_cap, + void *pEncodeCAP)) +{ + // Make sure the device supports Open Video extensions + amd::Device* device = as_amd(device_id); + if (!device->info().openVideo_) { + return CL_DEVICE_NOT_FOUND; + } + +// amd::VideoSession& session = *as_amd(video_session); + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, staticQueue->context(), 0, NULL); + amd::Command* command = new amd::SetVideoSessionCommand( + *staticQueue, eventWaitList, + amd::SetVideoSessionCommand::GetDeviceCapVCE, pEncodeCAP, encode_cap_total_size); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + command->release(); + staticQueue->finish(); + + return CL_SUCCESS; +} +RUNTIME_EXIT + +#if 1 +RUNTIME_ENTRY(cl_int, clEncodePictureAMD, ( + cl_video_session_amd video_session, + cl_uint number_of_encode_task_input_buffers, + void* encode_task_input_buffer_list, + void* picture_parameter, + cl_uint* pTaskID)) + +{ + cl_uint errcode_ret = CL_SUCCESS; + + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + + return errcode_ret; +} +RUNTIME_EXIT +#endif + +RUNTIME_ENTRY(cl_int, clEncodeQueryTaskDescriptionAMD, ( + cl_video_session_amd video_session, + cl_uint num_of_task_description_request, + cl_uint* num_of_task_description_return, + void * task_description_list)) +{ + cl_uint errcode_ret = CL_SUCCESS; + + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + + amd::VideoSession& session = *as_amd(video_session); + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, session.context(), 0, NULL); + amd::Command* command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::EncodeQueryTaskDescription, + num_of_task_description_request, task_description_list, num_of_task_description_return); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + command->release(); + session.queue().finish(); + + return errcode_ret; +} +RUNTIME_EXIT + + +RUNTIME_ENTRY(cl_int, clEncodeReleaseOutputResourceAMD, ( + cl_video_session_amd video_session, + cl_uint task_id)) +{ + cl_uint errcode_ret = CL_SUCCESS; + + if (!is_valid(video_session)) { + LogWarning("invalid parameter \"video_session\""); + return CL_INVALID_VIDEO_SESSION_AMD; + } + + amd::VideoSession& session = *as_amd(video_session); + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, session.context(), 0, NULL); + amd::Command* command = new amd::SetVideoSessionCommand( + session.queue(), eventWaitList, + amd::SetVideoSessionCommand::ReleaseOutputResource, + NULL, task_id); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + command->release(); + session.queue().finish(); + + return errcode_ret; +} +RUNTIME_EXIT + +namespace amd { + +cl_int +clEnqueueVideoDecodeAMD(VideoSession& session, + cl_video_decode_data_amd* data, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event) +{ + if ((NULL == data) || (data->video_type.type != CL_VIDEO_DECODE)) { + return CL_INVALID_OPERATION; + } + + if (!is_valid(data->output_surface)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Command::EventWaitList eventWaitList; + cl_int errcode = amd::clSetEventWaitList(eventWaitList, + session.context(), num_events_in_wait_list, event_wait_list); + if (errcode != CL_SUCCESS){ + return errcode; + } + + //! Now create command and enqueue + amd::RunVideoProgramCommand* command = + new amd::RunVideoProgramCommand(session.queue(), + eventWaitList, data, *as_amd(data->output_surface), + CL_COMMAND_VIDEO_DECODE_AMD); + if (command == NULL) { + LogError("Cannot create new RunVideoProgramCommand"); + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} + +cl_int +clEnqueueVideoEncodeAMD(VideoSession& session, + cl_video_encode_data_amd* data, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event) +{ + if ((NULL == data) || (data->video_type.type != CL_VIDEO_ENCODE)) { + return CL_INVALID_OPERATION; + } + + CAL_VID_BUFFER_DESCRIPTION *bufferList = + reinterpret_cast(data->pictureParam1); + cl_mem memory = static_cast(bufferList[0].buffer.pPicture); + + if (!is_valid(memory)) { + return CL_INVALID_MEM_OBJECT; + } + + // Start up the event queue + amd::Command::EventWaitList eventWaitList; + cl_int errcode = amd::clSetEventWaitList(eventWaitList, + session.context(), num_events_in_wait_list, event_wait_list); + if (errcode != CL_SUCCESS){ + return errcode; + } + + //! Now create command and enqueue + amd::RunVideoProgramCommand* command = + new amd::RunVideoProgramCommand(session.queue(), + eventWaitList, data, *as_amd(memory), + CL_COMMAND_VIDEO_ENCODE_AMD); + if (command == NULL) { + LogError("Cannot create new RunVideoProgramCommand"); + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_MEM_OBJECT_ALLOCATION_FAILURE; + } + + // Issue the command via the queue system. + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} + + +} // namespace amd + +#endif // cl_amd_video_session diff --git a/opencl/api/opencl/amdocl/cl_open_video_amd.h b/opencl/api/opencl/amdocl/cl_open_video_amd.h new file mode 100644 index 0000000000..9dfa4878e5 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_open_video_amd.h @@ -0,0 +1,362 @@ +/* ============================================================ + +Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance + +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +#ifndef __CL_OPEN_VIDEO_AMD_H +#define __CL_OPEN_VIDEO_AMD_H + +/****************************************** +* Private AMD extension cl_amd_open_video * +******************************************/ +#define cl_amd_open_video 1 + +#if cl_amd_open_video + +#ifdef __cplusplus +extern "C" { +#endif /*__cplusplus*/ + +/* cl_device_info*/ +#define CL_DEVICE_MAX_VIDEO_SESSIONS_AMD 0x10010 +#define CL_DEVICE_NUM_VIDEO_ATTRIBS_AMD 0x10011 +#define CL_DEVICE_VIDEO_ATTRIBS_AMD 0x10012 + +#define CL_DEVICE_VIDEO_ATTRIBS_ENC_AMD 0x10014 +#define CL_DEVICE_MAX_VIDEO_ENC_SESSIONS_AMD 0x10015 +#define CL_DEVICE_NUM_VIDEO_ENC_ATTRIBS_AMD 0x10016 + +/* cl_video_session_flags_amd */ +typedef cl_bitfield cl_video_session_flags_amd; + +/* cl_video_config_type_amd enumeration */ +// + // Open Encode Porting TBD check if the 0x4101 enum. that will result is OK + // + +/* cl_video_config_type_amd enumeration */ +typedef enum _cl_video_config_type_amd { + CL_VIDEO_DECODE_CONFIGURATION_AMD = 0x4100, + CL_VIDEO_ENCODE_CONFIGURATION_AMD = 0x4101 +} cl_video_config_type_amd; + +/* cl_video_session_info_amd enumeration */ +typedef enum _cl_video_session_info_amd { + CL_VIDEO_SESSION_CONFIG_SIZE_AMD = 0x10013, + CL_VIDEO_SESSION_CONFIG_TYPE_AMD, + CL_VIDEO_SESSION_CONFIG_BUDDEF_AMD +} cl_video_session_info_amd; + +typedef enum _cl_video_session_enc_info_amd { + CL_CONFIG_TYPE_NONE = 0, + CL_CONFIG_TYPE_PICTURE_CONTROL = 1, + CL_CONFIG_TYPE_RATE_CONTROL = 2, + CL_CONFIG_TYPE_MOTION_ESTIMATION = 3, + CL_CONFIG_TYPE_RDO = 4 +} cl_video_session_enc_info_amd; + +/* cl_amd_open_video typedefs */ +typedef struct _cl_video_session_amd* cl_video_session_amd; + +//typedef cl_bitfield cl_video_session_flags_amd; + +/* cl_video_decode_profile_amd enumeration */ +typedef enum _cl_video_decode_profile_amd +{ + CL_VIDEO_DECODE_PROFILE_H264_BASELINE_AMD = 1, + CL_VIDEO_DECODE_PROFILE_H264_MAIN_AMD, + CL_VIDEO_DECODE_PROFILE_H264_HIGH_AMD, + CL_VIDEO_DECODE_PROFILE_VC1_SIMPLE_AMD, + CL_VIDEO_DECODE_PROFILE_VC1_MAIN_AMD, + CL_VIDEO_DECODE_PROFILE_VC1_ADVANCED_AMD, + CL_VIDEO_DECODE_PROFILE_MPEG2_VLD_AMD, +} cl_video_decode_profile_amd; + +// +// Open Encode ported for encode +// +/* cl_video_encode_profile_amd enumeration */ +typedef enum _cl_video_encode_profile_amd +{ + CL_VIDEO_ENCODE_PROFILE_H264_BASELINE_AMD = 1, + CL_VIDEO_ENCODE_PROFILE_H264_MAIN_AMD, + CL_VIDEO_ENCODE_PROFILE_H264_HIGH_AMD, + +} cl_video_encode_profile_amd; + + +/* cl_video_format_amd enumeration */ +typedef enum _cl_video_format_amd +{ + CL_VIDEO_NV12_INTERLEAVED_AMD = 1, + CL_VIDEO_YV12_INTERLEAVED_AMD +} cl_video_format_amd; + +/* cl_video_attrib_amd */ +typedef struct _cl_video_attrib_amd { + cl_video_decode_profile_amd profile; + cl_video_format_amd format; +} cl_video_attrib_amd; + +/* cl_video_decode_desc_amd */ +typedef struct _cl_video_decode_desc_amd { + cl_video_attrib_amd attrib; + size_t image_width; + size_t image_height; +} cl_video_decode_desc_amd; + + +// +// Open Encode ported +// + +/* cl_video_attrib_amd */ +typedef struct _cl_video_attrib_encode_amd { + cl_video_encode_profile_amd codec_profile; + cl_uint profile; + cl_uint level; + cl_video_format_amd format; +} cl_video_attrib_encode_amd; + +/* cl_video_encode_desc_amd */ +typedef struct _cl_video_encode_desc_amd { + cl_video_attrib_encode_amd attrib; + size_t image_width; + size_t image_height; + cl_uint priority; + cl_uint encodeMode; + cl_uint frameRateNumerator; + cl_uint frameRateDenominator; + void * calVideoProperties; +} cl_video_encode_desc_amd; + +/* cl_video_program_type_amd enumeration*/ +typedef enum _cl_video_program_type_amd +{ + CL_VIDEO_DECODE = 1, + CL_VIDEO_ENCODE = 2, +} cl_video_program_type_amd; + +/* cl_video_program_data_amd */ +typedef struct _cl_video_program_data_amd { + cl_uint size; + cl_video_program_type_amd type; + cl_uint flags; +} cl_video_program_data_amd; + +/* cl_video_decode_data_amd */ +typedef struct _cl_video_decode_data_amd { + cl_video_program_data_amd video_type; + cl_mem output_surface; + void* picture_parameter_1; + void* picture_parameter_2; + cl_uint picture_parameter_2_size; + void* bitstream_data; + cl_uint bitstream_data_size; + void* slice_data_control; + cl_uint slice_data_control_size; +} cl_video_decode_data_amd; + +/* cl_video_encode_data_amd */ +typedef struct _cl_video_encode_data_amd { + cl_video_program_data_amd video_type; + cl_uint pictureParam1Size; + cl_uint pictureParam2Size; + void* pictureParam1; + void* pictureParam2; + cl_uint uiTaskID; +} cl_video_encode_data_amd; + +/* cl_command_type */ +#define CL_COMMAND_VIDEO_DECODE_AMD 0x4101 +#define CL_COMMAND_VIDEO_ENCODE_AMD 0x4102 +#define CL_COMMAND_VIDEO_QUERY_TASK_AMD 0x4103 + +/* cl_amd_open_video error codes */ +#define CL_INVALID_VIDEO_SESSION_FLAGS_AMD -1100 +#define CL_INVALID_VIDEO_SESSION_AMD -1101 +#define CL_INVALID_VIDEO_CONFIG_TYPE_AMD -1102 +#define CL_INVALID_VIDEO_CONFIG_BUFFER_AMD -1103 +#define CL_INVALID_VIDEO_DATA_AMD -1104 + +/* cl_video_session_flags_amd - bitfield */ +#define CL_VIDEO_DECODE_ACCELERATION_AMD (1<<0) +// Open Encode ported +#define CL_VIDEO_ENCODE_ACCELERATION_AMD (1<<0) + +// /* cl_video_session_info_amd enumeration */ +//typedef enum _cl_video_session_info_amd { +// CL_VIDEO_SESSION_CONFIG_SIZE_AMD = 0x10012, +// CL_VIDEO_SESSION_CONFIG_TYPE_AMD, +// CL_VIDEO_SESSION_CONFIG_BUDDEF_AMD +//} cl_video_session_info_amd; + +typedef CL_API_ENTRY cl_video_session_amd (CL_API_CALL* clCreateVideoSessionAMD_fn)( + cl_context /* context */, + cl_device_id /* device */, + cl_video_session_flags_amd /* flags */, + cl_video_config_type_amd /* config_buffer_type */, + cl_uint /* config_buffer_size */, + void* /* config_buffer */, + cl_int* /* errcode_ret */); + +// Open Encode Changes add Enc CreateSession +typedef CL_API_ENTRY cl_video_session_amd (CL_API_CALL* clCreateVideoEncSessionAMD_fn)( + cl_context /* context */, + cl_device_id /* device */, + cl_video_session_flags_amd /* flags */, + cl_video_config_type_amd /* config_buffer_type */, + cl_uint /* config_buffer_size */, + void* /* config_buffer */, + cl_int* /* errcode_ret */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clDestroyVideoEncSessionAMD_fn)( + cl_video_session_amd /*video_session*/); + +// Open Encode Changes + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clRetainVideoSessionAMD_fn)( + cl_video_session_amd /* video_session */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clReleaseVideoSessionAMD_fn)( + cl_video_session_amd /* video_session*/); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clGetVideoSessionInfoAMD_fn)( + cl_video_session_amd /* video_session */, + cl_video_session_info_amd /* param_name */, + size_t /* param_value_size */, + void* /* param_value */, + size_t* /* param_value_size_ret */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clGetVideoSessionEncInfoAMD_fn)( + cl_video_session_amd /* video_session */, + cl_video_session_enc_info_amd /* param_name */, + size_t /* param_value_size */, + void* /* param_value */, + size_t* /* param_value_size_ret */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clSendEncodeConfigInfoAMD_fn) ( + cl_video_session_amd /* video_session */, + size_t /* numBuffers */, + void* /* pConfigBuffers */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clEnqueueRunVideoProgramAMD_fn)( + cl_video_session_amd /* video_session */, + void* /* video_data_struct */, + cl_uint /* num_events_in_wait_list */, + const cl_event* /* event_wait_list */, + cl_event* /* event*/); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clEncodeGetDeviceCapAMD_fn)( + cl_device_id /* device_id */, + cl_uint /* encode_mode */, + cl_uint /* encode_cap_total_size */, + cl_uint* /* num_encode_cap */, + void* /* pEncodeCAP */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clEncodeQueryTaskDescriptionAMD_fn)( + cl_video_session_amd /* video_session */, + cl_uint /* num_of_task_description_request */, + cl_uint* /* num_of_task_description_return */, + void * /* task_description_list */ + ); + +typedef CL_API_ENTRY cl_int (CL_API_CALL* clEncodeReleaseOutputResourceAMD_fn)( + cl_video_session_amd /* video_session */, + cl_uint /* task_id */ + ); + +#ifdef __cplusplus +} +#endif /*__cplusplus*/ + +#endif /*cl_amd_open_video*/ + +#endif /*!__CL_OPEN_VIDEO_AMD_H*/ diff --git a/opencl/api/opencl/amdocl/cl_pipe.cpp b/opencl/api/opencl/amdocl/cl_pipe.cpp new file mode 100644 index 0000000000..efb9f6089f --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_pipe.cpp @@ -0,0 +1,187 @@ +// +// Copyright (c) 2013 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" +#include "platform/memory.hpp" +#include "platform/context.hpp" +#include "platform/command.hpp" + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_Pipes + * @{ + */ + +/*! \brief creates a pipe object. + * + * \param context is a valid OpenCL context used to create the pipe object. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information such as the memory arena that should be used to allocate the pipe + * object and how it will be used. Only CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY, + * CL_MEM_READ_WRITE and CL_MEM_HOST_NO_ACCESS can be specified when creating a + * pipe object. If value specified for flags is 0, the default is used which is + * CL_MEM_READ_WRITE. + * + * \param pipe_packet_size is the size in bytes of a pipe packet. + * + * \param pipe_max_packets specifies the pipe capacity by specifying the maximum + * number of packets the pipe can hold. + * + * \param properties specifies a list of properties for the pipe and their + * corresponding values. Each property name is immediately followed by the + * corresponding desired value. The list is terminated with 0. + * + * In OpenCL 2.0, properties must be NULL. + * + * \param errcode_ret will return an appropriate error code. + * If \a errcode_ret is NULL, no error code is returned. + * + * \return a valid non-zero pipe object and \a errcode_ret is set to CL_SUCCESS + * if the pipe object is created successfully. Otherwise, it returns a NULL + * value with one of the following error values returned in errcode_ret: + * - CL_INVALID_CONTEXT if context is not a valid context. + * - CL_INVALID_VALUE if values specified in flags are not as defined above. + * - CL_INVALID_VALUE if properties is not NULL. + * - CL_INVALID_PIPE_SIZE if pipe_packet_size is 0 or the pipe_packet_size + * exceeds CL_DEVICE_PIPE_MAX_PACKET_SIZE value for all devices in context + * or if pipe_max_packets is 0. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory + * for the pipe object. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r19 + */ +RUNTIME_ENTRY_RET(cl_mem, clCreatePipe, ( + cl_context context, + cl_mem_flags flags, + cl_uint pipe_packet_size, + cl_uint pipe_max_packets, + const cl_pipe_properties *properties, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return NULL; + } + + // check flags for validity + cl_bitfield temp = flags + & (CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY | CL_MEM_READ_ONLY | CL_MEM_HOST_NO_ACCESS); + + if(temp + && !(CL_MEM_READ_WRITE == temp + || CL_MEM_WRITE_ONLY == temp + || CL_MEM_READ_ONLY == temp + || CL_MEM_HOST_NO_ACCESS == temp)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid parameter \"flags\""); + return (cl_mem) 0; + } + + size_t size = sizeof(struct clk_pipe_t) + pipe_packet_size * pipe_max_packets; + + const std::vector& devices = as_amd(context)->devices(); + std::vector::const_iterator it; + bool sizePass = false; + for (it = devices.begin(); it != devices.end(); ++it) { + if (((*it)->info().maxMemAllocSize_ >= size)) { + sizePass = true; + break; + } + } + + // check size + if (pipe_packet_size == 0 || pipe_max_packets == 0 || !sizePass ) { + *not_null(errcode_ret) = CL_INVALID_PIPE_SIZE; + LogWarning("invalid parameter \"size = 0 or size > CL_DEVICE_PIPE_MAX_PACKET_SIZE\""); + return (cl_mem)0; + } + + amd::Context& amdContext = *as_amd(context); + amd::Memory* mem = new(amdContext) amd::Pipe(amdContext, flags, size, (size_t)pipe_packet_size, (size_t)pipe_max_packets); + if (mem == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_mem)0; + } + + if (!mem->create()) { + *not_null(errcode_ret) = CL_MEM_OBJECT_ALLOCATION_FAILURE; + mem->release(); + return NULL; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(mem); +} +RUNTIME_EXIT + +/*! \brief Get information specific to a pipe object created with clCreatePipe. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by param_value. If param_value_size_ret is NULL, it is ignored. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise, it + * returns one of the following errors: + * - CL_INVALID_VALUE if param_name is not valid, or if size in bytes specified + * by param_value_size is < size of return type and param_value is not NULL. + * - CL_INVALID_MEM_OBJECT if pipe is a not a valid pipe object. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r19 + */ +RUNTIME_ENTRY(cl_int, clGetPipeInfo, ( + cl_mem memobj, + cl_image_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(memobj)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Pipe* pipe = as_amd(memobj)->asPipe(); + if (pipe == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + switch (param_name) { + case CL_PIPE_PACKET_SIZE: { + cl_uint packetSize = pipe->getPacketSize(); + return amd::clGetInfo( + packetSize, param_value_size, param_value, param_value_size_ret); + } + case CL_PIPE_MAX_PACKETS: { + cl_uint count = pipe->getMaxNumPackets(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + + return CL_INVALID_VALUE; + +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_platform_amd.cpp b/opencl/api/opencl/amdocl/cl_platform_amd.cpp new file mode 100644 index 0000000000..66b2685ac7 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_platform_amd.cpp @@ -0,0 +1,28 @@ +// +// Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. +// +#include "cl_common.hpp" +#include "cl_platform_amd.h" +#include + +/*! \addtogroup API + * @{ + * + * \addtogroup AMD_Extensions + * @{ + * + */ + +RUNTIME_ENTRY(cl_int, clUnloadPlatformAMD, (cl_platform_id platform)) +{ + if (AMD_PLATFORM == platform) { + amd::Runtime::tearDown(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_platform_amd.h b/opencl/api/opencl/amdocl/cl_platform_amd.h new file mode 100644 index 0000000000..94929fa9a3 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_platform_amd.h @@ -0,0 +1,117 @@ +/* ============================================================ + +Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance + +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +// AMD-specific platform management extensions + +#ifndef __CL_PLATFORM_AMD_H +#define __CL_PLATFORM_AMD_H + +#include "CL/cl_platform.h" + +#ifdef __cplusplus +extern "C" { +#endif /*__cplusplus*/ + +/*! \brief Unloads the specified platform, handling all required cleanup. + * + * @todo This is still somewhat of a stub. It only works for the AMD + * platform and just forces shutdown of all devices (to get PM4 + * capture working). It should handle ICD unregistration as well. + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clUnloadPlatformAMD(cl_platform_id platform) CL_API_SUFFIX__VERSION_1_0; + +#ifdef __cplusplus +} /*extern "C"*/ +#endif /*__cplusplus*/ + +#endif /*__CL_AMD_PROFILE_H*/ diff --git a/opencl/api/opencl/amdocl/cl_profile_amd.cpp b/opencl/api/opencl/amdocl/cl_profile_amd.cpp new file mode 100644 index 0000000000..c8db627da6 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_profile_amd.cpp @@ -0,0 +1,382 @@ +// +// Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. +// +#include "cl_common.hpp" +#include "cl_profile_amd.h" +#include "platform/context.hpp" +#include "platform/command.hpp" +#include "platform/perfctr.hpp" +#include + +/*! \addtogroup API + * @{ + * + * \addtogroup AMD_Extensions + * @{ + * + */ + +/*! \brief Creates a new HW performance counter + * for the specified OpenCL context. + * + * \param device must be a valid OpenCL device. + * + * \param block_index index of the HW block to configure. + * + * \param counter_index index of the hardware counter + * within the block to configure. + * + * \param event_index Event you wish to count with + * the counter specified by block_index + counter_index + * + * \param perf_counter the created perfcounter object + * + * \param errcode_ret A non zero value if OpenCL failed to create PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_DEVICE if the specified context is invalid. + * - CL_INVALID_OPERATION if we couldn't create the object + * + * \return Created perfcounter object + */ +RUNTIME_ENTRY_RET(cl_perfcounter_amd, clCreatePerfCounterAMD, ( + cl_device_id device, + cl_perfcounter_property* properties, + cl_int* errcode_ret)) +{ + // Make sure we have a valid device object + if (!is_valid(device)) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + return NULL; + } + + // Make sure we have a valid pointer to the performance counter properties + if (NULL == properties) { + return NULL; + } + + amd::PerfCounter::Properties perfProperties; + size_t size = 0; + while (properties[size] != CL_PERFCOUNTER_NONE) { + if (properties[size] < CL_PERFCOUNTER_LAST) { + perfProperties[properties[size]] = + static_cast(properties[size+1]); + size += 2; + } + else { + return NULL; + } + } + + // Create the device perf counter + amd::PerfCounter* perfCounter = + new amd::PerfCounter(*as_amd(device), perfProperties); + + if (perfCounter == NULL) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(perfCounter); +} +RUNTIME_EXIT + +/*! \brief Destroy a performance counter object. + * + * \param perf_counter the perfcounter object for release + * + * \return A non zero value if OpenCL failed to release PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to release the object + */ +RUNTIME_ENTRY(cl_int, clReleasePerfCounterAMD, ( + cl_perfcounter_amd perf_counter)) +{ + if (!is_valid(perf_counter)) { + return CL_INVALID_OPERATION; + } + as_amd(perf_counter)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Increments the perfcounter object reference count. + * + * \param perf_counter the perfcounter object for retain + * + * \return A non zero value if OpenCL failed to retain PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to release the object + */ +RUNTIME_ENTRY(cl_int, clRetainPerfCounterAMD, ( + cl_perfcounter_amd perf_counter)) +{ + if (!is_valid(perf_counter)) { + return CL_INVALID_OPERATION; + } + as_amd(perf_counter)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueues the begin command for the specified counters. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param num_perf_counters the number of perfcounter objects in the array. + * + * \param perf_counters specifies an array of perfcounter objects. + * + * \param event_wait_list specify [is a pointer to] events that need to + * complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait + * on any event to complete. If \a event_wait_list is NULL, + * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, + * the list of events pointed to by \a event_wait_list must be valid and + * \a num_events_in_wait_list must be greater than 0. The events specified in + * \a event_wait_list act as synchronization points. + * + * \param num_events_in_wait_list specify the number of events in + * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be + * greater than 0 if \a event_wait_list is not NULL. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. + * + * \return A non zero value if OpenCL failed to release PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to enqueue the begin operation + * - CL_INVALID_COMMAND_QUEUE if the queue is + */ +RUNTIME_ENTRY(cl_int, clEnqueueBeginPerfCounterAMD, ( + cl_command_queue command_queue, + cl_uint num_perf_counters, + cl_perfcounter_amd* perf_counters, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if ((num_perf_counters == 0) || (perf_counters == NULL)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::PerfCounterCommand::PerfCounterList counters; + + // Place all counters into the list + for (cl_uint i = 0; i < num_perf_counters; ++i) { + amd::PerfCounter* amdPerf = as_amd(perf_counters[i]); + if (&hostQueue->device() == &amdPerf->device()) { + counters.push_back(amdPerf); + } + else { + return CL_INVALID_DEVICE; + } + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue->context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + // Create a new command for the performance counters + amd::PerfCounterCommand* command = + new amd::PerfCounterCommand(*hostQueue, eventWaitList, counters, + amd::PerfCounterCommand::Begin); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Submit the command to the device + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Enqueues the end command for the specified counters. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param num_perf_counters the number of perfcounter objects in the array. + * + * \param perf_counters specifies an array of perfcounter objects. + * + * \param event_wait_list specify [is a pointer to] events that need to + * complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait + * on any event to complete. If \a event_wait_list is NULL, + * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, + * the list of events pointed to by \a event_wait_list must be valid and + * \a num_events_in_wait_list must be greater than 0. The events specified in + * \a event_wait_list act as synchronization points. + * + * \param num_events_in_wait_list specify the number of events in + * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be + * greater than 0 if \a event_wait_list is not NULL. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. + * + * \return A non zero value if OpenCL failed to release PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to enqueue the end operation + */ +RUNTIME_ENTRY(cl_int, clEnqueueEndPerfCounterAMD, ( + cl_command_queue command_queue, + cl_uint num_perf_counters, + cl_perfcounter_amd* perf_counters, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if ((num_perf_counters == 0) || (perf_counters == NULL)) { + return CL_INVALID_OPERATION; + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + amd::PerfCounterCommand::PerfCounterList counters; + + // Place all counters into the list + for (cl_uint i = 0; i < num_perf_counters; ++i) { + amd::PerfCounter* amdPerf = as_amd(perf_counters[i]); + if (&hostQueue->device() == &amdPerf->device()) { + counters.push_back(amdPerf); + } + else { + return CL_INVALID_DEVICE; + } + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue->context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + // Create a new command for the performance counters + amd::PerfCounterCommand* command = + new amd::PerfCounterCommand(*hostQueue, eventWaitList, counters, + amd::PerfCounterCommand::End); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Submit the command to the device + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Retrieves the results from the counter objects. + * + * \param num_perf_counter the perfcounter object for the information query. + * + * \param perf_counters specifies an array of perfcounter objects. + * + * \param wait_event specifies the wait event, returned in + * the clEnqueueEndPerfCounterAMD. + * + * \param wait true if OpenCL should wait for the perfcounter data. + * + * \param values must be a valid pointer to an array of 64-bit values + * and the array size must be equal to num_perf_counters. + * + * \return + * - CL_SUCCESS if the function is executed successfully. + * - CL_PROFILING_INFO_NOT_AVAILABLE if event isn't finished. + * - CL_INVALID_OPERATION if we failed to get the data + */ +RUNTIME_ENTRY(cl_int, clGetPerfCounterInfoAMD, ( + cl_perfcounter_amd perf_counter, + cl_perfcounter_info param_name, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret)) +{ + // Check if we have a valid performance counter + if (!is_valid(perf_counter)) { + return CL_INVALID_OPERATION; + } + + // Find the kernel, associated with the specified device + const device::PerfCounter* devCounter = + as_amd(perf_counter)->getDeviceCounter(); + + // Make sure we found a valid performance counter + if (devCounter == NULL) { + return CL_INVALID_OPERATION; + } + + // Get the corresponded parameters + switch (param_name) { + case CL_PERFCOUNTER_REFERENCE_COUNT: { + cl_uint count = as_amd(perf_counter)->referenceCount(); + // Return the reference counter + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_PERFCOUNTER_GPU_BLOCK_INDEX: + case CL_PERFCOUNTER_GPU_COUNTER_INDEX: + case CL_PERFCOUNTER_GPU_EVENT_INDEX: { + cl_ulong data = devCounter->getInfo(param_name); + // Return the device performance counter information + return amd::clGetInfo(data, + param_value_size, param_value, param_value_size_ret); + } + case CL_PERFCOUNTER_DATA: { + cl_ulong data = devCounter->getInfo(param_name); + if (static_cast(0xffffffffffffffffULL) == data) { + return CL_PROFILING_INFO_NOT_AVAILABLE; + } + // Return the device performance counter result + return amd::clGetInfo(data, + param_value_size, param_value, param_value_size_ret); + } + default: + return CL_INVALID_VALUE; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_profile_amd.h b/opencl/api/opencl/amdocl/cl_profile_amd.h new file mode 100644 index 0000000000..1c6e281e60 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_profile_amd.h @@ -0,0 +1,247 @@ +/* ============================================================ + +Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance + +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +#ifndef __CL_PROFILE_AMD_H +#define __CL_PROFILE_AMD_H + +#include "CL/cl_platform.h" + +#ifdef __cplusplus +extern "C" { +#endif /*__cplusplus*/ + +typedef struct _cl_perfcounter_amd * cl_perfcounter_amd; +typedef cl_ulong cl_perfcounter_property; +typedef cl_uint cl_perfcounter_info; + +/* cl_perfcounter_info */ +enum PerfcounterInfo +{ + CL_PERFCOUNTER_NONE = 0x0, + CL_PERFCOUNTER_REFERENCE_COUNT = 0x1, + CL_PERFCOUNTER_DATA = 0x2, + CL_PERFCOUNTER_GPU_BLOCK_INDEX = 0x3, + CL_PERFCOUNTER_GPU_COUNTER_INDEX = 0x4, + CL_PERFCOUNTER_GPU_EVENT_INDEX = 0x5, + CL_PERFCOUNTER_LAST +}; + +/*! \brief Creates a new HW performance counter + * for the specified OpenCL context. + * + * \param device must be a valid OpenCL device. + * + * \param properties the list of properties of the hardware counter + * + * \param errcode_ret A non zero value if OpenCL failed to create PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_CONTEXT if the specified context is invalid. + * - CL_OUT_OF_RESOURCES if we couldn't create the object + * + * \return the created perfcounter object + */ +extern CL_API_ENTRY cl_perfcounter_amd CL_API_CALL +clCreatePerfCounterAMD( + cl_device_id /* device */, + cl_perfcounter_property* /* properties */, + cl_int* /* errcode_ret */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Destroy a performance counter object. + * + * \param perf_counter the perfcounter object for release + * + * \return A non zero value if OpenCL failed to release PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to release the object + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clReleasePerfCounterAMD( + cl_perfcounter_amd /* perf_counter */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Increments the perfcounter object reference count. + * + * \param perf_counter the perfcounter object for retain + * + * \return A non zero value if OpenCL failed to retain PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to release the object + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainPerfCounterAMD( + cl_perfcounter_amd /* perf_counter */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Enqueues the begin command for the specified counters. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param num_perf_counters the number of perfcounter objects in the array. + * + * \param perf_counters specifies an array of perfcounter objects. + * + * \return A non zero value if OpenCL failed to release PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to enqueue the begin operation + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueBeginPerfCounterAMD( + cl_command_queue /* command_queue */, + cl_uint /* num_perf_counters */, + cl_perfcounter_amd* /* perf_counters */, + cl_uint /* num_events_in_wait_list */, + const cl_event* /* event_wait_list */, + cl_event* /* event */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Enqueues the end command for the specified counters. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param num_perf_counters the number of perfcounter objects in the array. + * + * \param perf_counters specifies an array of perfcounter objects. + * + * \param event the event object associated with the end operation. + * + * \return A non zero value if OpenCL failed to release PerfCounter + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_OPERATION if we failed to enqueue the end operation + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueEndPerfCounterAMD( + cl_command_queue /* command_queue */, + cl_uint /* num_perf_counters */, + cl_perfcounter_amd* /* perf_counters */, + cl_uint /* num_events_in_wait_list */, + const cl_event* /* event_wait_list */, + cl_event* /* event */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Retrieves the results from the counter objects. + * + * \param perf_counter specifies a perfcounter objects for query. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \param values must be a valid pointer to an array of 64-bit values + * and the array size must be equal to num_perf_counters. + * + * \return + * - CL_SUCCESS if the function is executed successfully. + * - CL_PROFILING_INFO_NOT_AVAILABLE if event isn't finished. + * - CL_INVALID_OPERATION if we failed to get the data + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetPerfCounterInfoAMD( + cl_perfcounter_amd /* perf_counter */, + cl_perfcounter_info /* param_name */, + size_t /* param_value_size */, + void* /* param_value */, + size_t* /* param_value_size_ret */ +) CL_API_SUFFIX__VERSION_1_0; + +#ifdef __cplusplus +} /*extern "C"*/ +#endif /*__cplusplus*/ + +#endif /*__CL_PROFILE_AMD_H*/ diff --git a/opencl/api/opencl/amdocl/cl_program.cpp b/opencl/api/opencl/amdocl/cl_program.cpp new file mode 100644 index 0000000000..313e18ab2a --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_program.cpp @@ -0,0 +1,1895 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/context.hpp" +#include "platform/program.hpp" +#include "platform/kernel.hpp" +#include "platform/sampler.hpp" +#include "cl_semaphore_amd.h" + +#include + +static amd::Program* createProgram( + cl_context context, + cl_uint num_devices, + const cl_device_id *device_list, + cl_int *errcode_ret) +{ + // Create the program + amd::Program* program = new amd::Program(*as_amd(context)); + if (program == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return NULL; + } + + // Add programs for all devices in the context. + if (device_list == NULL) { + const std::vector& devices = as_amd(context)->devices(); + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + if (program->addDeviceProgram(**it) == CL_OUT_OF_HOST_MEMORY) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + program->release(); + return NULL; + } + } + return program; + } + + *not_null(errcode_ret) = CL_SUCCESS; + for (cl_uint i = 0; i < num_devices; ++i) { + cl_device_id device = device_list[i]; + + if (!is_valid(device) + || !as_amd(context)->containsDevice(as_amd(device))) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + program->release(); + return NULL; + } + + cl_int status = program->addDeviceProgram(*as_amd(device)); + if (status == CL_OUT_OF_HOST_MEMORY) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + program->release(); + return NULL; + } + } + return program; +} + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_Programs + * + * An OpenCL program consists of a set of kernels that are identified as + * functions declared with the __kernel qualifier in the program source. + * OpenCL programs may also contain auxiliary functions and constant data that + * can be used by __kernel functions. The program executable can be generated + * online or offline by the OpenCL compiler for the appropriate + * target device(s). + * + * @{ + * + * \addtogroup CL_CreatingPrograms + * @{ + */ + +/*! \brief Create a program object for a context, and loads the source code + * specified by the text strings in the strings array into the program object. + * + * \param context must be a valid OpenCL context. + * + * \param count is the number of pointers in \a strings + * + * \param strings is an array of \a count pointers to optionally + * null-terminated character strings that make up the source code. + * + * \param lengths is an array with the number of chars in each string (the + * string length). If an element in lengths is zero, its accompanying string + * is null-terminated. If lengths is NULL, all strings in the strings argument + * are considered null-terminated. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero program object and errcode_ret is set to + * \a CL_SUCCESS if the program object is created successfully. It returns a + * NULL value with one of the following error values returned in + * \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if \a count is zero or if \a strings or any entry in + * \a strings is NULL. + * - CL_COMPILER_NOT_AVAILABLE if a compiler is not available. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_program, clCreateProgramWithSource, ( + cl_context context, + cl_uint count, + const char **strings, + const size_t *lengths, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return (cl_program)0; + } + if (count == 0 || strings == NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + + std::string sourceCode; + for (cl_uint i = 0; i < count; ++i) { + if (strings[i] == NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + if (lengths && lengths[i] != 0) { + sourceCode.append(strings[i], lengths[i]); + } + else { + sourceCode.append(strings[i]); + } + } + if (sourceCode.empty()) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + + // Create the program + amd::Program* program = new amd::Program(*as_amd(context), sourceCode); + if (program == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_program)0; + } + + // Add programs for all devices in the context. + const std::vector& devices = as_amd(context)->devices(); + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + if (program->addDeviceProgram(**it) == CL_OUT_OF_HOST_MEMORY) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + program->release(); + return (cl_program)0; + } + } + +#if defined(DEBUG) + if (OCL_STRESS_BINARY_IMAGE) { + size_t num_devices = devices.size(); + cl_int ret; + + cl_device_id* device_list = new cl_device_id[num_devices]; + memset(device_list, (int) (num_devices * sizeof(cl_device_id)), '\0'); + clGetProgramInfo(as_cl(program), CL_PROGRAM_DEVICES, + num_devices * sizeof(cl_device_id), device_list, NULL); + + clBuildProgram(as_cl(program), (int) num_devices, device_list, NULL, NULL, NULL); + + size_t* lengths = new size_t[num_devices]; + memset(lengths, (int) (num_devices * sizeof(size_t)), '\0'); + clGetProgramInfo(as_cl(program), CL_PROGRAM_BINARY_SIZES, + num_devices * sizeof(size_t), lengths, NULL); + + unsigned char** binaries = new unsigned char*[num_devices]; + for (size_t i = 0; i < num_devices; ++i) { + binaries[i] = new unsigned char[lengths[i]]; + memset(binaries[i], (int) lengths[i], '\0'); + } + clGetProgramInfo(as_cl(program), CL_PROGRAM_BINARIES, + num_devices * sizeof(unsigned char*), binaries, NULL); + + program->release(); + + cl_program new_prg = clCreateProgramWithBinary( + context, (int) num_devices, device_list, lengths, + (const unsigned char **) binaries, NULL, &ret); + + + for (size_t i = 0; i < num_devices; ++i) { + delete[] binaries[i]; + } + delete[] binaries; + delete[] lengths; + delete[] device_list; + + if (ret != CL_SUCCESS) { + *not_null(errcode_ret) = ret; + return NULL; + } + + program = as_amd(new_prg); + } +#endif // DEBUG + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(program); +} +RUNTIME_EXIT + +/*! \brief Create a program object for a context, and loads the binary images + * into the program object. + * + * \param context must be a valid OpenCL context. + * + * \param device_list is a pointer to a list of devices that are in context. + * \a device_list must be a non-NULL value. The binaries are loaded for devices + * specified in this list. + * + * \param num_devices is the number of devices listed in \a device_list. + * + * \param device_list The devices associated with the program object. The + * list of devices specified by \a device_list must be devices associated with + * \a context. + * + * \param lengths is an array of the size in bytes of the program binaries to + * be loaded for devices specified by \a device_list. + * + * \param binaries is an array of pointers to program binaries to be loaded + * for devices specified by \a device_list. For each device given by + * \a device_list[i], the pointer to the program binary for that device is + * given by \a binaries[i] and the length of this corresponding binary is given + * by \a lengths[i]. \a lengths[i] cannot be zero and \a binaries[i] cannot be + * a NULL pointer. The program binaries specified by binaries contain the bits + * that describe the program executable that will be run on the device(s) + * associated with context. The program binary can consist of either or both: + * - Device-specific executable(s) + * - Implementation specific intermediate representation (IR) which will be + * converted to the device-specific executable. + * + * \param binary_status returns whether the program binary for each device + * specified in \a device_list was loaded successfully or not. It is an array + * of \a num_devices entries and returns CL_SUCCESS in \a binary_status[i] if + * binary was successfully loaded for device specified by \a device_list[i]; + * otherwise returns CL_INVALID_VALUE if \a lengths[i] is zero or if + * \a binaries[i] is a NULL value or CL_INVALID_BINARY in \a binary_status[i] + * if program binary is not a valid binary for the specified device. + * If \a binary_status is NULL, it is ignored. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero program object and \a errcode_ret is set to + * CL_SUCCESS if the program object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if \a device_list is NULL or \a num_devices is zero. + * - CL_INVALID_DEVICE if OpenCL devices listed in \a device_list are not in + * the list of devices associated with \a context + * - CL_INVALID_VALUE if \a lengths or \a binaries are NULL or if any entry + * in \a lengths[i] is zero or \a binaries[i] is NULL. + * - CL_INVALID_BINARY if an invalid program binary was encountered for any + * device. \a binary_status will return specific status for each device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_program, clCreateProgramWithBinary, ( + cl_context context, + cl_uint num_devices, + const cl_device_id *device_list, + const size_t *lengths, + const unsigned char **binaries, + cl_int *binary_status, + cl_int *errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return (cl_program)0; + } + if (num_devices == 0 || device_list == NULL + || binaries == NULL || lengths == NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + + amd::Program* program = new amd::Program(*as_amd(context)); + if (program == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_program)0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + for (cl_uint i = 0; i < num_devices; ++i) { + cl_device_id device = device_list[i]; + + if (!is_valid(device) + || !as_amd(context)->containsDevice(as_amd(device))) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + program->release(); + return (cl_program)0; + } + if (binaries[i] == NULL || lengths[i] == 0) { + if (binary_status != NULL) { + binary_status[i] = CL_INVALID_VALUE; + } + *not_null(errcode_ret) = CL_INVALID_VALUE; + continue; + } + + cl_int status = program->addDeviceProgram( + *as_amd(device), binaries[i], lengths[i]); + + *not_null(errcode_ret) = status; + + if (status == CL_OUT_OF_HOST_MEMORY) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + program->release(); + return (cl_program)0; + } + + if (binary_status != NULL) { + binary_status[i] = status; + } + } + return as_cl(program); +} +RUNTIME_EXIT + +/*! \brief Increment the program reference count. + * + * clCreateProgram does an implicit retain. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_PROGRAM if \a program is not a valid program object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainProgram, (cl_program program)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + as_amd(program)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the program reference count. + * + * The program object is deleted after all kernel objects associated with + * \a program have been deleted and the program reference count becomes zero. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_PROGRAM if \a program is not a valid program object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseProgram, (cl_program program)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + as_amd(program)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_Build + * @{ + */ + +/*! \brief Build (compile & link) a program executable from the program source + * or binary for all the devices or a specific device(s) in the OpenCL context + * associated with program. + * + * OpenCL allows program executables to be built using the sources or binaries. + * + * \param program is the program object. + * + * \param device_list is a pointer to a list of devices associated with + * \a program. If \a device_list is a NULL value, the program executable is + * built for all devices associated with \a program for which a source or + * binary has been loaded. If \a device_list is a non-NULL value, the program + * executable is built for devices specified in this list for which a source + * or binary has been loaded. + * + * \param num_devices is the number of devices listed in \a device_list. + * + * \param options is a pointer to a string that describes the build options to + * be used for building the program executable. + * + * \param pfn_notify is a function pointer to a notification routine. The + * notification routine allows an application to register a callback function + * which will be called when the program executable has been built + * (successfully or unsuccessfully). If \a pfn_notify is not NULL, + * clBuildProgram does not need to wait for the build to complete and can + * return immediately. If \a pfn_notify is NULL, clBuildProgram does not + * return until the build has completed. This callback function may be called + * asynchronously by the OpenCL implementation. It is the application's + * responsibility to ensure that the callback function is thread-safe. + * + * \param user_data will be passed as the argument when \a pfn_notify is + * called. \a user_data can be NULL. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_PROGRAM if \a program is not a valid program object + * - CL_INVALID_VALUE if \a device_list is NULL and \a num_devices is greater + * than zero, or if \a device_list is not NULL and \a num_devices is zero, + * - CL_INVALID_DEVICE if OpenCL devices listed in \a device_list are not in + * the list of devices associated with \a program + * - CL_INVALID_BINARY if \a program is created with clCreateWithProgramBinary + * and devices listed in \a device_list do not have a valid program binary + * loaded + * - CL_INVALID_BUILD_OPTIONS if the build options specified by \a options are + * invalid + * - CL_INVALID_OPERATION if the build of a program executable for any of the + * devices listed in \a device_list by a previous call to clBuildProgram for + * \a program has not completed + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clBuildProgram, ( + cl_program program, + cl_uint num_devices, + const cl_device_id *device_list, + const char *options, + void (CL_CALLBACK * pfn_notify)(cl_program program, void *user_data), + void *user_data)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + if ((num_devices > 0 && device_list == NULL) + || (num_devices == 0 && device_list != NULL)) { + return CL_INVALID_VALUE; + } + + amd::Program* amdProgram = as_amd(program); + + if (device_list == NULL) { + // build for all devices in the context. + return amdProgram->build(amdProgram->context().devices(), + options, pfn_notify, user_data); + } + + std::vector devices(num_devices); + while (num_devices--) { + amd::Device* device = as_amd(*device_list++); + if (!amdProgram->context().containsDevice(device)) { + return CL_INVALID_DEVICE; + } + devices[num_devices] = device; + } + return amdProgram->build(devices, options, pfn_notify, user_data); +} +RUNTIME_EXIT + +/*! \brief compiles a program’s source for all the devices or a specific + * device(s) in the OpenCL context associated with program. The pre-processor + * runs before the program sources are compiled. + * The compiled binary is built for all devices associated with program or + * the list of devices specified. The compiled binary can be queried using + * \a clGetProgramInfo(program, CL_PROGRAM_BINARIES, …) and can be specified + * to \a clCreateProgramWithBinary to create a new program object. + * + * \param program is the program object that is the compilation target. + * + * \param device_list is a pointer to a list of devices associated with program. + * If device_list is a NULL value, the compile is performed for all devices + * associated with program. If device_list is a non-NULL value, the compile is + * performed for devices specified in this list. + * + * \param num_devices is the number of devices listed in \a device_list. + * + * \param options is a pointer to a null-terminated string of characters that + * describes the compilation options to be used for building the program + * executable. The list of supported options is as described in section 5.6.4. + * + * \param num_input_headers specifies the number of programs that describe + * headers in the array referenced by input_headers. + * + * \param input_headers is an array of program embedded headers created with + * \a clCreateProgramWithSource. + * + * \param header_include_names is an array that has a one to one correspondence + * with input_headers. + * Each entry in \a header_include_names specifies the include name used by + * source in program that comes from an embedded header. The corresponding entry + * in input_headers identifies the program object which contains the header + * source to be used. The embedded headers are first searched before the headers + * in the list of directories specified by the –I compile option (as described in + * section 5.6.4.1). If multiple entries in header_include_names refer to the same + * header name, the first one encountered will be used. + * + * \param pfn_notify is a function pointer to a notification routine. The + * notification routine is a callback function that an application can register + * and which will be called when the program executable has been built + * (successfully or unsuccessfully). If pfn_notify is not NULL, + * \a clCompileProgram does not need to wait for the compiler to complete and can + * return immediately. If \a pfn_notify is NULL, \a clCompileProgram does not + * return until the compiler has completed. This callback function may be called + * asynchronously by the OpenCL implementation. It is the application’s + * responsibility to ensure that the callback function is thread-safe. + * + * \param user_data will be passed as an argument when pfn_notify is called. + * \a user_data can be NULL. + * + * \return CL_SUCCESS if the function is executed successfully. Otherwise, it + * returns one of the following errors: + * - CL_INVALID_PROGRAM if program is not a valid program object. + * - CL_INVALID_VALUE if device_list is NULL and num_devices is greater than + * zero, or if \a device_list is not NULL and \a num_devices is zero. + * - CL_INVALID_VALUE if num_input_headers is zero and \a header_include_names + * or input_headers are not NULL or if num_input_headers is not zero and + * \a header_include_names or input_headers are NULL. + * - CL_INVALID_VALUE if \a pfn_notify is NULL but \a user_data is not NULL. + * - CL_INVALID_DEVICE if OpenCL devices listed in device_list are not in the + * list of devices associated with program + * - CL_INVALID_COMPILER_OPTIONS if the compiler options specified by options + * are invalid. + * - CL_INVALID_OPERATION if the compilation or build of a program executable + * for any of the devices listed in device_list by a previous call to + * \a clCompileProgram or \a clBuildProgram for program has not completed. + * - CL_COMPILER_NOT_AVAILABLE if a compiler is not available i.e. + * - CL_DEVICE_COMPILER_AVAILABLE specified in table 4.3 is set to CL_FALSE. + * - CL_COMPILE_PROGRAM_FAILURE if there is a failure to compile the program + * source. This error will be returned if clCompileProgram does not return + * until the compile has completed. + * - CL_INVALID_OPERATION if there are kernel objects attached to program. + * - CL_INVALID_OPERATION if program has no source i.e. it has not been created + * with \a clCreateProgramWithSource. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clCompileProgram, ( + cl_program program, + cl_uint num_devices, + const cl_device_id *device_list, + const char *options, + cl_uint num_input_headers, + const cl_program *input_headers, + const char **header_include_names, + void (CL_CALLBACK *pfn_notify)(cl_program program, void *user_data), + void *user_data)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + if ((num_devices > 0 && device_list == NULL) + || (num_devices == 0 && device_list != NULL)) { + return CL_INVALID_VALUE; + } + if ((num_input_headers > 0 + && (input_headers == NULL || header_include_names == NULL)) + || (num_input_headers == 0 + && (input_headers != NULL || header_include_names != NULL))) { + return CL_INVALID_VALUE; + } + if (pfn_notify == NULL && user_data != NULL) { + return CL_INVALID_VALUE; + } + + amd::Program* amdProgram = as_amd(program); + if (amdProgram->referenceCount() > 1) { + return CL_INVALID_OPERATION; + } + + std::vector headerPrograms(num_input_headers); + for (cl_uint i = 0; i < num_input_headers; ++i) { + if (!is_valid(input_headers[i])) { + return CL_INVALID_OPERATION; + } + const amd::Program* headerProgram = as_amd(input_headers[i]); + headerPrograms[i] = headerProgram; + } + + if (device_list == NULL) { + // compile for all devices in the context. + return amdProgram->compile(amdProgram->context().devices(), + num_input_headers, headerPrograms, + header_include_names, options, + pfn_notify, user_data); + } + + std::vector devices(num_devices); + + while (num_devices--) { + amd::Device* device = as_amd(*device_list++); + if (!amdProgram->context().containsDevice(device)) { + return CL_INVALID_DEVICE; + } + devices[num_devices] = device; + } + + return amdProgram->compile(devices, num_input_headers, + headerPrograms, header_include_names, options, + pfn_notify, user_data); +} +RUNTIME_EXIT + +/*! \brief links a set of compiled program objects and libraries for all + * the devices or a specific device(s) in the OpenCL context and creates + * an executable. clLinkProgram creates a new program object which contains + * this executable. The executable binary can be queried using + * \a clGetProgramInfo(program, CL_PROGRAM_BINARIES, …) and can be specified + * to \a clCreateProgramWithBinary to create a new program object. + * The devices associated with the returned program object will be the list + * of devices specified by device_list or if device_list is NULL it will be + * the list of devices associated with context. + * + * \param context must be a valid OpenCL context. + * + * \param device_list is a pointer to a list of devices that are in context. + * If device_list is a NULL value, the link is performed for all devices + * associated with context for which a compiled object is available. + * If device_list is a non-NULL value, the compile is performed for devices + * specified in this list for which a source has been loaded. + * + * \param num_devices is the number of devices listed in device_list. + * + * \param options is a pointer to a null-terminated string of characters + * that describes the link options to be used for building the program + * executable. The list of supported options is as described in section 5.6.5. + * + * \param num_input_programs specifies the number of programs in array + * referenced by input_programs. + * + * \param input_programs is an array of program objects that are compiled + * binaries or libraries that are to be linked to create the program executable. + * For each device in device_list or if device_list is NULL the list of devices + * associated with context, the following cases occur: + * All programs specified by input_programs contain a compiled binary or + * library for the device. In this case, a link is performed to generate + * a program executable for this device. None of the programs contain + * a compiled binary or library for that device. In this case, no link is + * performed and there will be no program executable generated for this device. + * All other cases will return a CL_INVALID_OPERATION error. + * + * \param pfn_notify is a function pointer to a notification routine. + * The notification routine is a callback function that an application can + * register and which will be called when the program executable has been built + * (successfully or unsuccessfully). If \a pfn_notify is not NULL, + * \a clLinkProgram does not need to wait for the linker to complete and can + * return immediately. Once the linker has completed, the \a pfn_notify + * callback function is called with a valid program object (if the link was + * successful) or NULL (if the link encountered a failure). This callback + * function may be called asynchronously by the OpenCL implementation. It is + * the application’s responsibility to ensure that the callback function is + * thread-safe. If \a pfn_notify is NULL, \a clLinkProgram does not return + * until the linker has completed. clLinkProgram returns a valid non-zero + * program object (if the link was successful) or NULL (if the link + * encountered a failure). + * + * \a user_data will be passed as an argument when \a pfn_notify is called. + * user_data can be NULL. + * + * \return a valid non-zero program object and errcode_ret is set to CL_SUCCESS + * if the link was successful in generating a program executable for at least + * one device and the program object was created successfully. If \a pfn_notify + * is not NULL, \a clLinkProgram returns a NULL program object and + * \a errcode_ret is set to CL_SUCCESS if the function was executed + * successfully. Otherwise, it returns one of the following errors: + * - CL_INVALID_CONTEXT if context is not a valid context. + * - CL_INVALID_VALUE if device_list is NULL and num_devices is greater than + * zero, or if \a device_list is not NULL and \a num_devices is zero. + * - CL_INVALID_VALUE if \a num_input_programs is zero and \a input_programs + * is NULL or if \a num_input_programs is zero and \a input_programs is not + * NULL or if \a num_input_programs is not zero and \a input_programs is NULL. + * - CL_INVALID_PROGRAM if programs specified in \a input_programs are not + * valid program objects. + * - CL_INVALID_VALUE if \a pfn_notify is NULL but \a user_data is not NULL. + * - CL_INVALID_DEVICE if OpenCL devices listed in \a device_list are not in + * the list of devices associated with context + * - CL_INVALID_LINKER_OPTIONS if the linker options specified by options are + * invalid. + * - CL_INVALID_OPERATION if the compilation or build of a program executable + * for any of the devices listed in \a device_list by a previous call to + * clCompileProgram or clBuildProgram for program has not completed. + * - CL_INVALID_OPERATION if the rules for devices containing compiled binaries + * or libraries as described in \a input_programs argument above are + * not followed. + * - CL_LINKER_NOT_AVAILABLE if a linker is not available i.e. + * - CL_DEVICE_LINKER_AVAILABLE specified in table 4.3 is set to CL_FALSE. + * - CL_LINK_PROGRAM_FAILURE if there is a failure to link the compiled + * binaries and/or libraries. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY_RET(cl_program, clLinkProgram, ( + cl_context context, + cl_uint num_devices, + const cl_device_id* device_list, + const char* options, + cl_uint num_input_programs, + const cl_program* input_programs, + void (CL_CALLBACK *pfn_notify)(cl_program program, void* user_data), + void* user_data, + cl_int* errcode_ret)) +{ + if (!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + return (cl_program)0; + } + + if ((num_devices > 0 && device_list == NULL) || + (num_devices == 0 && device_list != NULL)) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + + if (num_input_programs == 0 || input_programs == NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + + if (pfn_notify == NULL && user_data != NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_program)0; + } + + std::vector inputPrograms(num_input_programs); + for (cl_uint i = 0; i < num_input_programs; ++i) { + if (!is_valid(input_programs[i])) { + *not_null(errcode_ret) = CL_INVALID_PROGRAM; + return (cl_program)0; + } + amd::Program* inputProgram = as_amd(input_programs[i]); + inputPrograms[i] = inputProgram; + } + + amd::Program* program = + createProgram( context, num_devices, device_list, errcode_ret); + if (program == NULL) + return (cl_program)0; + + *not_null(errcode_ret) = CL_SUCCESS; + cl_int status; + + if (device_list == NULL) { + // compile for all devices in the context. + status = program->link(as_amd(context)->devices(), + num_input_programs, inputPrograms, + options, pfn_notify, user_data); + } + else { + std::vector devices(num_devices); + + while (num_devices--) { + amd::Device* device = as_amd(*device_list++); + if (!as_amd(context)->containsDevice(device)) { + program->release(); + *not_null(errcode_ret) = CL_INVALID_DEVICE; + return (cl_program)0; + } + devices[num_devices] = device; + } + + status = program->link(devices, num_input_programs, inputPrograms, + options, pfn_notify, user_data); + } + *not_null(errcode_ret) = status; + if (status == CL_SUCCESS) { + return as_cl(program); + } + + program->release(); + return (cl_program)0; +} +RUNTIME_EXIT + +/*! \brief creates a program object for a context, and loads the information + * related to the built-in kernels into a program object. + * + * \param context must be a valid OpenCL context. + * + * \param num_devices is the number of devices listed in device_list. + * + * \param device_list is a pointer to a list of devices that are in context. + * \a device_list must be a non-NULL value. The built-in kernels are loaded + * for devices specified in this list. The devices associated with the + * program object will be the list of devices specified by \a device_list. + * The list of devices specified by \a device_list must be devices associated + * with context. + * + * \param kernel_names is a semi-colon separated list of built-in kernel names. + * + * \return a valid non-zero program object and \a errcode_ret is set to + * CL_SUCCESS if the program object is created successfully. Otherwise, it + * returns a NULL value with one of the following error values returned + * in errcode_ret: + * - CL_INVALID_CONTEXT if context is not a valid context. + * - CL_INVALID_VALUE if device_list is NULL or num_devices is zero. + * - CL_INVALID_VALUE if kernel_names is NULL or kernel_names contains a kernel + * name that is not supported by any of the devices in \a device_list. + * - CL_INVALID_DEVICE if devices listed in device_list are not in the list + * of devices associated with context. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY_RET(cl_program, clCreateProgramWithBuiltInKernels, ( + cl_context context, + cl_uint num_devices, + const cl_device_id *device_list, + const char *kernel_names, + cl_int *errcode_ret)) +{ + //!@todo Add implementation + amd::Program* program = NULL; + Unimplemented(); + return as_cl(program); +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_Unloading + * @{ + */ + +/*! \brief Allows the implementation to release the resources allocated by + * the OpenCL compiler for platform. This is a hint from the application + * and does not guarantee that the compiler will not be used in the future + * or that the compiler will actually be unloaded by the implementation. + * Calls to \a clBuildProgram, \a clCompileProgram or \a clLinkProgram after + * \a clUnloadPlatformCompiler will reload the compiler, + * if necessary, to build the appropriate program executable. + * + * \return CL_SUCCESS if the function is executed successfully. + * Otherwise, it returns one of the following errors: + * - CL_INVALID_PLATFORM if platform is not a valid platform. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clUnloadPlatformCompiler, (cl_platform_id platform)) +{ + if (platform != NULL && platform != AMD_PLATFORM) { + return CL_INVALID_PLATFORM; + } + + //! @todo: Implement Compiler::unload() + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Allow to runtime to release the resources allocated by the OpenCL + * compiler. + * + * This is a hint from the application and does not guarantee that the compiler + * will not be used in the future or that the compiler will actually be + * unloaded by the implementation. + * + * Calls to clBuildProgram after clUnloadCompiler may reload the compiler, + * if necessary, to build the appropriate program executable. + * + * \return This call currently always returns CL_SUCCESS + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clUnloadCompiler, (void)) +{ + //! @todo: Implement Compiler::unload() + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_ProgramQueries + * @{ + */ + +/*! \brief Return information about the program object. + * + * \param program specifies the program object being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_PROGRAM_EXECUTABLE if param_name is + * CL_PROGRAM_NUM_KERNELS or CL_PROGRAM_KERNEL_NAMES and a successful + * program executable has not been built for at least one device in the list + * of devices associated with program. + * - CL_INVALID_PROGRAM if \a program is a not a valid program object + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clGetProgramInfo, ( + cl_program program, + cl_program_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + + switch (param_name) { + case CL_PROGRAM_REFERENCE_COUNT: { + cl_uint count = as_amd(program)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_CONTEXT: { + cl_context context = const_cast( + as_cl(&as_amd(program)->context())); + return amd::clGetInfo( + context, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_NUM_DEVICES: { + cl_uint numDevices = (cl_uint)as_amd(program)->deviceList().size(); + return amd::clGetInfo( + numDevices, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_DEVICES: { + const amd::Program::devicelist_t& devices = + as_amd(program)->deviceList(); + const size_t numDevices = devices.size(); + const size_t valueSize = numDevices * sizeof(cl_device_id); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + cl_device_id* device_list = (cl_device_id*) param_value; + amd::Program::devicelist_t::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + *device_list++ = const_cast(as_cl(*it)); + } + if (param_value_size > valueSize) { + ::memset(static_cast
(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_PROGRAM_SOURCE: { + const char* source = as_amd(program)->sourceCode().c_str(); + return amd::clGetInfo( + source, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_BINARY_SIZES: { + amd::Program* amdProgram = as_amd(program); + const amd::Program::devicelist_t& devices = amdProgram->deviceList(); + const size_t numBinaries = devices.size(); + const size_t valueSize = numBinaries * sizeof(size_t); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + size_t* binary_sizes = (size_t*) param_value; + amd::Program::devicelist_t::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + *binary_sizes++ = + amdProgram->getDeviceProgram(**it)->binary().second; + } + if (param_value_size > valueSize) { + ::memset(static_cast
(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_PROGRAM_BINARIES: { + amd::Program* amdProgram = as_amd(program); + const amd::Program::devicelist_t& devices = amdProgram->deviceList(); + const size_t numBinaries = devices.size(); + const size_t valueSize = numBinaries * sizeof(char*); + + if (param_value != NULL && param_value_size < valueSize) { + return CL_INVALID_VALUE; + } + *not_null(param_value_size_ret) = valueSize; + if (param_value != NULL) { + char** binaries = (char**) param_value; + amd::Program::devicelist_t::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + const device::Program::binary_t& binary = + amdProgram->getDeviceProgram(**it)->binary(); + // If an entry value in the array is NULL, + // then runtime should skip copying the program binary + if (*binaries != NULL) { + ::memcpy(*binaries, binary.first, binary.second); + } + binaries++; + } + if (param_value_size > valueSize) { + ::memset(static_cast
(param_value) + valueSize, + '\0', param_value_size - valueSize); + } + } + return CL_SUCCESS; + } + case CL_PROGRAM_NUM_KERNELS: { + size_t numKernels = as_amd(program)->symbols().size(); + return amd::clGetInfo( + numKernels, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_KERNEL_NAMES: { + const char* kernelNames = as_amd(program)->kernelNames().c_str(); + return amd::clGetInfo( + kernelNames, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! \brief Return build information for each device in the program object. + * + * \param program specifies the program object being queried. + * + * \param device specifies the device for which build information is being + * queried. device must be a valid device associated with \a program. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_DEVICE if \a device is not in the list of devices associated + * with \a program + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_PROGRAM if \a program is a not a valid program object + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetProgramBuildInfo, ( + cl_program program, + cl_device_id device, + cl_program_build_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + + const device::Program* devProgram = + as_amd(program)->getDeviceProgram(*as_amd(device)); + if (devProgram == NULL) { + return CL_INVALID_DEVICE; + } + + switch (param_name) { + case CL_PROGRAM_BUILD_STATUS: { + cl_build_status status = devProgram->buildStatus(); + return amd::clGetInfo( + status, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_BUILD_OPTIONS: { + const std::string optionsStr = devProgram->lastBuildOptionsArg(); + const char* options = optionsStr.c_str(); + return amd::clGetInfo( + options, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_BUILD_LOG: { + const std::string logstr = + as_amd(program)->programLog() + devProgram->buildLog().c_str(); + const char* log = logstr.c_str(); + return amd::clGetInfo( + log, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_BINARY_TYPE: { + const device::Program::type_t devProgramType = devProgram->type(); + cl_uint type; + switch (devProgramType) { + case device::Program::TYPE_NONE: + { + type = CL_PROGRAM_BINARY_TYPE_NONE; + break; + } + case device::Program::TYPE_COMPILED: + { + type = CL_PROGRAM_BINARY_TYPE_COMPILED_OBJECT; + break; + } + case device::Program::TYPE_LIBRARY: + { + type = CL_PROGRAM_BINARY_TYPE_LIBRARY; + break; + } + case device::Program::TYPE_EXECUTABLE: + { + type = CL_PROGRAM_BINARY_TYPE_EXECUTABLE; + break; + } + case device::Program::TYPE_INTERMEDIATE: + { + type = CL_PROGRAM_BINARY_TYPE_INTERMEDIATE; + break; + } + default: + return CL_INVALID_VALUE; + } + return amd::clGetInfo( + type, param_value_size, param_value, param_value_size_ret); + } + case CL_PROGRAM_BUILD_GLOBAL_VARIABLE_TOTAL_SIZE: { + size_t size = devProgram->globalVariableTotalSize(); + return amd::clGetInfo( + size, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! @} + * @} + * + * \addtogroup CL_Kernels + * + * A kernel is a function declared in a program. A kernel is identified by the + * __kernel qualifier applied to any function in a program. A kernel object + * encapsulates the specific __kernel function declared in a program and + * the argument values to be used when executing this __kernel function. + * + * @{ + * + * \addtogroup CL_CreateKernel + * @{ + */ + +/*! \brief Create a kernel object. + * + * \param program is a program object with a successfully built executable. + * + * \param kernel_name is a function name in the program declared with the + * __kernel qualifier. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero kernel object and \a errcode_ret is set to + * CL_SUCCESS if the kernel object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_PROGRAM if \a program is not a valid program object + * - CL_INVALID_PROGRAM_EXECUTABLE if there is no successfully built executable + * for \a program. + * - CL_INVALID_KERNEL_NAME if \a kernel_name is not found in \a program. + * - CL_INVALID_KERNEL_DEFINITION if the function definition for __kernel + * function given by \a kernel_name such as the number of arguments, the + * argument types are not the same for all devices for which the program + * executable has been built. + * - CL_INVALID_VALUE if \a kernel_name is NULL. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_kernel, clCreateKernel, ( + cl_program program, + const char *kernel_name, + cl_int *errcode_ret)) +{ + if (!is_valid(program)) { + *not_null(errcode_ret) = CL_INVALID_PROGRAM; + return (cl_kernel) 0; + } + if (kernel_name == NULL) { + *not_null(errcode_ret) = CL_INVALID_VALUE; + return (cl_kernel) 0; + } + /* FIXME_lmoriche, FIXME_spec: What are we supposed to do here? + * if (!as_amd(program)->containsOneSuccesfullyBuiltProgram()) + * { + * *NotNull(errcode) = CL_INVALID_PROGRAM_EXECUTABLE; + * return (cl_kernel) 0; + * } + */ + amd::Program* amd_program = as_amd(program); + const amd::Symbol* symbol = amd_program->findSymbol(kernel_name); + if (symbol == NULL) { + *not_null(errcode_ret) = CL_INVALID_KERNEL_NAME; + return (cl_kernel) 0; + } + + amd::Kernel* kernel = + new amd::Kernel(*amd_program, *symbol, kernel_name); + if (kernel == NULL) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + return (cl_kernel)0; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(kernel); +} +RUNTIME_EXIT + +/*! \brief Create kernel objects for all kernel functions in program. + * + * Kernel objects may not be created for any __kernel functions in program + * that do not have the same function definition across all devices for which + * a program executable has been successfully built. + * + * \param program is a program object with a successfully built executable. + * + * \param num_kernels is the size of memory pointed to by \a kernels specified + * as the number of cl_kernel entries. + * + * \param kernels is the buffer where the kernel objects for kernels in + * \a program will be returned. If \a kernels is NULL, it is ignored. + * If \a kernels is not NULL, \a num_kernels must be greater than or equal + * to the number of kernels in program. + * + * \param num_kernels_ret is the number of kernels in program. If + * \a num_kernels_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the kernel objects were successfully allocated + * - CL_INVALID_PROGRAM if \a program is not a valid program object + * - CL_INVALID_PROGRAM_EXECUTABLE if there is no successfully built executable + * for any device in \a program + * - CL_INVALID_VALUE if \a kernels is not NULL and \a num_kernels is less + * than the number of kernels in program + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * Kernel objects can only be created once you have a program object with a + * valid program source or binary loaded into the program object and the + * program executable has been successfully built for one or more devices + * associated with \a program. No changes to the program executable are + * allowed while there are kernel objects associated with a program object. + * This means that calls to clBuildProgram return CL_INVALID_OPERATION if there + * are kernel objects attached to a program object. The OpenCL context + * associated with program will be the context associated with kernel. + * Devices associated with a program object for which a valid program + * executable has been built can be used to execute kernels declared in the + * program object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clCreateKernelsInProgram, ( + cl_program program, + cl_uint num_kernels, + cl_kernel *kernels, + cl_uint *num_kernels_ret)) +{ + if (!is_valid(program)) { + return CL_INVALID_PROGRAM; + } + + cl_uint numKernels = (cl_uint) as_amd(program)->symbols().size(); + + if (kernels != NULL && num_kernels < numKernels) { + return CL_INVALID_VALUE; + } + *not_null(num_kernels_ret) = numKernels; + if (kernels == NULL) { + return CL_SUCCESS; + } + + const amd::Program::symbols_t& symbols = as_amd(program)->symbols(); + cl_kernel* result = kernels; + + amd::Program::symbols_t::const_iterator it; + for (it = symbols.begin(); it != symbols.end(); ++it) { + + amd::Kernel* kernel = new amd::Kernel( + *as_amd(program), it->second, it->first); + if (kernel == NULL) { + while (--result >= kernels) { + as_amd(*result)->release(); + } + return CL_OUT_OF_HOST_MEMORY; + } + *result++ = as_cl(kernel); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Increment the kernel reference count. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_KERNEL if \a kernel is not a valid kernel object. + * + * clCreateKernel or clCreateKernelsInProgram do an implicit retain. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainKernel, (cl_kernel kernel)) +{ + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + as_amd(kernel)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the kernel reference count. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_KERNEL if \a kernel is not a valid kernel object. + * + * The kernel object is deleted once the number of instances that are retained + * to \a kernel become zero and after all queued execution instances of + * \a kernel have finished. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseKernel, (cl_kernel kernel)) +{ + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + as_amd(kernel)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_SettingArgs + * @{ + */ + +/*! \brief Set the argument value for a specific argument of a kernel. + * + * \param kernel is a valid kernel object. + * + * \param arg_index is the argument index. Arguments to the kernel are referred + * by indices that go from 0 for the leftmost argument to n - 1, where n is the + * total number of arguments declared by a kernel. + * + * \param arg_value is a pointer to data that should be used as the argument + * value for argument specified by \a arg_index. The argument data pointed to + * by \a arg_value is copied and the \a arg_value pointer can therefore be + * reused by the application after clSetKernelArg returns. If the argument is + * a memory object (buffer or image), the \a arg_value entry will be a pointer + * to the appropriate buffer or image object. The memory object must be created + * with the context associated with the kernel object. If the argument is + * declared with the __local qualifier, the \a arg_value entry must be NULL. + * For all other kernel arguments, the \a arg_value entry must be a pointer to + * the actual data to be used as argument value. The memory object specified + * as argument value must be a buffer object if the argument is declared to be + * a pointer of a built-in or user defined type with the __global or __constant + * qualifier. If the argument is declared with the __constant qualifier, the + * size in bytes of the memory object cannot exceed + * CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE and the number of arguments declared + * with the __constant qualifier cannot exceed CL_DEVICE_MAX_CONSTANT_ARGS. The + * memory object specified as argument value must be a 2D image object if the + * argument is declared to be of type image2d_t. The memory object specified as + * argument value must be a 3D image object if argument is declared to be of + * type image3d_t. If the argument is of type sampler_t, the arg_value entry + * must be a pointer to the sampler object. + * + * \param arg_size specifies the size of the argument value. If the argument is + * a memory object, the size is the size of the buffer or image object type. + * For arguments declared with the __local qualifier, the size specified will + * be the size in bytes of the buffer that must be allocated for the __local + * argument. If the argument is of type sampler_t, the arg_size value must be + * equal to sizeof(cl_sampler). For all other arguments, the size will be the + * size of argument type. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_KERNEL if \a kernel is not a valid kernel object. + * - CL_INVALID_ARG_INDEX if \a arg_index is not a valid argument index. + * - CL_INVALID_ARG_VALUE if \a arg_value specified is NULL for an argument + * that is not declared with the __local qualifier or vice-versa. + * - CL_INVALID_MEM_OBJECT for an argument declared to be a memory object but + * the specified \a arg_value is not a valid memory object. + * - CL_INVALID_SAMPLER for an argument declared to be of type sampler_t but + * the specified \a arg_value is not a valid sampler object. + * - CL_INVALID_ARG_SIZE if \a arg_size does not match the size of the data + * type for an argument that is not a memory object or if the argument is a + * memory object and \a arg_size != sizeof(cl_mem) or if \a arg_size is zero + * and the argument is declared with the __local qualifier or if the + * argument is a sampler and arg_size != sizeof(cl_sampler). + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clSetKernelArg, ( + cl_kernel kernel, + cl_uint arg_index, + size_t arg_size, + const void *arg_value)) +{ + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + const amd::KernelSignature& signature = as_amd(kernel)->signature(); + if (arg_index >= signature.numParameters()) { + return CL_INVALID_ARG_INDEX; + } + + as_amd(kernel)->parameters().reset(static_cast(arg_index)); + + const amd::KernelParameterDescriptor& desc = signature.at(arg_index); + const bool is_local = desc.size_ == 0; + if (((arg_value == NULL) && !is_local && (desc.type_ != T_POINTER)) || + ((arg_value != NULL) && is_local)) { + return CL_INVALID_ARG_VALUE; + } + if (!is_local && (desc.type_ == T_POINTER) && (arg_value != NULL)) + { + cl_mem memObj = *static_cast(arg_value); + amd::RuntimeObject * pObject = as_amd(memObj); + if (NULL != memObj && + amd::RuntimeObject::ObjectTypeMemory != pObject->objectType()) { + return CL_INVALID_MEM_OBJECT; + } + } + else if ((desc.type_ == T_SAMPLER) && + !is_valid(*static_cast(arg_value))) { + return CL_INVALID_SAMPLER; + } + else if (desc.type_ == T_QUEUE) { + cl_command_queue queue = *static_cast(arg_value); + if (!is_valid(queue)) { + return CL_INVALID_DEVICE_QUEUE; + } + if (NULL == as_amd(queue)->asDeviceQueue()) { + return CL_INVALID_DEVICE_QUEUE; + } + } + if ((!is_local && (arg_size != desc.size_)) || (is_local && (arg_size == 0))) { + return CL_INVALID_ARG_SIZE; + } + + as_amd(kernel)->parameters().set( + static_cast(arg_index), arg_size, arg_value); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * \addtogroup CL_KernelQuery + * @{ + */ + +/*! \brief Return information about the kernel object. + * + * \param kernel specifies the kernel object being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_KERNEL if \a kernel is a not a valid kernel object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetKernelInfo, ( + cl_kernel kernel, + cl_kernel_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + // Check if we have a valid kernel + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + const amd::Kernel* amdKernel = as_amd(kernel); + + // Get the corresponded parameters + switch (param_name) { + case CL_KERNEL_FUNCTION_NAME: { + const char* name = amdKernel->name().c_str(); + // Return the kernel's name + return amd::clGetInfo( + name, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_NUM_ARGS: { + cl_uint numParam = + static_cast(amdKernel->signature().numParameters()); + // Return the number of kernel's parameters + return amd::clGetInfo( + numParam, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_REFERENCE_COUNT: { + cl_uint count = amdKernel->referenceCount(); + // Return the reference counter + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_CONTEXT: { + cl_context context = const_cast( + as_cl(&amdKernel->program().context())); + // Return the context, associated with the program + return amd::clGetInfo( + context, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_PROGRAM: { + cl_program program = const_cast( + as_cl(&amdKernel->program())); + // Return the program, associated with the kernel + return amd::clGetInfo( + program, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_ATTRIBUTES: { + const char* name = amdKernel->signature().attributes().c_str(); + // Return the kernel attributes + return amd::clGetInfo( + name, param_value_size, param_value, param_value_size_ret); + } + default: + return CL_INVALID_VALUE; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Returns information about the arguments of a kernel. Kernel + * argument information is only available if the program object associated + * with kernel is created with \a clCreateProgramWithSource and the program + * executable is built with the -cl-kernel-arg-info option specified in + * options argument to clBuildProgram or clCompileProgram. + * + * \param kernel specifies the kernel object being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_KERNEL if \a kernel is a not a valid kernel object. + * + * \version 1.2r07 + */ +RUNTIME_ENTRY(cl_int, clGetKernelArgInfo, ( + cl_kernel kernel, + cl_uint arg_indx, + cl_kernel_arg_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + // Check if we have a valid kernel + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + amd::Kernel* amdKernel = as_amd(kernel); + + const amd::KernelSignature& signature = amdKernel->signature(); + if (arg_indx >= signature.numParameters()) { + return CL_INVALID_ARG_INDEX; + } + + const amd::KernelParameterDescriptor& desc = signature.at(arg_indx); + + // Get the corresponded parameters + switch (param_name) { + case CL_KERNEL_ARG_ADDRESS_QUALIFIER: { + cl_kernel_arg_address_qualifier qualifier = desc.addressQualifier_; + return amd::clGetInfo( + qualifier, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_ARG_ACCESS_QUALIFIER: { + cl_kernel_arg_access_qualifier qualifier = desc.accessQualifier_; + return amd::clGetInfo( + qualifier, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_ARG_TYPE_NAME: { + // Return the argument's type name + return amd::clGetInfo( + desc.typeName_, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_ARG_TYPE_QUALIFIER: { + cl_kernel_arg_type_qualifier qualifier = desc.typeQualifier_; + return amd::clGetInfo( + qualifier, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_ARG_NAME: { + // Return the argument's name + return amd::clGetInfo( + desc.name_, param_value_size, param_value, param_value_size_ret); + } + default: + return CL_INVALID_VALUE; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Return information about the kernel object that may be specific + * to a device. + * + * \param kernel specifies the kernel object being queried. + * + * \param device identifies a specific device in the list of devices associated + * with \a kernel. The list of devices is the list of devices in the OpenCL + * context that is associated with \a kernel. If the list of devices associated + * with kernel is a single device, \a device can be a NULL value. + * + * \param param_name specifies the information to query + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully, + * - CL_INVALID_DEVICE if \a device is not in the list of devices associated + * with \a kernel or if \a device is NULL but there are more than one + * devices in the associated with \a kernel + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_KERNEL if \a kernel is a not a valid kernel object. + * + * \version 1.2r15 + */ +RUNTIME_ENTRY(cl_int, clGetKernelWorkGroupInfo, ( + cl_kernel kernel, + cl_device_id device, + cl_kernel_work_group_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + // Check if we have a valid device + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + + // Check if we have a valid kernel + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + + const amd::Device& amdDevice = *as_amd(device); + // Find the kernel, associated with the specified device + const device::Kernel* devKernel = + as_amd(kernel)->getDeviceKernel(amdDevice); + + // Make sure we found a valid kernel + if (devKernel == NULL) { + return CL_INVALID_KERNEL; + } + + // Get the corresponded parameters + switch (param_name) { + case CL_KERNEL_WORK_GROUP_SIZE: { + // Return workgroup size + return amd::clGetInfo(devKernel->workGroupInfo()->size_, + param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_COMPILE_WORK_GROUP_SIZE: { + // Return the compile workgroup size + return amd::clGetInfo(devKernel->workGroupInfo()->compileSize_, + param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_LOCAL_MEM_SIZE: { + // Return the amount of used local memory + const size_t align = amdDevice.info().minDataTypeAlignSize_; + cl_ulong memSize = as_amd(kernel)->parameters().localMemSize(align) + + amd::alignUp(devKernel->workGroupInfo()->localMemSize_, align); + return amd::clGetInfo( + memSize, param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE: { + // Return the compile workgroup size + return amd::clGetInfo( + devKernel->workGroupInfo()->preferredSizeMultiple_, + param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_PRIVATE_MEM_SIZE: { + // Return the compile workgroup size + return amd::clGetInfo(devKernel->workGroupInfo()->privateMemSize_, + param_value_size, param_value, param_value_size_ret); + } + case CL_KERNEL_GLOBAL_WORK_SIZE: { + return CL_INVALID_VALUE; + } + case CL_KERNEL_MAX_SEMAPHORE_SIZE_AMD: { + return amd::clGetInfo(amdDevice.info().maxSemaphoreSize_, + param_value_size, param_value, param_value_size_ret); + } + default: + return CL_INVALID_VALUE; + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Returns information about the kernel object. + * + * \param kernel specifies the kernel object being queried. + * + * \param device identifies a specific device in the list of devices associated + * with kernel. The list of devices is the list of devices in the OpenCL context + * that is associated with kernel. If the list of devices associated with kernel + * is a single device, device can be a NULL value. + * + * \param param_name specifies the information to query. The list of supported + * param_name types and the information returned in param_value by + * clGetKernelSubGroupInfoKHR is described in the table below. + * + * \param input_value_size is used to specify the size in bytes of memory + * pointed to by input_value. This size must be == size of input type as + * described in the table below. + * + * \param input_value is a pointer to memory where the appropriate + * parameterization of the query is passed from. If input_value is NULL, it is + * ignored. + * + * \param param_value is a pointer to memory where the appropriate result being + * queried is returned. If param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by param_value. This size must be >= size of return type as + * described in the table below. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to param_value. If param_value_size_ret is NULL, it is ignored. + * + * \return CL_SUCCESS if the function is executed successfully. + * Otherwise, it returns one of the following errors: + * + * - CL_INVALID_DEVICE if device is not in the list of devices associated with + * kernel or if device is NULL but there is more than one device associated + * with kernel. + * - CL_INVALID_VALUE if param_name is not valid, or if size in bytes specified + * by param_value_size is < size of return type as described in the table + * above and param_value is not NULL. + * - CL_INVALID_VALUE if param_name is CL_KERNEL_SUB_GROUP_SIZE_FOR_NDRANGE and + * the size in bytes specified by input_value_size is not valid or if + * input_value is NULL. + * - CL_INVALID_KERNEL if kernel is a not a valid kernel object. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by + * the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r12 + */ +RUNTIME_ENTRY(cl_int, clGetKernelSubGroupInfoKHR, ( + cl_kernel kernel, + cl_device_id device, + cl_kernel_sub_group_info param_name, + size_t input_value_size, + const void * input_value, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret)) +{ + // Check if we have a valid device + if (!is_valid(device)) { + return CL_INVALID_DEVICE; + } + + // Check if we have a valid kernel + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + + const amd::Device& amdDevice = *as_amd(device); + // Find the kernel, associated with the specified device + const device::Kernel* devKernel = + as_amd(kernel)->getDeviceKernel(amdDevice); + + // Make sure we found a valid kernel + if (devKernel == NULL) { + return CL_INVALID_KERNEL; + } + + // Get the corresponded parameters + switch (param_name) { + case CL_KERNEL_MAX_SUB_GROUP_SIZE_FOR_NDRANGE: + case CL_KERNEL_SUB_GROUP_COUNT_FOR_NDRANGE: { + + // Infer the number of dimensions from 'input_value_size' + size_t dims = input_value_size / sizeof(size_t); + if (dims == 0 || dims > 3 || input_value_size != dims * sizeof(size_t)) { + return CL_INVALID_VALUE; + } + + // Get the linear workgroup size + size_t workGroupSize = ((size_t*)input_value)[0]; + for (size_t i = 1; i < dims; ++i) { + workGroupSize *= ((size_t*)input_value)[i]; + } + + // Get the subgroup size. CPU devices only have one subgroup + // per workgroup. GPU devices sub-groups are wavefronts. + size_t subGroupSize = as_amd(device)->type() == CL_DEVICE_TYPE_CPU + ? workGroupSize : as_amd(device)->info().wavefrontWidth_; + + size_t numSubGroups = (workGroupSize + subGroupSize - 1) / subGroupSize; + + + return amd::clGetInfo( + (param_name == CL_KERNEL_MAX_SUB_GROUP_SIZE_FOR_NDRANGE) + ? subGroupSize : numSubGroups, + param_value_size, param_value, param_value_size_ret); + } + default: + return CL_INVALID_VALUE; + } + + return CL_SUCCESS;} +RUNTIME_EXIT + +/*! @} + * @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_sampler.cpp b/opencl/api/opencl/amdocl/cl_sampler.cpp new file mode 100644 index 0000000000..2a76a33293 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_sampler.cpp @@ -0,0 +1,328 @@ +// +// Copyright (c) 2008 Advanced Micro Devices, Inc. All rights reserved. +// + +#include "cl_common.hpp" + +#include "platform/context.hpp" +#include "platform/sampler.hpp" + + +/*! \addtogroup API + * @{ + * + * \addtogroup CL_Samplers + * + * A sampler object describes how to sample an image when the image is read + * in the kernel. The built-in functions to read from an image in a kernel + * take a sampler as an argument. The sampler arguments to the image read + * function can be sampler objects created using OpenCL functions and passed + * as argument values to the kernel or can be samplers declared inside + * a kernel. + * + * @{ + */ + +/*! \brief Create a sampler object. + * + * \param context must be a valid OpenCL context. + * + * \param specifies a list of sampler property names and their corresponding + * values. Each sampler property name is immediately followed by the + * corresponding desired value. The list is terminated with 0. If a supported + * property and its value is not specified in sampler_properties, its default + * value will be used. sampler_properties can be NULL in which case the default + * values for supported sampler properties will be used. + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero sampler object and \a errcode_ret is set to + * CL_SUCCESS if the sampler object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if the property name in sampler_properties is not a + * supported property name, if the value specified for a supported property + * name is not valid, or if the same property name is specified more than + * once + * - CL_INVALID_OPERATION if images are not supported by any device associated + * with context + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 2.0r19 + */ +RUNTIME_ENTRY_RET(cl_sampler, clCreateSamplerWithProperties, ( + cl_context context, + const cl_sampler_properties *sampler_properties, + cl_int *errcode_ret)) +{ + if(!is_valid(context)) { + *not_null(errcode_ret) = CL_INVALID_CONTEXT; + LogWarning("invalid parameter \"context\""); + return (cl_sampler) 0; + } + + cl_bool normalized_coords = CL_TRUE; + cl_addressing_mode addressing_mode = CL_ADDRESS_CLAMP; + cl_filter_mode filter_mode = CL_FILTER_NEAREST; + + const struct SamplerProperty { + cl_sampler_properties name; + union { + cl_sampler_properties raw; + cl_bool normalized_coords; + cl_addressing_mode addressing_mode; + cl_filter_mode filter_mode; + } value; + } *p = reinterpret_cast(sampler_properties); + + if(p != NULL) while(p->name != 0) { + switch(p->name) { + case CL_SAMPLER_NORMALIZED_COORDS: + normalized_coords = p->value.normalized_coords; + break; + case CL_SAMPLER_ADDRESSING_MODE: + addressing_mode = p->value.addressing_mode; + break; + case CL_SAMPLER_FILTER_MODE: + filter_mode = p->value.filter_mode; + break; + default: + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid property name"); + return (cl_sampler) 0; + } + ++p; + } + + // Check sampler validity + // Check addressing mode + switch(addressing_mode) { + case CL_ADDRESS_NONE: + case CL_ADDRESS_CLAMP_TO_EDGE: + case CL_ADDRESS_CLAMP: + break; + + case CL_ADDRESS_REPEAT: + if(!normalized_coords) { + // repeat mode cannot be used with unnormalized coordinates + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid combination for sampler"); + return (cl_sampler) 0; + } + break; + + case CL_ADDRESS_MIRRORED_REPEAT: + if(!normalized_coords) { + // repeat mode cannot be used with unnormalized coordinates + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid combination for sampler"); + return (cl_sampler) 0; + } + break; + + default: + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid addressing mode"); + return (cl_sampler) 0; + } + // Check filter mode + switch(filter_mode) { + case CL_FILTER_NEAREST: + case CL_FILTER_LINEAR: + break; + + default: + *not_null(errcode_ret) = CL_INVALID_VALUE; + LogWarning("invalid filter mode"); + return (cl_sampler) 0; + } + // Create instance of Sampler + amd::Sampler* sampler = new amd::Sampler( + *as_amd(context), + normalized_coords != 0, // To get rid of VS warning C4800 + addressing_mode, + filter_mode); + if (!sampler) { + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("not enough host memory"); + return (cl_sampler) 0; + } + + if (!sampler->create()) { + delete sampler; + *not_null(errcode_ret) = CL_OUT_OF_HOST_MEMORY; + LogWarning("Runtime failed sampler creation!"); + return as_cl(0); + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(sampler); +} +RUNTIME_EXIT + +/*! \brief Create a sampler object. + * + * \param context must be a valid OpenCL context. + * + * \param addressing_mode specifies how out of range image coordinates are + * handled when reading from an image. This can be set to CL_ADDRESS_REPEAT, + * CL_ADDRESS_CLAMP_TO_EDGE, CL_ADDRESS_CLAMP and CL_ADDRESS_NONE. + * + * \param filter_mode specifies the type of filter that must be applied when + * reading an image. This can be CL_FILTER_NEAREST or CL_FILTER_LINEAR. + * + * \param normalized_coords determines if the image coordinates specified are + * normalized (if \a normalized_coords is not zero) or not (if + * \a normalized_coords is zero). + * + * \param errcode_ret will return an appropriate error code. If \a errcode_ret + * is NULL, no error code is returned. + * + * \return A valid non-zero sampler object and \a errcode_ret is set to + * CL_SUCCESS if the sampler object is created successfully. It returns a NULL + * value with one of the following error values returned in \a errcode_ret: + * - CL_INVALID_CONTEXT if \a context is not a valid context. + * - CL_INVALID_VALUE if \a addressing_mode, \a filter_mode or + * \a normalized_coords or combination of these argument values are not + * valid. + * - CL_INVALID_OPERATION if images are not supported by any device associated + * with context + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the runtime. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY_RET(cl_sampler, clCreateSampler, ( + cl_context context, + cl_bool normalized_coords, + cl_addressing_mode addressing_mode, + cl_filter_mode filter_mode, + cl_int *errcode_ret)) +{ + const cl_sampler_properties sprops[] = { + CL_SAMPLER_NORMALIZED_COORDS, + static_cast(normalized_coords), + CL_SAMPLER_ADDRESSING_MODE, + static_cast(addressing_mode), + CL_SAMPLER_FILTER_MODE, + static_cast(filter_mode), + 0 }; + return clCreateSamplerWithProperties(context, sprops, errcode_ret); +} +RUNTIME_EXIT + +/*! \brief Increment the sampler reference count. + * + * clCreateSampler does an implicit retain. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_SAMPLER if \a sampler is not a valid sampler object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clRetainSampler, (cl_sampler sampler)) +{ + if (!is_valid(sampler)) { + return CL_INVALID_SAMPLER; + } + as_amd(sampler)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Decrement the sampler reference count. + * + * The sampler object is deleted after the reference count becomes zero and + * commands queued for execution on a command-queue(s) that use sampler have + * finished. + * + * \return CL_SUCCESS if the function is executed successfully. It returns + * CL_INVALID_SAMPLER if \a sampler is not a valid sampler object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clReleaseSampler, (cl_sampler sampler)) +{ + if (!is_valid(sampler)) { + return CL_INVALID_SAMPLER; + } + as_amd(sampler)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Return information about the sampler object. + * + * \param sampler specifies the sampler being queried. + * + * \param param_name specifies the information to query. + * + * \param param_value is a pointer to memory where the appropriate result + * being queried is returned. If \a param_value is NULL, it is ignored. + * + * \param param_value_size is used to specify the size in bytes of memory + * pointed to by \a param_value. This size must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data copied + * to \a param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully + * - CL_INVALID_VALUE if \a param_name is not valid, or if size in bytes + * specified by \a param_value_size is < size of return type and + * \a param_value is not NULL + * - CL_INVALID_SAMPLER if \a sampler is a not a valid sampler object. + * + * \version 1.0r33 + */ +RUNTIME_ENTRY(cl_int, clGetSamplerInfo, ( + cl_sampler sampler, + cl_sampler_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret)) +{ + if (!is_valid(sampler)) { + return CL_INVALID_SAMPLER; + } + + switch (param_name) + { + case CL_SAMPLER_REFERENCE_COUNT: { + cl_uint count = as_amd(sampler)->referenceCount(); + return amd::clGetInfo( + count, param_value_size, param_value, param_value_size_ret); + } + case CL_SAMPLER_CONTEXT: { + cl_context context = as_cl(&as_amd(sampler)->context()); + return amd::clGetInfo( + context, param_value_size, param_value, param_value_size_ret); + } + case CL_SAMPLER_ADDRESSING_MODE: { + cl_addressing_mode addressing = as_amd(sampler)->addressingMode(); + return amd::clGetInfo( + addressing, param_value_size, param_value, param_value_size_ret); + } + case CL_SAMPLER_FILTER_MODE: { + cl_filter_mode filter = as_amd(sampler)->filterMode(); + return amd::clGetInfo( + filter, param_value_size, param_value, param_value_size_ret); + } + case CL_SAMPLER_NORMALIZED_COORDS: { + cl_bool normalized = as_amd(sampler)->normalizedCoords(); + return amd::clGetInfo( + normalized, param_value_size, param_value, param_value_size_ret); + } + default: + break; + } + + return CL_INVALID_VALUE; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_sdi_amd.cpp b/opencl/api/opencl/amdocl/cl_sdi_amd.cpp new file mode 100644 index 0000000000..ddbe34129c --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_sdi_amd.cpp @@ -0,0 +1,259 @@ +#include "cl_common.hpp" +#include "cl_sdi_amd.h" +#include "platform/context.hpp" +#include "platform/command.hpp" +#include "platform/memory.hpp" +#include + + +RUNTIME_ENTRY(cl_int, clEnqueueWaitSignalAMD, ( + cl_command_queue command_queue, + cl_mem mem_object, + cl_uint value, + cl_uint num_events, + const cl_event * event_wait_list, + cl_event * event) ) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(mem_object)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Buffer* buffer = as_amd(mem_object)->asBuffer(); + if (buffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (!(buffer->getMemFlags() & CL_MEM_BUS_ADDRESSABLE_AMD)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != buffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList( eventWaitList, + hostQueue.context(), + num_events, + event_wait_list); + + if (err != CL_SUCCESS) { + return err; + } + + amd::SignalCommand * command = new amd::SignalCommand ( + hostQueue, + CL_COMMAND_WAIT_SIGNAL_AMD, + eventWaitList, + *buffer, + value); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_OUT_OF_RESOURCES; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; + +} +RUNTIME_EXIT + + +RUNTIME_ENTRY(cl_int, clEnqueueWriteSignalAMD, ( + cl_command_queue command_queue, + cl_mem mem_object, + cl_uint value, + cl_ulong offset, + cl_uint num_events, + const cl_event * event_wait_list, + cl_event * event)) + +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (!is_valid(mem_object)) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Buffer* buffer = as_amd(mem_object)->asBuffer(); + if (buffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (!(buffer->getMemFlags() & CL_MEM_EXTERNAL_PHYSICAL_AMD)) { + return CL_INVALID_MEM_OBJECT; + } + + if ((offset + sizeof(value)) > (buffer->getSize()+ amd::Os::pageSize())) { + return CL_INVALID_BUFFER_SIZE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + if (hostQueue.context() != buffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList( eventWaitList, + hostQueue.context(), + num_events, + event_wait_list); + + if (err != CL_SUCCESS) { + return err; + } + + amd::SignalCommand * command = new amd::SignalCommand ( + hostQueue, + CL_COMMAND_WRITE_SIGNAL_AMD, + eventWaitList, + *buffer, + value, + offset); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_OUT_OF_RESOURCES; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + + +RUNTIME_ENTRY(cl_int, clEnqueueMakeBuffersResidentAMD, ( + cl_command_queue command_queue, + cl_uint num_mem_objs, + cl_mem * mem_objects, + cl_bool blocking_make_resident, + cl_bus_address_amd * bus_addresses, + cl_uint num_events, + const cl_event * event_wait_list, + cl_event * event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (mem_objects == 0) { + return CL_INVALID_MEM_OBJECT; + } + + if (bus_addresses == 0 || num_mem_objs == 0) { + return CL_INVALID_VALUE; + } + + memset(bus_addresses,0, sizeof(cl_bus_address_amd)*num_mem_objs); + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + std::vector memObjects; + for (unsigned int i = 0; i < num_mem_objs; ++i) { + + if (!is_valid(mem_objects[i])) { + return CL_INVALID_MEM_OBJECT; + } + + amd::Buffer* buffer = as_amd(mem_objects[i])->asBuffer(); + if (buffer == NULL) { + return CL_INVALID_MEM_OBJECT; + } + + if (!(buffer->getMemFlags() & CL_MEM_BUS_ADDRESSABLE_AMD)) { + return CL_INVALID_MEM_OBJECT; + } + + if (hostQueue.context() != buffer->getContext()) { + return CL_INVALID_CONTEXT; + } + + memObjects.push_back(buffer); + + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList( eventWaitList, + hostQueue.context(), + num_events, + event_wait_list); + + if (err != CL_SUCCESS) { + return err; + } + + amd::MakeBuffersResidentCommand * command = + new amd::MakeBuffersResidentCommand ( hostQueue, + CL_COMMAND_MAKE_BUFFERS_RESIDENT_AMD, + eventWaitList, + memObjects, + bus_addresses); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_OUT_OF_RESOURCES; + } + + command->enqueue(); + + if (blocking_make_resident) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT \ No newline at end of file diff --git a/opencl/api/opencl/amdocl/cl_sdi_amd.h b/opencl/api/opencl/amdocl/cl_sdi_amd.h new file mode 100644 index 0000000000..d17d085c25 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_sdi_amd.h @@ -0,0 +1,48 @@ +#ifndef __CL_SDI_AMD_H +#define __CL_SDI_AMD_H + +#include "CL/cl_ext.h" + +#ifdef __cplusplus +extern "C" { +#endif /*__cplusplus*/ + + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWaitSignalAMD( + cl_command_queue command_queue, + cl_mem mem_object, + cl_uint value, + cl_uint num_events, + const cl_event * event_wait_list, + cl_event * event) CL_EXT_SUFFIX__VERSION_1_2; + + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteSignalAMD( + cl_command_queue command_queue, + cl_mem mem_object, + cl_uint value, + cl_ulong offset, + cl_uint num_events, + const cl_event * event_list, + cl_event * event) CL_EXT_SUFFIX__VERSION_1_2; + + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMakeBuffersResidentAMD( + cl_command_queue command_queue, + cl_uint num_mem_objs, + cl_mem * mem_objects, + cl_bool blocking_make_resident, + cl_bus_address_amd * bus_addresses, + cl_uint num_events, + const cl_event * event_list, + cl_event * event) CL_EXT_SUFFIX__VERSION_1_2; + + +#ifdef __cplusplus +} /*extern "C"*/ +#endif /*__cplusplus*/ + +#endif \ No newline at end of file diff --git a/opencl/api/opencl/amdocl/cl_semaphore_amd.h b/opencl/api/opencl/amdocl/cl_semaphore_amd.h new file mode 100644 index 0000000000..818e47928c --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_semaphore_amd.h @@ -0,0 +1,118 @@ +/* ============================================================ + +Copyright (c) 2012 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance + +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ +#ifndef __CL_SEMAPHORE_AMD_H +#define __CL_SEMAPHORE_AMD_H +/******************************************* + * AMD Extension cl_amd_semaphore + *******************************************/ +#define cl_amd_semaphore 1 + +#if cl_amd_semaphore + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + + /* cl_device_info */ +#define CL_DEVICE_MAX_SEMAPHORES_AMD 0xF050 +#define CL_DEVICE_MAX_SEMAPHORE_SIZE_AMD 0xF051 + + /* cl_kernel_work_group_info */ +#define CL_KERNEL_MAX_SEMAPHORE_SIZE_AMD 0xF052 + +#ifdef __cplusplus +} +#endif /* __cplusplus */ + +#endif /* cl_amd_semaphore */ + +#endif /* __CL_SEMAPHORE_AMD_H */ diff --git a/opencl/api/opencl/amdocl/cl_svm.cpp b/opencl/api/opencl/amdocl/cl_svm.cpp new file mode 100644 index 0000000000..728b548b18 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_svm.cpp @@ -0,0 +1,1091 @@ +// +// Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. +// +#include "cl_common.hpp" +#include "platform/command.hpp" +#include "platform/kernel.hpp" +#include "platform/program.hpp" + +/*! \brief Helper function to validate SVM allocation flags + * + * \return true if flags are valid, otherwise - false + */ +static bool +validateSvmFlags(cl_svm_mem_flags flags) +{ + if (!flags) { + // coarse-grained allocation + return true; + } + const cl_svm_mem_flags rwFlags = CL_MEM_READ_WRITE | CL_MEM_WRITE_ONLY | + CL_MEM_READ_ONLY; + const cl_svm_mem_flags setFlags = flags & (rwFlags | + CL_MEM_SVM_ATOMICS | CL_MEM_SVM_FINE_GRAIN_BUFFER); + if (flags != setFlags) { + // invalid flags value + return false; + } + + if (amd::countBitsSet(flags & rwFlags) > 1) { + // contradictory R/W flags + return false; + } + + if ((flags & CL_MEM_SVM_ATOMICS) + && !(flags & CL_MEM_SVM_FINE_GRAIN_BUFFER)) { + return false; + } + + return true; +} + +/*! \brief Helper function to validate cl_map_flags + * + * \return true if flags are valid, otherwise - false + */ +static bool +validateMapFlags(cl_map_flags flags) +{ + const cl_map_flags maxFlag = CL_MAP_WRITE_INVALIDATE_REGION; + if (flags >= (maxFlag << 1)) { + // at least one flag is out-of-range + return false; + } else if ((flags & CL_MAP_WRITE_INVALIDATE_REGION) && + (flags & (CL_MAP_READ | CL_MAP_WRITE))) { + // CL_MAP_READ or CL_MAP_WRITE and CL_MAP_WRITE_INVALIDATE_REGION are + // mutually exclusive. + return false; + } + return true; +} + +/*! \addtogroup API + * @{ + * + * \addtogroup SVM + * @{ + * + */ + +/*! \brief Allocate a shared virtual memory buffer that can be shared by the + * host and all devices in an OpenCL context. + * + * \param context is a valid OpenCL context used to create the SVM buffer. + * + * \param flags is a bit-field that is used to specify allocation and usage + * information. If CL_MEM_SVM_FINE_GRAIN_BUFFER is not specified, the + * buffer is created as a coarse grained SVM allocation. Similarly, if + * CL_MEM_SVM_ATOMICS is not specified, the buffer is created without + * support for SVM atomic operations. + * + * \param size is the size in bytes of the SVM buffer to be allocated. + * + * \param alignment is the minimum alignment in bytes that is required for the + * newly created buffer’s memory region. It must be a power of two up to the + * largest data type supported by the OpenCL device. For the full profile, the + * largest data type is long16. For the embedded profile, it is long16 if the + * device supports 64-bit integers; otherwise it is int16. If alignment is 0, a + * default alignment will be used that is equal to the size of largest data + * type supported by the OpenCL implementation. + * + * \return A valid non-NULL shared virtual memory address if the SVM buffer + * is successfully allocated. Otherwise, like malloc, it returns a NULL pointer + * value. clSVMAlloc will fail if + * - \a context is not a valid context. + * - \a flags does not contain CL_MEM_SVM_FINE_GRAIN_BUFFER but does + * contain CL_MEM_SVM_ATOMICS. + * - Values specified in \a flags do not follow rules for that particular type. + * - CL_MEM_SVM_FINE_GRAIN_BUFFER or CL_MEM_SVM_ATOMICS is specified + * in \a flags and these are not supported by at least one device in + * \a context. + * - The values specified in \a flags are not valid. + * - \a size is 0 or > CL_DEVICE_MAX_MEM_ALLOC_SIZE value for any device in + * \a context. + * - \a alignment is not a power of two or the OpenCL implementation cannot + * support the specified alignment for at least one device in \a context. + * - There was a failure to allocate resources. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY_RET_NOERRCODE(void*, clSVMAlloc, ( + cl_context context, + cl_svm_mem_flags flags, + size_t size, + unsigned int alignment)) +{ + if (!is_valid(context)) { + LogWarning("invalid parameter \"context\""); + return NULL; + } + + if (size == 0) { + LogWarning("invalid parameter \"size = 0\""); + return NULL; + } + + if (!validateSvmFlags(flags)) { + LogWarning("invalid parameter \"flags\""); + return NULL; + } + + if (!amd::isPowerOfTwo(alignment)) { + LogWarning("invalid parameter \"alignment\""); + return NULL; + } + + const std::vector& devices = as_amd(context)->devices(); + bool sizePass = false; + cl_device_svm_capabilities combinedSvmCapabilities = 0; + const cl_uint hostAddressBits = LP64_SWITCH(32, 64); + cl_uint minContextAlignment = UINT_MAX; + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + cl_device_svm_capabilities svmCapabilities = + (*it)->info().svmCapabilities_; + combinedSvmCapabilities |= svmCapabilities; + + if (((*it)->info().maxMemAllocSize_ >= size) || + (*it)->settings().largeHostMemAlloc_) { + sizePass = true; + } + + if ((*it)->info().addressBits_ < hostAddressBits) { + LogWarning("address mode mismatch between host and device"); + return NULL; + } + + // maximum alignment for a device is given in bits. + cl_uint baseAlignment = (*it)->info().memBaseAddrAlign_ >> 3; + if (alignment > baseAlignment) { + LogWarning("invalid parameter \"alignment\""); + return NULL; + } + + minContextAlignment = std::min(minContextAlignment, baseAlignment); + } + if ((flags & CL_MEM_SVM_FINE_GRAIN_BUFFER) && + !(combinedSvmCapabilities & CL_DEVICE_SVM_FINE_GRAIN_BUFFER)) { + LogWarning("No device in context supports SVM fine grained buffers"); + return NULL; + } + if ((flags & CL_MEM_SVM_ATOMICS) && + !(combinedSvmCapabilities & CL_DEVICE_SVM_ATOMICS)) { + LogWarning("No device in context supports SVM atomics"); + return NULL; + } + if (!sizePass) { + LogWarning("invalid parameter \"size\""); + return NULL; + } + + // if alignment not specified, use largest data type alignment supported + if (alignment == 0) { + alignment = minContextAlignment; + LogPrintfInfo("Assumed alignment %d\n", alignment); + } + + amd::Context& amdContext = *as_amd(context); + return amd::SvmBuffer::malloc(amdContext, flags, size, alignment); +} +RUNTIME_EXIT + +/*! \brief Free a shared virtual memory buffer allocated using clSVMAlloc. + * + * \param context is a valid OpenCL context used to create the SVM buffer. + * + * \param svm_pointer must be the value returned by a call to clSVMAlloc. If a + * NULL pointer is passed in \a svm_pointer, no action occurs. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY_VOID(void, clSVMFree, ( + cl_context context, void* svm_pointer)) +{ + if (!is_valid(context)) { + LogWarning("invalid parameter \"context\""); + return; + } + + if (svm_pointer == NULL) { + return; + } + + amd::Context& amdContext = *as_amd(context); + amd::SvmBuffer::free(amdContext, svm_pointer); +} +RUNTIME_EXIT + +/*! \brief enqueues a command to free shared virtual memory allocated using + * clSVMAlloc or a shared system memory pointer. + * + * \param command_queue is a valid host command-queue. + * + * \param num_svm_pointers specifies the number of elements in \a svm_pointers. + * + * \param svm_pointers is a list of shared virtual memory pointers to + * be freed. Each pointer in \a svm_pointers that was allocated using SVMAlloc + * must have been allocated from the same context from which \a command_queue + * was created. The memory associated with \a svm_pointers can be reused or + * freed after the function returns. + * + * \param pfn_free_func specifies the callback function to be called to free + * the SVM pointers. \a pfn_free_func takes four arguments: \a queue which is + * the command queue in which clEnqueueSVMFree was enqueued, the count and list + * of SVM pointers to free and \a user_data which is a pointer to user + * specified data. If \a pfn_free_func is NULL, all the pointers specified in + * \a svm_pointers array must be allocated using clSVMAlloc. \a pfn_free_func + * must however be a valid callback function if any SVM pointer to be freed is + * a shared system memory pointer i.e. not allocated using clSVMAlloc. + * + * \param user_data will be passed as the user_data argument when + * \a pfn_free_func is called. \a user_data can be NULL. + * + * \param even_wait_list specifies the events that need to complete before + * this particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The memory + * associated with \a event_wait_list can be reused or freed after the function + * returns. + * + * \param num_events_in_wait_list specifies the number of elements in + * \a even_wait_list + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. If the \a event_wait_list and the \a event arguments + * are not NULL, the \a event argument should not refer to an element of the + * \a event_wait_list array. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid host + * command-queue + * - CL_INVALID_VALUE if \a num_svm_pointers is 0 or if \a svm_pointers is + * NULL or if any of the pointers specified in \a svm_pointers array is NULL + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * events in \a event_wait_list are not the same + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clEnqueueSVMFree, ( + cl_command_queue command_queue, + cl_uint num_svm_pointers, + void* svm_pointers[], + void (CL_CALLBACK *pfn_free_func)( + cl_command_queue queue, + cl_uint num_svm_pointers, + void* svm_pointers[], + void* user_data), + void* user_data, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (num_svm_pointers == 0) { + LogWarning("invalid parameter \"num_svm_pointers = 0\""); + return CL_INVALID_VALUE; + } + + if (svm_pointers == NULL) { + LogWarning("invalid parameter \"svm_pointers = NULL\""); + return CL_INVALID_VALUE; + } + + //!@todo why are NULL pointers disallowed here but not in clSVMFree? + for (cl_uint i = 0; i < num_svm_pointers; i++) { + if (svm_pointers[i] == NULL) { + LogWarning("Null pointers are not allowed"); + return CL_INVALID_VALUE; + } + } + + //!@todo what if the callback is NULL but \a user_data is not? + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::Command *command = new amd::SvmFreeMemoryCommand( + hostQueue, + eventWaitList, + num_svm_pointers, + svm_pointers, + pfn_free_func, + user_data); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief enqueues a command to do a memcpy operation. + * + * \param command_queue refers to the host command-queue in which the read/ + * write commands will be queued. + * + * \param blocking_copy indicates if the copy operation is blocking or + * non-blocking. If \a blocking_copy is CL_TRUE i.e. the copy command is + * blocking, clEnqueueSVMMemcpy does not return until the buffer data has been + * copied into memory pointed to by \a dst_ptr. If \a blocking_copy is CL_FALSE + * i.e. the copy command is non-blocking, clEnqueueSVMMemcpy queues a + * non-blocking copy command and returns. The contents of the buffer that + * \a dst_ptr point to cannot be used until the copy command has completed. + * The \a event argument returns an event object which can be used to query the + * execution status of the read command. When the copy command has completed, + * the contents of the buffer that \a dst_ptr points to can be used by the + * application. + * + * \param dst_ptr is the pointer to a memory region where data is copied to. + * + * \param src_ptr is the pointer to a memory region where data is copied from. + * If \a dst_ptr and/or \a src_ptr are allocated using clSVMAlloc then they + * must be allocated from the same context from which \a command_queue was + * created. Otherwise the behavior is undefined. + * + * \param size is the size in bytes of data being copied. + * + * \param even_wait_list specifies the events that need to complete before + * this particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The memory + * associated with \a event_wait_list can be reused or freed after the function + * returns. + * + * \param num_events_in_wait_list specifies the number of elements in + * \a even_wait_list + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. If the \a event_wait_list and the \a event arguments + * are not NULL, the \a event argument should not refer to an element of the + * \a event_wait_list array. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid host + * command-queue + * - CL_INVALID_CONTEXT if the context associated with \a command_queue and + * events in \a event_wait_list are not the same + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST if the operation is + * blocking and the execution status of any of the events in + * \a event_wait_list is a negative integer value. + * - CL_INVALID_VALUE if \a dst_ptr or \a src_ptr are NULL. + * - CL_INVALID_VALUE if \a size is 0. + * - CL_MEM_COPY_OVERLAP if the values specified for \a dst_ptr, \a src_ptr + * and \a size result in an overlapping copy. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clEnqueueSVMMemcpy, ( + cl_command_queue command_queue, + cl_bool blocking_copy, + void* dst_ptr, + const void* src_ptr, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (dst_ptr == NULL || src_ptr == NULL) { + return CL_INVALID_VALUE; + } + + if (size == 0) { + return CL_INVALID_VALUE; + } + + char* dst = reinterpret_cast(dst_ptr); + const char* src = reinterpret_cast(src_ptr); + if ((dst > src - size) && (dst < src + size)) { + return CL_MEM_COPY_OVERLAP; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::Command *command = new amd::SvmCopyMemoryCommand( + hostQueue, + eventWaitList, + dst_ptr, + src_ptr, + size); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + + if (blocking_copy) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief enqueues a command to fill a region in memory with a pattern of a + * given pattern size. + * + * \param command_queue refers to the host command-queue in which the fill + * command will be queued. The OpenCL context associated with \a command_queue + * and SVM pointer referred to by \a svm_ptr must be the same.. + * + * \param svm_ptr is a pointer to a memory region that will be filled with + * \a pattern. It must be aligned to \a pattern_size bytes. If \a svm_ptr is + * allocated using clSVMAlloc then it must be allocated from the same context + * from which \a command_queue was created. Otherwise the behavior is + * undefined. + * + * \a pattern is a pointer to the data pattern of size \a pattern_size in + * bytes. \a pattern will be used to fill a region in buffer starting at + * \a svm_ptr and is \a size bytes in size. The data pattern must be a scalar + * or vector integer or floating-point data type supported by OpenCL. For + * example, if region pointed to by \a svm_ptr is to be filled with a pattern + * of float4 values, then \a pattern will be a pointer to a cl_float4 value + * and \a pattern_size will be sizeof(cl_float4). The maximum value of + * \a pattern_size is the size of the largest integer or floating-point vector + * data type supported by the OpenCL device. The memory associated with + * \a pattern can be reused or freed after the function returns. + * + * \param size is the size in bytes of region being filled starting with + * \a svm_ptr and must be a multiple of \a pattern_size. + * + * \param even_wait_list specifies the events that need to complete before + * this particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The memory + * associated with \a event_wait_list can be reused or freed after the function + * returns. + * + * \param num_events_in_wait_list specifies the number of elements in + * \a even_wait_list + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. clEnqueueBarrierWithWaitList can be used instead. If + * the \a event_wait_list and the \a event arguments are not NULL, the \a event + * argument should not refer to an element of the \a event_wait_list array. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid host + * command-queue + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * events in \a event_wait_list are not the same + * - CL_INVALID_VALUE if \a svm_ptr is NULL. + * - CL_INVALID_VALUE if \a svm_ptr is not aligned to \a pattern_size bytes. + * - CL_INVALID_VALUE if \a pattern is NULL or if \a pattern_size is 0 or if + * \a pattern_size is not one of {1, 2, 4, 8, 16, 32, 64, 128}. + * - CL_INVALID_VALUE if \a size is 0 or is not a multiple of \a pattern_size. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clEnqueueSVMMemFill, ( + cl_command_queue command_queue, + void* svm_ptr, + const void* pattern, + size_t pattern_size, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (svm_ptr == NULL) { + return CL_INVALID_VALUE; + } + + char* dst = reinterpret_cast(svm_ptr); + if (!amd::isMultipleOf(dst, pattern_size)) { + return CL_INVALID_VALUE; + } + + if (pattern == NULL) { + return CL_INVALID_VALUE; + } + + if (!amd::isPowerOfTwo(pattern_size) || pattern_size == 0 + || pattern_size > amd::FillMemoryCommand::MaxFillPatterSize) { + return CL_INVALID_VALUE; + } + + if (size == 0 || !amd::isMultipleOf(size, pattern_size)) { + return CL_INVALID_VALUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amd::Command *command = new amd::SvmFillMemoryCommand( + hostQueue, + eventWaitList, + svm_ptr, + pattern, + pattern_size, + size); + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief enqueues a command that will allow the host to update a region of a + * SVM buffer + * + * \param command_queue is a valid host command-queue. + * + * \param blocking_map indicates if the map operation is blocking or + * non-blocking. If \a blocking_map is CL_TRUE, clEnqueueSVMMap does not return + * until the application can access the contents of the SVM region specified by + * \a svm_ptr and \a size on the host. If blocking_map is CL_FALSE i.e. map + * operation is non-blocking, the region specified by \a svm_ptr and \a size + * cannot be used until the map command has completed. The \a event argument + * returns an event object which can be used to query the execution status of + * the map command. When the map command is completed, the application can + * access the contents of the region specified by \a svm_ptr and \a size. + * + * \param maps_flag is a valid cl_map_flags flag. + * + * \param svm_ptr is a pointer to a memory region that will be updated by the + * host. If \a svm_ptr is allocated using clSVMAlloc then it must be allocated + * from the same context from which \a command_queue was created. Otherwise + * the behavior is undefined. + * + * \param size is the size in bytes of the memory region that will be updated + * by the host. + * + * \param even_wait_list specifies the events that need to complete before + * this particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The memory + * associated with \a event_wait_list can be reused or freed after the function + * returns. + * + * \param num_events_in_wait_list specifies the number of elements in + * \a even_wait_list + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. clEnqueueBarrierWithWaitList can be used instead. If + * the \a event_wait_list and the \a event arguments are not NULL, the \a event + * argument should not refer to an element of the \a event_wait_list array. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid host + * command-queue + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * events in \a event_wait_list are not the same + * - CL_INVALID_VALUE if \a svm_ptr is NULL. + * - CL_INVALID_VALUE if \a size is 0 or if values specified in \a map_flags + * are not valid. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST if the operation is + * blocking and the execution status of any of the events in + * \a event_wait_list is a negative integer value. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clEnqueueSVMMap, ( + cl_command_queue command_queue, + cl_bool blocking_map, + cl_map_flags map_flags, + void* svm_ptr, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (svm_ptr == NULL) { + return CL_INVALID_VALUE; + } + + if (size == 0) { + return CL_INVALID_VALUE; + } + + if (!validateMapFlags(map_flags)) { + return CL_INVALID_VALUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + size_t offset = 0; + + //make sure the context is the same as the context of creation of svm space + amd::Memory * svmMem = amd::SvmManager::FindSvmBuffer(svm_ptr); + if (NULL != svmMem) { + if (hostQueue.context() != svmMem->getContext()) { + LogWarning("different contexts"); + return CL_INVALID_CONTEXT; + } + + svmMem->commitSvmMemory(); + offset = static_cast
(svm_ptr) - static_cast
(svmMem->getSvmPtr()); + if (offset < 0 || offset + size > svmMem->getSize()) { + LogWarning("wrong svm address "); + return CL_INVALID_VALUE; + } + amd::Buffer* srcBuffer = svmMem->asBuffer(); + + amd::Coord3D srcSize(size); + amd::Coord3D srcOffset(offset); + if (NULL != srcBuffer) { + if (!srcBuffer->validateRegion(srcOffset, srcSize)) { + return CL_INVALID_VALUE; + } + } + + // Make sure we have memory for the command execution + device::Memory* mem = svmMem->getDeviceMemory(queue->device()); + if (NULL == mem) { + LogPrintfError("Can't allocate memory size - 0x%08X bytes!", + svmMem->getSize()); + return CL_OUT_OF_RESOURCES; + } + // Attempt to allocate the map target now (whether blocking or non-blocking) + void* mapPtr = (queue->device()).allocMapTarget(*svmMem, srcOffset, srcSize); + if (NULL == mapPtr || mapPtr != svm_ptr) { + return CL_OUT_OF_RESOURCES; + } + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::Command* command = new amd::SvmMapMemoryCommand( + hostQueue, eventWaitList, svmMem, size, offset, map_flags); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + + if (blocking_map) { + command->awaitCompletion(); + } + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief enqueues a command to indicate that the host has completed updating + * a memory region which was specified in a previous call to clEnqueueSVMUnmap. + * + * \param command_queue is a valid host command-queue. + * + * \param svm_ptr is a pointer that was specified in a previous call to + * clEnqueueSVMMap. If \a svm_ptr is allocated using clSVMAlloc then it must be + * allocated from the same context from which \a command_queue was created. + * Otherwise the behavior is undefined. + * + * \param even_wait_list specifies the events that need to complete before + * this particular command can be executed. If \a event_wait_list is NULL, then + * this particular command does not wait on any event to complete. If + * \a event_wait_list is NULL, \a num_events_in_wait_list must be 0. If + * \a event_wait_list is not NULL, the list of events pointed to by + * \a event_wait_list must be valid and \a num_events_in_wait_list must be + * greater than 0. The events specified in \a event_wait_list act as + * synchronization points. The context associated with events in + * \a event_wait_list and \a command_queue must be the same. The memory + * associated with \a event_wait_list can be reused or freed after the function + * returns. + * + * \param num_events_in_wait_list specifies the number of elements in + * \a even_wait_list + * + * \param event returns an event object that identifies this particular command + * and can be used to query or queue a wait for this particular command to + * complete. \a event can be NULL in which case it will not be possible for the + * application to query the status of this command or queue a wait for this + * command to complete. clEnqueueBarrierWithWaitList can be used instead. If + * the \a event_wait_list and the \a event arguments are not NULL, the \a event + * argument should not refer to an element of the \a event_wait_list array. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_COMMAND_QUEUE if \a command_queue is not a valid host + * command-queue + * - CL_INVALID_CONTEXT if context associated with \a command_queue and + * events in \a event_wait_list are not the same + * - CL_INVALID_VALUE if \a svm_ptr is NULL. + * - CL_INVALID_EVENT_WAIT_LIST if \a event_wait_list is NULL and + * \a num_events_in_wait_list > 0, or \a event_wait_list is not NULL and + * \a num_events_in_wait_list is 0, or if event objects in \a event_wait_list + * are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clEnqueueSVMUnmap, ( + cl_command_queue command_queue, + void* svm_ptr, + cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, + cl_event* event)) +{ + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + if (svm_ptr == NULL) { + return CL_INVALID_VALUE; + } + + amd::HostQueue* queue = as_amd(command_queue)->asHostQueue(); + if (NULL == queue) { + return CL_INVALID_COMMAND_QUEUE; + } + amd::HostQueue& hostQueue = *queue; + + //check if the ptr is in the svm space + amd::Memory * svmMem = amd::SvmManager::FindSvmBuffer(svm_ptr); + // Make sure we have memory for the command execution + if (NULL != svmMem) { + // Make sure we have memory for the command execution + device::Memory* mem = svmMem->getDeviceMemory(queue->device()); + if (NULL == mem) { + LogPrintfError("Can't allocate memory size - 0x%08X bytes!", + svmMem->getSize()); + return CL_INVALID_VALUE; + } + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue.context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS) { + return err; + } + + amd::Command* command = new amd::SvmUnmapMemoryCommand( + hostQueue, eventWaitList, svmMem); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Set the argument value for a specific argument of a kernel to be + * a SVM pointer. + * + * \param kernel is a valid kernel object. + * + * \param arg_index is the argument index. Arguments to the kernel are referred + * by indices that go from 0 for the leftmost argument to n - 1, where n is the + * total number of arguments declared by a kernel. + * + * \param arg_value is the SVM pointer that should be used as the argument + * value for argument specified by \a arg_index. The SVM pointer pointed to by + * \a arg_value is copied and the \a arg_value pointer can therefore be reused + * by the application after clSetKernelArgSVMPointer returns. The SVM pointer + * specified is the value used by all API calls that enqueue kernel + * (clEnqueueNDRangeKernel) until the argument value is changed by a call to + * clSetKernelArgSVMPointer for \a kernel. The SVM pointer can only be used for + * arguments that are declared to be a pointer to global or constant memory. + * The SVM pointer value must be aligned according to the argument’s type. For + * example, if the argument is declared to be global float4 *p, the SVM pointer + * value passed for p must be at a minimum aligned to a float4. The SVM pointer + * value specified as the argument value can be the pointer returned by + * clSVMAlloc or can be a pointer + offset into the SVM region. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_KERNEL if \a kernel is not a valid kernel object + * - CL_INVALID_ARG_INDEX if \a arg_index is not a valid argument index + * - CL_INVALID_ARG_VALUE if \a arg_value is not a valid value + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clSetKernelArgSVMPointer, ( + cl_kernel kernel, + cl_uint arg_index, + const void * arg_value)) +{ + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + const amd::KernelSignature& signature = as_amd(kernel)->signature(); + if (arg_index >= signature.numParameters()) { + return CL_INVALID_ARG_INDEX; + } + + const amd::KernelParameterDescriptor& desc = signature.at(arg_index); + if (desc.type_ != T_POINTER || !(desc.addressQualifier_ & + (CL_KERNEL_ARG_ADDRESS_GLOBAL | CL_KERNEL_ARG_ADDRESS_CONSTANT))) { + return CL_INVALID_ARG_VALUE; + } + + as_amd(kernel)->parameters().reset(static_cast(arg_index)); + + //! @todo We need to check that the alignment of \a arg_value. For instance, + // if the argument is of type 'global float4*', then \a arg_value must be + // aligned to sizeof(float4*). Note that desc.size_ contains the size of the + // pointer type itself and the size of the pointed type. + + + // We do not perform additional pointer validations: + // -verifying pointers returned by SVMAlloc would imply keeping track + // of every allocation range and then matching the pointer against that + // range. Note that even if the pointer would look correct, nothing + // prevents the user from using an offset within the kernel that would + // result on an invalid access. + // -verifying system pointers (if supported) requires matching the pointer + // against the address space of the current process. + + as_amd(kernel)->parameters().set( + static_cast(arg_index), sizeof(arg_value), arg_value, true); + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Pass additional information other than argument values to a kernel. + * + * \param kernel is a valid kernel object. + * + * \param param_name specifies the information to be passed to \a kernel. It + * must be a cl_kernel_exec_info value. + * + * \param param_value_size specifies the size in bytes of the memory pointed to + * by \a param_value. + * + * \param param_value is a pointer to memory where the appropiate values + * determined by \a param_name are specified. + * + * \return One of the following values: + * - CL_SUCCESS if the function was executed successfully + * - CL_INVALID_KERNEL if \a kernel is not a valid kernel object. + * - CL_INVALID_VALUE if \a param_name is not valid, if \a param_value is + * NULL or if the size specified by \a param_value_size is not valid + * - CL_INVALID_OPERATION if \a param_name is + * CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM and \a param_value = CL_TRUE + * but no devices in context associated with \a kernel support fine-grained + * system SVM allocations + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required + * by the OpenCL implementation on the device + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required + * by the OpenCL implementation on the host. + * + * \version 2.0r15 + */ +RUNTIME_ENTRY(cl_int, clSetKernelExecInfo, ( + cl_kernel kernel, + cl_kernel_exec_info param_name, + size_t param_value_size, + const void* param_value)) +{ + if (!is_valid(kernel)) { + return CL_INVALID_KERNEL; + } + + if (param_name != CL_KERNEL_EXEC_INFO_SVM_PTRS && + param_name != CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM) { + return CL_INVALID_VALUE; + } + + if (param_value == NULL) { + return CL_INVALID_VALUE; + } + + const amd::Kernel* amdKernel = as_amd(kernel); + + if (param_name == CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM) { + if (param_value_size != sizeof(cl_bool)) { + return CL_INVALID_VALUE; + } + + const bool flag = *(static_cast(param_value)); + const amd::Context* amdContext = &amdKernel->program().context(); + bool foundFineGrainedSystemDevice = false; + const std::vector& devices = amdContext->devices(); + std::vector::const_iterator it; + for (it = devices.begin(); it != devices.end(); ++it) { + if ((*it)->info().svmCapabilities_ & + CL_DEVICE_SVM_FINE_GRAIN_SYSTEM) { + foundFineGrainedSystemDevice = true; + break; + } + } + if (flag && !foundFineGrainedSystemDevice) { + return CL_INVALID_OPERATION; + } + amdKernel->parameters().setSvmSystemPointersSupport(flag ? FGS_YES : FGS_NO); + } + else { + if (param_value_size == 0 || !amd::isMultipleOf(param_value_size, + sizeof(void*))) { + return CL_INVALID_VALUE; + } + + size_t count = param_value_size/sizeof(void*); + void* const* execInfoArray = reinterpret_cast(param_value); + for (size_t i = 0; i < count; i++) { + if (NULL == execInfoArray[i]) { + return CL_INVALID_VALUE; + } + } + amdKernel->parameters().addSvmPtr(execInfoArray, count); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_thread_trace_amd.cpp b/opencl/api/opencl/amdocl/cl_thread_trace_amd.cpp new file mode 100644 index 0000000000..f1e1a4ca91 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_thread_trace_amd.cpp @@ -0,0 +1,498 @@ +// +// Copyright (c) 2009 Advanced Micro Devices, Inc. All rights reserved. +// +#include "cl_common.hpp" +#include "cl_thread_trace_amd.h" +#include "platform/context.hpp" +#include "platform/command.hpp" +#include "platform/threadtrace.hpp" +#include +#include + +/*! \addtogroup API + * @{ + * + * \addtogroup AMD_Extensions + * @{ + * + */ + +/*! \brief Creates a new HW threadTrace + * + * \param device must be a valid OpenCL device. + * + * \param threadTrace the created cl_threadtrace_amd object + * + * \param errcode_ret A non zero value if OpenCL failed to create cl_threadtrace_amd + * - CL_SUCCESS if the function is executed successfully. + * - CL_INVALID_DEVICE if the specified context is invalid. + * - CL_INVALID_OPERATION if we couldn't create the object + * + * \return Created cl_threadtrace_amd object + */ +RUNTIME_ENTRY_RET(cl_threadtrace_amd, clCreateThreadTraceAMD, ( + cl_device_id device, + cl_int* errcode_ret)) +{ + // Make sure we have a valid device object + if (!is_valid(device)) { + *not_null(errcode_ret) = CL_INVALID_DEVICE; + return NULL; + } + + // Create the device thread trace object + amd::ThreadTrace* threadTrace = + new amd::ThreadTrace(*as_amd(device)); + + if (threadTrace == NULL) { + *not_null(errcode_ret) = CL_INVALID_OPERATION; + return NULL; + } + + *not_null(errcode_ret) = CL_SUCCESS; + return as_cl(threadTrace); +} +RUNTIME_EXIT + +///*! \brief Destroy a threadTrace object. +// * +// * \param threadTrace the cl_threadtrace_amd object for release +// * +// * \return A non zero value if OpenCL failed to release cl_threadtrace_amd +// * - CL_SUCCESS if the function is executed successfully. +// * - CL_INVALID_OPERATION if we failed to release the object +// */ +RUNTIME_ENTRY(cl_int, clReleaseThreadTraceAMD, ( + cl_threadtrace_amd threadTrace)) +{ + if (!is_valid(threadTrace)) { + return CL_INVALID_OPERATION; + } + as_amd(threadTrace)->release(); + return CL_SUCCESS; +} +RUNTIME_EXIT +// +// *! \brief Increments the cl_threadtrace_amd object reference count. +// * +// * \param threadTrace the cl_threadtrace_amd object for retain +// * +// * \return A non zero value if OpenCL failed to retain cl_threadtrace_amd +// * - CL_SUCCESS if the function is executed successfully. +// * - CL_INVALID_OPERATION if we failed to release the object +// */ +RUNTIME_ENTRY(cl_int, clRetainThreadTraceAMD, ( + cl_threadtrace_amd threadTrace)) +{ + if (!is_valid(threadTrace)) { + return CL_INVALID_OPERATION; + } + as_amd(threadTrace)->retain(); + return CL_SUCCESS; +} +RUNTIME_EXIT + +// +// *! \brief Sets the cl_threadtrace_amd object configuration parameter. +// * +// * \param thread_trace the cl_threadtrace_amd object to set configuration parameter +// * +// * \param config_param the cl_thread_trace_param +// * +// * \param param_value corresponding to configParam +// * +// * \return A non zero value if OpenCL failed to set threadTrace buffer parameter +// * - CL_INVALID_VALUE if the thread_trace is invalid thread trace object. +// * - CL_INVALID_VALUE if the invalid config_param or param_value enum values , are used. +// * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and num_events_in_wait_list > 0, or event_wait_list is not NULL and num_events_in_wait_list is 0, +// * - or if event objects in event_wait_list are not valid events. +// * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the OpenCL implementation on the device. +// * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the +// OpenCL implementation on the host. +// */ +RUNTIME_ENTRY(cl_int, clSetThreadTraceParamAMD,(cl_threadtrace_amd thread_trace , + cl_thread_trace_param config_param , + cl_uint param_value )) +{ + if (!is_valid(thread_trace)) { + return CL_INVALID_OPERATION; + } + switch (config_param) { + case CL_THREAD_TRACE_PARAM_TOKEN_MASK: + if (param_value > CL_THREAD_TRACE_TOKEN_MASK_ALL_SI) { + return CL_INVALID_VALUE; + } + as_amd(thread_trace)->setTokenMask(param_value); + break; + case CL_THREAD_TRACE_PARAM_REG_MASK: + if (param_value > CL_THREAD_TRACE_REG_MASK_ALL_SI) { + return CL_INVALID_VALUE; + } + as_amd(thread_trace)->setRegMask(param_value); + break; + case CL_THREAD_TRACE_PARAM_VM_ID_MASK: + if (param_value > CL_THREAD_TRACE_VM_ID_MASK_SINGLE_DETAIL) { + return CL_INVALID_VALUE; + } + as_amd(thread_trace)->setVmIdMask(param_value); + break; + case CL_THREAD_TRACE_PARAM_INSTRUCTION_MASK: + if (param_value > CL_THREAD_TRACE_INST_MASK_IMMEDIATE_CI) { + return CL_INVALID_VALUE; + } + as_amd(thread_trace)->setInstMask(param_value); + break; + case CL_THREAD_TRACE_PARAM_COMPUTE_UNIT_TARGET: + as_amd(thread_trace)->setCU(param_value); + break; + case CL_THREAD_TRACE_PARAM_SHADER_ARRAY_TARGET: + as_amd(thread_trace)->setSH(param_value); + break; + case CL_THREAD_TRACE_PARAM_SIMD_MASK: + as_amd(thread_trace)->setSIMD(param_value); + break; + case CL_THREAD_TRACE_PARAM_USER_DATA: + as_amd(thread_trace)->setUserData(param_value); + break; + case CL_THREAD_TRACE_PARAM_CAPTURE_MODE: + if (param_value > CL_THREAD_TRACE_CAPTURE_SELECT_DETAIL) { + return CL_INVALID_VALUE; + } + as_amd(thread_trace)->setCaptureMode(param_value); + break; + case CL_THREAD_TRACE_PARAM_IS_WRAPPED: + as_amd(thread_trace)->setIsWrapped(true); + break; + case CL_THREAD_TRACE_PARAM_RANDOM_SEED: + as_amd(thread_trace)->setRandomSeed(param_value); + break; +} + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! \brief Get specific information about the OpenCL Thread Trace. + * + * \param threadTrace_info_param is an enum that identifies the Thread Trace information being + * queried. + * + * \param param_value is a pointer to memory location where appropriate values + * for a given \a threadTrace_info_param will be returned. If \a param_value is NULL, + * it is ignored. + * + * \param param_value_size specifies the size in bytes of memory pointed to by + * \a param_value. This size in bytes must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * CL_INVALID_OPERATION if cl_threadtrace_amd object is not valid + * - CL_INVALID_VALUE if \a param_name is not one of the supported + * values or if size in bytes specified by \a param_value_size is < size of + * return type and \a param_value is not a NULL value. + * - CL_SUCCESS if the function is executed successfully. + * + */ +RUNTIME_ENTRY(cl_int, clGetThreadTraceInfoAMD, ( + cl_threadtrace_amd thread_trace /* threadTrace */, + cl_threadtrace_info thread_trace_info_param, + size_t param_value_size, + void* param_value, + size_t* param_value_size_ret)) +{ + if (!is_valid(thread_trace)) { + return CL_INVALID_OPERATION; + } + + // Find the thread trace object, associated with the specified device + const device::ThreadTrace* devThreadTrace = + as_amd(thread_trace)->getDeviceThreadTrace(); + + const size_t seNum = as_amd(thread_trace)->deviceSeNumThreadTrace(); + switch (thread_trace_info_param) { + case CL_THREAD_TRACE_SE: + { + return amd::clGetInfo(seNum, + param_value_size, param_value, param_value_size_ret); + break; + } + case CL_THREAD_TRACE_BUFFERS_SIZE: + { + // Make sure we found a valid thread trace object + if (devThreadTrace == NULL) { + return CL_INVALID_OPERATION; + } + unsigned int bufSize2Se[2] = {0}; + + if (!devThreadTrace->info(thread_trace_info_param,bufSize2Se,seNum)) { + return CL_INVALID_VALUE; + } + return amd::clGetInfo(bufSize2Se, + param_value_size, param_value, param_value_size_ret); + break; + } + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +/* \brief Enqueues the command for the specified thread trace object. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param thread_trace specifies the cl_threadtrace_amd object. + * + * \param event_wait_list specify [is a pointer to] events that need to + * complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait + * on any event to complete. If \a event_wait_list is NULL, + * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, + * the list of events pointed to by \a event_wait_list must be valid and + * \a num_events_in_wait_list must be greater than 0. The events specified in + * \a event_wait_list act as synchronization points. + * + * \param num_events_in_wait_list specify the number of events in + * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be + * greater than 0 if \a event_wait_list is not NULL. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. + * \return A non zero value if OpenCL failed to release threadTrace + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with command_queue and events in event_wait_list are not the same. + * - CL_INVALID_VALUE if the thread_trace is invalid thread trace object . + * - CL_INVALID_VALUE if the invalid command name enum value , not described in the cl_threadtrace_command_name_amd, is used. + * - CL_INVALID_OPERATION if the command enqueue failed. It can happen in the following cases: + * o BEGIN_COMMAND is queued for thread trace object for which memory object/s was/were not bound.. + * o END_COMMAND is queued for thread trace object, for which BEGIN_COMMAND was not queued. + * o PAUSE_COMMAND is queued for thread trace object, for which BEGIN_COMMAND was not queued. + * o RESUME_COMMAND is queued for thread trace object, for which PAUSE_COMMAND was not queued. + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and num_events_in_wait_list > 0, or event_wait_list is not NULL and num_events_in_wait_list is 0, or if event objects in event_wait_list are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the OpenCL implementation on the host. + */ + +RUNTIME_ENTRY(cl_int, clEnqueueThreadTraceCommandAMD ,( + cl_command_queue command_queue , + cl_threadtrace_amd thread_trace , + cl_threadtrace_command_name_amd command_name , + cl_uint num_events_in_wait_list , + const cl_event* event_wait_list , + cl_event* event )) +{ + // Check if command queue is valid + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + // Check if thread trace is valid + if (!is_valid(thread_trace)) { + return CL_INVALID_OPERATION; + } + + amd::ThreadTrace* amdThreadTrace = as_amd(thread_trace); + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + // Check that device associated with the command queue is the same as with thread trace + if (&hostQueue->device() != &amdThreadTrace->device()) { + return CL_INVALID_DEVICE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue->context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + // Create a new command for the threadTraces + amd::ThreadTraceCommand* command = NULL; + switch (command_name) { + case CL_THREAD_TRACE_BEGIN_COMMAND : + if ((amdThreadTrace->getState() != amd::ThreadTrace::MemoryBound) + &&(amdThreadTrace->getState() != amd::ThreadTrace::End)) { + return CL_INVALID_OPERATION; + } + amdThreadTrace->setState(amd::ThreadTrace::Begin); + command = new amd::ThreadTraceCommand(*hostQueue, eventWaitList,static_cast(&amdThreadTrace->threadTraceConfig()), + *amdThreadTrace,amd::ThreadTraceCommand::Begin,CL_COMMAND_THREAD_TRACE); + break; + case CL_THREAD_TRACE_END_COMMAND : + if ((amdThreadTrace->getState() != amd::ThreadTrace::Begin) + &&(amdThreadTrace->getState() != amd::ThreadTrace::Pause)) { + return CL_INVALID_OPERATION; + } + amdThreadTrace->setState(amd::ThreadTrace::End); + command = new amd::ThreadTraceCommand(*hostQueue, eventWaitList,&amdThreadTrace->threadTraceConfig(), + *amdThreadTrace,amd::ThreadTraceCommand::End,CL_COMMAND_THREAD_TRACE); + break; + case CL_THREAD_TRACE_PAUSE_COMMAND : + if (amdThreadTrace->getState() != amd::ThreadTrace::Begin) { + return CL_INVALID_OPERATION; + } + amdThreadTrace->setState(amd::ThreadTrace::Pause); + command = new amd::ThreadTraceCommand(*hostQueue, eventWaitList,&amdThreadTrace->threadTraceConfig(), + *amdThreadTrace,amd::ThreadTraceCommand::Pause,CL_COMMAND_THREAD_TRACE); + break; + case CL_THREAD_TRACE_RESUME_COMMAND : + if (amdThreadTrace->getState() != amd::ThreadTrace::Pause) { + return CL_INVALID_OPERATION; + } + amdThreadTrace->setState(amd::ThreadTrace::Begin); + command = new amd::ThreadTraceCommand(*hostQueue, eventWaitList,&amdThreadTrace->threadTraceConfig(), + *amdThreadTrace,amd::ThreadTraceCommand::Resume,CL_COMMAND_THREAD_TRACE); + break; + } + + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Submit the command to the device + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + + return CL_SUCCESS; +} +RUNTIME_EXIT + +// +///*! \brief Enqueues the binding command to bind cl_threadtrace_amd to cl_mem object for trace recording.. +// * +// * \param command_queue must be a valid OpenCL command queue. +// * +// * \param thread_trace specifies the cl_threadtrace_amd object. +// * +// * \param mem_objects the cl_mem objects for trace recording +// * +// * \param mem_objects_num the number of cl_mem objects in the mem_objects +// * +// * \param buffer_size the size of each cl_mem object from mem_objects +// * +// * \param event_wait_list specify [is a pointer to] events that need to +// * complete before this particular command can be executed. +// * If \a event_wait_list is NULL, then this particular command does not wait +// * on any event to complete. If \a event_wait_list is NULL, +// * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, +// * the list of events pointed to by \a event_wait_list must be valid and +// * \a num_events_in_wait_list must be greater than 0. The events specified in +// * \a event_wait_list act as synchronization points. +// * +// * \param num_events_in_wait_list specify the number of events in +// * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be +// * greater than 0 if \a event_wait_list is not NULL. +// * +// * \param event returns an event object that identifies this particular +// * command and can be used to query or queue a wait for this particular +// * command to complete. \a event can be NULL in which case it will not be +// * possible for the application to query the status of this command or queue a +// * wait for this command to complete. +// * \return A non zero value if OpenCL failed to set threadTrace buffer parameter +// * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. +// * - CL_INVALID_CONTEXT if the context associated with command_queue and events in event_wait_list are not the same. +// * - CL_INVALID_VALUE if the thread_trace is invalid thread trace object. +// * - CL_INVALID_VALUE if the buffer_size is negative or zero. +// * - CL_INVALID_VALUE if the sub_buffers_num I less than 1. +// * - CL_INVALID_OPERATION if the mem_objects_num is not equal to the number of Shader Engines of the [GPU] device. +// * - CL_INVALID_MEM_OBJECT if one on memory objects in the mem_objects array is not a valid memory object or memory_objects is NULL. +// * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory for the data store associated from the memory objects of the mem_objects array. +// * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and num_events_in_wait_list > 0, or event_wait_list is not NULL and num_events_in_wait_list is 0, or if event objects in event_wait_list are not valid events. +// * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the OpenCL implementation on the device. +// * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the +// * OpenCL implementation on the host. +// */ +RUNTIME_ENTRY(cl_int,clEnqueueBindThreadTraceBufferAMD ,(cl_command_queue command_queue , + cl_threadtrace_amd thread_trace , + cl_mem* mem_objects , + cl_uint mem_objects_num , + cl_uint buffer_size , + cl_uint num_events_in_wait_list , + const cl_event* event_wait_list , + cl_event* event )) +{ + // Check if command queue is valid + if (!is_valid(command_queue)) { + return CL_INVALID_COMMAND_QUEUE; + } + + // Check if thread trace is valid + if (!is_valid(thread_trace)) { + return CL_INVALID_OPERATION; + } + + // Check if input values are valid + if ((mem_objects == NULL) + || (buffer_size <= 0)) { + return CL_INVALID_VALUE; + } + + amd::ThreadTrace* amdThreadTrace = as_amd(thread_trace); + + // Check if the number of bound memory objects is the same as the number of SEs + if (amdThreadTrace->deviceSeNumThreadTrace() != mem_objects_num) { + return CL_INVALID_OPERATION; + } + // Check if memory objects ,bound the thread trace,are valid + for (size_t i = 0; i < mem_objects_num; ++i) { + cl_mem obj = mem_objects[i]; + if (!is_valid(obj)) { + return CL_INVALID_MEM_OBJECT; + } + } + + amd::HostQueue* hostQueue = as_amd(command_queue)->asHostQueue(); + if (NULL == hostQueue) { + return CL_INVALID_COMMAND_QUEUE; + } + + // Check that device associated with the command queue is the same as with thread trace + if (&hostQueue->device() != &amdThreadTrace->device()) { + return CL_INVALID_DEVICE; + } + + amd::Command::EventWaitList eventWaitList; + cl_int err = amd::clSetEventWaitList(eventWaitList, + hostQueue->context(), num_events_in_wait_list, event_wait_list); + if (err != CL_SUCCESS){ + return err; + } + + amdThreadTrace->setState(amd::ThreadTrace::MemoryBound); + // Create a new ThreadTraceMemObjectsCommand command + amd::ThreadTraceMemObjectsCommand* command = + new amd::ThreadTraceMemObjectsCommand(*hostQueue, eventWaitList, mem_objects_num,mem_objects,buffer_size,*amdThreadTrace,CL_COMMAND_THREAD_TRACE_MEM); + if (command == NULL) { + return CL_OUT_OF_HOST_MEMORY; + } + + // Make sure we have memory for the command execution + if (!command->validateMemory()) { + delete command; + return CL_OUT_OF_RESOURCES; + } + // Submit the command to the device + command->enqueue(); + + *not_null(event) = as_cl(&command->event()); + if (event == NULL) { + command->release(); + } + return CL_SUCCESS; +} +RUNTIME_EXIT + +/*! @} + * @} + */ diff --git a/opencl/api/opencl/amdocl/cl_thread_trace_amd.h b/opencl/api/opencl/amdocl/cl_thread_trace_amd.h new file mode 100644 index 0000000000..5a430080a6 --- /dev/null +++ b/opencl/api/opencl/amdocl/cl_thread_trace_amd.h @@ -0,0 +1,439 @@ +/* ============================================================ + +Copyright (c) 2012 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERATION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance + +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +#ifndef __CL_THREAD_TRACE_AMD_H +#define __CL_THREAD_TRACE_AMD_H + +#include "CL/cl_platform.h" + +#ifdef __cplusplus +extern "C" { +#endif /*__cplusplus*/ + +typedef struct _cl_threadtrace_amd * cl_threadtrace_amd; +typedef cl_uint cl_thread_trace_param; +typedef cl_uint cl_threadtrace_info; + +/* cl_command_type */ +#define CL_COMMAND_THREAD_TRACE_MEM 0x4500 +#define CL_COMMAND_THREAD_TRACE 0x4501 + +/* cl_threadtrace_command_name_amd enumeration */ +typedef enum _cl_threadtrace_command_name_amd { + CL_THREAD_TRACE_BEGIN_COMMAND, + CL_THREAD_TRACE_END_COMMAND, + CL_THREAD_TRACE_PAUSE_COMMAND, + CL_THREAD_TRACE_RESUME_COMMAND +} cl_threadtrace_command_name_amd; + +// Thread trace parameters +enum ThreadTraceParameter +{ + CL_THREAD_TRACE_PARAM_TOKEN_MASK, + CL_THREAD_TRACE_PARAM_REG_MASK, + CL_THREAD_TRACE_PARAM_COMPUTE_UNIT_TARGET, + CL_THREAD_TRACE_PARAM_SHADER_ARRAY_TARGET, + CL_THREAD_TRACE_PARAM_SIMD_MASK, + CL_THREAD_TRACE_PARAM_VM_ID_MASK, + CL_THREAD_TRACE_PARAM_RANDOM_SEED, + CL_THREAD_TRACE_PARAM_CAPTURE_MODE, + CL_THREAD_TRACE_PARAM_INSTRUCTION_MASK, + CL_THREAD_TRACE_PARAM_USER_DATA, + CL_THREAD_TRACE_PARAM_IS_WRAPPED +}; + +// CL_THREAD_TRACE_PARAM_TOKEN_MASK data selects for SI +enum CL_THREAD_TRACE_TOKEN_MASK +{ + //Time passed + CL_THREAD_TRACE_TOKEN_MASK_TIME_SI = 0x00000001, + //Resync the timestamp + CL_THREAD_TRACE_TOKEN_MASK_TIMESTAMP_SI = 0x00000002, + //A register write has occurred + CL_THREAD_TRACE_TOKEN_MASK_REG_SI = 0x00000004, + //A wavefront has started + CL_THREAD_TRACE_TOKEN_MASK_WAVE_START_SI = 0x00000008, + //Output space has been allocated for color/Z [Should be used for cl-gl] + CL_THREAD_TRACE_TOKEN_MASK_WAVE_PS_ALLOC_SI = 0x00000010, + //Output space has been allocated for vertex position [Should be used for cl-gl] + CL_THREAD_TRACE_TOKEN_MASK_WAVE_VS_ALLOC_SI = 0x00000020, + //Wavefront completion + CL_THREAD_TRACE_TOKEN_MASK_WAVE_END_SI = 0x00000040, + //An event has reached the top of a shader stage. In-order with WAVE_START + CL_THREAD_TRACE_TOKEN_MASK_EVENT_SI = 0x00000080, + //An event has reached the top of a compute shader stage. In-order with WAVE_START + CL_THREAD_TRACE_TOKEN_MASK_EVENT_CS_SI = 0x00000100, + //An event has reached the top of a shader stage for the second GFX pipe. In-order with WAVE_START. + //[Should be used for cl-gl] + CL_THREAD_TRACE_TOKEN_MASK_EVENT_GFX_SI = 0x00000200, + //The kernel has executed an instruction + CL_THREAD_TRACE_TOKEN_MASK_INST_SI = 0x00000400, + //The kernel has explicitly written the PC value + CL_THREAD_TRACE_TOKEN_MASK_INST_PC_SI = 0x00000800, + //The kernel has written user data into the thread trace buffer + CL_THREAD_TRACE_TOKEN_MASK_INST_USERDATA_SI = 0x00001000, + //Provides information about instruction scheduling + CL_THREAD_TRACE_TOKEN_MASK_ISSUE_SI = 0x00002000, + //The performance counter delta has been updated + CL_THREAD_TRACE_TOKEN_MASK_PERF_SI = 0x00004000, + //A miscellaneous event has been sent + CL_THREAD_TRACE_TOKEN_MASK_MISC_SI = 0x00008000, + //All possible tokens + CL_THREAD_TRACE_TOKEN_MASK_ALL_SI = 0x0000ffff, +}; + +//CL_THREAD_TRACE_PARAM_REG_MASK data selects +enum CL_THREAD_TRACE_REG_MASK +{ + //Event initiator + CL_THREAD_TRACE_REG_MASK_EVENT_SI = 0x00000001, + //Draw initiator [Should be used for cl-gl] + CL_THREAD_TRACE_REG_MASK_DRAW_SI = 0x00000002, + //Dispatch initiator + CL_THREAD_TRACE_REG_MASK_DISPATCH_SI = 0x00000004, + //User data from host + CL_THREAD_TRACE_REG_MASK_USERDATA_SI = 0x00000008, + //GFXDEC register (8-state) [Should be used for cl-gl] + CL_THREAD_TRACE_REG_MASK_GFXDEC_SI = 0x00000020, + //SHDEC register (many state) + CL_THREAD_TRACE_REG_MASK_SHDEC_SI = 0x00000040, + //Other registers + CL_THREAD_TRACE_REG_MASK_OTHER_SI = 0x00000080, + //All possible registers types + CL_THREAD_TRACE_REG_MASK_ALL_SI = 0x000000ff, +}; + +//CL_THREAD_TRACE_PARAM_VM_ID_MASK data selects +enum CL_THREAD_TRACE_VM_ID_MASK +{ + //Capture only data from the VM_ID used to write {SQTT}_BASE + CL_THREAD_TRACE_VM_ID_MASK_SINGLE = 0, + //Capture all data from all VM_IDs + CL_THREAD_TRACE_VM_ID_MASK_ALL = 1, + //Capture all data but only get target (a.k.a. detail) data from VM_ID used to write {SQTT}_BASE + CL_THREAD_TRACE_VM_ID_MASK_SINGLE_DETAIL = 2 +}; + +//CL_THREAD_TRACE_PARAM_CAPTURE_MODE data +enum CL_THREAD_TRACE_CAPTURE_MODE +{ + //Capture all data in the thread trace buffer + CL_THREAD_TRACE_CAPTURE_ALL = 0, + //Capture only data between THREAD_TRACE_START and THREAD_TRACE_STOP events + CL_THREAD_TRACE_CAPTURE_SELECT = 1, + //Capture data between THREAD_TRACE_START and THREAD_TRACE_/STOP events, + //and global/reference data at all times + CL_THREAD_TRACE_CAPTURE_SELECT_DETAIL = 2 +}; + +//CL_THREAD_TRACE_PARAM_INSTRUCTION_MASK data selects +enum CL_THREAD_TRACE_INSTRUCTION_MASK +{ + //Generate {SQTT}_TOKEN_INST tokens for all instructions + CL_THREAD_TRACE_INST_MASK_ALL, + //Generate {SQTT}_TOKEN_INST tokens for stalled instructions only + CL_THREAD_TRACE_INST_MASK_STALLED, + //Generate {SQTT}_TOKEN_INST messages for stalled and other (no op/wait/set prio/etc) instructions + CL_THREAD_TRACE_INST_MASK_STALLED_AND_IMMEDIATE, + //Generate {SQTT}_TOKEN_INST messages for immediate instructions only only [ Should be used only for CI] + CL_THREAD_TRACE_INST_MASK_IMMEDIATE_CI, +}; + +enum ThreadTraceInfo +{ + CL_THREAD_TRACE_SE, + CL_THREAD_TRACE_BUFFERS_FILLED, + CL_THREAD_TRACE_BUFFERS_SIZE +}; + + +/*! \brief Creates a new cl_threadtrace_amd object + * + * \param device must be a valid OpenCL device. + * + * \param errcode_ret A non zero value if OpenCL failed to create threadTrace + * -CL_INVALID_DEVICE if devices contains an invalid device. + * -CL_DEVICE_NOT_AVAILABLE if a device is currently not available even + * though the device was returned by clGetDeviceIDs. + * -CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the + * OpenCL implementation on the device. + * -CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the + OpenCL implementation on the host. + * + * \return the created threadTrace object + */ +extern CL_API_ENTRY cl_threadtrace_amd CL_API_CALL +clCreateThreadTraceAMD( + cl_device_id /* device */, + cl_int* /* errcode_ret */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Destroys a cl_threadtrace_amd object. + * + * \param threadTrace the cl_threadtrace_amd object for release + * + * \return A non zero value if OpenCL failed to release threadTrace + * -CL_INVALID_VALUE if the thread_trace is not a valid OpenCL thread trace object (cl_threadtrace_amd) . + * -CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the + * OpenCL implementation on the device. + * -CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the + OpenCL implementation on the host. + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseThreadTraceAMD( + cl_threadtrace_amd /* threadTrace */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Increments the cl_threadtrace_amd object reference count. + * + * \param threadTrace the cl_threadtrace_amd object for retain + * + * \return A non zero value if OpenCL failed to retain threadTrace + * -CL_INVALID_VALUE if the thread_trace is not a valid thread trace object (cl_threadtrace_amd) . + * -CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the + OpenCL implementation on the device. + * -CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the + OpenCL implementation on the host. + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainThreadTraceAMD( + cl_threadtrace_amd /* threadTrace */ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Sets the cl_threadtrace_amd object configuration parameter. + * + * \param thread_trace the cl_threadtrace_amd object to set configuration parameter + * + * \param config_param the cl_thread_trace_param + * + * \param param_value corresponding to configParam + * + * \return A non zero value if OpenCL failed to set threadTrace buffer parameter + * - CL_INVALID_VALUE if the thread_trace is invalid thread trace object. + * - CL_INVALID_VALUE if the invalid config_param or param_value enum values , are used. + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and num_events_in_wait_list > 0, or event_wait_list is not NULL and num_events_in_wait_list is 0, + * - or if event objects in event_wait_list are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the + OpenCL implementation on the host. + */ + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetThreadTraceParamAMD(cl_threadtrace_amd /*thread_trace*/ , + cl_thread_trace_param /*config_param*/ , + cl_uint /*param_value*/ +)CL_API_SUFFIX__VERSION_1_0; + +/* \brief Enqueues the binding command to bind cl_threadtrace_amd to cl_mem object for trace recording.. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param thread_trace specifies the cl_threadtrace_amd object. + * + * \param mem_objects the cl_mem objects for trace recording + * + * \param mem_objects_num the number of cl_mem objects in the mem_objects + * + * \param buffer_size the size of each cl_mem object from mem_objects + * + * \param event_wait_list specify [is a pointer to] events that need to + * complete before this particular command can be executed. + * If \a event_wait_list is NULL, then this particular command does not wait + * on any event to complete. If \a event_wait_list is NULL, + * \a num_events_in_wait_list must be 0. If \a event_wait_list is not NULL, + * the list of events pointed to by \a event_wait_list must be valid and + * \a num_events_in_wait_list must be greater than 0. The events specified in + * \a event_wait_list act as synchronization points. + * + * \param num_events_in_wait_list specify the number of events in + * \a event_wait_list. It must be 0 if \a event_wait_list is NULL. It must be + * greater than 0 if \a event_wait_list is not NULL. + * + * \param event returns an event object that identifies this particular + * command and can be used to query or queue a wait for this particular + * command to complete. \a event can be NULL in which case it will not be + * possible for the application to query the status of this command or queue a + * wait for this command to complete. + * \return A non zero value if OpenCL failed to set threadTrace buffer parameter + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with command_queue and events in event_wait_list are not the same. + * - CL_INVALID_VALUE if the thread_trace is invalid thread trace object. + * - CL_INVALID_VALUE if the buffer_size is negative or zero. + * - CL_INVALID_VALUE if the sub_buffers_num I less than 1. + * - CL_INVALID_OPERATION if the mem_objects_num is not equal to the number of Shader Engines of the [GPU] device. + * - CL_INVALID_MEM_OBJECT if one on memory objects in the mem_objects array is not a valid memory object or memory_objects is NULL. + * - CL_MEM_OBJECT_ALLOCATION_FAILURE if there is a failure to allocate memory for the data store associated from the memory objects of the mem_objects array. + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and num_events_in_wait_list > 0, or event_wait_list is not NULL and num_events_in_wait_list is 0, or if event objects in event_wait_list are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the + * OpenCL implementation on the host. + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueBindThreadTraceBufferAMD(cl_command_queue command_queue , + cl_threadtrace_amd /*thread_trace*/ , + cl_mem* /*mem_objects*/ , + cl_uint /*mem_objects_num*/ , + cl_uint /*buffer_size*/ , + cl_uint /*num_events_in_wait_list*/ , + const cl_event* /*event_wait_list*/ , + cl_event* /*event*/ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Get specific information about the OpenCL Thread Trace. + * + * \param thread_trace_info_param is an enum that identifies the Thread Trace information being + * queried. + * + * \param param_value is a pointer to memory location where appropriate values + * for a given \a threadTrace_info_param will be returned. If \a param_value is NULL, + * it is ignored. + * + * \param param_value_size specifies the size in bytes of memory pointed to by + * \a param_value. This size in bytes must be >= size of return type. + * + * \param param_value_size_ret returns the actual size in bytes of data being + * queried by param_value. If \a param_value_size_ret is NULL, it is ignored. + * + * \return One of the following values: + * CL_INVALID_OPERATION if cl_threadtrace_amd object is not valid + * - CL_INVALID_VALUE if \a param_name is not one of the supported + * values or if size in bytes specified by \a param_value_size is < size of + * return type and \a param_value is not a NULL value. + * CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the + * OpenCL implementation on the host. + * CL_SUCCESS if the function is executed successfully. + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetThreadTraceInfoAMD( + cl_threadtrace_amd /* thread_trace */, + cl_threadtrace_info /*thread_trace_info_param*/, + size_t /*param_value_size*/, + void* /*param_value*/, + size_t* /*param_value_size_ret*/ +) CL_API_SUFFIX__VERSION_1_0; + +/*! \brief Enqueues the thread trace command for the specified thread trace object. + * + * \param command_queue must be a valid OpenCL command queue. + * + * \param threadTraces specifies an array of cl_threadtrace_amd objects. + * + * \return A non zero value if OpenCL failed to release threadTrace + * - CL_INVALID_COMMAND_QUEUE if command_queue is not a valid command-queue. + * - CL_INVALID_CONTEXT if the context associated with command_queue and events in event_wait_list are not the same. + * - CL_INVALID_VALUE if the thread_trace is invalid thread trace object . + * - CL_INVALID_VALUE if the invalid command name enum value , not described in the cl_threadtrace_command_name_amd, is used. + * - CL_INVALID_OPERATION if the command enqueue failed. It can happen in the following cases: + * o BEGIN_COMMAND is queued for thread trace object for which memory object/s was/were not bound.. + * o END_COMMAND is queued for thread trace object, for which BEGIN_COMMAND was not queued. + * o PAUSE_COMMAND is queued for thread trace object, for which BEGIN_COMMAND was not queued. + * o RESUME_COMMAND is queued for thread trace object, for which PAUSE_COMMAND was not queued. + * - CL_INVALID_EVENT_WAIT_LIST if event_wait_list is NULL and num_events_in_wait_list > 0, or event_wait_list is not NULL and num_events_in_wait_list is 0, or if event objects in event_wait_list are not valid events. + * - CL_OUT_OF_RESOURCES if there is a failure to allocate resources required by the OpenCL implementation on the device. + * - CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the OpenCL implementation on the host. + */ +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueThreadTraceCommandAMD( + cl_command_queue /*command_queue*/ , + cl_threadtrace_amd /*thread_trace*/ , + cl_threadtrace_command_name_amd /*command_name*/ , + cl_uint /*num_events_in_wait_list*/ , + const cl_event* /*event_wait_list*/, + cl_event* /*event*/ +)CL_API_SUFFIX__VERSION_1_0; + + +#ifdef __cplusplus +} /*extern "C"*/ +#endif /*__cplusplus*/ + +#endif /*__CL_THREAD_TRACE_AMD_H*/ diff --git a/opencl/api/opencl/amdocl/gl_functions.hpp b/opencl/api/opencl/amdocl/gl_functions.hpp new file mode 100644 index 0000000000..2f223ce68b --- /dev/null +++ b/opencl/api/opencl/amdocl/gl_functions.hpp @@ -0,0 +1,56 @@ +// +// Copyright 2010 Advanced Micro Devices, Inc. All rights reserved. +// +GLPREFIX(GLubyte*, glGetString, (GLenum name)) + +GLPREFIX(void, glBindBuffer, (GLenum target, GLuint buffer)) +//GLPREFIX(void, glBindFramebufferEXT, (GLenum target, GLuint framebuffer)) +GLPREFIX(void, glBindRenderbuffer, (GLenum target, GLuint renderbuffer)) +GLPREFIX(void, glBindTexture, (GLenum target, GLuint texture)) +GLPREFIX(void, glBufferData, (GLenum target, GLsizeiptr size, const GLvoid* data, GLenum usage)) + +GLPREFIX(GLenum, glCheckFramebufferStatusEXT, (GLenum target)) + +GLPREFIX(void, glDeleteBuffers, (GLsizei n, const GLuint* buffers)) +GLPREFIX(void, glDrawPixels, (GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *pixels)) + +//GLPREFIX(void, glFramebufferRenderbufferEXT, (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer)) + +GLPREFIX(void, glGenBuffers, (GLsizei n, GLuint* buffers)) +//GLPREFIX(void, glGenFramebuffersEXT, (GLsizei n, GLuint* framebuffers)) +//10 +GLPREFIX(void, glGetBufferParameteriv, (GLenum target, GLenum pname, GLint* params)) +GLPREFIX(GLenum, glGetError, (void)) +GLPREFIX(void, glFinish, (void)) +GLPREFIX(void, glFlush, (void)) +GLPREFIX(GLenum, glClientWaitSync, (GLsync sync, GLbitfield flags, GLuint64 timeout)) +GLPREFIX(void, glGetIntegerv, (GLenum pname, GLint *params)) +GLPREFIX(void, glGetRenderbufferParameterivEXT, (GLenum target, GLenum pname, GLint* params)) +//GLPREFIX(GLubyte*, glGetString, (GLenum name)) +GLPREFIX(void, glGetTexImage, (GLenum target, GLint level, GLenum format, GLenum type, GLvoid *pixels)) +GLPREFIX(void, glGetTexLevelParameteriv, (GLenum target, GLint level, GLenum pname, GLint *params)) +GLPREFIX(void, glGetTexParameteriv, (GLenum target, GLenum pname, GLint *params)) + +GLPREFIX(GLboolean, glIsBuffer, (GLuint buffer)) +GLPREFIX(GLboolean, glIsRenderbufferEXT, (GLuint renderbuffer)) +GLPREFIX(GLboolean, glIsTexture, (GLuint texture)) +//20 +GLPREFIX(GLvoid*, glMapBuffer, (GLenum target, GLenum access)) + +GLPREFIX(void, glReadPixels, (GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, GLvoid *pixels)) + +GLPREFIX(void, glTexImage2D, (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid *pixels)) +GLPREFIX(void, glTexImage3D, (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid *pixels)) + +GLPREFIX(GLboolean, glUnmapBuffer, (GLenum target)) + +#ifdef _WIN32 +GLPREFIX(HGLRC, wglCreateContext, (HDC hdc)) +GLPREFIX(HGLRC, wglGetCurrentContext, (void)) +GLPREFIX(HDC, wglGetCurrentDC, (void)) +GLPREFIX(BOOL, wglDeleteContext, (HGLRC hglrc)) +GLPREFIX(BOOL, wglMakeCurrent, (HDC hdc, HGLRC hglrc)) +GLPREFIX(BOOL, wglShareLists, (HGLRC hglrc1, HGLRC hglrc2)) +#endif //_WIN32 + +#undef GLPREFIX diff --git a/opencl/api/opencl/config/amdocl32.icd b/opencl/api/opencl/config/amdocl32.icd new file mode 100644 index 0000000000..53fb988802 --- /dev/null +++ b/opencl/api/opencl/config/amdocl32.icd @@ -0,0 +1 @@ +libamdocl32.so diff --git a/opencl/api/opencl/config/amdocl64.icd b/opencl/api/opencl/config/amdocl64.icd new file mode 100644 index 0000000000..ec1be031ec --- /dev/null +++ b/opencl/api/opencl/config/amdocl64.icd @@ -0,0 +1 @@ +libamdocl64.so diff --git a/opencl/api/opencl/khronos/headers/EGL/egl.h b/opencl/api/opencl/khronos/headers/EGL/egl.h new file mode 100644 index 0000000000..99ea342a47 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/EGL/egl.h @@ -0,0 +1,329 @@ +/* -*- mode: c; tab-width: 8; -*- */ +/* vi: set sw=4 ts=8: */ +/* Reference version of egl.h for EGL 1.4. + * $Revision: 9356 $ on $Date: 2009-10-21 02:52:25 -0700 (Wed, 21 Oct 2009) $ + */ + +/* +** Copyright (c) 2007-2009 The Khronos Group Inc. +** +** Permission is hereby granted, free of charge, to any person obtaining a +** copy of this software and/or associated documentation files (the +** "Materials"), to deal in the Materials without restriction, including +** without limitation the rights to use, copy, modify, merge, publish, +** distribute, sublicense, and/or sell copies of the Materials, and to +** permit persons to whom the Materials are furnished to do so, subject to +** the following conditions: +** +** The above copyright notice and this permission notice shall be included +** in all copies or substantial portions of the Materials. +** +** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY +** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. +*/ + +#ifndef __egl_h_ +#define __egl_h_ + +/* All platform-dependent types and macro boilerplate (such as EGLAPI + * and EGLAPIENTRY) should go in eglplatform.h. + */ +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/* EGL Types */ +/* EGLint is defined in eglplatform.h */ +typedef unsigned int EGLBoolean; +typedef unsigned int EGLenum; +typedef void *EGLConfig; +typedef void *EGLContext; +typedef void *EGLDisplay; +typedef void *EGLSurface; +typedef void *EGLClientBuffer; + +/* EGL Versioning */ +#define EGL_VERSION_1_0 1 +#define EGL_VERSION_1_1 1 +#define EGL_VERSION_1_2 1 +#define EGL_VERSION_1_3 1 +#define EGL_VERSION_1_4 1 + +/* EGL Enumerants. Bitmasks and other exceptional cases aside, most + * enums are assigned unique values starting at 0x3000. + */ + +/* EGL aliases */ +#define EGL_FALSE 0 +#define EGL_TRUE 1 + +/* Out-of-band handle values */ +#define EGL_DEFAULT_DISPLAY ((EGLNativeDisplayType)0) +#define EGL_NO_CONTEXT ((EGLContext)0) +#define EGL_NO_DISPLAY ((EGLDisplay)0) +#define EGL_NO_SURFACE ((EGLSurface)0) + +/* Out-of-band attribute value */ +#define EGL_DONT_CARE ((EGLint)-1) + +/* Errors / GetError return values */ +#define EGL_SUCCESS 0x3000 +#define EGL_NOT_INITIALIZED 0x3001 +#define EGL_BAD_ACCESS 0x3002 +#define EGL_BAD_ALLOC 0x3003 +#define EGL_BAD_ATTRIBUTE 0x3004 +#define EGL_BAD_CONFIG 0x3005 +#define EGL_BAD_CONTEXT 0x3006 +#define EGL_BAD_CURRENT_SURFACE 0x3007 +#define EGL_BAD_DISPLAY 0x3008 +#define EGL_BAD_MATCH 0x3009 +#define EGL_BAD_NATIVE_PIXMAP 0x300A +#define EGL_BAD_NATIVE_WINDOW 0x300B +#define EGL_BAD_PARAMETER 0x300C +#define EGL_BAD_SURFACE 0x300D +#define EGL_CONTEXT_LOST 0x300E /* EGL 1.1 - IMG_power_management */ + +/* Reserved 0x300F-0x301F for additional errors */ + +/* Config attributes */ +#define EGL_BUFFER_SIZE 0x3020 +#define EGL_ALPHA_SIZE 0x3021 +#define EGL_BLUE_SIZE 0x3022 +#define EGL_GREEN_SIZE 0x3023 +#define EGL_RED_SIZE 0x3024 +#define EGL_DEPTH_SIZE 0x3025 +#define EGL_STENCIL_SIZE 0x3026 +#define EGL_CONFIG_CAVEAT 0x3027 +#define EGL_CONFIG_ID 0x3028 +#define EGL_LEVEL 0x3029 +#define EGL_MAX_PBUFFER_HEIGHT 0x302A +#define EGL_MAX_PBUFFER_PIXELS 0x302B +#define EGL_MAX_PBUFFER_WIDTH 0x302C +#define EGL_NATIVE_RENDERABLE 0x302D +#define EGL_NATIVE_VISUAL_ID 0x302E +#define EGL_NATIVE_VISUAL_TYPE 0x302F +#define EGL_SAMPLES 0x3031 +#define EGL_SAMPLE_BUFFERS 0x3032 +#define EGL_SURFACE_TYPE 0x3033 +#define EGL_TRANSPARENT_TYPE 0x3034 +#define EGL_TRANSPARENT_BLUE_VALUE 0x3035 +#define EGL_TRANSPARENT_GREEN_VALUE 0x3036 +#define EGL_TRANSPARENT_RED_VALUE 0x3037 +#define EGL_NONE 0x3038 /* Attrib list terminator */ +#define EGL_BIND_TO_TEXTURE_RGB 0x3039 +#define EGL_BIND_TO_TEXTURE_RGBA 0x303A +#define EGL_MIN_SWAP_INTERVAL 0x303B +#define EGL_MAX_SWAP_INTERVAL 0x303C +#define EGL_LUMINANCE_SIZE 0x303D +#define EGL_ALPHA_MASK_SIZE 0x303E +#define EGL_COLOR_BUFFER_TYPE 0x303F +#define EGL_RENDERABLE_TYPE 0x3040 +#define EGL_MATCH_NATIVE_PIXMAP 0x3041 /* Pseudo-attribute (not queryable) */ +#define EGL_CONFORMANT 0x3042 + +/* Reserved 0x3041-0x304F for additional config attributes */ + +/* Config attribute values */ +#define EGL_SLOW_CONFIG 0x3050 /* EGL_CONFIG_CAVEAT value */ +#define EGL_NON_CONFORMANT_CONFIG 0x3051 /* EGL_CONFIG_CAVEAT value */ +#define EGL_TRANSPARENT_RGB 0x3052 /* EGL_TRANSPARENT_TYPE value */ +#define EGL_RGB_BUFFER 0x308E /* EGL_COLOR_BUFFER_TYPE value */ +#define EGL_LUMINANCE_BUFFER 0x308F /* EGL_COLOR_BUFFER_TYPE value */ + +/* More config attribute values, for EGL_TEXTURE_FORMAT */ +#define EGL_NO_TEXTURE 0x305C +#define EGL_TEXTURE_RGB 0x305D +#define EGL_TEXTURE_RGBA 0x305E +#define EGL_TEXTURE_2D 0x305F + +/* Config attribute mask bits */ +#define EGL_PBUFFER_BIT 0x0001 /* EGL_SURFACE_TYPE mask bits */ +#define EGL_PIXMAP_BIT 0x0002 /* EGL_SURFACE_TYPE mask bits */ +#define EGL_WINDOW_BIT 0x0004 /* EGL_SURFACE_TYPE mask bits */ +#define EGL_VG_COLORSPACE_LINEAR_BIT 0x0020 /* EGL_SURFACE_TYPE mask bits */ +#define EGL_VG_ALPHA_FORMAT_PRE_BIT 0x0040 /* EGL_SURFACE_TYPE mask bits */ +#define EGL_MULTISAMPLE_RESOLVE_BOX_BIT 0x0200 /* EGL_SURFACE_TYPE mask bits */ +#define EGL_SWAP_BEHAVIOR_PRESERVED_BIT 0x0400 /* EGL_SURFACE_TYPE mask bits */ + +#define EGL_OPENGL_ES_BIT 0x0001 /* EGL_RENDERABLE_TYPE mask bits */ +#define EGL_OPENVG_BIT 0x0002 /* EGL_RENDERABLE_TYPE mask bits */ +#define EGL_OPENGL_ES2_BIT 0x0004 /* EGL_RENDERABLE_TYPE mask bits */ +#define EGL_OPENGL_BIT 0x0008 /* EGL_RENDERABLE_TYPE mask bits */ + +/* QueryString targets */ +#define EGL_VENDOR 0x3053 +#define EGL_VERSION 0x3054 +#define EGL_EXTENSIONS 0x3055 +#define EGL_CLIENT_APIS 0x308D + +/* QuerySurface / SurfaceAttrib / CreatePbufferSurface targets */ +#define EGL_HEIGHT 0x3056 +#define EGL_WIDTH 0x3057 +#define EGL_LARGEST_PBUFFER 0x3058 +#define EGL_TEXTURE_FORMAT 0x3080 +#define EGL_TEXTURE_TARGET 0x3081 +#define EGL_MIPMAP_TEXTURE 0x3082 +#define EGL_MIPMAP_LEVEL 0x3083 +#define EGL_RENDER_BUFFER 0x3086 +#define EGL_VG_COLORSPACE 0x3087 +#define EGL_VG_ALPHA_FORMAT 0x3088 +#define EGL_HORIZONTAL_RESOLUTION 0x3090 +#define EGL_VERTICAL_RESOLUTION 0x3091 +#define EGL_PIXEL_ASPECT_RATIO 0x3092 +#define EGL_SWAP_BEHAVIOR 0x3093 +#define EGL_MULTISAMPLE_RESOLVE 0x3099 + +/* EGL_RENDER_BUFFER values / BindTexImage / ReleaseTexImage buffer targets */ +#define EGL_BACK_BUFFER 0x3084 +#define EGL_SINGLE_BUFFER 0x3085 + +/* OpenVG color spaces */ +#define EGL_VG_COLORSPACE_sRGB 0x3089 /* EGL_VG_COLORSPACE value */ +#define EGL_VG_COLORSPACE_LINEAR 0x308A /* EGL_VG_COLORSPACE value */ + +/* OpenVG alpha formats */ +#define EGL_VG_ALPHA_FORMAT_NONPRE 0x308B /* EGL_ALPHA_FORMAT value */ +#define EGL_VG_ALPHA_FORMAT_PRE 0x308C /* EGL_ALPHA_FORMAT value */ + +/* Constant scale factor by which fractional display resolutions & + * aspect ratio are scaled when queried as integer values. + */ +#define EGL_DISPLAY_SCALING 10000 + +/* Unknown display resolution/aspect ratio */ +#define EGL_UNKNOWN ((EGLint)-1) + +/* Back buffer swap behaviors */ +#define EGL_BUFFER_PRESERVED 0x3094 /* EGL_SWAP_BEHAVIOR value */ +#define EGL_BUFFER_DESTROYED 0x3095 /* EGL_SWAP_BEHAVIOR value */ + +/* CreatePbufferFromClientBuffer buffer types */ +#define EGL_OPENVG_IMAGE 0x3096 + +/* QueryContext targets */ +#define EGL_CONTEXT_CLIENT_TYPE 0x3097 + +/* CreateContext attributes */ +#define EGL_CONTEXT_CLIENT_VERSION 0x3098 + +/* Multisample resolution behaviors */ +#define EGL_MULTISAMPLE_RESOLVE_DEFAULT 0x309A /* EGL_MULTISAMPLE_RESOLVE value */ +#define EGL_MULTISAMPLE_RESOLVE_BOX 0x309B /* EGL_MULTISAMPLE_RESOLVE value */ + +/* BindAPI/QueryAPI targets */ +#define EGL_OPENGL_ES_API 0x30A0 +#define EGL_OPENVG_API 0x30A1 +#define EGL_OPENGL_API 0x30A2 + +/* GetCurrentSurface targets */ +#define EGL_DRAW 0x3059 +#define EGL_READ 0x305A + +/* WaitNative engines */ +#define EGL_CORE_NATIVE_ENGINE 0x305B + +/* EGL 1.2 tokens renamed for consistency in EGL 1.3 */ +#define EGL_COLORSPACE EGL_VG_COLORSPACE +#define EGL_ALPHA_FORMAT EGL_VG_ALPHA_FORMAT +#define EGL_COLORSPACE_sRGB EGL_VG_COLORSPACE_sRGB +#define EGL_COLORSPACE_LINEAR EGL_VG_COLORSPACE_LINEAR +#define EGL_ALPHA_FORMAT_NONPRE EGL_VG_ALPHA_FORMAT_NONPRE +#define EGL_ALPHA_FORMAT_PRE EGL_VG_ALPHA_FORMAT_PRE + +/* EGL extensions must request enum blocks from the Khronos + * API Registrar, who maintains the enumerant registry. Submit + * a bug in Khronos Bugzilla against task "Registry". + */ + + + +/* EGL Functions */ + +EGLAPI EGLint EGLAPIENTRY eglGetError(void); + +EGLAPI EGLDisplay EGLAPIENTRY eglGetDisplay(EGLNativeDisplayType display_id); +EGLAPI EGLBoolean EGLAPIENTRY eglInitialize(EGLDisplay dpy, EGLint *major, EGLint *minor); +EGLAPI EGLBoolean EGLAPIENTRY eglTerminate(EGLDisplay dpy); + +EGLAPI const char * EGLAPIENTRY eglQueryString(EGLDisplay dpy, EGLint name); + +EGLAPI EGLBoolean EGLAPIENTRY eglGetConfigs(EGLDisplay dpy, EGLConfig *configs, + EGLint config_size, EGLint *num_config); +EGLAPI EGLBoolean EGLAPIENTRY eglChooseConfig(EGLDisplay dpy, const EGLint *attrib_list, + EGLConfig *configs, EGLint config_size, + EGLint *num_config); +EGLAPI EGLBoolean EGLAPIENTRY eglGetConfigAttrib(EGLDisplay dpy, EGLConfig config, + EGLint attribute, EGLint *value); + +EGLAPI EGLSurface EGLAPIENTRY eglCreateWindowSurface(EGLDisplay dpy, EGLConfig config, + EGLNativeWindowType win, + const EGLint *attrib_list); +EGLAPI EGLSurface EGLAPIENTRY eglCreatePbufferSurface(EGLDisplay dpy, EGLConfig config, + const EGLint *attrib_list); +EGLAPI EGLSurface EGLAPIENTRY eglCreatePixmapSurface(EGLDisplay dpy, EGLConfig config, + EGLNativePixmapType pixmap, + const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglDestroySurface(EGLDisplay dpy, EGLSurface surface); +EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurface(EGLDisplay dpy, EGLSurface surface, + EGLint attribute, EGLint *value); + +EGLAPI EGLBoolean EGLAPIENTRY eglBindAPI(EGLenum api); +EGLAPI EGLenum EGLAPIENTRY eglQueryAPI(void); + +EGLAPI EGLBoolean EGLAPIENTRY eglWaitClient(void); + +EGLAPI EGLBoolean EGLAPIENTRY eglReleaseThread(void); + +EGLAPI EGLSurface EGLAPIENTRY eglCreatePbufferFromClientBuffer( + EGLDisplay dpy, EGLenum buftype, EGLClientBuffer buffer, + EGLConfig config, const EGLint *attrib_list); + +EGLAPI EGLBoolean EGLAPIENTRY eglSurfaceAttrib(EGLDisplay dpy, EGLSurface surface, + EGLint attribute, EGLint value); +EGLAPI EGLBoolean EGLAPIENTRY eglBindTexImage(EGLDisplay dpy, EGLSurface surface, EGLint buffer); +EGLAPI EGLBoolean EGLAPIENTRY eglReleaseTexImage(EGLDisplay dpy, EGLSurface surface, EGLint buffer); + + +EGLAPI EGLBoolean EGLAPIENTRY eglSwapInterval(EGLDisplay dpy, EGLint interval); + + +EGLAPI EGLContext EGLAPIENTRY eglCreateContext(EGLDisplay dpy, EGLConfig config, + EGLContext share_context, + const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglDestroyContext(EGLDisplay dpy, EGLContext ctx); +EGLAPI EGLBoolean EGLAPIENTRY eglMakeCurrent(EGLDisplay dpy, EGLSurface draw, + EGLSurface read, EGLContext ctx); + +EGLAPI EGLContext EGLAPIENTRY eglGetCurrentContext(void); +EGLAPI EGLSurface EGLAPIENTRY eglGetCurrentSurface(EGLint readdraw); +EGLAPI EGLDisplay EGLAPIENTRY eglGetCurrentDisplay(void); +EGLAPI EGLBoolean EGLAPIENTRY eglQueryContext(EGLDisplay dpy, EGLContext ctx, + EGLint attribute, EGLint *value); + +EGLAPI EGLBoolean EGLAPIENTRY eglWaitGL(void); +EGLAPI EGLBoolean EGLAPIENTRY eglWaitNative(EGLint engine); +EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffers(EGLDisplay dpy, EGLSurface surface); +EGLAPI EGLBoolean EGLAPIENTRY eglCopyBuffers(EGLDisplay dpy, EGLSurface surface, + EGLNativePixmapType target); + +/* This is a generic function pointer type, whose name indicates it must + * be cast to the proper type *and calling convention* before use. + */ +typedef void (*__eglMustCastToProperFunctionPointerType)(void); + +/* Now, define eglGetProcAddress using the generic function ptr. type */ +EGLAPI __eglMustCastToProperFunctionPointerType EGLAPIENTRY + eglGetProcAddress(const char *procname); + +#ifdef __cplusplus +} +#endif + +#endif /* __egl_h_ */ diff --git a/opencl/api/opencl/khronos/headers/EGL/eglext.h b/opencl/api/opencl/khronos/headers/EGL/eglext.h new file mode 100644 index 0000000000..2317b0cf45 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/EGL/eglext.h @@ -0,0 +1,645 @@ +#ifndef __eglext_h_ +#define __eglext_h_ 1 + +#ifdef __cplusplus +extern "C" { +#endif + +/* +** Copyright (c) 2013 The Khronos Group Inc. +** +** Permission is hereby granted, free of charge, to any person obtaining a +** copy of this software and/or associated documentation files (the +** "Materials"), to deal in the Materials without restriction, including +** without limitation the rights to use, copy, modify, merge, publish, +** distribute, sublicense, and/or sell copies of the Materials, and to +** permit persons to whom the Materials are furnished to do so, subject to +** the following conditions: +** +** The above copyright notice and this permission notice shall be included +** in all copies or substantial portions of the Materials. +** +** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY +** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. +*/ +/* +** This header is generated from the Khronos OpenGL / OpenGL ES XML +** API Registry. The current version of the Registry, generator scripts +** used to make the header, and the header can be found at +** http://www.opengl.org/registry/ +** +** Khronos $Revision: 24350 $ on $Date: 2013-12-04 12:46:23 -0800 (Wed, 04 Dec 2013) $ +*/ + +#include + +#define EGL_EGLEXT_VERSION 20131204 + +/* Generated C header for: + * API: egl + * Versions considered: .* + * Versions emitted: _nomatch_^ + * Default extensions included: egl + * Additional extensions included: _nomatch_^ + * Extensions removed: _nomatch_^ + */ + +#ifndef EGL_KHR_cl_event +#define EGL_KHR_cl_event 1 +#define EGL_CL_EVENT_HANDLE_KHR 0x309C +#define EGL_SYNC_CL_EVENT_KHR 0x30FE +#define EGL_SYNC_CL_EVENT_COMPLETE_KHR 0x30FF +#endif /* EGL_KHR_cl_event */ + +#ifndef EGL_KHR_cl_event2 +#define EGL_KHR_cl_event2 1 +typedef void *EGLSyncKHR; +typedef intptr_t EGLAttribKHR; +typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESYNC64KHRPROC) (EGLDisplay dpy, EGLenum type, const EGLAttribKHR *attrib_list); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateSync64KHR (EGLDisplay dpy, EGLenum type, const EGLAttribKHR *attrib_list); +#endif +#endif /* EGL_KHR_cl_event2 */ + +#ifndef EGL_KHR_client_get_all_proc_addresses +#define EGL_KHR_client_get_all_proc_addresses 1 +#endif /* EGL_KHR_client_get_all_proc_addresses */ + +#ifndef EGL_KHR_config_attribs +#define EGL_KHR_config_attribs 1 +#define EGL_CONFORMANT_KHR 0x3042 +#define EGL_VG_COLORSPACE_LINEAR_BIT_KHR 0x0020 +#define EGL_VG_ALPHA_FORMAT_PRE_BIT_KHR 0x0040 +#endif /* EGL_KHR_config_attribs */ + +#ifndef EGL_KHR_create_context +#define EGL_KHR_create_context 1 +#define EGL_CONTEXT_MAJOR_VERSION_KHR 0x3098 +#define EGL_CONTEXT_MINOR_VERSION_KHR 0x30FB +#define EGL_CONTEXT_FLAGS_KHR 0x30FC +#define EGL_CONTEXT_OPENGL_PROFILE_MASK_KHR 0x30FD +#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY_KHR 0x31BD +#define EGL_NO_RESET_NOTIFICATION_KHR 0x31BE +#define EGL_LOSE_CONTEXT_ON_RESET_KHR 0x31BF +#define EGL_CONTEXT_OPENGL_DEBUG_BIT_KHR 0x00000001 +#define EGL_CONTEXT_OPENGL_FORWARD_COMPATIBLE_BIT_KHR 0x00000002 +#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS_BIT_KHR 0x00000004 +#define EGL_CONTEXT_OPENGL_CORE_PROFILE_BIT_KHR 0x00000001 +#define EGL_CONTEXT_OPENGL_COMPATIBILITY_PROFILE_BIT_KHR 0x00000002 +#define EGL_OPENGL_ES3_BIT_KHR 0x00000040 +#endif /* EGL_KHR_create_context */ + +#ifndef EGL_KHR_fence_sync +#define EGL_KHR_fence_sync 1 +#ifdef KHRONOS_SUPPORT_INT64 +#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE_KHR 0x30F0 +#define EGL_SYNC_CONDITION_KHR 0x30F8 +#define EGL_SYNC_FENCE_KHR 0x30F9 +#endif /* KHRONOS_SUPPORT_INT64 */ +#endif /* EGL_KHR_fence_sync */ + +#ifndef EGL_KHR_get_all_proc_addresses +#define EGL_KHR_get_all_proc_addresses 1 +#endif /* EGL_KHR_get_all_proc_addresses */ + +#ifndef EGL_KHR_gl_renderbuffer_image +#define EGL_KHR_gl_renderbuffer_image 1 +#define EGL_GL_RENDERBUFFER_KHR 0x30B9 +#endif /* EGL_KHR_gl_renderbuffer_image */ + +#ifndef EGL_KHR_gl_texture_2D_image +#define EGL_KHR_gl_texture_2D_image 1 +#define EGL_GL_TEXTURE_2D_KHR 0x30B1 +#define EGL_GL_TEXTURE_LEVEL_KHR 0x30BC +#endif /* EGL_KHR_gl_texture_2D_image */ + +#ifndef EGL_KHR_gl_texture_3D_image +#define EGL_KHR_gl_texture_3D_image 1 +#define EGL_GL_TEXTURE_3D_KHR 0x30B2 +#define EGL_GL_TEXTURE_ZOFFSET_KHR 0x30BD +#endif /* EGL_KHR_gl_texture_3D_image */ + +#ifndef EGL_KHR_gl_texture_cubemap_image +#define EGL_KHR_gl_texture_cubemap_image 1 +#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_X_KHR 0x30B3 +#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_X_KHR 0x30B4 +#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Y_KHR 0x30B5 +#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_KHR 0x30B6 +#define EGL_GL_TEXTURE_CUBE_MAP_POSITIVE_Z_KHR 0x30B7 +#define EGL_GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_KHR 0x30B8 +#endif /* EGL_KHR_gl_texture_cubemap_image */ + +#ifndef EGL_KHR_image +#define EGL_KHR_image 1 +typedef void *EGLImageKHR; +#define EGL_NATIVE_PIXMAP_KHR 0x30B0 +#define EGL_NO_IMAGE_KHR ((EGLImageKHR)0) +typedef EGLImageKHR (EGLAPIENTRYP PFNEGLCREATEIMAGEKHRPROC) (EGLDisplay dpy, EGLContext ctx, EGLenum target, EGLClientBuffer buffer, const EGLint *attrib_list); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYIMAGEKHRPROC) (EGLDisplay dpy, EGLImageKHR image); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLImageKHR EGLAPIENTRY eglCreateImageKHR (EGLDisplay dpy, EGLContext ctx, EGLenum target, EGLClientBuffer buffer, const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglDestroyImageKHR (EGLDisplay dpy, EGLImageKHR image); +#endif +#endif /* EGL_KHR_image */ + +#ifndef EGL_KHR_image_base +#define EGL_KHR_image_base 1 +#define EGL_IMAGE_PRESERVED_KHR 0x30D2 +#endif /* EGL_KHR_image_base */ + +#ifndef EGL_KHR_image_pixmap +#define EGL_KHR_image_pixmap 1 +#endif /* EGL_KHR_image_pixmap */ + +#ifndef EGL_KHR_lock_surface +#define EGL_KHR_lock_surface 1 +#define EGL_READ_SURFACE_BIT_KHR 0x0001 +#define EGL_WRITE_SURFACE_BIT_KHR 0x0002 +#define EGL_LOCK_SURFACE_BIT_KHR 0x0080 +#define EGL_OPTIMAL_FORMAT_BIT_KHR 0x0100 +#define EGL_MATCH_FORMAT_KHR 0x3043 +#define EGL_FORMAT_RGB_565_EXACT_KHR 0x30C0 +#define EGL_FORMAT_RGB_565_KHR 0x30C1 +#define EGL_FORMAT_RGBA_8888_EXACT_KHR 0x30C2 +#define EGL_FORMAT_RGBA_8888_KHR 0x30C3 +#define EGL_MAP_PRESERVE_PIXELS_KHR 0x30C4 +#define EGL_LOCK_USAGE_HINT_KHR 0x30C5 +#define EGL_BITMAP_POINTER_KHR 0x30C6 +#define EGL_BITMAP_PITCH_KHR 0x30C7 +#define EGL_BITMAP_ORIGIN_KHR 0x30C8 +#define EGL_BITMAP_PIXEL_RED_OFFSET_KHR 0x30C9 +#define EGL_BITMAP_PIXEL_GREEN_OFFSET_KHR 0x30CA +#define EGL_BITMAP_PIXEL_BLUE_OFFSET_KHR 0x30CB +#define EGL_BITMAP_PIXEL_ALPHA_OFFSET_KHR 0x30CC +#define EGL_BITMAP_PIXEL_LUMINANCE_OFFSET_KHR 0x30CD +#define EGL_LOWER_LEFT_KHR 0x30CE +#define EGL_UPPER_LEFT_KHR 0x30CF +typedef EGLBoolean (EGLAPIENTRYP PFNEGLLOCKSURFACEKHRPROC) (EGLDisplay dpy, EGLSurface surface, const EGLint *attrib_list); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLUNLOCKSURFACEKHRPROC) (EGLDisplay dpy, EGLSurface surface); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglLockSurfaceKHR (EGLDisplay dpy, EGLSurface surface, const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglUnlockSurfaceKHR (EGLDisplay dpy, EGLSurface surface); +#endif +#endif /* EGL_KHR_lock_surface */ + +#ifndef EGL_KHR_lock_surface2 +#define EGL_KHR_lock_surface2 1 +#define EGL_BITMAP_PIXEL_SIZE_KHR 0x3110 +#endif /* EGL_KHR_lock_surface2 */ + +#ifndef EGL_KHR_lock_surface3 +#define EGL_KHR_lock_surface3 1 +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSURFACE64KHRPROC) (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLAttribKHR *value); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurface64KHR (EGLDisplay dpy, EGLSurface surface, EGLint attribute, EGLAttribKHR *value); +#endif +#endif /* EGL_KHR_lock_surface3 */ + +#ifndef EGL_KHR_reusable_sync +#define EGL_KHR_reusable_sync 1 +typedef khronos_utime_nanoseconds_t EGLTimeKHR; +#ifdef KHRONOS_SUPPORT_INT64 +#define EGL_SYNC_STATUS_KHR 0x30F1 +#define EGL_SIGNALED_KHR 0x30F2 +#define EGL_UNSIGNALED_KHR 0x30F3 +#define EGL_TIMEOUT_EXPIRED_KHR 0x30F5 +#define EGL_CONDITION_SATISFIED_KHR 0x30F6 +#define EGL_SYNC_TYPE_KHR 0x30F7 +#define EGL_SYNC_REUSABLE_KHR 0x30FA +#define EGL_SYNC_FLUSH_COMMANDS_BIT_KHR 0x0001 +#define EGL_FOREVER_KHR 0xFFFFFFFFFFFFFFFFull +#define EGL_NO_SYNC_KHR ((EGLSyncKHR)0) +typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESYNCKHRPROC) (EGLDisplay dpy, EGLenum type, const EGLint *attrib_list); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync); +typedef EGLint (EGLAPIENTRYP PFNEGLCLIENTWAITSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR timeout); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSIGNALSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLGETSYNCATTRIBKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *value); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateSyncKHR (EGLDisplay dpy, EGLenum type, const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglDestroySyncKHR (EGLDisplay dpy, EGLSyncKHR sync); +EGLAPI EGLint EGLAPIENTRY eglClientWaitSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR timeout); +EGLAPI EGLBoolean EGLAPIENTRY eglSignalSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode); +EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttribKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *value); +#endif +#endif /* KHRONOS_SUPPORT_INT64 */ +#endif /* EGL_KHR_reusable_sync */ + +#ifndef EGL_KHR_stream +#define EGL_KHR_stream 1 +typedef void *EGLStreamKHR; +typedef khronos_uint64_t EGLuint64KHR; +#ifdef KHRONOS_SUPPORT_INT64 +#define EGL_NO_STREAM_KHR ((EGLStreamKHR)0) +#define EGL_CONSUMER_LATENCY_USEC_KHR 0x3210 +#define EGL_PRODUCER_FRAME_KHR 0x3212 +#define EGL_CONSUMER_FRAME_KHR 0x3213 +#define EGL_STREAM_STATE_KHR 0x3214 +#define EGL_STREAM_STATE_CREATED_KHR 0x3215 +#define EGL_STREAM_STATE_CONNECTING_KHR 0x3216 +#define EGL_STREAM_STATE_EMPTY_KHR 0x3217 +#define EGL_STREAM_STATE_NEW_FRAME_AVAILABLE_KHR 0x3218 +#define EGL_STREAM_STATE_OLD_FRAME_AVAILABLE_KHR 0x3219 +#define EGL_STREAM_STATE_DISCONNECTED_KHR 0x321A +#define EGL_BAD_STREAM_KHR 0x321B +#define EGL_BAD_STATE_KHR 0x321C +typedef EGLStreamKHR (EGLAPIENTRYP PFNEGLCREATESTREAMKHRPROC) (EGLDisplay dpy, const EGLint *attrib_list); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSTREAMKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMATTRIBKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint value); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint *value); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMU64KHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLuint64KHR *value); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLStreamKHR EGLAPIENTRY eglCreateStreamKHR (EGLDisplay dpy, const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglDestroyStreamKHR (EGLDisplay dpy, EGLStreamKHR stream); +EGLAPI EGLBoolean EGLAPIENTRY eglStreamAttribKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint value); +EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLint *value); +EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamu64KHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLuint64KHR *value); +#endif +#endif /* KHRONOS_SUPPORT_INT64 */ +#endif /* EGL_KHR_stream */ + +#ifndef EGL_KHR_stream_consumer_gltexture +#define EGL_KHR_stream_consumer_gltexture 1 +#ifdef EGL_KHR_stream +#define EGL_CONSUMER_ACQUIRE_TIMEOUT_USEC_KHR 0x321E +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERGLTEXTUREEXTERNALKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERACQUIREKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSTREAMCONSUMERRELEASEKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerGLTextureExternalKHR (EGLDisplay dpy, EGLStreamKHR stream); +EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerAcquireKHR (EGLDisplay dpy, EGLStreamKHR stream); +EGLAPI EGLBoolean EGLAPIENTRY eglStreamConsumerReleaseKHR (EGLDisplay dpy, EGLStreamKHR stream); +#endif +#endif /* EGL_KHR_stream */ +#endif /* EGL_KHR_stream_consumer_gltexture */ + +#ifndef EGL_KHR_stream_cross_process_fd +#define EGL_KHR_stream_cross_process_fd 1 +typedef int EGLNativeFileDescriptorKHR; +#ifdef EGL_KHR_stream +#define EGL_NO_FILE_DESCRIPTOR_KHR ((EGLNativeFileDescriptorKHR)(-1)) +typedef EGLNativeFileDescriptorKHR (EGLAPIENTRYP PFNEGLGETSTREAMFILEDESCRIPTORKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream); +typedef EGLStreamKHR (EGLAPIENTRYP PFNEGLCREATESTREAMFROMFILEDESCRIPTORKHRPROC) (EGLDisplay dpy, EGLNativeFileDescriptorKHR file_descriptor); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLNativeFileDescriptorKHR EGLAPIENTRY eglGetStreamFileDescriptorKHR (EGLDisplay dpy, EGLStreamKHR stream); +EGLAPI EGLStreamKHR EGLAPIENTRY eglCreateStreamFromFileDescriptorKHR (EGLDisplay dpy, EGLNativeFileDescriptorKHR file_descriptor); +#endif +#endif /* EGL_KHR_stream */ +#endif /* EGL_KHR_stream_cross_process_fd */ + +#ifndef EGL_KHR_stream_fifo +#define EGL_KHR_stream_fifo 1 +#ifdef EGL_KHR_stream +#define EGL_STREAM_FIFO_LENGTH_KHR 0x31FC +#define EGL_STREAM_TIME_NOW_KHR 0x31FD +#define EGL_STREAM_TIME_CONSUMER_KHR 0x31FE +#define EGL_STREAM_TIME_PRODUCER_KHR 0x31FF +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSTREAMTIMEKHRPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLTimeKHR *value); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglQueryStreamTimeKHR (EGLDisplay dpy, EGLStreamKHR stream, EGLenum attribute, EGLTimeKHR *value); +#endif +#endif /* EGL_KHR_stream */ +#endif /* EGL_KHR_stream_fifo */ + +#ifndef EGL_KHR_stream_producer_aldatalocator +#define EGL_KHR_stream_producer_aldatalocator 1 +#ifdef EGL_KHR_stream +#endif /* EGL_KHR_stream */ +#endif /* EGL_KHR_stream_producer_aldatalocator */ + +#ifndef EGL_KHR_stream_producer_eglsurface +#define EGL_KHR_stream_producer_eglsurface 1 +#ifdef EGL_KHR_stream +#define EGL_STREAM_BIT_KHR 0x0800 +typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATESTREAMPRODUCERSURFACEKHRPROC) (EGLDisplay dpy, EGLConfig config, EGLStreamKHR stream, const EGLint *attrib_list); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLSurface EGLAPIENTRY eglCreateStreamProducerSurfaceKHR (EGLDisplay dpy, EGLConfig config, EGLStreamKHR stream, const EGLint *attrib_list); +#endif +#endif /* EGL_KHR_stream */ +#endif /* EGL_KHR_stream_producer_eglsurface */ + +#ifndef EGL_KHR_surfaceless_context +#define EGL_KHR_surfaceless_context 1 +#endif /* EGL_KHR_surfaceless_context */ + +#ifndef EGL_KHR_vg_parent_image +#define EGL_KHR_vg_parent_image 1 +#define EGL_VG_PARENT_IMAGE_KHR 0x30BA +#endif /* EGL_KHR_vg_parent_image */ + +#ifndef EGL_KHR_wait_sync +#define EGL_KHR_wait_sync 1 +typedef EGLint (EGLAPIENTRYP PFNEGLWAITSYNCKHRPROC) (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLint EGLAPIENTRY eglWaitSyncKHR (EGLDisplay dpy, EGLSyncKHR sync, EGLint flags); +#endif +#endif /* EGL_KHR_wait_sync */ + +#ifndef EGL_ANDROID_blob_cache +#define EGL_ANDROID_blob_cache 1 +typedef khronos_ssize_t EGLsizeiANDROID; +typedef void (*EGLSetBlobFuncANDROID) (const void *key, EGLsizeiANDROID keySize, const void *value, EGLsizeiANDROID valueSize); +typedef EGLsizeiANDROID (*EGLGetBlobFuncANDROID) (const void *key, EGLsizeiANDROID keySize, void *value, EGLsizeiANDROID valueSize); +typedef void (EGLAPIENTRYP PFNEGLSETBLOBCACHEFUNCSANDROIDPROC) (EGLDisplay dpy, EGLSetBlobFuncANDROID set, EGLGetBlobFuncANDROID get); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI void EGLAPIENTRY eglSetBlobCacheFuncsANDROID (EGLDisplay dpy, EGLSetBlobFuncANDROID set, EGLGetBlobFuncANDROID get); +#endif +#endif /* EGL_ANDROID_blob_cache */ + +#ifndef EGL_ANDROID_framebuffer_target +#define EGL_ANDROID_framebuffer_target 1 +#define EGL_FRAMEBUFFER_TARGET_ANDROID 0x3147 +#endif /* EGL_ANDROID_framebuffer_target */ + +#ifndef EGL_ANDROID_image_native_buffer +#define EGL_ANDROID_image_native_buffer 1 +#define EGL_NATIVE_BUFFER_ANDROID 0x3140 +#endif /* EGL_ANDROID_image_native_buffer */ + +#ifndef EGL_ANDROID_native_fence_sync +#define EGL_ANDROID_native_fence_sync 1 +#define EGL_SYNC_NATIVE_FENCE_ANDROID 0x3144 +#define EGL_SYNC_NATIVE_FENCE_FD_ANDROID 0x3145 +#define EGL_SYNC_NATIVE_FENCE_SIGNALED_ANDROID 0x3146 +#define EGL_NO_NATIVE_FENCE_FD_ANDROID -1 +typedef EGLint (EGLAPIENTRYP PFNEGLDUPNATIVEFENCEFDANDROIDPROC) (EGLDisplay dpy, EGLSyncKHR sync); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLint EGLAPIENTRY eglDupNativeFenceFDANDROID (EGLDisplay dpy, EGLSyncKHR sync); +#endif +#endif /* EGL_ANDROID_native_fence_sync */ + +#ifndef EGL_ANDROID_recordable +#define EGL_ANDROID_recordable 1 +#define EGL_RECORDABLE_ANDROID 0x3142 +#endif /* EGL_ANDROID_recordable */ + +#ifndef EGL_ANGLE_d3d_share_handle_client_buffer +#define EGL_ANGLE_d3d_share_handle_client_buffer 1 +#define EGL_D3D_TEXTURE_2D_SHARE_HANDLE_ANGLE 0x3200 +#endif /* EGL_ANGLE_d3d_share_handle_client_buffer */ + +#ifndef EGL_ANGLE_query_surface_pointer +#define EGL_ANGLE_query_surface_pointer 1 +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYSURFACEPOINTERANGLEPROC) (EGLDisplay dpy, EGLSurface surface, EGLint attribute, void **value); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglQuerySurfacePointerANGLE (EGLDisplay dpy, EGLSurface surface, EGLint attribute, void **value); +#endif +#endif /* EGL_ANGLE_query_surface_pointer */ + +#ifndef EGL_ANGLE_surface_d3d_texture_2d_share_handle +#define EGL_ANGLE_surface_d3d_texture_2d_share_handle 1 +#endif /* EGL_ANGLE_surface_d3d_texture_2d_share_handle */ + +#ifndef EGL_ARM_pixmap_multisample_discard +#define EGL_ARM_pixmap_multisample_discard 1 +#define EGL_DISCARD_SAMPLES_ARM 0x3286 +#endif /* EGL_ARM_pixmap_multisample_discard */ + +#ifndef EGL_EXT_buffer_age +#define EGL_EXT_buffer_age 1 +#define EGL_BUFFER_AGE_EXT 0x313D +#endif /* EGL_EXT_buffer_age */ + +#ifndef EGL_EXT_client_extensions +#define EGL_EXT_client_extensions 1 +#endif /* EGL_EXT_client_extensions */ + +#ifndef EGL_EXT_create_context_robustness +#define EGL_EXT_create_context_robustness 1 +#define EGL_CONTEXT_OPENGL_ROBUST_ACCESS_EXT 0x30BF +#define EGL_CONTEXT_OPENGL_RESET_NOTIFICATION_STRATEGY_EXT 0x3138 +#define EGL_NO_RESET_NOTIFICATION_EXT 0x31BE +#define EGL_LOSE_CONTEXT_ON_RESET_EXT 0x31BF +#endif /* EGL_EXT_create_context_robustness */ + +#ifndef EGL_EXT_image_dma_buf_import +#define EGL_EXT_image_dma_buf_import 1 +#define EGL_LINUX_DMA_BUF_EXT 0x3270 +#define EGL_LINUX_DRM_FOURCC_EXT 0x3271 +#define EGL_DMA_BUF_PLANE0_FD_EXT 0x3272 +#define EGL_DMA_BUF_PLANE0_OFFSET_EXT 0x3273 +#define EGL_DMA_BUF_PLANE0_PITCH_EXT 0x3274 +#define EGL_DMA_BUF_PLANE1_FD_EXT 0x3275 +#define EGL_DMA_BUF_PLANE1_OFFSET_EXT 0x3276 +#define EGL_DMA_BUF_PLANE1_PITCH_EXT 0x3277 +#define EGL_DMA_BUF_PLANE2_FD_EXT 0x3278 +#define EGL_DMA_BUF_PLANE2_OFFSET_EXT 0x3279 +#define EGL_DMA_BUF_PLANE2_PITCH_EXT 0x327A +#define EGL_YUV_COLOR_SPACE_HINT_EXT 0x327B +#define EGL_SAMPLE_RANGE_HINT_EXT 0x327C +#define EGL_YUV_CHROMA_HORIZONTAL_SITING_HINT_EXT 0x327D +#define EGL_YUV_CHROMA_VERTICAL_SITING_HINT_EXT 0x327E +#define EGL_ITU_REC601_EXT 0x327F +#define EGL_ITU_REC709_EXT 0x3280 +#define EGL_ITU_REC2020_EXT 0x3281 +#define EGL_YUV_FULL_RANGE_EXT 0x3282 +#define EGL_YUV_NARROW_RANGE_EXT 0x3283 +#define EGL_YUV_CHROMA_SITING_0_EXT 0x3284 +#define EGL_YUV_CHROMA_SITING_0_5_EXT 0x3285 +#endif /* EGL_EXT_image_dma_buf_import */ + +#ifndef EGL_EXT_multiview_window +#define EGL_EXT_multiview_window 1 +#define EGL_MULTIVIEW_VIEW_COUNT_EXT 0x3134 +#endif /* EGL_EXT_multiview_window */ + +#ifndef EGL_EXT_platform_base +#define EGL_EXT_platform_base 1 +typedef EGLDisplay (EGLAPIENTRYP PFNEGLGETPLATFORMDISPLAYEXTPROC) (EGLenum platform, void *native_display, const EGLint *attrib_list); +typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPLATFORMWINDOWSURFACEEXTPROC) (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLint *attrib_list); +typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPLATFORMPIXMAPSURFACEEXTPROC) (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLint *attrib_list); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLDisplay EGLAPIENTRY eglGetPlatformDisplayEXT (EGLenum platform, void *native_display, const EGLint *attrib_list); +EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformWindowSurfaceEXT (EGLDisplay dpy, EGLConfig config, void *native_window, const EGLint *attrib_list); +EGLAPI EGLSurface EGLAPIENTRY eglCreatePlatformPixmapSurfaceEXT (EGLDisplay dpy, EGLConfig config, void *native_pixmap, const EGLint *attrib_list); +#endif +#endif /* EGL_EXT_platform_base */ + +#ifndef EGL_EXT_platform_wayland +#define EGL_EXT_platform_wayland 1 +#define EGL_PLATFORM_WAYLAND_EXT 0x31D8 +#endif /* EGL_EXT_platform_wayland */ + +#ifndef EGL_EXT_platform_x11 +#define EGL_EXT_platform_x11 1 +#define EGL_PLATFORM_X11_EXT 0x31D5 +#define EGL_PLATFORM_X11_SCREEN_EXT 0x31D6 +#endif /* EGL_EXT_platform_x11 */ + +#ifndef EGL_EXT_swap_buffers_with_damage +#define EGL_EXT_swap_buffers_with_damage 1 +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSWAPBUFFERSWITHDAMAGEEXTPROC) (EGLDisplay dpy, EGLSurface surface, EGLint *rects, EGLint n_rects); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglSwapBuffersWithDamageEXT (EGLDisplay dpy, EGLSurface surface, EGLint *rects, EGLint n_rects); +#endif +#endif /* EGL_EXT_swap_buffers_with_damage */ + +#ifndef EGL_HI_clientpixmap +#define EGL_HI_clientpixmap 1 +struct EGLClientPixmapHI { + void *pData; + EGLint iWidth; + EGLint iHeight; + EGLint iStride; +}; +#define EGL_CLIENT_PIXMAP_POINTER_HI 0x8F74 +typedef EGLSurface (EGLAPIENTRYP PFNEGLCREATEPIXMAPSURFACEHIPROC) (EGLDisplay dpy, EGLConfig config, struct EGLClientPixmapHI *pixmap); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLSurface EGLAPIENTRY eglCreatePixmapSurfaceHI (EGLDisplay dpy, EGLConfig config, struct EGLClientPixmapHI *pixmap); +#endif +#endif /* EGL_HI_clientpixmap */ + +#ifndef EGL_HI_colorformats +#define EGL_HI_colorformats 1 +#define EGL_COLOR_FORMAT_HI 0x8F70 +#define EGL_COLOR_RGB_HI 0x8F71 +#define EGL_COLOR_RGBA_HI 0x8F72 +#define EGL_COLOR_ARGB_HI 0x8F73 +#endif /* EGL_HI_colorformats */ + +#ifndef EGL_IMG_context_priority +#define EGL_IMG_context_priority 1 +#define EGL_CONTEXT_PRIORITY_LEVEL_IMG 0x3100 +#define EGL_CONTEXT_PRIORITY_HIGH_IMG 0x3101 +#define EGL_CONTEXT_PRIORITY_MEDIUM_IMG 0x3102 +#define EGL_CONTEXT_PRIORITY_LOW_IMG 0x3103 +#endif /* EGL_IMG_context_priority */ + +#ifndef EGL_MESA_drm_image +#define EGL_MESA_drm_image 1 +#define EGL_DRM_BUFFER_FORMAT_MESA 0x31D0 +#define EGL_DRM_BUFFER_USE_MESA 0x31D1 +#define EGL_DRM_BUFFER_FORMAT_ARGB32_MESA 0x31D2 +#define EGL_DRM_BUFFER_MESA 0x31D3 +#define EGL_DRM_BUFFER_STRIDE_MESA 0x31D4 +#define EGL_DRM_BUFFER_USE_SCANOUT_MESA 0x00000001 +#define EGL_DRM_BUFFER_USE_SHARE_MESA 0x00000002 +typedef EGLImageKHR (EGLAPIENTRYP PFNEGLCREATEDRMIMAGEMESAPROC) (EGLDisplay dpy, const EGLint *attrib_list); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLEXPORTDRMIMAGEMESAPROC) (EGLDisplay dpy, EGLImageKHR image, EGLint *name, EGLint *handle, EGLint *stride); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLImageKHR EGLAPIENTRY eglCreateDRMImageMESA (EGLDisplay dpy, const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglExportDRMImageMESA (EGLDisplay dpy, EGLImageKHR image, EGLint *name, EGLint *handle, EGLint *stride); +#endif +#endif /* EGL_MESA_drm_image */ + +#ifndef EGL_MESA_platform_gbm +#define EGL_MESA_platform_gbm 1 +#define EGL_PLATFORM_GBM_MESA 0x31D7 +#endif /* EGL_MESA_platform_gbm */ + +#ifndef EGL_NV_3dvision_surface +#define EGL_NV_3dvision_surface 1 +#define EGL_AUTO_STEREO_NV 0x3136 +#endif /* EGL_NV_3dvision_surface */ + +#ifndef EGL_NV_coverage_sample +#define EGL_NV_coverage_sample 1 +#define EGL_COVERAGE_BUFFERS_NV 0x30E0 +#define EGL_COVERAGE_SAMPLES_NV 0x30E1 +#endif /* EGL_NV_coverage_sample */ + +#ifndef EGL_NV_coverage_sample_resolve +#define EGL_NV_coverage_sample_resolve 1 +#define EGL_COVERAGE_SAMPLE_RESOLVE_NV 0x3131 +#define EGL_COVERAGE_SAMPLE_RESOLVE_DEFAULT_NV 0x3132 +#define EGL_COVERAGE_SAMPLE_RESOLVE_NONE_NV 0x3133 +#endif /* EGL_NV_coverage_sample_resolve */ + +#ifndef EGL_NV_depth_nonlinear +#define EGL_NV_depth_nonlinear 1 +#define EGL_DEPTH_ENCODING_NV 0x30E2 +#define EGL_DEPTH_ENCODING_NONE_NV 0 +#define EGL_DEPTH_ENCODING_NONLINEAR_NV 0x30E3 +#endif /* EGL_NV_depth_nonlinear */ + +#ifndef EGL_NV_native_query +#define EGL_NV_native_query 1 +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEDISPLAYNVPROC) (EGLDisplay dpy, EGLNativeDisplayType *display_id); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEWINDOWNVPROC) (EGLDisplay dpy, EGLSurface surf, EGLNativeWindowType *window); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLQUERYNATIVEPIXMAPNVPROC) (EGLDisplay dpy, EGLSurface surf, EGLNativePixmapType *pixmap); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativeDisplayNV (EGLDisplay dpy, EGLNativeDisplayType *display_id); +EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativeWindowNV (EGLDisplay dpy, EGLSurface surf, EGLNativeWindowType *window); +EGLAPI EGLBoolean EGLAPIENTRY eglQueryNativePixmapNV (EGLDisplay dpy, EGLSurface surf, EGLNativePixmapType *pixmap); +#endif +#endif /* EGL_NV_native_query */ + +#ifndef EGL_NV_post_convert_rounding +#define EGL_NV_post_convert_rounding 1 +#endif /* EGL_NV_post_convert_rounding */ + +#ifndef EGL_NV_post_sub_buffer +#define EGL_NV_post_sub_buffer 1 +#define EGL_POST_SUB_BUFFER_SUPPORTED_NV 0x30BE +typedef EGLBoolean (EGLAPIENTRYP PFNEGLPOSTSUBBUFFERNVPROC) (EGLDisplay dpy, EGLSurface surface, EGLint x, EGLint y, EGLint width, EGLint height); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLBoolean EGLAPIENTRY eglPostSubBufferNV (EGLDisplay dpy, EGLSurface surface, EGLint x, EGLint y, EGLint width, EGLint height); +#endif +#endif /* EGL_NV_post_sub_buffer */ + +#ifndef EGL_NV_stream_sync +#define EGL_NV_stream_sync 1 +#define EGL_SYNC_NEW_FRAME_NV 0x321F +typedef EGLSyncKHR (EGLAPIENTRYP PFNEGLCREATESTREAMSYNCNVPROC) (EGLDisplay dpy, EGLStreamKHR stream, EGLenum type, const EGLint *attrib_list); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLSyncKHR EGLAPIENTRY eglCreateStreamSyncNV (EGLDisplay dpy, EGLStreamKHR stream, EGLenum type, const EGLint *attrib_list); +#endif +#endif /* EGL_NV_stream_sync */ + +#ifndef EGL_NV_sync +#define EGL_NV_sync 1 +typedef void *EGLSyncNV; +typedef khronos_utime_nanoseconds_t EGLTimeNV; +#ifdef KHRONOS_SUPPORT_INT64 +#define EGL_SYNC_PRIOR_COMMANDS_COMPLETE_NV 0x30E6 +#define EGL_SYNC_STATUS_NV 0x30E7 +#define EGL_SIGNALED_NV 0x30E8 +#define EGL_UNSIGNALED_NV 0x30E9 +#define EGL_SYNC_FLUSH_COMMANDS_BIT_NV 0x0001 +#define EGL_FOREVER_NV 0xFFFFFFFFFFFFFFFFull +#define EGL_ALREADY_SIGNALED_NV 0x30EA +#define EGL_TIMEOUT_EXPIRED_NV 0x30EB +#define EGL_CONDITION_SATISFIED_NV 0x30EC +#define EGL_SYNC_TYPE_NV 0x30ED +#define EGL_SYNC_CONDITION_NV 0x30EE +#define EGL_SYNC_FENCE_NV 0x30EF +#define EGL_NO_SYNC_NV ((EGLSyncNV)0) +typedef EGLSyncNV (EGLAPIENTRYP PFNEGLCREATEFENCESYNCNVPROC) (EGLDisplay dpy, EGLenum condition, const EGLint *attrib_list); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLDESTROYSYNCNVPROC) (EGLSyncNV sync); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLFENCENVPROC) (EGLSyncNV sync); +typedef EGLint (EGLAPIENTRYP PFNEGLCLIENTWAITSYNCNVPROC) (EGLSyncNV sync, EGLint flags, EGLTimeNV timeout); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLSIGNALSYNCNVPROC) (EGLSyncNV sync, EGLenum mode); +typedef EGLBoolean (EGLAPIENTRYP PFNEGLGETSYNCATTRIBNVPROC) (EGLSyncNV sync, EGLint attribute, EGLint *value); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLSyncNV EGLAPIENTRY eglCreateFenceSyncNV (EGLDisplay dpy, EGLenum condition, const EGLint *attrib_list); +EGLAPI EGLBoolean EGLAPIENTRY eglDestroySyncNV (EGLSyncNV sync); +EGLAPI EGLBoolean EGLAPIENTRY eglFenceNV (EGLSyncNV sync); +EGLAPI EGLint EGLAPIENTRY eglClientWaitSyncNV (EGLSyncNV sync, EGLint flags, EGLTimeNV timeout); +EGLAPI EGLBoolean EGLAPIENTRY eglSignalSyncNV (EGLSyncNV sync, EGLenum mode); +EGLAPI EGLBoolean EGLAPIENTRY eglGetSyncAttribNV (EGLSyncNV sync, EGLint attribute, EGLint *value); +#endif +#endif /* KHRONOS_SUPPORT_INT64 */ +#endif /* EGL_NV_sync */ + +#ifndef EGL_NV_system_time +#define EGL_NV_system_time 1 +typedef khronos_utime_nanoseconds_t EGLuint64NV; +#ifdef KHRONOS_SUPPORT_INT64 +typedef EGLuint64NV (EGLAPIENTRYP PFNEGLGETSYSTEMTIMEFREQUENCYNVPROC) (void); +typedef EGLuint64NV (EGLAPIENTRYP PFNEGLGETSYSTEMTIMENVPROC) (void); +#ifdef EGL_EGLEXT_PROTOTYPES +EGLAPI EGLuint64NV EGLAPIENTRY eglGetSystemTimeFrequencyNV (void); +EGLAPI EGLuint64NV EGLAPIENTRY eglGetSystemTimeNV (void); +#endif +#endif /* KHRONOS_SUPPORT_INT64 */ +#endif /* EGL_NV_system_time */ + +#ifdef __cplusplus +} +#endif + +#endif diff --git a/opencl/api/opencl/khronos/headers/EGL/eglplatform.h b/opencl/api/opencl/khronos/headers/EGL/eglplatform.h new file mode 100644 index 0000000000..3ab8844f09 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/EGL/eglplatform.h @@ -0,0 +1,125 @@ +#ifndef __eglplatform_h_ +#define __eglplatform_h_ + +/* +** Copyright (c) 2007-2013 The Khronos Group Inc. +** +** Permission is hereby granted, free of charge, to any person obtaining a +** copy of this software and/or associated documentation files (the +** "Materials"), to deal in the Materials without restriction, including +** without limitation the rights to use, copy, modify, merge, publish, +** distribute, sublicense, and/or sell copies of the Materials, and to +** permit persons to whom the Materials are furnished to do so, subject to +** the following conditions: +** +** The above copyright notice and this permission notice shall be included +** in all copies or substantial portions of the Materials. +** +** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY +** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. +*/ + +/* Platform-specific types and definitions for egl.h + * $Revision: 23432 $ on $Date: 2013-10-09 00:57:24 -0700 (Wed, 09 Oct 2013) $ + * + * Adopters may modify khrplatform.h and this file to suit their platform. + * You are encouraged to submit all modifications to the Khronos group so that + * they can be included in future versions of this file. Please submit changes + * by sending them to the public Khronos Bugzilla (http://khronos.org/bugzilla) + * by filing a bug against product "EGL" component "Registry". + */ + +#include + +/* Macros used in EGL function prototype declarations. + * + * EGL functions should be prototyped as: + * + * EGLAPI return-type EGLAPIENTRY eglFunction(arguments); + * typedef return-type (EXPAPIENTRYP PFNEGLFUNCTIONPROC) (arguments); + * + * KHRONOS_APICALL and KHRONOS_APIENTRY are defined in KHR/khrplatform.h + */ + +#ifndef EGLAPI +#define EGLAPI KHRONOS_APICALL +#endif + +#ifndef EGLAPIENTRY +#define EGLAPIENTRY KHRONOS_APIENTRY +#endif +#define EGLAPIENTRYP EGLAPIENTRY* + +/* The types NativeDisplayType, NativeWindowType, and NativePixmapType + * are aliases of window-system-dependent types, such as X Display * or + * Windows Device Context. They must be defined in platform-specific + * code below. The EGL-prefixed versions of Native*Type are the same + * types, renamed in EGL 1.3 so all types in the API start with "EGL". + * + * Khronos STRONGLY RECOMMENDS that you use the default definitions + * provided below, since these changes affect both binary and source + * portability of applications using EGL running on different EGL + * implementations. + */ + +#if defined(_WIN32) || defined(__VC32__) && !defined(__CYGWIN__) && !defined(__SCITECH_SNAP__) /* Win32 and WinCE */ +#ifndef WIN32_LEAN_AND_MEAN +#define WIN32_LEAN_AND_MEAN 1 +#endif +#include + +typedef HDC EGLNativeDisplayType; +typedef HBITMAP EGLNativePixmapType; +typedef HWND EGLNativeWindowType; + +#elif defined(__WINSCW__) || defined(__SYMBIAN32__) /* Symbian */ + +typedef int EGLNativeDisplayType; +typedef void *EGLNativeWindowType; +typedef void *EGLNativePixmapType; + +#elif defined(__ANDROID__) || defined(ANDROID) + +#include + +struct egl_native_pixmap_t; + +typedef struct ANativeWindow* EGLNativeWindowType; +typedef struct egl_native_pixmap_t* EGLNativePixmapType; +typedef void* EGLNativeDisplayType; + +#elif defined(__unix__) + +/* X11 (tentative) */ +#include +#include + +typedef Display *EGLNativeDisplayType; +typedef Pixmap EGLNativePixmapType; +typedef Window EGLNativeWindowType; + +#else +#error "Platform not recognized" +#endif + +/* EGL 1.2 types, renamed for consistency in EGL 1.3 */ +typedef EGLNativeDisplayType NativeDisplayType; +typedef EGLNativePixmapType NativePixmapType; +typedef EGLNativeWindowType NativeWindowType; + + +/* Define EGLint. This must be a signed integral type large enough to contain + * all legal attribute names and values passed into and out of EGL, whether + * their type is boolean, bitmask, enumerant (symbolic constant), integer, + * handle, or other. While in general a 32-bit integer will suffice, if + * handles are 64 bit types, then EGLint should be defined as a signed 64-bit + * integer type. + */ +typedef khronos_int32_t EGLint; + +#endif /* __eglplatform_h */ diff --git a/opencl/api/opencl/khronos/headers/GL/glext.h b/opencl/api/opencl/khronos/headers/GL/glext.h new file mode 100644 index 0000000000..c09e23b29b --- /dev/null +++ b/opencl/api/opencl/khronos/headers/GL/glext.h @@ -0,0 +1,9253 @@ +#ifndef __glext_h_ +#define __glext_h_ + +#ifdef __cplusplus +extern "C" { +#endif + +/* +** Copyright (c) 2007-2009 The Khronos Group Inc. +** +** Permission is hereby granted, free of charge, to any person obtaining a +** copy of this software and/or associated documentation files (the +** "Materials"), to deal in the Materials without restriction, including +** without limitation the rights to use, copy, modify, merge, publish, +** distribute, sublicense, and/or sell copies of the Materials, and to +** permit persons to whom the Materials are furnished to do so, subject to +** the following conditions: +** +** The above copyright notice and this permission notice shall be included +** in all copies or substantial portions of the Materials. +** +** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY +** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. +*/ + +/* Header file version number, required by OpenGL ABI for Linux */ +/* glext.h last updated $Date: 2009-08-03 02:13:51 -0700 (Mon, 03 Aug 2009) $ */ +/* Current version at http://www.opengl.org/registry/ */ +#define GL_GLEXT_VERSION 54 + +/* Function declaration macros - to move into glplatform.h */ + +#if defined(_WIN32) && !defined(APIENTRY) && !defined(__CYGWIN__) && !defined(__SCITECH_SNAP__) +#define WIN32_LEAN_AND_MEAN 1 +#include +#endif + +#ifndef APIENTRY +#define APIENTRY +#endif +#ifndef APIENTRYP +#define APIENTRYP APIENTRY * +#endif +#ifndef GLAPI +#define GLAPI extern +#endif + +/*************************************************************/ + +#ifndef GL_VERSION_1_2 +#define GL_UNSIGNED_BYTE_3_3_2 0x8032 +#define GL_UNSIGNED_SHORT_4_4_4_4 0x8033 +#define GL_UNSIGNED_SHORT_5_5_5_1 0x8034 +#define GL_UNSIGNED_INT_8_8_8_8 0x8035 +#define GL_UNSIGNED_INT_10_10_10_2 0x8036 +#define GL_TEXTURE_BINDING_3D 0x806A +#define GL_PACK_SKIP_IMAGES 0x806B +#define GL_PACK_IMAGE_HEIGHT 0x806C +#define GL_UNPACK_SKIP_IMAGES 0x806D +#define GL_UNPACK_IMAGE_HEIGHT 0x806E +#define GL_TEXTURE_3D 0x806F +#define GL_PROXY_TEXTURE_3D 0x8070 +#define GL_TEXTURE_DEPTH 0x8071 +#define GL_TEXTURE_WRAP_R 0x8072 +#define GL_MAX_3D_TEXTURE_SIZE 0x8073 +#define GL_UNSIGNED_BYTE_2_3_3_REV 0x8362 +#define GL_UNSIGNED_SHORT_5_6_5 0x8363 +#define GL_UNSIGNED_SHORT_5_6_5_REV 0x8364 +#define GL_UNSIGNED_SHORT_4_4_4_4_REV 0x8365 +#define GL_UNSIGNED_SHORT_1_5_5_5_REV 0x8366 +#define GL_UNSIGNED_INT_8_8_8_8_REV 0x8367 +#define GL_UNSIGNED_INT_2_10_10_10_REV 0x8368 +#define GL_BGR 0x80E0 +#define GL_BGRA 0x80E1 +#define GL_MAX_ELEMENTS_VERTICES 0x80E8 +#define GL_MAX_ELEMENTS_INDICES 0x80E9 +#define GL_CLAMP_TO_EDGE 0x812F +#define GL_TEXTURE_MIN_LOD 0x813A +#define GL_TEXTURE_MAX_LOD 0x813B +#define GL_TEXTURE_BASE_LEVEL 0x813C +#define GL_TEXTURE_MAX_LEVEL 0x813D +#define GL_SMOOTH_POINT_SIZE_RANGE 0x0B12 +#define GL_SMOOTH_POINT_SIZE_GRANULARITY 0x0B13 +#define GL_SMOOTH_LINE_WIDTH_RANGE 0x0B22 +#define GL_SMOOTH_LINE_WIDTH_GRANULARITY 0x0B23 +#define GL_ALIASED_LINE_WIDTH_RANGE 0x846E +#endif + +#ifndef GL_VERSION_1_2_DEPRECATED +#define GL_RESCALE_NORMAL 0x803A +#define GL_LIGHT_MODEL_COLOR_CONTROL 0x81F8 +#define GL_SINGLE_COLOR 0x81F9 +#define GL_SEPARATE_SPECULAR_COLOR 0x81FA +#define GL_ALIASED_POINT_SIZE_RANGE 0x846D +#endif + +#ifndef GL_ARB_imaging +#define GL_CONSTANT_COLOR 0x8001 +#define GL_ONE_MINUS_CONSTANT_COLOR 0x8002 +#define GL_CONSTANT_ALPHA 0x8003 +#define GL_ONE_MINUS_CONSTANT_ALPHA 0x8004 +#define GL_BLEND_COLOR 0x8005 +#define GL_FUNC_ADD 0x8006 +#define GL_MIN 0x8007 +#define GL_MAX 0x8008 +#define GL_BLEND_EQUATION 0x8009 +#define GL_FUNC_SUBTRACT 0x800A +#define GL_FUNC_REVERSE_SUBTRACT 0x800B +#endif + +#ifndef GL_ARB_imaging_DEPRECATED +#define GL_CONVOLUTION_1D 0x8010 +#define GL_CONVOLUTION_2D 0x8011 +#define GL_SEPARABLE_2D 0x8012 +#define GL_CONVOLUTION_BORDER_MODE 0x8013 +#define GL_CONVOLUTION_FILTER_SCALE 0x8014 +#define GL_CONVOLUTION_FILTER_BIAS 0x8015 +#define GL_REDUCE 0x8016 +#define GL_CONVOLUTION_FORMAT 0x8017 +#define GL_CONVOLUTION_WIDTH 0x8018 +#define GL_CONVOLUTION_HEIGHT 0x8019 +#define GL_MAX_CONVOLUTION_WIDTH 0x801A +#define GL_MAX_CONVOLUTION_HEIGHT 0x801B +#define GL_POST_CONVOLUTION_RED_SCALE 0x801C +#define GL_POST_CONVOLUTION_GREEN_SCALE 0x801D +#define GL_POST_CONVOLUTION_BLUE_SCALE 0x801E +#define GL_POST_CONVOLUTION_ALPHA_SCALE 0x801F +#define GL_POST_CONVOLUTION_RED_BIAS 0x8020 +#define GL_POST_CONVOLUTION_GREEN_BIAS 0x8021 +#define GL_POST_CONVOLUTION_BLUE_BIAS 0x8022 +#define GL_POST_CONVOLUTION_ALPHA_BIAS 0x8023 +#define GL_HISTOGRAM 0x8024 +#define GL_PROXY_HISTOGRAM 0x8025 +#define GL_HISTOGRAM_WIDTH 0x8026 +#define GL_HISTOGRAM_FORMAT 0x8027 +#define GL_HISTOGRAM_RED_SIZE 0x8028 +#define GL_HISTOGRAM_GREEN_SIZE 0x8029 +#define GL_HISTOGRAM_BLUE_SIZE 0x802A +#define GL_HISTOGRAM_ALPHA_SIZE 0x802B +#define GL_HISTOGRAM_LUMINANCE_SIZE 0x802C +#define GL_HISTOGRAM_SINK 0x802D +#define GL_MINMAX 0x802E +#define GL_MINMAX_FORMAT 0x802F +#define GL_MINMAX_SINK 0x8030 +#define GL_TABLE_TOO_LARGE 0x8031 +#define GL_COLOR_MATRIX 0x80B1 +#define GL_COLOR_MATRIX_STACK_DEPTH 0x80B2 +#define GL_MAX_COLOR_MATRIX_STACK_DEPTH 0x80B3 +#define GL_POST_COLOR_MATRIX_RED_SCALE 0x80B4 +#define GL_POST_COLOR_MATRIX_GREEN_SCALE 0x80B5 +#define GL_POST_COLOR_MATRIX_BLUE_SCALE 0x80B6 +#define GL_POST_COLOR_MATRIX_ALPHA_SCALE 0x80B7 +#define GL_POST_COLOR_MATRIX_RED_BIAS 0x80B8 +#define GL_POST_COLOR_MATRIX_GREEN_BIAS 0x80B9 +#define GL_POST_COLOR_MATRIX_BLUE_BIAS 0x80BA +#define GL_POST_COLOR_MATRIX_ALPHA_BIAS 0x80BB +#define GL_COLOR_TABLE 0x80D0 +#define GL_POST_CONVOLUTION_COLOR_TABLE 0x80D1 +#define GL_POST_COLOR_MATRIX_COLOR_TABLE 0x80D2 +#define GL_PROXY_COLOR_TABLE 0x80D3 +#define GL_PROXY_POST_CONVOLUTION_COLOR_TABLE 0x80D4 +#define GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE 0x80D5 +#define GL_COLOR_TABLE_SCALE 0x80D6 +#define GL_COLOR_TABLE_BIAS 0x80D7 +#define GL_COLOR_TABLE_FORMAT 0x80D8 +#define GL_COLOR_TABLE_WIDTH 0x80D9 +#define GL_COLOR_TABLE_RED_SIZE 0x80DA +#define GL_COLOR_TABLE_GREEN_SIZE 0x80DB +#define GL_COLOR_TABLE_BLUE_SIZE 0x80DC +#define GL_COLOR_TABLE_ALPHA_SIZE 0x80DD +#define GL_COLOR_TABLE_LUMINANCE_SIZE 0x80DE +#define GL_COLOR_TABLE_INTENSITY_SIZE 0x80DF +#define GL_CONSTANT_BORDER 0x8151 +#define GL_REPLICATE_BORDER 0x8153 +#define GL_CONVOLUTION_BORDER_COLOR 0x8154 +#endif + +#ifndef GL_VERSION_1_3 +#define GL_TEXTURE0 0x84C0 +#define GL_TEXTURE1 0x84C1 +#define GL_TEXTURE2 0x84C2 +#define GL_TEXTURE3 0x84C3 +#define GL_TEXTURE4 0x84C4 +#define GL_TEXTURE5 0x84C5 +#define GL_TEXTURE6 0x84C6 +#define GL_TEXTURE7 0x84C7 +#define GL_TEXTURE8 0x84C8 +#define GL_TEXTURE9 0x84C9 +#define GL_TEXTURE10 0x84CA +#define GL_TEXTURE11 0x84CB +#define GL_TEXTURE12 0x84CC +#define GL_TEXTURE13 0x84CD +#define GL_TEXTURE14 0x84CE +#define GL_TEXTURE15 0x84CF +#define GL_TEXTURE16 0x84D0 +#define GL_TEXTURE17 0x84D1 +#define GL_TEXTURE18 0x84D2 +#define GL_TEXTURE19 0x84D3 +#define GL_TEXTURE20 0x84D4 +#define GL_TEXTURE21 0x84D5 +#define GL_TEXTURE22 0x84D6 +#define GL_TEXTURE23 0x84D7 +#define GL_TEXTURE24 0x84D8 +#define GL_TEXTURE25 0x84D9 +#define GL_TEXTURE26 0x84DA +#define GL_TEXTURE27 0x84DB +#define GL_TEXTURE28 0x84DC +#define GL_TEXTURE29 0x84DD +#define GL_TEXTURE30 0x84DE +#define GL_TEXTURE31 0x84DF +#define GL_ACTIVE_TEXTURE 0x84E0 +#define GL_MULTISAMPLE 0x809D +#define GL_SAMPLE_ALPHA_TO_COVERAGE 0x809E +#define GL_SAMPLE_ALPHA_TO_ONE 0x809F +#define GL_SAMPLE_COVERAGE 0x80A0 +#define GL_SAMPLE_BUFFERS 0x80A8 +#define GL_SAMPLES 0x80A9 +#define GL_SAMPLE_COVERAGE_VALUE 0x80AA +#define GL_SAMPLE_COVERAGE_INVERT 0x80AB +#define GL_TEXTURE_CUBE_MAP 0x8513 +#define GL_TEXTURE_BINDING_CUBE_MAP 0x8514 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_X 0x8515 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X 0x8516 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y 0x8517 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y 0x8518 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z 0x8519 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z 0x851A +#define GL_PROXY_TEXTURE_CUBE_MAP 0x851B +#define GL_MAX_CUBE_MAP_TEXTURE_SIZE 0x851C +#define GL_COMPRESSED_RGB 0x84ED +#define GL_COMPRESSED_RGBA 0x84EE +#define GL_TEXTURE_COMPRESSION_HINT 0x84EF +#define GL_TEXTURE_COMPRESSED_IMAGE_SIZE 0x86A0 +#define GL_TEXTURE_COMPRESSED 0x86A1 +#define GL_NUM_COMPRESSED_TEXTURE_FORMATS 0x86A2 +#define GL_COMPRESSED_TEXTURE_FORMATS 0x86A3 +#define GL_CLAMP_TO_BORDER 0x812D +#endif + +#ifndef GL_VERSION_1_3_DEPRECATED +#define GL_CLIENT_ACTIVE_TEXTURE 0x84E1 +#define GL_MAX_TEXTURE_UNITS 0x84E2 +#define GL_TRANSPOSE_MODELVIEW_MATRIX 0x84E3 +#define GL_TRANSPOSE_PROJECTION_MATRIX 0x84E4 +#define GL_TRANSPOSE_TEXTURE_MATRIX 0x84E5 +#define GL_TRANSPOSE_COLOR_MATRIX 0x84E6 +#define GL_MULTISAMPLE_BIT 0x20000000 +#define GL_NORMAL_MAP 0x8511 +#define GL_REFLECTION_MAP 0x8512 +#define GL_COMPRESSED_ALPHA 0x84E9 +#define GL_COMPRESSED_LUMINANCE 0x84EA +#define GL_COMPRESSED_LUMINANCE_ALPHA 0x84EB +#define GL_COMPRESSED_INTENSITY 0x84EC +#define GL_COMBINE 0x8570 +#define GL_COMBINE_RGB 0x8571 +#define GL_COMBINE_ALPHA 0x8572 +#define GL_SOURCE0_RGB 0x8580 +#define GL_SOURCE1_RGB 0x8581 +#define GL_SOURCE2_RGB 0x8582 +#define GL_SOURCE0_ALPHA 0x8588 +#define GL_SOURCE1_ALPHA 0x8589 +#define GL_SOURCE2_ALPHA 0x858A +#define GL_OPERAND0_RGB 0x8590 +#define GL_OPERAND1_RGB 0x8591 +#define GL_OPERAND2_RGB 0x8592 +#define GL_OPERAND0_ALPHA 0x8598 +#define GL_OPERAND1_ALPHA 0x8599 +#define GL_OPERAND2_ALPHA 0x859A +#define GL_RGB_SCALE 0x8573 +#define GL_ADD_SIGNED 0x8574 +#define GL_INTERPOLATE 0x8575 +#define GL_SUBTRACT 0x84E7 +#define GL_CONSTANT 0x8576 +#define GL_PRIMARY_COLOR 0x8577 +#define GL_PREVIOUS 0x8578 +#define GL_DOT3_RGB 0x86AE +#define GL_DOT3_RGBA 0x86AF +#endif + +#ifndef GL_VERSION_1_4 +#define GL_BLEND_DST_RGB 0x80C8 +#define GL_BLEND_SRC_RGB 0x80C9 +#define GL_BLEND_DST_ALPHA 0x80CA +#define GL_BLEND_SRC_ALPHA 0x80CB +#define GL_POINT_FADE_THRESHOLD_SIZE 0x8128 +#define GL_DEPTH_COMPONENT16 0x81A5 +#define GL_DEPTH_COMPONENT24 0x81A6 +#define GL_DEPTH_COMPONENT32 0x81A7 +#define GL_MIRRORED_REPEAT 0x8370 +#define GL_MAX_TEXTURE_LOD_BIAS 0x84FD +#define GL_TEXTURE_LOD_BIAS 0x8501 +#define GL_INCR_WRAP 0x8507 +#define GL_DECR_WRAP 0x8508 +#define GL_TEXTURE_DEPTH_SIZE 0x884A +#define GL_TEXTURE_COMPARE_MODE 0x884C +#define GL_TEXTURE_COMPARE_FUNC 0x884D +#endif + +#ifndef GL_VERSION_1_4_DEPRECATED +#define GL_POINT_SIZE_MIN 0x8126 +#define GL_POINT_SIZE_MAX 0x8127 +#define GL_POINT_DISTANCE_ATTENUATION 0x8129 +#define GL_GENERATE_MIPMAP 0x8191 +#define GL_GENERATE_MIPMAP_HINT 0x8192 +#define GL_FOG_COORDINATE_SOURCE 0x8450 +#define GL_FOG_COORDINATE 0x8451 +#define GL_FRAGMENT_DEPTH 0x8452 +#define GL_CURRENT_FOG_COORDINATE 0x8453 +#define GL_FOG_COORDINATE_ARRAY_TYPE 0x8454 +#define GL_FOG_COORDINATE_ARRAY_STRIDE 0x8455 +#define GL_FOG_COORDINATE_ARRAY_POINTER 0x8456 +#define GL_FOG_COORDINATE_ARRAY 0x8457 +#define GL_COLOR_SUM 0x8458 +#define GL_CURRENT_SECONDARY_COLOR 0x8459 +#define GL_SECONDARY_COLOR_ARRAY_SIZE 0x845A +#define GL_SECONDARY_COLOR_ARRAY_TYPE 0x845B +#define GL_SECONDARY_COLOR_ARRAY_STRIDE 0x845C +#define GL_SECONDARY_COLOR_ARRAY_POINTER 0x845D +#define GL_SECONDARY_COLOR_ARRAY 0x845E +#define GL_TEXTURE_FILTER_CONTROL 0x8500 +#define GL_DEPTH_TEXTURE_MODE 0x884B +#define GL_COMPARE_R_TO_TEXTURE 0x884E +#endif + +#ifndef GL_VERSION_1_5 +#define GL_BUFFER_SIZE 0x8764 +#define GL_BUFFER_USAGE 0x8765 +#define GL_QUERY_COUNTER_BITS 0x8864 +#define GL_CURRENT_QUERY 0x8865 +#define GL_QUERY_RESULT 0x8866 +#define GL_QUERY_RESULT_AVAILABLE 0x8867 +#define GL_ARRAY_BUFFER 0x8892 +#define GL_ELEMENT_ARRAY_BUFFER 0x8893 +#define GL_ARRAY_BUFFER_BINDING 0x8894 +#define GL_ELEMENT_ARRAY_BUFFER_BINDING 0x8895 +#define GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING 0x889F +#define GL_READ_ONLY 0x88B8 +#define GL_WRITE_ONLY 0x88B9 +#define GL_READ_WRITE 0x88BA +#define GL_BUFFER_ACCESS 0x88BB +#define GL_BUFFER_MAPPED 0x88BC +#define GL_BUFFER_MAP_POINTER 0x88BD +#define GL_STREAM_DRAW 0x88E0 +#define GL_STREAM_READ 0x88E1 +#define GL_STREAM_COPY 0x88E2 +#define GL_STATIC_DRAW 0x88E4 +#define GL_STATIC_READ 0x88E5 +#define GL_STATIC_COPY 0x88E6 +#define GL_DYNAMIC_DRAW 0x88E8 +#define GL_DYNAMIC_READ 0x88E9 +#define GL_DYNAMIC_COPY 0x88EA +#define GL_SAMPLES_PASSED 0x8914 +#endif + +#ifndef GL_VERSION_1_5_DEPRECATED +#define GL_VERTEX_ARRAY_BUFFER_BINDING 0x8896 +#define GL_NORMAL_ARRAY_BUFFER_BINDING 0x8897 +#define GL_COLOR_ARRAY_BUFFER_BINDING 0x8898 +#define GL_INDEX_ARRAY_BUFFER_BINDING 0x8899 +#define GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING 0x889A +#define GL_EDGE_FLAG_ARRAY_BUFFER_BINDING 0x889B +#define GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING 0x889C +#define GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING 0x889D +#define GL_WEIGHT_ARRAY_BUFFER_BINDING 0x889E +#define GL_FOG_COORD_SRC 0x8450 +#define GL_FOG_COORD 0x8451 +#define GL_CURRENT_FOG_COORD 0x8453 +#define GL_FOG_COORD_ARRAY_TYPE 0x8454 +#define GL_FOG_COORD_ARRAY_STRIDE 0x8455 +#define GL_FOG_COORD_ARRAY_POINTER 0x8456 +#define GL_FOG_COORD_ARRAY 0x8457 +#define GL_FOG_COORD_ARRAY_BUFFER_BINDING 0x889D +#define GL_SRC0_RGB 0x8580 +#define GL_SRC1_RGB 0x8581 +#define GL_SRC2_RGB 0x8582 +#define GL_SRC0_ALPHA 0x8588 +#define GL_SRC1_ALPHA 0x8589 +#define GL_SRC2_ALPHA 0x858A +#endif + +#ifndef GL_VERSION_2_0 +#define GL_BLEND_EQUATION_RGB 0x8009 +#define GL_VERTEX_ATTRIB_ARRAY_ENABLED 0x8622 +#define GL_VERTEX_ATTRIB_ARRAY_SIZE 0x8623 +#define GL_VERTEX_ATTRIB_ARRAY_STRIDE 0x8624 +#define GL_VERTEX_ATTRIB_ARRAY_TYPE 0x8625 +#define GL_CURRENT_VERTEX_ATTRIB 0x8626 +#define GL_VERTEX_PROGRAM_POINT_SIZE 0x8642 +#define GL_VERTEX_ATTRIB_ARRAY_POINTER 0x8645 +#define GL_STENCIL_BACK_FUNC 0x8800 +#define GL_STENCIL_BACK_FAIL 0x8801 +#define GL_STENCIL_BACK_PASS_DEPTH_FAIL 0x8802 +#define GL_STENCIL_BACK_PASS_DEPTH_PASS 0x8803 +#define GL_MAX_DRAW_BUFFERS 0x8824 +#define GL_DRAW_BUFFER0 0x8825 +#define GL_DRAW_BUFFER1 0x8826 +#define GL_DRAW_BUFFER2 0x8827 +#define GL_DRAW_BUFFER3 0x8828 +#define GL_DRAW_BUFFER4 0x8829 +#define GL_DRAW_BUFFER5 0x882A +#define GL_DRAW_BUFFER6 0x882B +#define GL_DRAW_BUFFER7 0x882C +#define GL_DRAW_BUFFER8 0x882D +#define GL_DRAW_BUFFER9 0x882E +#define GL_DRAW_BUFFER10 0x882F +#define GL_DRAW_BUFFER11 0x8830 +#define GL_DRAW_BUFFER12 0x8831 +#define GL_DRAW_BUFFER13 0x8832 +#define GL_DRAW_BUFFER14 0x8833 +#define GL_DRAW_BUFFER15 0x8834 +#define GL_BLEND_EQUATION_ALPHA 0x883D +#define GL_MAX_VERTEX_ATTRIBS 0x8869 +#define GL_VERTEX_ATTRIB_ARRAY_NORMALIZED 0x886A +#define GL_MAX_TEXTURE_IMAGE_UNITS 0x8872 +#define GL_FRAGMENT_SHADER 0x8B30 +#define GL_VERTEX_SHADER 0x8B31 +#define GL_MAX_FRAGMENT_UNIFORM_COMPONENTS 0x8B49 +#define GL_MAX_VERTEX_UNIFORM_COMPONENTS 0x8B4A +#define GL_MAX_VARYING_FLOATS 0x8B4B +#define GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS 0x8B4C +#define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS 0x8B4D +#define GL_SHADER_TYPE 0x8B4F +#define GL_FLOAT_VEC2 0x8B50 +#define GL_FLOAT_VEC3 0x8B51 +#define GL_FLOAT_VEC4 0x8B52 +#define GL_INT_VEC2 0x8B53 +#define GL_INT_VEC3 0x8B54 +#define GL_INT_VEC4 0x8B55 +#define GL_BOOL 0x8B56 +#define GL_BOOL_VEC2 0x8B57 +#define GL_BOOL_VEC3 0x8B58 +#define GL_BOOL_VEC4 0x8B59 +#define GL_FLOAT_MAT2 0x8B5A +#define GL_FLOAT_MAT3 0x8B5B +#define GL_FLOAT_MAT4 0x8B5C +#define GL_SAMPLER_1D 0x8B5D +#define GL_SAMPLER_2D 0x8B5E +#define GL_SAMPLER_3D 0x8B5F +#define GL_SAMPLER_CUBE 0x8B60 +#define GL_SAMPLER_1D_SHADOW 0x8B61 +#define GL_SAMPLER_2D_SHADOW 0x8B62 +#define GL_DELETE_STATUS 0x8B80 +#define GL_COMPILE_STATUS 0x8B81 +#define GL_LINK_STATUS 0x8B82 +#define GL_VALIDATE_STATUS 0x8B83 +#define GL_INFO_LOG_LENGTH 0x8B84 +#define GL_ATTACHED_SHADERS 0x8B85 +#define GL_ACTIVE_UNIFORMS 0x8B86 +#define GL_ACTIVE_UNIFORM_MAX_LENGTH 0x8B87 +#define GL_SHADER_SOURCE_LENGTH 0x8B88 +#define GL_ACTIVE_ATTRIBUTES 0x8B89 +#define GL_ACTIVE_ATTRIBUTE_MAX_LENGTH 0x8B8A +#define GL_FRAGMENT_SHADER_DERIVATIVE_HINT 0x8B8B +#define GL_SHADING_LANGUAGE_VERSION 0x8B8C +#define GL_CURRENT_PROGRAM 0x8B8D +#define GL_POINT_SPRITE_COORD_ORIGIN 0x8CA0 +#define GL_LOWER_LEFT 0x8CA1 +#define GL_UPPER_LEFT 0x8CA2 +#define GL_STENCIL_BACK_REF 0x8CA3 +#define GL_STENCIL_BACK_VALUE_MASK 0x8CA4 +#define GL_STENCIL_BACK_WRITEMASK 0x8CA5 +#endif + +#ifndef GL_VERSION_2_0_DEPRECATED +#define GL_VERTEX_PROGRAM_TWO_SIDE 0x8643 +#define GL_POINT_SPRITE 0x8861 +#define GL_COORD_REPLACE 0x8862 +#define GL_MAX_TEXTURE_COORDS 0x8871 +#endif + +#ifndef GL_VERSION_2_1 +#define GL_PIXEL_PACK_BUFFER 0x88EB +#define GL_PIXEL_UNPACK_BUFFER 0x88EC +#define GL_PIXEL_PACK_BUFFER_BINDING 0x88ED +#define GL_PIXEL_UNPACK_BUFFER_BINDING 0x88EF +#define GL_FLOAT_MAT2x3 0x8B65 +#define GL_FLOAT_MAT2x4 0x8B66 +#define GL_FLOAT_MAT3x2 0x8B67 +#define GL_FLOAT_MAT3x4 0x8B68 +#define GL_FLOAT_MAT4x2 0x8B69 +#define GL_FLOAT_MAT4x3 0x8B6A +#define GL_SRGB 0x8C40 +#define GL_SRGB8 0x8C41 +#define GL_SRGB_ALPHA 0x8C42 +#define GL_SRGB8_ALPHA8 0x8C43 +#define GL_COMPRESSED_SRGB 0x8C48 +#define GL_COMPRESSED_SRGB_ALPHA 0x8C49 +#endif + +#ifndef GL_VERSION_2_1_DEPRECATED +#define GL_CURRENT_RASTER_SECONDARY_COLOR 0x845F +#define GL_SLUMINANCE_ALPHA 0x8C44 +#define GL_SLUMINANCE8_ALPHA8 0x8C45 +#define GL_SLUMINANCE 0x8C46 +#define GL_SLUMINANCE8 0x8C47 +#define GL_COMPRESSED_SLUMINANCE 0x8C4A +#define GL_COMPRESSED_SLUMINANCE_ALPHA 0x8C4B +#endif + +#ifndef GL_VERSION_3_0 +#define GL_COMPARE_REF_TO_TEXTURE 0x884E +#define GL_CLIP_DISTANCE0 0x3000 +#define GL_CLIP_DISTANCE1 0x3001 +#define GL_CLIP_DISTANCE2 0x3002 +#define GL_CLIP_DISTANCE3 0x3003 +#define GL_CLIP_DISTANCE4 0x3004 +#define GL_CLIP_DISTANCE5 0x3005 +#define GL_CLIP_DISTANCE6 0x3006 +#define GL_CLIP_DISTANCE7 0x3007 +#define GL_MAX_CLIP_DISTANCES 0x0D32 +#define GL_MAJOR_VERSION 0x821B +#define GL_MINOR_VERSION 0x821C +#define GL_NUM_EXTENSIONS 0x821D +#define GL_CONTEXT_FLAGS 0x821E +#define GL_DEPTH_BUFFER 0x8223 +#define GL_STENCIL_BUFFER 0x8224 +#define GL_COMPRESSED_RED 0x8225 +#define GL_COMPRESSED_RG 0x8226 +#define GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT 0x0001 +#define GL_RGBA32F 0x8814 +#define GL_RGB32F 0x8815 +#define GL_RGBA16F 0x881A +#define GL_RGB16F 0x881B +#define GL_VERTEX_ATTRIB_ARRAY_INTEGER 0x88FD +#define GL_MAX_ARRAY_TEXTURE_LAYERS 0x88FF +#define GL_MIN_PROGRAM_TEXEL_OFFSET 0x8904 +#define GL_MAX_PROGRAM_TEXEL_OFFSET 0x8905 +#define GL_CLAMP_READ_COLOR 0x891C +#define GL_FIXED_ONLY 0x891D +#define GL_MAX_VARYING_COMPONENTS 0x8B4B +#define GL_TEXTURE_1D_ARRAY 0x8C18 +#define GL_PROXY_TEXTURE_1D_ARRAY 0x8C19 +#define GL_TEXTURE_2D_ARRAY 0x8C1A +#define GL_PROXY_TEXTURE_2D_ARRAY 0x8C1B +#define GL_TEXTURE_BINDING_1D_ARRAY 0x8C1C +#define GL_TEXTURE_BINDING_2D_ARRAY 0x8C1D +#define GL_R11F_G11F_B10F 0x8C3A +#define GL_UNSIGNED_INT_10F_11F_11F_REV 0x8C3B +#define GL_RGB9_E5 0x8C3D +#define GL_UNSIGNED_INT_5_9_9_9_REV 0x8C3E +#define GL_TEXTURE_SHARED_SIZE 0x8C3F +#define GL_TRANSFORM_FEEDBACK_VARYING_MAX_LENGTH 0x8C76 +#define GL_TRANSFORM_FEEDBACK_BUFFER_MODE 0x8C7F +#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS 0x8C80 +#define GL_TRANSFORM_FEEDBACK_VARYINGS 0x8C83 +#define GL_TRANSFORM_FEEDBACK_BUFFER_START 0x8C84 +#define GL_TRANSFORM_FEEDBACK_BUFFER_SIZE 0x8C85 +#define GL_PRIMITIVES_GENERATED 0x8C87 +#define GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN 0x8C88 +#define GL_RASTERIZER_DISCARD 0x8C89 +#define GL_MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS 0x8C8A +#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS 0x8C8B +#define GL_INTERLEAVED_ATTRIBS 0x8C8C +#define GL_SEPARATE_ATTRIBS 0x8C8D +#define GL_TRANSFORM_FEEDBACK_BUFFER 0x8C8E +#define GL_TRANSFORM_FEEDBACK_BUFFER_BINDING 0x8C8F +#define GL_RGBA32UI 0x8D70 +#define GL_RGB32UI 0x8D71 +#define GL_RGBA16UI 0x8D76 +#define GL_RGB16UI 0x8D77 +#define GL_RGBA8UI 0x8D7C +#define GL_RGB8UI 0x8D7D +#define GL_RGBA32I 0x8D82 +#define GL_RGB32I 0x8D83 +#define GL_RGBA16I 0x8D88 +#define GL_RGB16I 0x8D89 +#define GL_RGBA8I 0x8D8E +#define GL_RGB8I 0x8D8F +#define GL_RED_INTEGER 0x8D94 +#define GL_GREEN_INTEGER 0x8D95 +#define GL_BLUE_INTEGER 0x8D96 +#define GL_RGB_INTEGER 0x8D98 +#define GL_RGBA_INTEGER 0x8D99 +#define GL_BGR_INTEGER 0x8D9A +#define GL_BGRA_INTEGER 0x8D9B +#define GL_SAMPLER_1D_ARRAY 0x8DC0 +#define GL_SAMPLER_2D_ARRAY 0x8DC1 +#define GL_SAMPLER_1D_ARRAY_SHADOW 0x8DC3 +#define GL_SAMPLER_2D_ARRAY_SHADOW 0x8DC4 +#define GL_SAMPLER_CUBE_SHADOW 0x8DC5 +#define GL_UNSIGNED_INT_VEC2 0x8DC6 +#define GL_UNSIGNED_INT_VEC3 0x8DC7 +#define GL_UNSIGNED_INT_VEC4 0x8DC8 +#define GL_INT_SAMPLER_1D 0x8DC9 +#define GL_INT_SAMPLER_2D 0x8DCA +#define GL_INT_SAMPLER_3D 0x8DCB +#define GL_INT_SAMPLER_CUBE 0x8DCC +#define GL_INT_SAMPLER_1D_ARRAY 0x8DCE +#define GL_INT_SAMPLER_2D_ARRAY 0x8DCF +#define GL_UNSIGNED_INT_SAMPLER_1D 0x8DD1 +#define GL_UNSIGNED_INT_SAMPLER_2D 0x8DD2 +#define GL_UNSIGNED_INT_SAMPLER_3D 0x8DD3 +#define GL_UNSIGNED_INT_SAMPLER_CUBE 0x8DD4 +#define GL_UNSIGNED_INT_SAMPLER_1D_ARRAY 0x8DD6 +#define GL_UNSIGNED_INT_SAMPLER_2D_ARRAY 0x8DD7 +#define GL_QUERY_WAIT 0x8E13 +#define GL_QUERY_NO_WAIT 0x8E14 +#define GL_QUERY_BY_REGION_WAIT 0x8E15 +#define GL_QUERY_BY_REGION_NO_WAIT 0x8E16 +#define GL_BUFFER_ACCESS_FLAGS 0x911F +#define GL_BUFFER_MAP_LENGTH 0x9120 +#define GL_BUFFER_MAP_OFFSET 0x9121 +/* Reuse tokens from ARB_depth_buffer_float */ +/* reuse GL_DEPTH_COMPONENT32F */ +/* reuse GL_DEPTH32F_STENCIL8 */ +/* reuse GL_FLOAT_32_UNSIGNED_INT_24_8_REV */ +/* Reuse tokens from ARB_framebuffer_object */ +/* reuse GL_INVALID_FRAMEBUFFER_OPERATION */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */ +/* reuse GL_FRAMEBUFFER_DEFAULT */ +/* reuse GL_FRAMEBUFFER_UNDEFINED */ +/* reuse GL_DEPTH_STENCIL_ATTACHMENT */ +/* reuse GL_INDEX */ +/* reuse GL_MAX_RENDERBUFFER_SIZE */ +/* reuse GL_DEPTH_STENCIL */ +/* reuse GL_UNSIGNED_INT_24_8 */ +/* reuse GL_DEPTH24_STENCIL8 */ +/* reuse GL_TEXTURE_STENCIL_SIZE */ +/* reuse GL_TEXTURE_RED_TYPE */ +/* reuse GL_TEXTURE_GREEN_TYPE */ +/* reuse GL_TEXTURE_BLUE_TYPE */ +/* reuse GL_TEXTURE_ALPHA_TYPE */ +/* reuse GL_TEXTURE_DEPTH_TYPE */ +/* reuse GL_UNSIGNED_NORMALIZED */ +/* reuse GL_FRAMEBUFFER_BINDING */ +/* reuse GL_DRAW_FRAMEBUFFER_BINDING */ +/* reuse GL_RENDERBUFFER_BINDING */ +/* reuse GL_READ_FRAMEBUFFER */ +/* reuse GL_DRAW_FRAMEBUFFER */ +/* reuse GL_READ_FRAMEBUFFER_BINDING */ +/* reuse GL_RENDERBUFFER_SAMPLES */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ +/* reuse GL_FRAMEBUFFER_COMPLETE */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER */ +/* reuse GL_FRAMEBUFFER_UNSUPPORTED */ +/* reuse GL_MAX_COLOR_ATTACHMENTS */ +/* reuse GL_COLOR_ATTACHMENT0 */ +/* reuse GL_COLOR_ATTACHMENT1 */ +/* reuse GL_COLOR_ATTACHMENT2 */ +/* reuse GL_COLOR_ATTACHMENT3 */ +/* reuse GL_COLOR_ATTACHMENT4 */ +/* reuse GL_COLOR_ATTACHMENT5 */ +/* reuse GL_COLOR_ATTACHMENT6 */ +/* reuse GL_COLOR_ATTACHMENT7 */ +/* reuse GL_COLOR_ATTACHMENT8 */ +/* reuse GL_COLOR_ATTACHMENT9 */ +/* reuse GL_COLOR_ATTACHMENT10 */ +/* reuse GL_COLOR_ATTACHMENT11 */ +/* reuse GL_COLOR_ATTACHMENT12 */ +/* reuse GL_COLOR_ATTACHMENT13 */ +/* reuse GL_COLOR_ATTACHMENT14 */ +/* reuse GL_COLOR_ATTACHMENT15 */ +/* reuse GL_DEPTH_ATTACHMENT */ +/* reuse GL_STENCIL_ATTACHMENT */ +/* reuse GL_FRAMEBUFFER */ +/* reuse GL_RENDERBUFFER */ +/* reuse GL_RENDERBUFFER_WIDTH */ +/* reuse GL_RENDERBUFFER_HEIGHT */ +/* reuse GL_RENDERBUFFER_INTERNAL_FORMAT */ +/* reuse GL_STENCIL_INDEX1 */ +/* reuse GL_STENCIL_INDEX4 */ +/* reuse GL_STENCIL_INDEX8 */ +/* reuse GL_STENCIL_INDEX16 */ +/* reuse GL_RENDERBUFFER_RED_SIZE */ +/* reuse GL_RENDERBUFFER_GREEN_SIZE */ +/* reuse GL_RENDERBUFFER_BLUE_SIZE */ +/* reuse GL_RENDERBUFFER_ALPHA_SIZE */ +/* reuse GL_RENDERBUFFER_DEPTH_SIZE */ +/* reuse GL_RENDERBUFFER_STENCIL_SIZE */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */ +/* reuse GL_MAX_SAMPLES */ +/* Reuse tokens from ARB_framebuffer_sRGB */ +/* reuse GL_FRAMEBUFFER_SRGB */ +/* Reuse tokens from ARB_half_float_vertex */ +/* reuse GL_HALF_FLOAT */ +/* Reuse tokens from ARB_map_buffer_range */ +/* reuse GL_MAP_READ_BIT */ +/* reuse GL_MAP_WRITE_BIT */ +/* reuse GL_MAP_INVALIDATE_RANGE_BIT */ +/* reuse GL_MAP_INVALIDATE_BUFFER_BIT */ +/* reuse GL_MAP_FLUSH_EXPLICIT_BIT */ +/* reuse GL_MAP_UNSYNCHRONIZED_BIT */ +/* Reuse tokens from ARB_texture_compression_rgtc */ +/* reuse GL_COMPRESSED_RED_RGTC1 */ +/* reuse GL_COMPRESSED_SIGNED_RED_RGTC1 */ +/* reuse GL_COMPRESSED_RG_RGTC2 */ +/* reuse GL_COMPRESSED_SIGNED_RG_RGTC2 */ +/* Reuse tokens from ARB_texture_rg */ +/* reuse GL_RG */ +/* reuse GL_RG_INTEGER */ +/* reuse GL_R8 */ +/* reuse GL_R16 */ +/* reuse GL_RG8 */ +/* reuse GL_RG16 */ +/* reuse GL_R16F */ +/* reuse GL_R32F */ +/* reuse GL_RG16F */ +/* reuse GL_RG32F */ +/* reuse GL_R8I */ +/* reuse GL_R8UI */ +/* reuse GL_R16I */ +/* reuse GL_R16UI */ +/* reuse GL_R32I */ +/* reuse GL_R32UI */ +/* reuse GL_RG8I */ +/* reuse GL_RG8UI */ +/* reuse GL_RG16I */ +/* reuse GL_RG16UI */ +/* reuse GL_RG32I */ +/* reuse GL_RG32UI */ +/* Reuse tokens from ARB_vertex_array_object */ +/* reuse GL_VERTEX_ARRAY_BINDING */ +#endif + +#ifndef GL_VERSION_3_0_DEPRECATED +#define GL_CLAMP_VERTEX_COLOR 0x891A +#define GL_CLAMP_FRAGMENT_COLOR 0x891B +#define GL_ALPHA_INTEGER 0x8D97 +/* Reuse tokens from ARB_framebuffer_object */ +/* reuse GL_TEXTURE_LUMINANCE_TYPE */ +/* reuse GL_TEXTURE_INTENSITY_TYPE */ +#endif + +#ifndef GL_VERSION_3_1 +#define GL_SAMPLER_2D_RECT 0x8B63 +#define GL_SAMPLER_2D_RECT_SHADOW 0x8B64 +#define GL_SAMPLER_BUFFER 0x8DC2 +#define GL_INT_SAMPLER_2D_RECT 0x8DCD +#define GL_INT_SAMPLER_BUFFER 0x8DD0 +#define GL_UNSIGNED_INT_SAMPLER_2D_RECT 0x8DD5 +#define GL_UNSIGNED_INT_SAMPLER_BUFFER 0x8DD8 +#define GL_TEXTURE_BUFFER 0x8C2A +#define GL_MAX_TEXTURE_BUFFER_SIZE 0x8C2B +#define GL_TEXTURE_BINDING_BUFFER 0x8C2C +#define GL_TEXTURE_BUFFER_DATA_STORE_BINDING 0x8C2D +#define GL_TEXTURE_BUFFER_FORMAT 0x8C2E +#define GL_TEXTURE_RECTANGLE 0x84F5 +#define GL_TEXTURE_BINDING_RECTANGLE 0x84F6 +#define GL_PROXY_TEXTURE_RECTANGLE 0x84F7 +#define GL_MAX_RECTANGLE_TEXTURE_SIZE 0x84F8 +#define GL_RED_SNORM 0x8F90 +#define GL_RG_SNORM 0x8F91 +#define GL_RGB_SNORM 0x8F92 +#define GL_RGBA_SNORM 0x8F93 +#define GL_R8_SNORM 0x8F94 +#define GL_RG8_SNORM 0x8F95 +#define GL_RGB8_SNORM 0x8F96 +#define GL_RGBA8_SNORM 0x8F97 +#define GL_R16_SNORM 0x8F98 +#define GL_RG16_SNORM 0x8F99 +#define GL_RGB16_SNORM 0x8F9A +#define GL_RGBA16_SNORM 0x8F9B +#define GL_SIGNED_NORMALIZED 0x8F9C +#define GL_PRIMITIVE_RESTART 0x8F9D +#define GL_PRIMITIVE_RESTART_INDEX 0x8F9E +/* Reuse tokens from ARB_copy_buffer */ +/* reuse GL_COPY_READ_BUFFER */ +/* reuse GL_COPY_WRITE_BUFFER */ +/* Would reuse tokens from ARB_draw_instanced, but it has none */ +/* Reuse tokens from ARB_uniform_buffer_object */ +/* reuse GL_UNIFORM_BUFFER */ +/* reuse GL_UNIFORM_BUFFER_BINDING */ +/* reuse GL_UNIFORM_BUFFER_START */ +/* reuse GL_UNIFORM_BUFFER_SIZE */ +/* reuse GL_MAX_VERTEX_UNIFORM_BLOCKS */ +/* reuse GL_MAX_FRAGMENT_UNIFORM_BLOCKS */ +/* reuse GL_MAX_COMBINED_UNIFORM_BLOCKS */ +/* reuse GL_MAX_UNIFORM_BUFFER_BINDINGS */ +/* reuse GL_MAX_UNIFORM_BLOCK_SIZE */ +/* reuse GL_MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS */ +/* reuse GL_MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS */ +/* reuse GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT */ +/* reuse GL_ACTIVE_UNIFORM_BLOCK_MAX_NAME_LENGTH */ +/* reuse GL_ACTIVE_UNIFORM_BLOCKS */ +/* reuse GL_UNIFORM_TYPE */ +/* reuse GL_UNIFORM_SIZE */ +/* reuse GL_UNIFORM_NAME_LENGTH */ +/* reuse GL_UNIFORM_BLOCK_INDEX */ +/* reuse GL_UNIFORM_OFFSET */ +/* reuse GL_UNIFORM_ARRAY_STRIDE */ +/* reuse GL_UNIFORM_MATRIX_STRIDE */ +/* reuse GL_UNIFORM_IS_ROW_MAJOR */ +/* reuse GL_UNIFORM_BLOCK_BINDING */ +/* reuse GL_UNIFORM_BLOCK_DATA_SIZE */ +/* reuse GL_UNIFORM_BLOCK_NAME_LENGTH */ +/* reuse GL_UNIFORM_BLOCK_ACTIVE_UNIFORMS */ +/* reuse GL_UNIFORM_BLOCK_ACTIVE_UNIFORM_INDICES */ +/* reuse GL_UNIFORM_BLOCK_REFERENCED_BY_VERTEX_SHADER */ +/* reuse GL_UNIFORM_BLOCK_REFERENCED_BY_FRAGMENT_SHADER */ +/* reuse GL_INVALID_INDEX */ +#endif + +#ifndef GL_VERSION_3_2 +#define GL_CONTEXT_CORE_PROFILE_BIT 0x00000001 +#define GL_CONTEXT_COMPATIBILITY_PROFILE_BIT 0x00000002 +#define GL_LINES_ADJACENCY 0x000A +#define GL_LINE_STRIP_ADJACENCY 0x000B +#define GL_TRIANGLES_ADJACENCY 0x000C +#define GL_TRIANGLE_STRIP_ADJACENCY 0x000D +#define GL_PROGRAM_POINT_SIZE 0x8642 +#define GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS 0x8C29 +#define GL_FRAMEBUFFER_ATTACHMENT_LAYERED 0x8DA7 +#define GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS 0x8DA8 +#define GL_GEOMETRY_SHADER 0x8DD9 +#define GL_GEOMETRY_VERTICES_OUT 0x8916 +#define GL_GEOMETRY_INPUT_TYPE 0x8917 +#define GL_GEOMETRY_OUTPUT_TYPE 0x8918 +#define GL_MAX_GEOMETRY_UNIFORM_COMPONENTS 0x8DDF +#define GL_MAX_GEOMETRY_OUTPUT_VERTICES 0x8DE0 +#define GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS 0x8DE1 +#define GL_MAX_VERTEX_OUTPUT_COMPONENTS 0x9122 +#define GL_MAX_GEOMETRY_INPUT_COMPONENTS 0x9123 +#define GL_MAX_GEOMETRY_OUTPUT_COMPONENTS 0x9124 +#define GL_MAX_FRAGMENT_INPUT_COMPONENTS 0x9125 +#define GL_CONTEXT_PROFILE_MASK 0x9126 +/* reuse GL_MAX_VARYING_COMPONENTS */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ +/* Reuse tokens from ARB_depth_clamp */ +/* reuse GL_DEPTH_CLAMP */ +/* Would reuse tokens from ARB_draw_elements_base_vertex, but it has none */ +/* Would reuse tokens from ARB_fragment_coord_conventions, but it has none */ +/* Reuse tokens from ARB_provoking_vertex */ +/* reuse GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */ +/* reuse GL_FIRST_VERTEX_CONVENTION */ +/* reuse GL_LAST_VERTEX_CONVENTION */ +/* reuse GL_PROVOKING_VERTEX */ +/* Reuse tokens from ARB_seamless_cube_map */ +/* reuse GL_TEXTURE_CUBE_MAP_SEAMLESS */ +/* Reuse tokens from ARB_sync */ +/* reuse GL_MAX_SERVER_WAIT_TIMEOUT */ +/* reuse GL_OBJECT_TYPE */ +/* reuse GL_SYNC_CONDITION */ +/* reuse GL_SYNC_STATUS */ +/* reuse GL_SYNC_FLAGS */ +/* reuse GL_SYNC_FENCE */ +/* reuse GL_SYNC_GPU_COMMANDS_COMPLETE */ +/* reuse GL_UNSIGNALED */ +/* reuse GL_SIGNALED */ +/* reuse GL_ALREADY_SIGNALED */ +/* reuse GL_TIMEOUT_EXPIRED */ +/* reuse GL_CONDITION_SATISFIED */ +/* reuse GL_WAIT_FAILED */ +/* reuse GL_TIMEOUT_IGNORED */ +/* reuse GL_SYNC_FLUSH_COMMANDS_BIT */ +/* reuse GL_TIMEOUT_IGNORED */ +/* Reuse tokens from ARB_texture_multisample */ +/* reuse GL_SAMPLE_POSITION */ +/* reuse GL_SAMPLE_MASK */ +/* reuse GL_SAMPLE_MASK_VALUE */ +/* reuse GL_MAX_SAMPLE_MASK_WORDS */ +/* reuse GL_TEXTURE_2D_MULTISAMPLE */ +/* reuse GL_PROXY_TEXTURE_2D_MULTISAMPLE */ +/* reuse GL_TEXTURE_2D_MULTISAMPLE_ARRAY */ +/* reuse GL_PROXY_TEXTURE_2D_MULTISAMPLE_ARRAY */ +/* reuse GL_TEXTURE_BINDING_2D_MULTISAMPLE */ +/* reuse GL_TEXTURE_BINDING_2D_MULTISAMPLE_ARRAY */ +/* reuse GL_TEXTURE_SAMPLES */ +/* reuse GL_TEXTURE_FIXED_SAMPLE_LOCATIONS */ +/* reuse GL_SAMPLER_2D_MULTISAMPLE */ +/* reuse GL_INT_SAMPLER_2D_MULTISAMPLE */ +/* reuse GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE */ +/* reuse GL_SAMPLER_2D_MULTISAMPLE_ARRAY */ +/* reuse GL_INT_SAMPLER_2D_MULTISAMPLE_ARRAY */ +/* reuse GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE_ARRAY */ +/* reuse GL_MAX_COLOR_TEXTURE_SAMPLES */ +/* reuse GL_MAX_DEPTH_TEXTURE_SAMPLES */ +/* reuse GL_MAX_INTEGER_SAMPLES */ +/* Don't need to reuse tokens from ARB_vertex_array_bgra since they're already in 1.2 core */ +#endif + +#ifndef GL_ARB_multitexture +#define GL_TEXTURE0_ARB 0x84C0 +#define GL_TEXTURE1_ARB 0x84C1 +#define GL_TEXTURE2_ARB 0x84C2 +#define GL_TEXTURE3_ARB 0x84C3 +#define GL_TEXTURE4_ARB 0x84C4 +#define GL_TEXTURE5_ARB 0x84C5 +#define GL_TEXTURE6_ARB 0x84C6 +#define GL_TEXTURE7_ARB 0x84C7 +#define GL_TEXTURE8_ARB 0x84C8 +#define GL_TEXTURE9_ARB 0x84C9 +#define GL_TEXTURE10_ARB 0x84CA +#define GL_TEXTURE11_ARB 0x84CB +#define GL_TEXTURE12_ARB 0x84CC +#define GL_TEXTURE13_ARB 0x84CD +#define GL_TEXTURE14_ARB 0x84CE +#define GL_TEXTURE15_ARB 0x84CF +#define GL_TEXTURE16_ARB 0x84D0 +#define GL_TEXTURE17_ARB 0x84D1 +#define GL_TEXTURE18_ARB 0x84D2 +#define GL_TEXTURE19_ARB 0x84D3 +#define GL_TEXTURE20_ARB 0x84D4 +#define GL_TEXTURE21_ARB 0x84D5 +#define GL_TEXTURE22_ARB 0x84D6 +#define GL_TEXTURE23_ARB 0x84D7 +#define GL_TEXTURE24_ARB 0x84D8 +#define GL_TEXTURE25_ARB 0x84D9 +#define GL_TEXTURE26_ARB 0x84DA +#define GL_TEXTURE27_ARB 0x84DB +#define GL_TEXTURE28_ARB 0x84DC +#define GL_TEXTURE29_ARB 0x84DD +#define GL_TEXTURE30_ARB 0x84DE +#define GL_TEXTURE31_ARB 0x84DF +#define GL_ACTIVE_TEXTURE_ARB 0x84E0 +#define GL_CLIENT_ACTIVE_TEXTURE_ARB 0x84E1 +#define GL_MAX_TEXTURE_UNITS_ARB 0x84E2 +#endif + +#ifndef GL_ARB_transpose_matrix +#define GL_TRANSPOSE_MODELVIEW_MATRIX_ARB 0x84E3 +#define GL_TRANSPOSE_PROJECTION_MATRIX_ARB 0x84E4 +#define GL_TRANSPOSE_TEXTURE_MATRIX_ARB 0x84E5 +#define GL_TRANSPOSE_COLOR_MATRIX_ARB 0x84E6 +#endif + +#ifndef GL_ARB_multisample +#define GL_MULTISAMPLE_ARB 0x809D +#define GL_SAMPLE_ALPHA_TO_COVERAGE_ARB 0x809E +#define GL_SAMPLE_ALPHA_TO_ONE_ARB 0x809F +#define GL_SAMPLE_COVERAGE_ARB 0x80A0 +#define GL_SAMPLE_BUFFERS_ARB 0x80A8 +#define GL_SAMPLES_ARB 0x80A9 +#define GL_SAMPLE_COVERAGE_VALUE_ARB 0x80AA +#define GL_SAMPLE_COVERAGE_INVERT_ARB 0x80AB +#define GL_MULTISAMPLE_BIT_ARB 0x20000000 +#endif + +#ifndef GL_ARB_texture_env_add +#endif + +#ifndef GL_ARB_texture_cube_map +#define GL_NORMAL_MAP_ARB 0x8511 +#define GL_REFLECTION_MAP_ARB 0x8512 +#define GL_TEXTURE_CUBE_MAP_ARB 0x8513 +#define GL_TEXTURE_BINDING_CUBE_MAP_ARB 0x8514 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB 0x8515 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB 0x8516 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB 0x8517 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB 0x8518 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB 0x8519 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB 0x851A +#define GL_PROXY_TEXTURE_CUBE_MAP_ARB 0x851B +#define GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB 0x851C +#endif + +#ifndef GL_ARB_texture_compression +#define GL_COMPRESSED_ALPHA_ARB 0x84E9 +#define GL_COMPRESSED_LUMINANCE_ARB 0x84EA +#define GL_COMPRESSED_LUMINANCE_ALPHA_ARB 0x84EB +#define GL_COMPRESSED_INTENSITY_ARB 0x84EC +#define GL_COMPRESSED_RGB_ARB 0x84ED +#define GL_COMPRESSED_RGBA_ARB 0x84EE +#define GL_TEXTURE_COMPRESSION_HINT_ARB 0x84EF +#define GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB 0x86A0 +#define GL_TEXTURE_COMPRESSED_ARB 0x86A1 +#define GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB 0x86A2 +#define GL_COMPRESSED_TEXTURE_FORMATS_ARB 0x86A3 +#endif + +#ifndef GL_ARB_texture_border_clamp +#define GL_CLAMP_TO_BORDER_ARB 0x812D +#endif + +#ifndef GL_ARB_point_parameters +#define GL_POINT_SIZE_MIN_ARB 0x8126 +#define GL_POINT_SIZE_MAX_ARB 0x8127 +#define GL_POINT_FADE_THRESHOLD_SIZE_ARB 0x8128 +#define GL_POINT_DISTANCE_ATTENUATION_ARB 0x8129 +#endif + +#ifndef GL_ARB_vertex_blend +#define GL_MAX_VERTEX_UNITS_ARB 0x86A4 +#define GL_ACTIVE_VERTEX_UNITS_ARB 0x86A5 +#define GL_WEIGHT_SUM_UNITY_ARB 0x86A6 +#define GL_VERTEX_BLEND_ARB 0x86A7 +#define GL_CURRENT_WEIGHT_ARB 0x86A8 +#define GL_WEIGHT_ARRAY_TYPE_ARB 0x86A9 +#define GL_WEIGHT_ARRAY_STRIDE_ARB 0x86AA +#define GL_WEIGHT_ARRAY_SIZE_ARB 0x86AB +#define GL_WEIGHT_ARRAY_POINTER_ARB 0x86AC +#define GL_WEIGHT_ARRAY_ARB 0x86AD +#define GL_MODELVIEW0_ARB 0x1700 +#define GL_MODELVIEW1_ARB 0x850A +#define GL_MODELVIEW2_ARB 0x8722 +#define GL_MODELVIEW3_ARB 0x8723 +#define GL_MODELVIEW4_ARB 0x8724 +#define GL_MODELVIEW5_ARB 0x8725 +#define GL_MODELVIEW6_ARB 0x8726 +#define GL_MODELVIEW7_ARB 0x8727 +#define GL_MODELVIEW8_ARB 0x8728 +#define GL_MODELVIEW9_ARB 0x8729 +#define GL_MODELVIEW10_ARB 0x872A +#define GL_MODELVIEW11_ARB 0x872B +#define GL_MODELVIEW12_ARB 0x872C +#define GL_MODELVIEW13_ARB 0x872D +#define GL_MODELVIEW14_ARB 0x872E +#define GL_MODELVIEW15_ARB 0x872F +#define GL_MODELVIEW16_ARB 0x8730 +#define GL_MODELVIEW17_ARB 0x8731 +#define GL_MODELVIEW18_ARB 0x8732 +#define GL_MODELVIEW19_ARB 0x8733 +#define GL_MODELVIEW20_ARB 0x8734 +#define GL_MODELVIEW21_ARB 0x8735 +#define GL_MODELVIEW22_ARB 0x8736 +#define GL_MODELVIEW23_ARB 0x8737 +#define GL_MODELVIEW24_ARB 0x8738 +#define GL_MODELVIEW25_ARB 0x8739 +#define GL_MODELVIEW26_ARB 0x873A +#define GL_MODELVIEW27_ARB 0x873B +#define GL_MODELVIEW28_ARB 0x873C +#define GL_MODELVIEW29_ARB 0x873D +#define GL_MODELVIEW30_ARB 0x873E +#define GL_MODELVIEW31_ARB 0x873F +#endif + +#ifndef GL_ARB_matrix_palette +#define GL_MATRIX_PALETTE_ARB 0x8840 +#define GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB 0x8841 +#define GL_MAX_PALETTE_MATRICES_ARB 0x8842 +#define GL_CURRENT_PALETTE_MATRIX_ARB 0x8843 +#define GL_MATRIX_INDEX_ARRAY_ARB 0x8844 +#define GL_CURRENT_MATRIX_INDEX_ARB 0x8845 +#define GL_MATRIX_INDEX_ARRAY_SIZE_ARB 0x8846 +#define GL_MATRIX_INDEX_ARRAY_TYPE_ARB 0x8847 +#define GL_MATRIX_INDEX_ARRAY_STRIDE_ARB 0x8848 +#define GL_MATRIX_INDEX_ARRAY_POINTER_ARB 0x8849 +#endif + +#ifndef GL_ARB_texture_env_combine +#define GL_COMBINE_ARB 0x8570 +#define GL_COMBINE_RGB_ARB 0x8571 +#define GL_COMBINE_ALPHA_ARB 0x8572 +#define GL_SOURCE0_RGB_ARB 0x8580 +#define GL_SOURCE1_RGB_ARB 0x8581 +#define GL_SOURCE2_RGB_ARB 0x8582 +#define GL_SOURCE0_ALPHA_ARB 0x8588 +#define GL_SOURCE1_ALPHA_ARB 0x8589 +#define GL_SOURCE2_ALPHA_ARB 0x858A +#define GL_OPERAND0_RGB_ARB 0x8590 +#define GL_OPERAND1_RGB_ARB 0x8591 +#define GL_OPERAND2_RGB_ARB 0x8592 +#define GL_OPERAND0_ALPHA_ARB 0x8598 +#define GL_OPERAND1_ALPHA_ARB 0x8599 +#define GL_OPERAND2_ALPHA_ARB 0x859A +#define GL_RGB_SCALE_ARB 0x8573 +#define GL_ADD_SIGNED_ARB 0x8574 +#define GL_INTERPOLATE_ARB 0x8575 +#define GL_SUBTRACT_ARB 0x84E7 +#define GL_CONSTANT_ARB 0x8576 +#define GL_PRIMARY_COLOR_ARB 0x8577 +#define GL_PREVIOUS_ARB 0x8578 +#endif + +#ifndef GL_ARB_texture_env_crossbar +#endif + +#ifndef GL_ARB_texture_env_dot3 +#define GL_DOT3_RGB_ARB 0x86AE +#define GL_DOT3_RGBA_ARB 0x86AF +#endif + +#ifndef GL_ARB_texture_mirrored_repeat +#define GL_MIRRORED_REPEAT_ARB 0x8370 +#endif + +#ifndef GL_ARB_depth_texture +#define GL_DEPTH_COMPONENT16_ARB 0x81A5 +#define GL_DEPTH_COMPONENT24_ARB 0x81A6 +#define GL_DEPTH_COMPONENT32_ARB 0x81A7 +#define GL_TEXTURE_DEPTH_SIZE_ARB 0x884A +#define GL_DEPTH_TEXTURE_MODE_ARB 0x884B +#endif + +#ifndef GL_ARB_shadow +#define GL_TEXTURE_COMPARE_MODE_ARB 0x884C +#define GL_TEXTURE_COMPARE_FUNC_ARB 0x884D +#define GL_COMPARE_R_TO_TEXTURE_ARB 0x884E +#endif + +#ifndef GL_ARB_shadow_ambient +#define GL_TEXTURE_COMPARE_FAIL_VALUE_ARB 0x80BF +#endif + +#ifndef GL_ARB_window_pos +#endif + +#ifndef GL_ARB_vertex_program +#define GL_COLOR_SUM_ARB 0x8458 +#define GL_VERTEX_PROGRAM_ARB 0x8620 +#define GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB 0x8622 +#define GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB 0x8623 +#define GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB 0x8624 +#define GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB 0x8625 +#define GL_CURRENT_VERTEX_ATTRIB_ARB 0x8626 +#define GL_PROGRAM_LENGTH_ARB 0x8627 +#define GL_PROGRAM_STRING_ARB 0x8628 +#define GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB 0x862E +#define GL_MAX_PROGRAM_MATRICES_ARB 0x862F +#define GL_CURRENT_MATRIX_STACK_DEPTH_ARB 0x8640 +#define GL_CURRENT_MATRIX_ARB 0x8641 +#define GL_VERTEX_PROGRAM_POINT_SIZE_ARB 0x8642 +#define GL_VERTEX_PROGRAM_TWO_SIDE_ARB 0x8643 +#define GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB 0x8645 +#define GL_PROGRAM_ERROR_POSITION_ARB 0x864B +#define GL_PROGRAM_BINDING_ARB 0x8677 +#define GL_MAX_VERTEX_ATTRIBS_ARB 0x8869 +#define GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB 0x886A +#define GL_PROGRAM_ERROR_STRING_ARB 0x8874 +#define GL_PROGRAM_FORMAT_ASCII_ARB 0x8875 +#define GL_PROGRAM_FORMAT_ARB 0x8876 +#define GL_PROGRAM_INSTRUCTIONS_ARB 0x88A0 +#define GL_MAX_PROGRAM_INSTRUCTIONS_ARB 0x88A1 +#define GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB 0x88A2 +#define GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB 0x88A3 +#define GL_PROGRAM_TEMPORARIES_ARB 0x88A4 +#define GL_MAX_PROGRAM_TEMPORARIES_ARB 0x88A5 +#define GL_PROGRAM_NATIVE_TEMPORARIES_ARB 0x88A6 +#define GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB 0x88A7 +#define GL_PROGRAM_PARAMETERS_ARB 0x88A8 +#define GL_MAX_PROGRAM_PARAMETERS_ARB 0x88A9 +#define GL_PROGRAM_NATIVE_PARAMETERS_ARB 0x88AA +#define GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB 0x88AB +#define GL_PROGRAM_ATTRIBS_ARB 0x88AC +#define GL_MAX_PROGRAM_ATTRIBS_ARB 0x88AD +#define GL_PROGRAM_NATIVE_ATTRIBS_ARB 0x88AE +#define GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB 0x88AF +#define GL_PROGRAM_ADDRESS_REGISTERS_ARB 0x88B0 +#define GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB 0x88B1 +#define GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB 0x88B2 +#define GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB 0x88B3 +#define GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB 0x88B4 +#define GL_MAX_PROGRAM_ENV_PARAMETERS_ARB 0x88B5 +#define GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB 0x88B6 +#define GL_TRANSPOSE_CURRENT_MATRIX_ARB 0x88B7 +#define GL_MATRIX0_ARB 0x88C0 +#define GL_MATRIX1_ARB 0x88C1 +#define GL_MATRIX2_ARB 0x88C2 +#define GL_MATRIX3_ARB 0x88C3 +#define GL_MATRIX4_ARB 0x88C4 +#define GL_MATRIX5_ARB 0x88C5 +#define GL_MATRIX6_ARB 0x88C6 +#define GL_MATRIX7_ARB 0x88C7 +#define GL_MATRIX8_ARB 0x88C8 +#define GL_MATRIX9_ARB 0x88C9 +#define GL_MATRIX10_ARB 0x88CA +#define GL_MATRIX11_ARB 0x88CB +#define GL_MATRIX12_ARB 0x88CC +#define GL_MATRIX13_ARB 0x88CD +#define GL_MATRIX14_ARB 0x88CE +#define GL_MATRIX15_ARB 0x88CF +#define GL_MATRIX16_ARB 0x88D0 +#define GL_MATRIX17_ARB 0x88D1 +#define GL_MATRIX18_ARB 0x88D2 +#define GL_MATRIX19_ARB 0x88D3 +#define GL_MATRIX20_ARB 0x88D4 +#define GL_MATRIX21_ARB 0x88D5 +#define GL_MATRIX22_ARB 0x88D6 +#define GL_MATRIX23_ARB 0x88D7 +#define GL_MATRIX24_ARB 0x88D8 +#define GL_MATRIX25_ARB 0x88D9 +#define GL_MATRIX26_ARB 0x88DA +#define GL_MATRIX27_ARB 0x88DB +#define GL_MATRIX28_ARB 0x88DC +#define GL_MATRIX29_ARB 0x88DD +#define GL_MATRIX30_ARB 0x88DE +#define GL_MATRIX31_ARB 0x88DF +#endif + +#ifndef GL_ARB_fragment_program +#define GL_FRAGMENT_PROGRAM_ARB 0x8804 +#define GL_PROGRAM_ALU_INSTRUCTIONS_ARB 0x8805 +#define GL_PROGRAM_TEX_INSTRUCTIONS_ARB 0x8806 +#define GL_PROGRAM_TEX_INDIRECTIONS_ARB 0x8807 +#define GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB 0x8808 +#define GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB 0x8809 +#define GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB 0x880A +#define GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB 0x880B +#define GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB 0x880C +#define GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB 0x880D +#define GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB 0x880E +#define GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB 0x880F +#define GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB 0x8810 +#define GL_MAX_TEXTURE_COORDS_ARB 0x8871 +#define GL_MAX_TEXTURE_IMAGE_UNITS_ARB 0x8872 +#endif + +#ifndef GL_ARB_vertex_buffer_object +#define GL_BUFFER_SIZE_ARB 0x8764 +#define GL_BUFFER_USAGE_ARB 0x8765 +#define GL_ARRAY_BUFFER_ARB 0x8892 +#define GL_ELEMENT_ARRAY_BUFFER_ARB 0x8893 +#define GL_ARRAY_BUFFER_BINDING_ARB 0x8894 +#define GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB 0x8895 +#define GL_VERTEX_ARRAY_BUFFER_BINDING_ARB 0x8896 +#define GL_NORMAL_ARRAY_BUFFER_BINDING_ARB 0x8897 +#define GL_COLOR_ARRAY_BUFFER_BINDING_ARB 0x8898 +#define GL_INDEX_ARRAY_BUFFER_BINDING_ARB 0x8899 +#define GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB 0x889A +#define GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB 0x889B +#define GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB 0x889C +#define GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB 0x889D +#define GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB 0x889E +#define GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB 0x889F +#define GL_READ_ONLY_ARB 0x88B8 +#define GL_WRITE_ONLY_ARB 0x88B9 +#define GL_READ_WRITE_ARB 0x88BA +#define GL_BUFFER_ACCESS_ARB 0x88BB +#define GL_BUFFER_MAPPED_ARB 0x88BC +#define GL_BUFFER_MAP_POINTER_ARB 0x88BD +#define GL_STREAM_DRAW_ARB 0x88E0 +#define GL_STREAM_READ_ARB 0x88E1 +#define GL_STREAM_COPY_ARB 0x88E2 +#define GL_STATIC_DRAW_ARB 0x88E4 +#define GL_STATIC_READ_ARB 0x88E5 +#define GL_STATIC_COPY_ARB 0x88E6 +#define GL_DYNAMIC_DRAW_ARB 0x88E8 +#define GL_DYNAMIC_READ_ARB 0x88E9 +#define GL_DYNAMIC_COPY_ARB 0x88EA +#endif + +#ifndef GL_ARB_occlusion_query +#define GL_QUERY_COUNTER_BITS_ARB 0x8864 +#define GL_CURRENT_QUERY_ARB 0x8865 +#define GL_QUERY_RESULT_ARB 0x8866 +#define GL_QUERY_RESULT_AVAILABLE_ARB 0x8867 +#define GL_SAMPLES_PASSED_ARB 0x8914 +#endif + +#ifndef GL_ARB_shader_objects +#define GL_PROGRAM_OBJECT_ARB 0x8B40 +#define GL_SHADER_OBJECT_ARB 0x8B48 +#define GL_OBJECT_TYPE_ARB 0x8B4E +#define GL_OBJECT_SUBTYPE_ARB 0x8B4F +#define GL_FLOAT_VEC2_ARB 0x8B50 +#define GL_FLOAT_VEC3_ARB 0x8B51 +#define GL_FLOAT_VEC4_ARB 0x8B52 +#define GL_INT_VEC2_ARB 0x8B53 +#define GL_INT_VEC3_ARB 0x8B54 +#define GL_INT_VEC4_ARB 0x8B55 +#define GL_BOOL_ARB 0x8B56 +#define GL_BOOL_VEC2_ARB 0x8B57 +#define GL_BOOL_VEC3_ARB 0x8B58 +#define GL_BOOL_VEC4_ARB 0x8B59 +#define GL_FLOAT_MAT2_ARB 0x8B5A +#define GL_FLOAT_MAT3_ARB 0x8B5B +#define GL_FLOAT_MAT4_ARB 0x8B5C +#define GL_SAMPLER_1D_ARB 0x8B5D +#define GL_SAMPLER_2D_ARB 0x8B5E +#define GL_SAMPLER_3D_ARB 0x8B5F +#define GL_SAMPLER_CUBE_ARB 0x8B60 +#define GL_SAMPLER_1D_SHADOW_ARB 0x8B61 +#define GL_SAMPLER_2D_SHADOW_ARB 0x8B62 +#define GL_SAMPLER_2D_RECT_ARB 0x8B63 +#define GL_SAMPLER_2D_RECT_SHADOW_ARB 0x8B64 +#define GL_OBJECT_DELETE_STATUS_ARB 0x8B80 +#define GL_OBJECT_COMPILE_STATUS_ARB 0x8B81 +#define GL_OBJECT_LINK_STATUS_ARB 0x8B82 +#define GL_OBJECT_VALIDATE_STATUS_ARB 0x8B83 +#define GL_OBJECT_INFO_LOG_LENGTH_ARB 0x8B84 +#define GL_OBJECT_ATTACHED_OBJECTS_ARB 0x8B85 +#define GL_OBJECT_ACTIVE_UNIFORMS_ARB 0x8B86 +#define GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB 0x8B87 +#define GL_OBJECT_SHADER_SOURCE_LENGTH_ARB 0x8B88 +#endif + +#ifndef GL_ARB_vertex_shader +#define GL_VERTEX_SHADER_ARB 0x8B31 +#define GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB 0x8B4A +#define GL_MAX_VARYING_FLOATS_ARB 0x8B4B +#define GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB 0x8B4C +#define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB 0x8B4D +#define GL_OBJECT_ACTIVE_ATTRIBUTES_ARB 0x8B89 +#define GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB 0x8B8A +#endif + +#ifndef GL_ARB_fragment_shader +#define GL_FRAGMENT_SHADER_ARB 0x8B30 +#define GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB 0x8B49 +#define GL_FRAGMENT_SHADER_DERIVATIVE_HINT_ARB 0x8B8B +#endif + +#ifndef GL_ARB_shading_language_100 +#define GL_SHADING_LANGUAGE_VERSION_ARB 0x8B8C +#endif + +#ifndef GL_ARB_texture_non_power_of_two +#endif + +#ifndef GL_ARB_point_sprite +#define GL_POINT_SPRITE_ARB 0x8861 +#define GL_COORD_REPLACE_ARB 0x8862 +#endif + +#ifndef GL_ARB_fragment_program_shadow +#endif + +#ifndef GL_ARB_draw_buffers +#define GL_MAX_DRAW_BUFFERS_ARB 0x8824 +#define GL_DRAW_BUFFER0_ARB 0x8825 +#define GL_DRAW_BUFFER1_ARB 0x8826 +#define GL_DRAW_BUFFER2_ARB 0x8827 +#define GL_DRAW_BUFFER3_ARB 0x8828 +#define GL_DRAW_BUFFER4_ARB 0x8829 +#define GL_DRAW_BUFFER5_ARB 0x882A +#define GL_DRAW_BUFFER6_ARB 0x882B +#define GL_DRAW_BUFFER7_ARB 0x882C +#define GL_DRAW_BUFFER8_ARB 0x882D +#define GL_DRAW_BUFFER9_ARB 0x882E +#define GL_DRAW_BUFFER10_ARB 0x882F +#define GL_DRAW_BUFFER11_ARB 0x8830 +#define GL_DRAW_BUFFER12_ARB 0x8831 +#define GL_DRAW_BUFFER13_ARB 0x8832 +#define GL_DRAW_BUFFER14_ARB 0x8833 +#define GL_DRAW_BUFFER15_ARB 0x8834 +#endif + +#ifndef GL_ARB_texture_rectangle +#define GL_TEXTURE_RECTANGLE_ARB 0x84F5 +#define GL_TEXTURE_BINDING_RECTANGLE_ARB 0x84F6 +#define GL_PROXY_TEXTURE_RECTANGLE_ARB 0x84F7 +#define GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB 0x84F8 +#endif + +#ifndef GL_ARB_color_buffer_float +#define GL_RGBA_FLOAT_MODE_ARB 0x8820 +#define GL_CLAMP_VERTEX_COLOR_ARB 0x891A +#define GL_CLAMP_FRAGMENT_COLOR_ARB 0x891B +#define GL_CLAMP_READ_COLOR_ARB 0x891C +#define GL_FIXED_ONLY_ARB 0x891D +#endif + +#ifndef GL_ARB_half_float_pixel +#define GL_HALF_FLOAT_ARB 0x140B +#endif + +#ifndef GL_ARB_texture_float +#define GL_TEXTURE_RED_TYPE_ARB 0x8C10 +#define GL_TEXTURE_GREEN_TYPE_ARB 0x8C11 +#define GL_TEXTURE_BLUE_TYPE_ARB 0x8C12 +#define GL_TEXTURE_ALPHA_TYPE_ARB 0x8C13 +#define GL_TEXTURE_LUMINANCE_TYPE_ARB 0x8C14 +#define GL_TEXTURE_INTENSITY_TYPE_ARB 0x8C15 +#define GL_TEXTURE_DEPTH_TYPE_ARB 0x8C16 +#define GL_UNSIGNED_NORMALIZED_ARB 0x8C17 +#define GL_RGBA32F_ARB 0x8814 +#define GL_RGB32F_ARB 0x8815 +#define GL_ALPHA32F_ARB 0x8816 +#define GL_INTENSITY32F_ARB 0x8817 +#define GL_LUMINANCE32F_ARB 0x8818 +#define GL_LUMINANCE_ALPHA32F_ARB 0x8819 +#define GL_RGBA16F_ARB 0x881A +#define GL_RGB16F_ARB 0x881B +#define GL_ALPHA16F_ARB 0x881C +#define GL_INTENSITY16F_ARB 0x881D +#define GL_LUMINANCE16F_ARB 0x881E +#define GL_LUMINANCE_ALPHA16F_ARB 0x881F +#endif + +#ifndef GL_ARB_pixel_buffer_object +#define GL_PIXEL_PACK_BUFFER_ARB 0x88EB +#define GL_PIXEL_UNPACK_BUFFER_ARB 0x88EC +#define GL_PIXEL_PACK_BUFFER_BINDING_ARB 0x88ED +#define GL_PIXEL_UNPACK_BUFFER_BINDING_ARB 0x88EF +#endif + +#ifndef GL_ARB_depth_buffer_float +#define GL_DEPTH_COMPONENT32F 0x8CAC +#define GL_DEPTH32F_STENCIL8 0x8CAD +#define GL_FLOAT_32_UNSIGNED_INT_24_8_REV 0x8DAD +#endif + +#ifndef GL_ARB_draw_instanced +#endif + +#ifndef GL_ARB_framebuffer_object +#define GL_INVALID_FRAMEBUFFER_OPERATION 0x0506 +#define GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING 0x8210 +#define GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE 0x8211 +#define GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE 0x8212 +#define GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE 0x8213 +#define GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE 0x8214 +#define GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE 0x8215 +#define GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE 0x8216 +#define GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE 0x8217 +#define GL_FRAMEBUFFER_DEFAULT 0x8218 +#define GL_FRAMEBUFFER_UNDEFINED 0x8219 +#define GL_DEPTH_STENCIL_ATTACHMENT 0x821A +#define GL_MAX_RENDERBUFFER_SIZE 0x84E8 +#define GL_DEPTH_STENCIL 0x84F9 +#define GL_UNSIGNED_INT_24_8 0x84FA +#define GL_DEPTH24_STENCIL8 0x88F0 +#define GL_TEXTURE_STENCIL_SIZE 0x88F1 +#define GL_TEXTURE_RED_TYPE 0x8C10 +#define GL_TEXTURE_GREEN_TYPE 0x8C11 +#define GL_TEXTURE_BLUE_TYPE 0x8C12 +#define GL_TEXTURE_ALPHA_TYPE 0x8C13 +#define GL_TEXTURE_DEPTH_TYPE 0x8C16 +#define GL_UNSIGNED_NORMALIZED 0x8C17 +#define GL_FRAMEBUFFER_BINDING 0x8CA6 +#define GL_DRAW_FRAMEBUFFER_BINDING GL_FRAMEBUFFER_BINDING +#define GL_RENDERBUFFER_BINDING 0x8CA7 +#define GL_READ_FRAMEBUFFER 0x8CA8 +#define GL_DRAW_FRAMEBUFFER 0x8CA9 +#define GL_READ_FRAMEBUFFER_BINDING 0x8CAA +#define GL_RENDERBUFFER_SAMPLES 0x8CAB +#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE 0x8CD0 +#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME 0x8CD1 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL 0x8CD2 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE 0x8CD3 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER 0x8CD4 +#define GL_FRAMEBUFFER_COMPLETE 0x8CD5 +#define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT 0x8CD6 +#define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT 0x8CD7 +#define GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER 0x8CDB +#define GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER 0x8CDC +#define GL_FRAMEBUFFER_UNSUPPORTED 0x8CDD +#define GL_MAX_COLOR_ATTACHMENTS 0x8CDF +#define GL_COLOR_ATTACHMENT0 0x8CE0 +#define GL_COLOR_ATTACHMENT1 0x8CE1 +#define GL_COLOR_ATTACHMENT2 0x8CE2 +#define GL_COLOR_ATTACHMENT3 0x8CE3 +#define GL_COLOR_ATTACHMENT4 0x8CE4 +#define GL_COLOR_ATTACHMENT5 0x8CE5 +#define GL_COLOR_ATTACHMENT6 0x8CE6 +#define GL_COLOR_ATTACHMENT7 0x8CE7 +#define GL_COLOR_ATTACHMENT8 0x8CE8 +#define GL_COLOR_ATTACHMENT9 0x8CE9 +#define GL_COLOR_ATTACHMENT10 0x8CEA +#define GL_COLOR_ATTACHMENT11 0x8CEB +#define GL_COLOR_ATTACHMENT12 0x8CEC +#define GL_COLOR_ATTACHMENT13 0x8CED +#define GL_COLOR_ATTACHMENT14 0x8CEE +#define GL_COLOR_ATTACHMENT15 0x8CEF +#define GL_DEPTH_ATTACHMENT 0x8D00 +#define GL_STENCIL_ATTACHMENT 0x8D20 +#define GL_FRAMEBUFFER 0x8D40 +#define GL_RENDERBUFFER 0x8D41 +#define GL_RENDERBUFFER_WIDTH 0x8D42 +#define GL_RENDERBUFFER_HEIGHT 0x8D43 +#define GL_RENDERBUFFER_INTERNAL_FORMAT 0x8D44 +#define GL_STENCIL_INDEX1 0x8D46 +#define GL_STENCIL_INDEX4 0x8D47 +#define GL_STENCIL_INDEX8 0x8D48 +#define GL_STENCIL_INDEX16 0x8D49 +#define GL_RENDERBUFFER_RED_SIZE 0x8D50 +#define GL_RENDERBUFFER_GREEN_SIZE 0x8D51 +#define GL_RENDERBUFFER_BLUE_SIZE 0x8D52 +#define GL_RENDERBUFFER_ALPHA_SIZE 0x8D53 +#define GL_RENDERBUFFER_DEPTH_SIZE 0x8D54 +#define GL_RENDERBUFFER_STENCIL_SIZE 0x8D55 +#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE 0x8D56 +#define GL_MAX_SAMPLES 0x8D57 +#endif + +#ifndef GL_ARB_framebuffer_object_DEPRECATED +#define GL_INDEX 0x8222 +#define GL_TEXTURE_LUMINANCE_TYPE 0x8C14 +#define GL_TEXTURE_INTENSITY_TYPE 0x8C15 +#endif + +#ifndef GL_ARB_framebuffer_sRGB +#define GL_FRAMEBUFFER_SRGB 0x8DB9 +#endif + +#ifndef GL_ARB_geometry_shader4 +#define GL_LINES_ADJACENCY_ARB 0x000A +#define GL_LINE_STRIP_ADJACENCY_ARB 0x000B +#define GL_TRIANGLES_ADJACENCY_ARB 0x000C +#define GL_TRIANGLE_STRIP_ADJACENCY_ARB 0x000D +#define GL_PROGRAM_POINT_SIZE_ARB 0x8642 +#define GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_ARB 0x8C29 +#define GL_FRAMEBUFFER_ATTACHMENT_LAYERED_ARB 0x8DA7 +#define GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_ARB 0x8DA8 +#define GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_ARB 0x8DA9 +#define GL_GEOMETRY_SHADER_ARB 0x8DD9 +#define GL_GEOMETRY_VERTICES_OUT_ARB 0x8DDA +#define GL_GEOMETRY_INPUT_TYPE_ARB 0x8DDB +#define GL_GEOMETRY_OUTPUT_TYPE_ARB 0x8DDC +#define GL_MAX_GEOMETRY_VARYING_COMPONENTS_ARB 0x8DDD +#define GL_MAX_VERTEX_VARYING_COMPONENTS_ARB 0x8DDE +#define GL_MAX_GEOMETRY_UNIFORM_COMPONENTS_ARB 0x8DDF +#define GL_MAX_GEOMETRY_OUTPUT_VERTICES_ARB 0x8DE0 +#define GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS_ARB 0x8DE1 +/* reuse GL_MAX_VARYING_COMPONENTS */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */ +#endif + +#ifndef GL_ARB_half_float_vertex +#define GL_HALF_FLOAT 0x140B +#endif + +#ifndef GL_ARB_instanced_arrays +#define GL_VERTEX_ATTRIB_ARRAY_DIVISOR_ARB 0x88FE +#endif + +#ifndef GL_ARB_map_buffer_range +#define GL_MAP_READ_BIT 0x0001 +#define GL_MAP_WRITE_BIT 0x0002 +#define GL_MAP_INVALIDATE_RANGE_BIT 0x0004 +#define GL_MAP_INVALIDATE_BUFFER_BIT 0x0008 +#define GL_MAP_FLUSH_EXPLICIT_BIT 0x0010 +#define GL_MAP_UNSYNCHRONIZED_BIT 0x0020 +#endif + +#ifndef GL_ARB_texture_buffer_object +#define GL_TEXTURE_BUFFER_ARB 0x8C2A +#define GL_MAX_TEXTURE_BUFFER_SIZE_ARB 0x8C2B +#define GL_TEXTURE_BINDING_BUFFER_ARB 0x8C2C +#define GL_TEXTURE_BUFFER_DATA_STORE_BINDING_ARB 0x8C2D +#define GL_TEXTURE_BUFFER_FORMAT_ARB 0x8C2E +#endif + +#ifndef GL_ARB_texture_compression_rgtc +#define GL_COMPRESSED_RED_RGTC1 0x8DBB +#define GL_COMPRESSED_SIGNED_RED_RGTC1 0x8DBC +#define GL_COMPRESSED_RG_RGTC2 0x8DBD +#define GL_COMPRESSED_SIGNED_RG_RGTC2 0x8DBE +#endif + +#ifndef GL_ARB_texture_rg +#define GL_RG 0x8227 +#define GL_RG_INTEGER 0x8228 +#define GL_R8 0x8229 +#define GL_R16 0x822A +#define GL_RG8 0x822B +#define GL_RG16 0x822C +#define GL_R16F 0x822D +#define GL_R32F 0x822E +#define GL_RG16F 0x822F +#define GL_RG32F 0x8230 +#define GL_R8I 0x8231 +#define GL_R8UI 0x8232 +#define GL_R16I 0x8233 +#define GL_R16UI 0x8234 +#define GL_R32I 0x8235 +#define GL_R32UI 0x8236 +#define GL_RG8I 0x8237 +#define GL_RG8UI 0x8238 +#define GL_RG16I 0x8239 +#define GL_RG16UI 0x823A +#define GL_RG32I 0x823B +#define GL_RG32UI 0x823C +#endif + +#ifndef GL_ARB_vertex_array_object +#define GL_VERTEX_ARRAY_BINDING 0x85B5 +#endif + +#ifndef GL_ARB_uniform_buffer_object +#define GL_UNIFORM_BUFFER 0x8A11 +#define GL_UNIFORM_BUFFER_BINDING 0x8A28 +#define GL_UNIFORM_BUFFER_START 0x8A29 +#define GL_UNIFORM_BUFFER_SIZE 0x8A2A +#define GL_MAX_VERTEX_UNIFORM_BLOCKS 0x8A2B +#define GL_MAX_GEOMETRY_UNIFORM_BLOCKS 0x8A2C +#define GL_MAX_FRAGMENT_UNIFORM_BLOCKS 0x8A2D +#define GL_MAX_COMBINED_UNIFORM_BLOCKS 0x8A2E +#define GL_MAX_UNIFORM_BUFFER_BINDINGS 0x8A2F +#define GL_MAX_UNIFORM_BLOCK_SIZE 0x8A30 +#define GL_MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS 0x8A31 +#define GL_MAX_COMBINED_GEOMETRY_UNIFORM_COMPONENTS 0x8A32 +#define GL_MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS 0x8A33 +#define GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT 0x8A34 +#define GL_ACTIVE_UNIFORM_BLOCK_MAX_NAME_LENGTH 0x8A35 +#define GL_ACTIVE_UNIFORM_BLOCKS 0x8A36 +#define GL_UNIFORM_TYPE 0x8A37 +#define GL_UNIFORM_SIZE 0x8A38 +#define GL_UNIFORM_NAME_LENGTH 0x8A39 +#define GL_UNIFORM_BLOCK_INDEX 0x8A3A +#define GL_UNIFORM_OFFSET 0x8A3B +#define GL_UNIFORM_ARRAY_STRIDE 0x8A3C +#define GL_UNIFORM_MATRIX_STRIDE 0x8A3D +#define GL_UNIFORM_IS_ROW_MAJOR 0x8A3E +#define GL_UNIFORM_BLOCK_BINDING 0x8A3F +#define GL_UNIFORM_BLOCK_DATA_SIZE 0x8A40 +#define GL_UNIFORM_BLOCK_NAME_LENGTH 0x8A41 +#define GL_UNIFORM_BLOCK_ACTIVE_UNIFORMS 0x8A42 +#define GL_UNIFORM_BLOCK_ACTIVE_UNIFORM_INDICES 0x8A43 +#define GL_UNIFORM_BLOCK_REFERENCED_BY_VERTEX_SHADER 0x8A44 +#define GL_UNIFORM_BLOCK_REFERENCED_BY_GEOMETRY_SHADER 0x8A45 +#define GL_UNIFORM_BLOCK_REFERENCED_BY_FRAGMENT_SHADER 0x8A46 +#define GL_INVALID_INDEX 0xFFFFFFFFu +#endif + +#ifndef GL_ARB_compatibility +/* ARB_compatibility just defines tokens from core 3.0 */ +#endif + +#ifndef GL_ARB_copy_buffer +#define GL_COPY_READ_BUFFER 0x8F36 +#define GL_COPY_WRITE_BUFFER 0x8F37 +#endif + +#ifndef GL_ARB_shader_texture_lod +#endif + +#ifndef GL_ARB_depth_clamp +#define GL_DEPTH_CLAMP 0x864F +#endif + +#ifndef GL_ARB_draw_elements_base_vertex +#endif + +#ifndef GL_ARB_fragment_coord_conventions +#endif + +#ifndef GL_ARB_provoking_vertex +#define GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION 0x8E4C +#define GL_FIRST_VERTEX_CONVENTION 0x8E4D +#define GL_LAST_VERTEX_CONVENTION 0x8E4E +#define GL_PROVOKING_VERTEX 0x8E4F +#endif + +#ifndef GL_ARB_seamless_cube_map +#define GL_TEXTURE_CUBE_MAP_SEAMLESS 0x884F +#endif + +#ifndef GL_ARB_sync +#define GL_MAX_SERVER_WAIT_TIMEOUT 0x9111 +#define GL_OBJECT_TYPE 0x9112 +#define GL_SYNC_CONDITION 0x9113 +#define GL_SYNC_STATUS 0x9114 +#define GL_SYNC_FLAGS 0x9115 +#define GL_SYNC_FENCE 0x9116 +#define GL_SYNC_GPU_COMMANDS_COMPLETE 0x9117 +#define GL_UNSIGNALED 0x9118 +#define GL_SIGNALED 0x9119 +#define GL_ALREADY_SIGNALED 0x911A +#define GL_TIMEOUT_EXPIRED 0x911B +#define GL_CONDITION_SATISFIED 0x911C +#define GL_WAIT_FAILED 0x911D +#define GL_SYNC_FLUSH_COMMANDS_BIT 0x00000001 +#define GL_TIMEOUT_IGNORED 0xFFFFFFFFFFFFFFFFull +#endif + +#ifndef GL_ARB_texture_multisample +#define GL_SAMPLE_POSITION 0x8E50 +#define GL_SAMPLE_MASK 0x8E51 +#define GL_SAMPLE_MASK_VALUE 0x8E52 +#define GL_MAX_SAMPLE_MASK_WORDS 0x8E59 +#define GL_TEXTURE_2D_MULTISAMPLE 0x9100 +#define GL_PROXY_TEXTURE_2D_MULTISAMPLE 0x9101 +#define GL_TEXTURE_2D_MULTISAMPLE_ARRAY 0x9102 +#define GL_PROXY_TEXTURE_2D_MULTISAMPLE_ARRAY 0x9103 +#define GL_TEXTURE_BINDING_2D_MULTISAMPLE 0x9104 +#define GL_TEXTURE_BINDING_2D_MULTISAMPLE_ARRAY 0x9105 +#define GL_TEXTURE_SAMPLES 0x9106 +#define GL_TEXTURE_FIXED_SAMPLE_LOCATIONS 0x9107 +#define GL_SAMPLER_2D_MULTISAMPLE 0x9108 +#define GL_INT_SAMPLER_2D_MULTISAMPLE 0x9109 +#define GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE 0x910A +#define GL_SAMPLER_2D_MULTISAMPLE_ARRAY 0x910B +#define GL_INT_SAMPLER_2D_MULTISAMPLE_ARRAY 0x910C +#define GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE_ARRAY 0x910D +#define GL_MAX_COLOR_TEXTURE_SAMPLES 0x910E +#define GL_MAX_DEPTH_TEXTURE_SAMPLES 0x910F +#define GL_MAX_INTEGER_SAMPLES 0x9110 +#endif + +#ifndef GL_ARB_vertex_array_bgra +/* reuse GL_BGRA */ +#endif + +#ifndef GL_ARB_draw_buffers_blend +#endif + +#ifndef GL_ARB_sample_shading +#define GL_SAMPLE_SHADING 0x8C36 +#define GL_MIN_SAMPLE_SHADING_VALUE 0x8C37 +#endif + +#ifndef GL_ARB_texture_cube_map_array +#define GL_TEXTURE_CUBE_MAP_ARRAY 0x9009 +#define GL_TEXTURE_BINDING_CUBE_MAP_ARRAY 0x900A +#define GL_PROXY_TEXTURE_CUBE_MAP_ARRAY 0x900B +#define GL_SAMPLER_CUBE_MAP_ARRAY 0x900C +#define GL_SAMPLER_CUBE_MAP_ARRAY_SHADOW 0x900D +#define GL_INT_SAMPLER_CUBE_MAP_ARRAY 0x900E +#define GL_UNSIGNED_INT_SAMPLER_CUBE_MAP_ARRAY 0x900F +#endif + +#ifndef GL_ARB_texture_gather +#define GL_MIN_PROGRAM_TEXTURE_GATHER_OFFSET 0x8E5E +#define GL_MAX_PROGRAM_TEXTURE_GATHER_OFFSET 0x8E5F +#define GL_MAX_PROGRAM_TEXTURE_GATHER_COMPONENTS 0x8F9F +#endif + +#ifndef GL_ARB_texture_query_lod +#endif + +#ifndef GL_EXT_abgr +#define GL_ABGR_EXT 0x8000 +#endif + +#ifndef GL_EXT_blend_color +#define GL_CONSTANT_COLOR_EXT 0x8001 +#define GL_ONE_MINUS_CONSTANT_COLOR_EXT 0x8002 +#define GL_CONSTANT_ALPHA_EXT 0x8003 +#define GL_ONE_MINUS_CONSTANT_ALPHA_EXT 0x8004 +#define GL_BLEND_COLOR_EXT 0x8005 +#endif + +#ifndef GL_EXT_polygon_offset +#define GL_POLYGON_OFFSET_EXT 0x8037 +#define GL_POLYGON_OFFSET_FACTOR_EXT 0x8038 +#define GL_POLYGON_OFFSET_BIAS_EXT 0x8039 +#endif + +#ifndef GL_EXT_texture +#define GL_ALPHA4_EXT 0x803B +#define GL_ALPHA8_EXT 0x803C +#define GL_ALPHA12_EXT 0x803D +#define GL_ALPHA16_EXT 0x803E +#define GL_LUMINANCE4_EXT 0x803F +#define GL_LUMINANCE8_EXT 0x8040 +#define GL_LUMINANCE12_EXT 0x8041 +#define GL_LUMINANCE16_EXT 0x8042 +#define GL_LUMINANCE4_ALPHA4_EXT 0x8043 +#define GL_LUMINANCE6_ALPHA2_EXT 0x8044 +#define GL_LUMINANCE8_ALPHA8_EXT 0x8045 +#define GL_LUMINANCE12_ALPHA4_EXT 0x8046 +#define GL_LUMINANCE12_ALPHA12_EXT 0x8047 +#define GL_LUMINANCE16_ALPHA16_EXT 0x8048 +#define GL_INTENSITY_EXT 0x8049 +#define GL_INTENSITY4_EXT 0x804A +#define GL_INTENSITY8_EXT 0x804B +#define GL_INTENSITY12_EXT 0x804C +#define GL_INTENSITY16_EXT 0x804D +#define GL_RGB2_EXT 0x804E +#define GL_RGB4_EXT 0x804F +#define GL_RGB5_EXT 0x8050 +#define GL_RGB8_EXT 0x8051 +#define GL_RGB10_EXT 0x8052 +#define GL_RGB12_EXT 0x8053 +#define GL_RGB16_EXT 0x8054 +#define GL_RGBA2_EXT 0x8055 +#define GL_RGBA4_EXT 0x8056 +#define GL_RGB5_A1_EXT 0x8057 +#define GL_RGBA8_EXT 0x8058 +#define GL_RGB10_A2_EXT 0x8059 +#define GL_RGBA12_EXT 0x805A +#define GL_RGBA16_EXT 0x805B +#define GL_TEXTURE_RED_SIZE_EXT 0x805C +#define GL_TEXTURE_GREEN_SIZE_EXT 0x805D +#define GL_TEXTURE_BLUE_SIZE_EXT 0x805E +#define GL_TEXTURE_ALPHA_SIZE_EXT 0x805F +#define GL_TEXTURE_LUMINANCE_SIZE_EXT 0x8060 +#define GL_TEXTURE_INTENSITY_SIZE_EXT 0x8061 +#define GL_REPLACE_EXT 0x8062 +#define GL_PROXY_TEXTURE_1D_EXT 0x8063 +#define GL_PROXY_TEXTURE_2D_EXT 0x8064 +#define GL_TEXTURE_TOO_LARGE_EXT 0x8065 +#endif + +#ifndef GL_EXT_texture3D +#define GL_PACK_SKIP_IMAGES_EXT 0x806B +#define GL_PACK_IMAGE_HEIGHT_EXT 0x806C +#define GL_UNPACK_SKIP_IMAGES_EXT 0x806D +#define GL_UNPACK_IMAGE_HEIGHT_EXT 0x806E +#define GL_TEXTURE_3D_EXT 0x806F +#define GL_PROXY_TEXTURE_3D_EXT 0x8070 +#define GL_TEXTURE_DEPTH_EXT 0x8071 +#define GL_TEXTURE_WRAP_R_EXT 0x8072 +#define GL_MAX_3D_TEXTURE_SIZE_EXT 0x8073 +#endif + +#ifndef GL_SGIS_texture_filter4 +#define GL_FILTER4_SGIS 0x8146 +#define GL_TEXTURE_FILTER4_SIZE_SGIS 0x8147 +#endif + +#ifndef GL_EXT_subtexture +#endif + +#ifndef GL_EXT_copy_texture +#endif + +#ifndef GL_EXT_histogram +#define GL_HISTOGRAM_EXT 0x8024 +#define GL_PROXY_HISTOGRAM_EXT 0x8025 +#define GL_HISTOGRAM_WIDTH_EXT 0x8026 +#define GL_HISTOGRAM_FORMAT_EXT 0x8027 +#define GL_HISTOGRAM_RED_SIZE_EXT 0x8028 +#define GL_HISTOGRAM_GREEN_SIZE_EXT 0x8029 +#define GL_HISTOGRAM_BLUE_SIZE_EXT 0x802A +#define GL_HISTOGRAM_ALPHA_SIZE_EXT 0x802B +#define GL_HISTOGRAM_LUMINANCE_SIZE_EXT 0x802C +#define GL_HISTOGRAM_SINK_EXT 0x802D +#define GL_MINMAX_EXT 0x802E +#define GL_MINMAX_FORMAT_EXT 0x802F +#define GL_MINMAX_SINK_EXT 0x8030 +#define GL_TABLE_TOO_LARGE_EXT 0x8031 +#endif + +#ifndef GL_EXT_convolution +#define GL_CONVOLUTION_1D_EXT 0x8010 +#define GL_CONVOLUTION_2D_EXT 0x8011 +#define GL_SEPARABLE_2D_EXT 0x8012 +#define GL_CONVOLUTION_BORDER_MODE_EXT 0x8013 +#define GL_CONVOLUTION_FILTER_SCALE_EXT 0x8014 +#define GL_CONVOLUTION_FILTER_BIAS_EXT 0x8015 +#define GL_REDUCE_EXT 0x8016 +#define GL_CONVOLUTION_FORMAT_EXT 0x8017 +#define GL_CONVOLUTION_WIDTH_EXT 0x8018 +#define GL_CONVOLUTION_HEIGHT_EXT 0x8019 +#define GL_MAX_CONVOLUTION_WIDTH_EXT 0x801A +#define GL_MAX_CONVOLUTION_HEIGHT_EXT 0x801B +#define GL_POST_CONVOLUTION_RED_SCALE_EXT 0x801C +#define GL_POST_CONVOLUTION_GREEN_SCALE_EXT 0x801D +#define GL_POST_CONVOLUTION_BLUE_SCALE_EXT 0x801E +#define GL_POST_CONVOLUTION_ALPHA_SCALE_EXT 0x801F +#define GL_POST_CONVOLUTION_RED_BIAS_EXT 0x8020 +#define GL_POST_CONVOLUTION_GREEN_BIAS_EXT 0x8021 +#define GL_POST_CONVOLUTION_BLUE_BIAS_EXT 0x8022 +#define GL_POST_CONVOLUTION_ALPHA_BIAS_EXT 0x8023 +#endif + +#ifndef GL_SGI_color_matrix +#define GL_COLOR_MATRIX_SGI 0x80B1 +#define GL_COLOR_MATRIX_STACK_DEPTH_SGI 0x80B2 +#define GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI 0x80B3 +#define GL_POST_COLOR_MATRIX_RED_SCALE_SGI 0x80B4 +#define GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI 0x80B5 +#define GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI 0x80B6 +#define GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI 0x80B7 +#define GL_POST_COLOR_MATRIX_RED_BIAS_SGI 0x80B8 +#define GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI 0x80B9 +#define GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI 0x80BA +#define GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI 0x80BB +#endif + +#ifndef GL_SGI_color_table +#define GL_COLOR_TABLE_SGI 0x80D0 +#define GL_POST_CONVOLUTION_COLOR_TABLE_SGI 0x80D1 +#define GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI 0x80D2 +#define GL_PROXY_COLOR_TABLE_SGI 0x80D3 +#define GL_PROXY_POST_CONVOLUTION_COLOR_TABLE_SGI 0x80D4 +#define GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE_SGI 0x80D5 +#define GL_COLOR_TABLE_SCALE_SGI 0x80D6 +#define GL_COLOR_TABLE_BIAS_SGI 0x80D7 +#define GL_COLOR_TABLE_FORMAT_SGI 0x80D8 +#define GL_COLOR_TABLE_WIDTH_SGI 0x80D9 +#define GL_COLOR_TABLE_RED_SIZE_SGI 0x80DA +#define GL_COLOR_TABLE_GREEN_SIZE_SGI 0x80DB +#define GL_COLOR_TABLE_BLUE_SIZE_SGI 0x80DC +#define GL_COLOR_TABLE_ALPHA_SIZE_SGI 0x80DD +#define GL_COLOR_TABLE_LUMINANCE_SIZE_SGI 0x80DE +#define GL_COLOR_TABLE_INTENSITY_SIZE_SGI 0x80DF +#endif + +#ifndef GL_SGIS_pixel_texture +#define GL_PIXEL_TEXTURE_SGIS 0x8353 +#define GL_PIXEL_FRAGMENT_RGB_SOURCE_SGIS 0x8354 +#define GL_PIXEL_FRAGMENT_ALPHA_SOURCE_SGIS 0x8355 +#define GL_PIXEL_GROUP_COLOR_SGIS 0x8356 +#endif + +#ifndef GL_SGIX_pixel_texture +#define GL_PIXEL_TEX_GEN_SGIX 0x8139 +#define GL_PIXEL_TEX_GEN_MODE_SGIX 0x832B +#endif + +#ifndef GL_SGIS_texture4D +#define GL_PACK_SKIP_VOLUMES_SGIS 0x8130 +#define GL_PACK_IMAGE_DEPTH_SGIS 0x8131 +#define GL_UNPACK_SKIP_VOLUMES_SGIS 0x8132 +#define GL_UNPACK_IMAGE_DEPTH_SGIS 0x8133 +#define GL_TEXTURE_4D_SGIS 0x8134 +#define GL_PROXY_TEXTURE_4D_SGIS 0x8135 +#define GL_TEXTURE_4DSIZE_SGIS 0x8136 +#define GL_TEXTURE_WRAP_Q_SGIS 0x8137 +#define GL_MAX_4D_TEXTURE_SIZE_SGIS 0x8138 +#define GL_TEXTURE_4D_BINDING_SGIS 0x814F +#endif + +#ifndef GL_SGI_texture_color_table +#define GL_TEXTURE_COLOR_TABLE_SGI 0x80BC +#define GL_PROXY_TEXTURE_COLOR_TABLE_SGI 0x80BD +#endif + +#ifndef GL_EXT_cmyka +#define GL_CMYK_EXT 0x800C +#define GL_CMYKA_EXT 0x800D +#define GL_PACK_CMYK_HINT_EXT 0x800E +#define GL_UNPACK_CMYK_HINT_EXT 0x800F +#endif + +#ifndef GL_EXT_texture_object +#define GL_TEXTURE_PRIORITY_EXT 0x8066 +#define GL_TEXTURE_RESIDENT_EXT 0x8067 +#define GL_TEXTURE_1D_BINDING_EXT 0x8068 +#define GL_TEXTURE_2D_BINDING_EXT 0x8069 +#define GL_TEXTURE_3D_BINDING_EXT 0x806A +#endif + +#ifndef GL_SGIS_detail_texture +#define GL_DETAIL_TEXTURE_2D_SGIS 0x8095 +#define GL_DETAIL_TEXTURE_2D_BINDING_SGIS 0x8096 +#define GL_LINEAR_DETAIL_SGIS 0x8097 +#define GL_LINEAR_DETAIL_ALPHA_SGIS 0x8098 +#define GL_LINEAR_DETAIL_COLOR_SGIS 0x8099 +#define GL_DETAIL_TEXTURE_LEVEL_SGIS 0x809A +#define GL_DETAIL_TEXTURE_MODE_SGIS 0x809B +#define GL_DETAIL_TEXTURE_FUNC_POINTS_SGIS 0x809C +#endif + +#ifndef GL_SGIS_sharpen_texture +#define GL_LINEAR_SHARPEN_SGIS 0x80AD +#define GL_LINEAR_SHARPEN_ALPHA_SGIS 0x80AE +#define GL_LINEAR_SHARPEN_COLOR_SGIS 0x80AF +#define GL_SHARPEN_TEXTURE_FUNC_POINTS_SGIS 0x80B0 +#endif + +#ifndef GL_EXT_packed_pixels +#define GL_UNSIGNED_BYTE_3_3_2_EXT 0x8032 +#define GL_UNSIGNED_SHORT_4_4_4_4_EXT 0x8033 +#define GL_UNSIGNED_SHORT_5_5_5_1_EXT 0x8034 +#define GL_UNSIGNED_INT_8_8_8_8_EXT 0x8035 +#define GL_UNSIGNED_INT_10_10_10_2_EXT 0x8036 +#endif + +#ifndef GL_SGIS_texture_lod +#define GL_TEXTURE_MIN_LOD_SGIS 0x813A +#define GL_TEXTURE_MAX_LOD_SGIS 0x813B +#define GL_TEXTURE_BASE_LEVEL_SGIS 0x813C +#define GL_TEXTURE_MAX_LEVEL_SGIS 0x813D +#endif + +#ifndef GL_SGIS_multisample +#define GL_MULTISAMPLE_SGIS 0x809D +#define GL_SAMPLE_ALPHA_TO_MASK_SGIS 0x809E +#define GL_SAMPLE_ALPHA_TO_ONE_SGIS 0x809F +#define GL_SAMPLE_MASK_SGIS 0x80A0 +#define GL_1PASS_SGIS 0x80A1 +#define GL_2PASS_0_SGIS 0x80A2 +#define GL_2PASS_1_SGIS 0x80A3 +#define GL_4PASS_0_SGIS 0x80A4 +#define GL_4PASS_1_SGIS 0x80A5 +#define GL_4PASS_2_SGIS 0x80A6 +#define GL_4PASS_3_SGIS 0x80A7 +#define GL_SAMPLE_BUFFERS_SGIS 0x80A8 +#define GL_SAMPLES_SGIS 0x80A9 +#define GL_SAMPLE_MASK_VALUE_SGIS 0x80AA +#define GL_SAMPLE_MASK_INVERT_SGIS 0x80AB +#define GL_SAMPLE_PATTERN_SGIS 0x80AC +#endif + +#ifndef GL_EXT_rescale_normal +#define GL_RESCALE_NORMAL_EXT 0x803A +#endif + +#ifndef GL_EXT_vertex_array +#define GL_VERTEX_ARRAY_EXT 0x8074 +#define GL_NORMAL_ARRAY_EXT 0x8075 +#define GL_COLOR_ARRAY_EXT 0x8076 +#define GL_INDEX_ARRAY_EXT 0x8077 +#define GL_TEXTURE_COORD_ARRAY_EXT 0x8078 +#define GL_EDGE_FLAG_ARRAY_EXT 0x8079 +#define GL_VERTEX_ARRAY_SIZE_EXT 0x807A +#define GL_VERTEX_ARRAY_TYPE_EXT 0x807B +#define GL_VERTEX_ARRAY_STRIDE_EXT 0x807C +#define GL_VERTEX_ARRAY_COUNT_EXT 0x807D +#define GL_NORMAL_ARRAY_TYPE_EXT 0x807E +#define GL_NORMAL_ARRAY_STRIDE_EXT 0x807F +#define GL_NORMAL_ARRAY_COUNT_EXT 0x8080 +#define GL_COLOR_ARRAY_SIZE_EXT 0x8081 +#define GL_COLOR_ARRAY_TYPE_EXT 0x8082 +#define GL_COLOR_ARRAY_STRIDE_EXT 0x8083 +#define GL_COLOR_ARRAY_COUNT_EXT 0x8084 +#define GL_INDEX_ARRAY_TYPE_EXT 0x8085 +#define GL_INDEX_ARRAY_STRIDE_EXT 0x8086 +#define GL_INDEX_ARRAY_COUNT_EXT 0x8087 +#define GL_TEXTURE_COORD_ARRAY_SIZE_EXT 0x8088 +#define GL_TEXTURE_COORD_ARRAY_TYPE_EXT 0x8089 +#define GL_TEXTURE_COORD_ARRAY_STRIDE_EXT 0x808A +#define GL_TEXTURE_COORD_ARRAY_COUNT_EXT 0x808B +#define GL_EDGE_FLAG_ARRAY_STRIDE_EXT 0x808C +#define GL_EDGE_FLAG_ARRAY_COUNT_EXT 0x808D +#define GL_VERTEX_ARRAY_POINTER_EXT 0x808E +#define GL_NORMAL_ARRAY_POINTER_EXT 0x808F +#define GL_COLOR_ARRAY_POINTER_EXT 0x8090 +#define GL_INDEX_ARRAY_POINTER_EXT 0x8091 +#define GL_TEXTURE_COORD_ARRAY_POINTER_EXT 0x8092 +#define GL_EDGE_FLAG_ARRAY_POINTER_EXT 0x8093 +#endif + +#ifndef GL_EXT_misc_attribute +#endif + +#ifndef GL_SGIS_generate_mipmap +#define GL_GENERATE_MIPMAP_SGIS 0x8191 +#define GL_GENERATE_MIPMAP_HINT_SGIS 0x8192 +#endif + +#ifndef GL_SGIX_clipmap +#define GL_LINEAR_CLIPMAP_LINEAR_SGIX 0x8170 +#define GL_TEXTURE_CLIPMAP_CENTER_SGIX 0x8171 +#define GL_TEXTURE_CLIPMAP_FRAME_SGIX 0x8172 +#define GL_TEXTURE_CLIPMAP_OFFSET_SGIX 0x8173 +#define GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX 0x8174 +#define GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX 0x8175 +#define GL_TEXTURE_CLIPMAP_DEPTH_SGIX 0x8176 +#define GL_MAX_CLIPMAP_DEPTH_SGIX 0x8177 +#define GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX 0x8178 +#define GL_NEAREST_CLIPMAP_NEAREST_SGIX 0x844D +#define GL_NEAREST_CLIPMAP_LINEAR_SGIX 0x844E +#define GL_LINEAR_CLIPMAP_NEAREST_SGIX 0x844F +#endif + +#ifndef GL_SGIX_shadow +#define GL_TEXTURE_COMPARE_SGIX 0x819A +#define GL_TEXTURE_COMPARE_OPERATOR_SGIX 0x819B +#define GL_TEXTURE_LEQUAL_R_SGIX 0x819C +#define GL_TEXTURE_GEQUAL_R_SGIX 0x819D +#endif + +#ifndef GL_SGIS_texture_edge_clamp +#define GL_CLAMP_TO_EDGE_SGIS 0x812F +#endif + +#ifndef GL_SGIS_texture_border_clamp +#define GL_CLAMP_TO_BORDER_SGIS 0x812D +#endif + +#ifndef GL_EXT_blend_minmax +#define GL_FUNC_ADD_EXT 0x8006 +#define GL_MIN_EXT 0x8007 +#define GL_MAX_EXT 0x8008 +#define GL_BLEND_EQUATION_EXT 0x8009 +#endif + +#ifndef GL_EXT_blend_subtract +#define GL_FUNC_SUBTRACT_EXT 0x800A +#define GL_FUNC_REVERSE_SUBTRACT_EXT 0x800B +#endif + +#ifndef GL_EXT_blend_logic_op +#endif + +#ifndef GL_SGIX_interlace +#define GL_INTERLACE_SGIX 0x8094 +#endif + +#ifndef GL_SGIX_pixel_tiles +#define GL_PIXEL_TILE_BEST_ALIGNMENT_SGIX 0x813E +#define GL_PIXEL_TILE_CACHE_INCREMENT_SGIX 0x813F +#define GL_PIXEL_TILE_WIDTH_SGIX 0x8140 +#define GL_PIXEL_TILE_HEIGHT_SGIX 0x8141 +#define GL_PIXEL_TILE_GRID_WIDTH_SGIX 0x8142 +#define GL_PIXEL_TILE_GRID_HEIGHT_SGIX 0x8143 +#define GL_PIXEL_TILE_GRID_DEPTH_SGIX 0x8144 +#define GL_PIXEL_TILE_CACHE_SIZE_SGIX 0x8145 +#endif + +#ifndef GL_SGIS_texture_select +#define GL_DUAL_ALPHA4_SGIS 0x8110 +#define GL_DUAL_ALPHA8_SGIS 0x8111 +#define GL_DUAL_ALPHA12_SGIS 0x8112 +#define GL_DUAL_ALPHA16_SGIS 0x8113 +#define GL_DUAL_LUMINANCE4_SGIS 0x8114 +#define GL_DUAL_LUMINANCE8_SGIS 0x8115 +#define GL_DUAL_LUMINANCE12_SGIS 0x8116 +#define GL_DUAL_LUMINANCE16_SGIS 0x8117 +#define GL_DUAL_INTENSITY4_SGIS 0x8118 +#define GL_DUAL_INTENSITY8_SGIS 0x8119 +#define GL_DUAL_INTENSITY12_SGIS 0x811A +#define GL_DUAL_INTENSITY16_SGIS 0x811B +#define GL_DUAL_LUMINANCE_ALPHA4_SGIS 0x811C +#define GL_DUAL_LUMINANCE_ALPHA8_SGIS 0x811D +#define GL_QUAD_ALPHA4_SGIS 0x811E +#define GL_QUAD_ALPHA8_SGIS 0x811F +#define GL_QUAD_LUMINANCE4_SGIS 0x8120 +#define GL_QUAD_LUMINANCE8_SGIS 0x8121 +#define GL_QUAD_INTENSITY4_SGIS 0x8122 +#define GL_QUAD_INTENSITY8_SGIS 0x8123 +#define GL_DUAL_TEXTURE_SELECT_SGIS 0x8124 +#define GL_QUAD_TEXTURE_SELECT_SGIS 0x8125 +#endif + +#ifndef GL_SGIX_sprite +#define GL_SPRITE_SGIX 0x8148 +#define GL_SPRITE_MODE_SGIX 0x8149 +#define GL_SPRITE_AXIS_SGIX 0x814A +#define GL_SPRITE_TRANSLATION_SGIX 0x814B +#define GL_SPRITE_AXIAL_SGIX 0x814C +#define GL_SPRITE_OBJECT_ALIGNED_SGIX 0x814D +#define GL_SPRITE_EYE_ALIGNED_SGIX 0x814E +#endif + +#ifndef GL_SGIX_texture_multi_buffer +#define GL_TEXTURE_MULTI_BUFFER_HINT_SGIX 0x812E +#endif + +#ifndef GL_EXT_point_parameters +#define GL_POINT_SIZE_MIN_EXT 0x8126 +#define GL_POINT_SIZE_MAX_EXT 0x8127 +#define GL_POINT_FADE_THRESHOLD_SIZE_EXT 0x8128 +#define GL_DISTANCE_ATTENUATION_EXT 0x8129 +#endif + +#ifndef GL_SGIS_point_parameters +#define GL_POINT_SIZE_MIN_SGIS 0x8126 +#define GL_POINT_SIZE_MAX_SGIS 0x8127 +#define GL_POINT_FADE_THRESHOLD_SIZE_SGIS 0x8128 +#define GL_DISTANCE_ATTENUATION_SGIS 0x8129 +#endif + +#ifndef GL_SGIX_instruments +#define GL_INSTRUMENT_BUFFER_POINTER_SGIX 0x8180 +#define GL_INSTRUMENT_MEASUREMENTS_SGIX 0x8181 +#endif + +#ifndef GL_SGIX_texture_scale_bias +#define GL_POST_TEXTURE_FILTER_BIAS_SGIX 0x8179 +#define GL_POST_TEXTURE_FILTER_SCALE_SGIX 0x817A +#define GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX 0x817B +#define GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX 0x817C +#endif + +#ifndef GL_SGIX_framezoom +#define GL_FRAMEZOOM_SGIX 0x818B +#define GL_FRAMEZOOM_FACTOR_SGIX 0x818C +#define GL_MAX_FRAMEZOOM_FACTOR_SGIX 0x818D +#endif + +#ifndef GL_SGIX_tag_sample_buffer +#endif + +#ifndef GL_FfdMaskSGIX +#define GL_TEXTURE_DEFORMATION_BIT_SGIX 0x00000001 +#define GL_GEOMETRY_DEFORMATION_BIT_SGIX 0x00000002 +#endif + +#ifndef GL_SGIX_polynomial_ffd +#define GL_GEOMETRY_DEFORMATION_SGIX 0x8194 +#define GL_TEXTURE_DEFORMATION_SGIX 0x8195 +#define GL_DEFORMATIONS_MASK_SGIX 0x8196 +#define GL_MAX_DEFORMATION_ORDER_SGIX 0x8197 +#endif + +#ifndef GL_SGIX_reference_plane +#define GL_REFERENCE_PLANE_SGIX 0x817D +#define GL_REFERENCE_PLANE_EQUATION_SGIX 0x817E +#endif + +#ifndef GL_SGIX_flush_raster +#endif + +#ifndef GL_SGIX_depth_texture +#define GL_DEPTH_COMPONENT16_SGIX 0x81A5 +#define GL_DEPTH_COMPONENT24_SGIX 0x81A6 +#define GL_DEPTH_COMPONENT32_SGIX 0x81A7 +#endif + +#ifndef GL_SGIS_fog_function +#define GL_FOG_FUNC_SGIS 0x812A +#define GL_FOG_FUNC_POINTS_SGIS 0x812B +#define GL_MAX_FOG_FUNC_POINTS_SGIS 0x812C +#endif + +#ifndef GL_SGIX_fog_offset +#define GL_FOG_OFFSET_SGIX 0x8198 +#define GL_FOG_OFFSET_VALUE_SGIX 0x8199 +#endif + +#ifndef GL_HP_image_transform +#define GL_IMAGE_SCALE_X_HP 0x8155 +#define GL_IMAGE_SCALE_Y_HP 0x8156 +#define GL_IMAGE_TRANSLATE_X_HP 0x8157 +#define GL_IMAGE_TRANSLATE_Y_HP 0x8158 +#define GL_IMAGE_ROTATE_ANGLE_HP 0x8159 +#define GL_IMAGE_ROTATE_ORIGIN_X_HP 0x815A +#define GL_IMAGE_ROTATE_ORIGIN_Y_HP 0x815B +#define GL_IMAGE_MAG_FILTER_HP 0x815C +#define GL_IMAGE_MIN_FILTER_HP 0x815D +#define GL_IMAGE_CUBIC_WEIGHT_HP 0x815E +#define GL_CUBIC_HP 0x815F +#define GL_AVERAGE_HP 0x8160 +#define GL_IMAGE_TRANSFORM_2D_HP 0x8161 +#define GL_POST_IMAGE_TRANSFORM_COLOR_TABLE_HP 0x8162 +#define GL_PROXY_POST_IMAGE_TRANSFORM_COLOR_TABLE_HP 0x8163 +#endif + +#ifndef GL_HP_convolution_border_modes +#define GL_IGNORE_BORDER_HP 0x8150 +#define GL_CONSTANT_BORDER_HP 0x8151 +#define GL_REPLICATE_BORDER_HP 0x8153 +#define GL_CONVOLUTION_BORDER_COLOR_HP 0x8154 +#endif + +#ifndef GL_INGR_palette_buffer +#endif + +#ifndef GL_SGIX_texture_add_env +#define GL_TEXTURE_ENV_BIAS_SGIX 0x80BE +#endif + +#ifndef GL_EXT_color_subtable +#endif + +#ifndef GL_PGI_vertex_hints +#define GL_VERTEX_DATA_HINT_PGI 0x1A22A +#define GL_VERTEX_CONSISTENT_HINT_PGI 0x1A22B +#define GL_MATERIAL_SIDE_HINT_PGI 0x1A22C +#define GL_MAX_VERTEX_HINT_PGI 0x1A22D +#define GL_COLOR3_BIT_PGI 0x00010000 +#define GL_COLOR4_BIT_PGI 0x00020000 +#define GL_EDGEFLAG_BIT_PGI 0x00040000 +#define GL_INDEX_BIT_PGI 0x00080000 +#define GL_MAT_AMBIENT_BIT_PGI 0x00100000 +#define GL_MAT_AMBIENT_AND_DIFFUSE_BIT_PGI 0x00200000 +#define GL_MAT_DIFFUSE_BIT_PGI 0x00400000 +#define GL_MAT_EMISSION_BIT_PGI 0x00800000 +#define GL_MAT_COLOR_INDEXES_BIT_PGI 0x01000000 +#define GL_MAT_SHININESS_BIT_PGI 0x02000000 +#define GL_MAT_SPECULAR_BIT_PGI 0x04000000 +#define GL_NORMAL_BIT_PGI 0x08000000 +#define GL_TEXCOORD1_BIT_PGI 0x10000000 +#define GL_TEXCOORD2_BIT_PGI 0x20000000 +#define GL_TEXCOORD3_BIT_PGI 0x40000000 +#define GL_TEXCOORD4_BIT_PGI 0x80000000 +#define GL_VERTEX23_BIT_PGI 0x00000004 +#define GL_VERTEX4_BIT_PGI 0x00000008 +#endif + +#ifndef GL_PGI_misc_hints +#define GL_PREFER_DOUBLEBUFFER_HINT_PGI 0x1A1F8 +#define GL_CONSERVE_MEMORY_HINT_PGI 0x1A1FD +#define GL_RECLAIM_MEMORY_HINT_PGI 0x1A1FE +#define GL_NATIVE_GRAPHICS_HANDLE_PGI 0x1A202 +#define GL_NATIVE_GRAPHICS_BEGIN_HINT_PGI 0x1A203 +#define GL_NATIVE_GRAPHICS_END_HINT_PGI 0x1A204 +#define GL_ALWAYS_FAST_HINT_PGI 0x1A20C +#define GL_ALWAYS_SOFT_HINT_PGI 0x1A20D +#define GL_ALLOW_DRAW_OBJ_HINT_PGI 0x1A20E +#define GL_ALLOW_DRAW_WIN_HINT_PGI 0x1A20F +#define GL_ALLOW_DRAW_FRG_HINT_PGI 0x1A210 +#define GL_ALLOW_DRAW_MEM_HINT_PGI 0x1A211 +#define GL_STRICT_DEPTHFUNC_HINT_PGI 0x1A216 +#define GL_STRICT_LIGHTING_HINT_PGI 0x1A217 +#define GL_STRICT_SCISSOR_HINT_PGI 0x1A218 +#define GL_FULL_STIPPLE_HINT_PGI 0x1A219 +#define GL_CLIP_NEAR_HINT_PGI 0x1A220 +#define GL_CLIP_FAR_HINT_PGI 0x1A221 +#define GL_WIDE_LINE_HINT_PGI 0x1A222 +#define GL_BACK_NORMALS_HINT_PGI 0x1A223 +#endif + +#ifndef GL_EXT_paletted_texture +#define GL_COLOR_INDEX1_EXT 0x80E2 +#define GL_COLOR_INDEX2_EXT 0x80E3 +#define GL_COLOR_INDEX4_EXT 0x80E4 +#define GL_COLOR_INDEX8_EXT 0x80E5 +#define GL_COLOR_INDEX12_EXT 0x80E6 +#define GL_COLOR_INDEX16_EXT 0x80E7 +#define GL_TEXTURE_INDEX_SIZE_EXT 0x80ED +#endif + +#ifndef GL_EXT_clip_volume_hint +#define GL_CLIP_VOLUME_CLIPPING_HINT_EXT 0x80F0 +#endif + +#ifndef GL_SGIX_list_priority +#define GL_LIST_PRIORITY_SGIX 0x8182 +#endif + +#ifndef GL_SGIX_ir_instrument1 +#define GL_IR_INSTRUMENT1_SGIX 0x817F +#endif + +#ifndef GL_SGIX_calligraphic_fragment +#define GL_CALLIGRAPHIC_FRAGMENT_SGIX 0x8183 +#endif + +#ifndef GL_SGIX_texture_lod_bias +#define GL_TEXTURE_LOD_BIAS_S_SGIX 0x818E +#define GL_TEXTURE_LOD_BIAS_T_SGIX 0x818F +#define GL_TEXTURE_LOD_BIAS_R_SGIX 0x8190 +#endif + +#ifndef GL_SGIX_shadow_ambient +#define GL_SHADOW_AMBIENT_SGIX 0x80BF +#endif + +#ifndef GL_EXT_index_texture +#endif + +#ifndef GL_EXT_index_material +#define GL_INDEX_MATERIAL_EXT 0x81B8 +#define GL_INDEX_MATERIAL_PARAMETER_EXT 0x81B9 +#define GL_INDEX_MATERIAL_FACE_EXT 0x81BA +#endif + +#ifndef GL_EXT_index_func +#define GL_INDEX_TEST_EXT 0x81B5 +#define GL_INDEX_TEST_FUNC_EXT 0x81B6 +#define GL_INDEX_TEST_REF_EXT 0x81B7 +#endif + +#ifndef GL_EXT_index_array_formats +#define GL_IUI_V2F_EXT 0x81AD +#define GL_IUI_V3F_EXT 0x81AE +#define GL_IUI_N3F_V2F_EXT 0x81AF +#define GL_IUI_N3F_V3F_EXT 0x81B0 +#define GL_T2F_IUI_V2F_EXT 0x81B1 +#define GL_T2F_IUI_V3F_EXT 0x81B2 +#define GL_T2F_IUI_N3F_V2F_EXT 0x81B3 +#define GL_T2F_IUI_N3F_V3F_EXT 0x81B4 +#endif + +#ifndef GL_EXT_compiled_vertex_array +#define GL_ARRAY_ELEMENT_LOCK_FIRST_EXT 0x81A8 +#define GL_ARRAY_ELEMENT_LOCK_COUNT_EXT 0x81A9 +#endif + +#ifndef GL_EXT_cull_vertex +#define GL_CULL_VERTEX_EXT 0x81AA +#define GL_CULL_VERTEX_EYE_POSITION_EXT 0x81AB +#define GL_CULL_VERTEX_OBJECT_POSITION_EXT 0x81AC +#endif + +#ifndef GL_SGIX_ycrcb +#define GL_YCRCB_422_SGIX 0x81BB +#define GL_YCRCB_444_SGIX 0x81BC +#endif + +#ifndef GL_SGIX_fragment_lighting +#define GL_FRAGMENT_LIGHTING_SGIX 0x8400 +#define GL_FRAGMENT_COLOR_MATERIAL_SGIX 0x8401 +#define GL_FRAGMENT_COLOR_MATERIAL_FACE_SGIX 0x8402 +#define GL_FRAGMENT_COLOR_MATERIAL_PARAMETER_SGIX 0x8403 +#define GL_MAX_FRAGMENT_LIGHTS_SGIX 0x8404 +#define GL_MAX_ACTIVE_LIGHTS_SGIX 0x8405 +#define GL_CURRENT_RASTER_NORMAL_SGIX 0x8406 +#define GL_LIGHT_ENV_MODE_SGIX 0x8407 +#define GL_FRAGMENT_LIGHT_MODEL_LOCAL_VIEWER_SGIX 0x8408 +#define GL_FRAGMENT_LIGHT_MODEL_TWO_SIDE_SGIX 0x8409 +#define GL_FRAGMENT_LIGHT_MODEL_AMBIENT_SGIX 0x840A +#define GL_FRAGMENT_LIGHT_MODEL_NORMAL_INTERPOLATION_SGIX 0x840B +#define GL_FRAGMENT_LIGHT0_SGIX 0x840C +#define GL_FRAGMENT_LIGHT1_SGIX 0x840D +#define GL_FRAGMENT_LIGHT2_SGIX 0x840E +#define GL_FRAGMENT_LIGHT3_SGIX 0x840F +#define GL_FRAGMENT_LIGHT4_SGIX 0x8410 +#define GL_FRAGMENT_LIGHT5_SGIX 0x8411 +#define GL_FRAGMENT_LIGHT6_SGIX 0x8412 +#define GL_FRAGMENT_LIGHT7_SGIX 0x8413 +#endif + +#ifndef GL_IBM_rasterpos_clip +#define GL_RASTER_POSITION_UNCLIPPED_IBM 0x19262 +#endif + +#ifndef GL_HP_texture_lighting +#define GL_TEXTURE_LIGHTING_MODE_HP 0x8167 +#define GL_TEXTURE_POST_SPECULAR_HP 0x8168 +#define GL_TEXTURE_PRE_SPECULAR_HP 0x8169 +#endif + +#ifndef GL_EXT_draw_range_elements +#define GL_MAX_ELEMENTS_VERTICES_EXT 0x80E8 +#define GL_MAX_ELEMENTS_INDICES_EXT 0x80E9 +#endif + +#ifndef GL_WIN_phong_shading +#define GL_PHONG_WIN 0x80EA +#define GL_PHONG_HINT_WIN 0x80EB +#endif + +#ifndef GL_WIN_specular_fog +#define GL_FOG_SPECULAR_TEXTURE_WIN 0x80EC +#endif + +#ifndef GL_EXT_light_texture +#define GL_FRAGMENT_MATERIAL_EXT 0x8349 +#define GL_FRAGMENT_NORMAL_EXT 0x834A +#define GL_FRAGMENT_COLOR_EXT 0x834C +#define GL_ATTENUATION_EXT 0x834D +#define GL_SHADOW_ATTENUATION_EXT 0x834E +#define GL_TEXTURE_APPLICATION_MODE_EXT 0x834F +#define GL_TEXTURE_LIGHT_EXT 0x8350 +#define GL_TEXTURE_MATERIAL_FACE_EXT 0x8351 +#define GL_TEXTURE_MATERIAL_PARAMETER_EXT 0x8352 +/* reuse GL_FRAGMENT_DEPTH_EXT */ +#endif + +#ifndef GL_SGIX_blend_alpha_minmax +#define GL_ALPHA_MIN_SGIX 0x8320 +#define GL_ALPHA_MAX_SGIX 0x8321 +#endif + +#ifndef GL_SGIX_impact_pixel_texture +#define GL_PIXEL_TEX_GEN_Q_CEILING_SGIX 0x8184 +#define GL_PIXEL_TEX_GEN_Q_ROUND_SGIX 0x8185 +#define GL_PIXEL_TEX_GEN_Q_FLOOR_SGIX 0x8186 +#define GL_PIXEL_TEX_GEN_ALPHA_REPLACE_SGIX 0x8187 +#define GL_PIXEL_TEX_GEN_ALPHA_NO_REPLACE_SGIX 0x8188 +#define GL_PIXEL_TEX_GEN_ALPHA_LS_SGIX 0x8189 +#define GL_PIXEL_TEX_GEN_ALPHA_MS_SGIX 0x818A +#endif + +#ifndef GL_EXT_bgra +#define GL_BGR_EXT 0x80E0 +#define GL_BGRA_EXT 0x80E1 +#endif + +#ifndef GL_SGIX_async +#define GL_ASYNC_MARKER_SGIX 0x8329 +#endif + +#ifndef GL_SGIX_async_pixel +#define GL_ASYNC_TEX_IMAGE_SGIX 0x835C +#define GL_ASYNC_DRAW_PIXELS_SGIX 0x835D +#define GL_ASYNC_READ_PIXELS_SGIX 0x835E +#define GL_MAX_ASYNC_TEX_IMAGE_SGIX 0x835F +#define GL_MAX_ASYNC_DRAW_PIXELS_SGIX 0x8360 +#define GL_MAX_ASYNC_READ_PIXELS_SGIX 0x8361 +#endif + +#ifndef GL_SGIX_async_histogram +#define GL_ASYNC_HISTOGRAM_SGIX 0x832C +#define GL_MAX_ASYNC_HISTOGRAM_SGIX 0x832D +#endif + +#ifndef GL_INTEL_texture_scissor +#endif + +#ifndef GL_INTEL_parallel_arrays +#define GL_PARALLEL_ARRAYS_INTEL 0x83F4 +#define GL_VERTEX_ARRAY_PARALLEL_POINTERS_INTEL 0x83F5 +#define GL_NORMAL_ARRAY_PARALLEL_POINTERS_INTEL 0x83F6 +#define GL_COLOR_ARRAY_PARALLEL_POINTERS_INTEL 0x83F7 +#define GL_TEXTURE_COORD_ARRAY_PARALLEL_POINTERS_INTEL 0x83F8 +#endif + +#ifndef GL_HP_occlusion_test +#define GL_OCCLUSION_TEST_HP 0x8165 +#define GL_OCCLUSION_TEST_RESULT_HP 0x8166 +#endif + +#ifndef GL_EXT_pixel_transform +#define GL_PIXEL_TRANSFORM_2D_EXT 0x8330 +#define GL_PIXEL_MAG_FILTER_EXT 0x8331 +#define GL_PIXEL_MIN_FILTER_EXT 0x8332 +#define GL_PIXEL_CUBIC_WEIGHT_EXT 0x8333 +#define GL_CUBIC_EXT 0x8334 +#define GL_AVERAGE_EXT 0x8335 +#define GL_PIXEL_TRANSFORM_2D_STACK_DEPTH_EXT 0x8336 +#define GL_MAX_PIXEL_TRANSFORM_2D_STACK_DEPTH_EXT 0x8337 +#define GL_PIXEL_TRANSFORM_2D_MATRIX_EXT 0x8338 +#endif + +#ifndef GL_EXT_pixel_transform_color_table +#endif + +#ifndef GL_EXT_shared_texture_palette +#define GL_SHARED_TEXTURE_PALETTE_EXT 0x81FB +#endif + +#ifndef GL_EXT_separate_specular_color +#define GL_LIGHT_MODEL_COLOR_CONTROL_EXT 0x81F8 +#define GL_SINGLE_COLOR_EXT 0x81F9 +#define GL_SEPARATE_SPECULAR_COLOR_EXT 0x81FA +#endif + +#ifndef GL_EXT_secondary_color +#define GL_COLOR_SUM_EXT 0x8458 +#define GL_CURRENT_SECONDARY_COLOR_EXT 0x8459 +#define GL_SECONDARY_COLOR_ARRAY_SIZE_EXT 0x845A +#define GL_SECONDARY_COLOR_ARRAY_TYPE_EXT 0x845B +#define GL_SECONDARY_COLOR_ARRAY_STRIDE_EXT 0x845C +#define GL_SECONDARY_COLOR_ARRAY_POINTER_EXT 0x845D +#define GL_SECONDARY_COLOR_ARRAY_EXT 0x845E +#endif + +#ifndef GL_EXT_texture_perturb_normal +#define GL_PERTURB_EXT 0x85AE +#define GL_TEXTURE_NORMAL_EXT 0x85AF +#endif + +#ifndef GL_EXT_multi_draw_arrays +#endif + +#ifndef GL_EXT_fog_coord +#define GL_FOG_COORDINATE_SOURCE_EXT 0x8450 +#define GL_FOG_COORDINATE_EXT 0x8451 +#define GL_FRAGMENT_DEPTH_EXT 0x8452 +#define GL_CURRENT_FOG_COORDINATE_EXT 0x8453 +#define GL_FOG_COORDINATE_ARRAY_TYPE_EXT 0x8454 +#define GL_FOG_COORDINATE_ARRAY_STRIDE_EXT 0x8455 +#define GL_FOG_COORDINATE_ARRAY_POINTER_EXT 0x8456 +#define GL_FOG_COORDINATE_ARRAY_EXT 0x8457 +#endif + +#ifndef GL_REND_screen_coordinates +#define GL_SCREEN_COORDINATES_REND 0x8490 +#define GL_INVERTED_SCREEN_W_REND 0x8491 +#endif + +#ifndef GL_EXT_coordinate_frame +#define GL_TANGENT_ARRAY_EXT 0x8439 +#define GL_BINORMAL_ARRAY_EXT 0x843A +#define GL_CURRENT_TANGENT_EXT 0x843B +#define GL_CURRENT_BINORMAL_EXT 0x843C +#define GL_TANGENT_ARRAY_TYPE_EXT 0x843E +#define GL_TANGENT_ARRAY_STRIDE_EXT 0x843F +#define GL_BINORMAL_ARRAY_TYPE_EXT 0x8440 +#define GL_BINORMAL_ARRAY_STRIDE_EXT 0x8441 +#define GL_TANGENT_ARRAY_POINTER_EXT 0x8442 +#define GL_BINORMAL_ARRAY_POINTER_EXT 0x8443 +#define GL_MAP1_TANGENT_EXT 0x8444 +#define GL_MAP2_TANGENT_EXT 0x8445 +#define GL_MAP1_BINORMAL_EXT 0x8446 +#define GL_MAP2_BINORMAL_EXT 0x8447 +#endif + +#ifndef GL_EXT_texture_env_combine +#define GL_COMBINE_EXT 0x8570 +#define GL_COMBINE_RGB_EXT 0x8571 +#define GL_COMBINE_ALPHA_EXT 0x8572 +#define GL_RGB_SCALE_EXT 0x8573 +#define GL_ADD_SIGNED_EXT 0x8574 +#define GL_INTERPOLATE_EXT 0x8575 +#define GL_CONSTANT_EXT 0x8576 +#define GL_PRIMARY_COLOR_EXT 0x8577 +#define GL_PREVIOUS_EXT 0x8578 +#define GL_SOURCE0_RGB_EXT 0x8580 +#define GL_SOURCE1_RGB_EXT 0x8581 +#define GL_SOURCE2_RGB_EXT 0x8582 +#define GL_SOURCE0_ALPHA_EXT 0x8588 +#define GL_SOURCE1_ALPHA_EXT 0x8589 +#define GL_SOURCE2_ALPHA_EXT 0x858A +#define GL_OPERAND0_RGB_EXT 0x8590 +#define GL_OPERAND1_RGB_EXT 0x8591 +#define GL_OPERAND2_RGB_EXT 0x8592 +#define GL_OPERAND0_ALPHA_EXT 0x8598 +#define GL_OPERAND1_ALPHA_EXT 0x8599 +#define GL_OPERAND2_ALPHA_EXT 0x859A +#endif + +#ifndef GL_APPLE_specular_vector +#define GL_LIGHT_MODEL_SPECULAR_VECTOR_APPLE 0x85B0 +#endif + +#ifndef GL_APPLE_transform_hint +#define GL_TRANSFORM_HINT_APPLE 0x85B1 +#endif + +#ifndef GL_SGIX_fog_scale +#define GL_FOG_SCALE_SGIX 0x81FC +#define GL_FOG_SCALE_VALUE_SGIX 0x81FD +#endif + +#ifndef GL_SUNX_constant_data +#define GL_UNPACK_CONSTANT_DATA_SUNX 0x81D5 +#define GL_TEXTURE_CONSTANT_DATA_SUNX 0x81D6 +#endif + +#ifndef GL_SUN_global_alpha +#define GL_GLOBAL_ALPHA_SUN 0x81D9 +#define GL_GLOBAL_ALPHA_FACTOR_SUN 0x81DA +#endif + +#ifndef GL_SUN_triangle_list +#define GL_RESTART_SUN 0x0001 +#define GL_REPLACE_MIDDLE_SUN 0x0002 +#define GL_REPLACE_OLDEST_SUN 0x0003 +#define GL_TRIANGLE_LIST_SUN 0x81D7 +#define GL_REPLACEMENT_CODE_SUN 0x81D8 +#define GL_REPLACEMENT_CODE_ARRAY_SUN 0x85C0 +#define GL_REPLACEMENT_CODE_ARRAY_TYPE_SUN 0x85C1 +#define GL_REPLACEMENT_CODE_ARRAY_STRIDE_SUN 0x85C2 +#define GL_REPLACEMENT_CODE_ARRAY_POINTER_SUN 0x85C3 +#define GL_R1UI_V3F_SUN 0x85C4 +#define GL_R1UI_C4UB_V3F_SUN 0x85C5 +#define GL_R1UI_C3F_V3F_SUN 0x85C6 +#define GL_R1UI_N3F_V3F_SUN 0x85C7 +#define GL_R1UI_C4F_N3F_V3F_SUN 0x85C8 +#define GL_R1UI_T2F_V3F_SUN 0x85C9 +#define GL_R1UI_T2F_N3F_V3F_SUN 0x85CA +#define GL_R1UI_T2F_C4F_N3F_V3F_SUN 0x85CB +#endif + +#ifndef GL_SUN_vertex +#endif + +#ifndef GL_EXT_blend_func_separate +#define GL_BLEND_DST_RGB_EXT 0x80C8 +#define GL_BLEND_SRC_RGB_EXT 0x80C9 +#define GL_BLEND_DST_ALPHA_EXT 0x80CA +#define GL_BLEND_SRC_ALPHA_EXT 0x80CB +#endif + +#ifndef GL_INGR_color_clamp +#define GL_RED_MIN_CLAMP_INGR 0x8560 +#define GL_GREEN_MIN_CLAMP_INGR 0x8561 +#define GL_BLUE_MIN_CLAMP_INGR 0x8562 +#define GL_ALPHA_MIN_CLAMP_INGR 0x8563 +#define GL_RED_MAX_CLAMP_INGR 0x8564 +#define GL_GREEN_MAX_CLAMP_INGR 0x8565 +#define GL_BLUE_MAX_CLAMP_INGR 0x8566 +#define GL_ALPHA_MAX_CLAMP_INGR 0x8567 +#endif + +#ifndef GL_INGR_interlace_read +#define GL_INTERLACE_READ_INGR 0x8568 +#endif + +#ifndef GL_EXT_stencil_wrap +#define GL_INCR_WRAP_EXT 0x8507 +#define GL_DECR_WRAP_EXT 0x8508 +#endif + +#ifndef GL_EXT_422_pixels +#define GL_422_EXT 0x80CC +#define GL_422_REV_EXT 0x80CD +#define GL_422_AVERAGE_EXT 0x80CE +#define GL_422_REV_AVERAGE_EXT 0x80CF +#endif + +#ifndef GL_NV_texgen_reflection +#define GL_NORMAL_MAP_NV 0x8511 +#define GL_REFLECTION_MAP_NV 0x8512 +#endif + +#ifndef GL_EXT_texture_cube_map +#define GL_NORMAL_MAP_EXT 0x8511 +#define GL_REFLECTION_MAP_EXT 0x8512 +#define GL_TEXTURE_CUBE_MAP_EXT 0x8513 +#define GL_TEXTURE_BINDING_CUBE_MAP_EXT 0x8514 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_X_EXT 0x8515 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_X_EXT 0x8516 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_Y_EXT 0x8517 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT 0x8518 +#define GL_TEXTURE_CUBE_MAP_POSITIVE_Z_EXT 0x8519 +#define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT 0x851A +#define GL_PROXY_TEXTURE_CUBE_MAP_EXT 0x851B +#define GL_MAX_CUBE_MAP_TEXTURE_SIZE_EXT 0x851C +#endif + +#ifndef GL_SUN_convolution_border_modes +#define GL_WRAP_BORDER_SUN 0x81D4 +#endif + +#ifndef GL_EXT_texture_env_add +#endif + +#ifndef GL_EXT_texture_lod_bias +#define GL_MAX_TEXTURE_LOD_BIAS_EXT 0x84FD +#define GL_TEXTURE_FILTER_CONTROL_EXT 0x8500 +#define GL_TEXTURE_LOD_BIAS_EXT 0x8501 +#endif + +#ifndef GL_EXT_texture_filter_anisotropic +#define GL_TEXTURE_MAX_ANISOTROPY_EXT 0x84FE +#define GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT 0x84FF +#endif + +#ifndef GL_EXT_vertex_weighting +#define GL_MODELVIEW0_STACK_DEPTH_EXT GL_MODELVIEW_STACK_DEPTH +#define GL_MODELVIEW1_STACK_DEPTH_EXT 0x8502 +#define GL_MODELVIEW0_MATRIX_EXT GL_MODELVIEW_MATRIX +#define GL_MODELVIEW1_MATRIX_EXT 0x8506 +#define GL_VERTEX_WEIGHTING_EXT 0x8509 +#define GL_MODELVIEW0_EXT GL_MODELVIEW +#define GL_MODELVIEW1_EXT 0x850A +#define GL_CURRENT_VERTEX_WEIGHT_EXT 0x850B +#define GL_VERTEX_WEIGHT_ARRAY_EXT 0x850C +#define GL_VERTEX_WEIGHT_ARRAY_SIZE_EXT 0x850D +#define GL_VERTEX_WEIGHT_ARRAY_TYPE_EXT 0x850E +#define GL_VERTEX_WEIGHT_ARRAY_STRIDE_EXT 0x850F +#define GL_VERTEX_WEIGHT_ARRAY_POINTER_EXT 0x8510 +#endif + +#ifndef GL_NV_light_max_exponent +#define GL_MAX_SHININESS_NV 0x8504 +#define GL_MAX_SPOT_EXPONENT_NV 0x8505 +#endif + +#ifndef GL_NV_vertex_array_range +#define GL_VERTEX_ARRAY_RANGE_NV 0x851D +#define GL_VERTEX_ARRAY_RANGE_LENGTH_NV 0x851E +#define GL_VERTEX_ARRAY_RANGE_VALID_NV 0x851F +#define GL_MAX_VERTEX_ARRAY_RANGE_ELEMENT_NV 0x8520 +#define GL_VERTEX_ARRAY_RANGE_POINTER_NV 0x8521 +#endif + +#ifndef GL_NV_register_combiners +#define GL_REGISTER_COMBINERS_NV 0x8522 +#define GL_VARIABLE_A_NV 0x8523 +#define GL_VARIABLE_B_NV 0x8524 +#define GL_VARIABLE_C_NV 0x8525 +#define GL_VARIABLE_D_NV 0x8526 +#define GL_VARIABLE_E_NV 0x8527 +#define GL_VARIABLE_F_NV 0x8528 +#define GL_VARIABLE_G_NV 0x8529 +#define GL_CONSTANT_COLOR0_NV 0x852A +#define GL_CONSTANT_COLOR1_NV 0x852B +#define GL_PRIMARY_COLOR_NV 0x852C +#define GL_SECONDARY_COLOR_NV 0x852D +#define GL_SPARE0_NV 0x852E +#define GL_SPARE1_NV 0x852F +#define GL_DISCARD_NV 0x8530 +#define GL_E_TIMES_F_NV 0x8531 +#define GL_SPARE0_PLUS_SECONDARY_COLOR_NV 0x8532 +#define GL_UNSIGNED_IDENTITY_NV 0x8536 +#define GL_UNSIGNED_INVERT_NV 0x8537 +#define GL_EXPAND_NORMAL_NV 0x8538 +#define GL_EXPAND_NEGATE_NV 0x8539 +#define GL_HALF_BIAS_NORMAL_NV 0x853A +#define GL_HALF_BIAS_NEGATE_NV 0x853B +#define GL_SIGNED_IDENTITY_NV 0x853C +#define GL_SIGNED_NEGATE_NV 0x853D +#define GL_SCALE_BY_TWO_NV 0x853E +#define GL_SCALE_BY_FOUR_NV 0x853F +#define GL_SCALE_BY_ONE_HALF_NV 0x8540 +#define GL_BIAS_BY_NEGATIVE_ONE_HALF_NV 0x8541 +#define GL_COMBINER_INPUT_NV 0x8542 +#define GL_COMBINER_MAPPING_NV 0x8543 +#define GL_COMBINER_COMPONENT_USAGE_NV 0x8544 +#define GL_COMBINER_AB_DOT_PRODUCT_NV 0x8545 +#define GL_COMBINER_CD_DOT_PRODUCT_NV 0x8546 +#define GL_COMBINER_MUX_SUM_NV 0x8547 +#define GL_COMBINER_SCALE_NV 0x8548 +#define GL_COMBINER_BIAS_NV 0x8549 +#define GL_COMBINER_AB_OUTPUT_NV 0x854A +#define GL_COMBINER_CD_OUTPUT_NV 0x854B +#define GL_COMBINER_SUM_OUTPUT_NV 0x854C +#define GL_MAX_GENERAL_COMBINERS_NV 0x854D +#define GL_NUM_GENERAL_COMBINERS_NV 0x854E +#define GL_COLOR_SUM_CLAMP_NV 0x854F +#define GL_COMBINER0_NV 0x8550 +#define GL_COMBINER1_NV 0x8551 +#define GL_COMBINER2_NV 0x8552 +#define GL_COMBINER3_NV 0x8553 +#define GL_COMBINER4_NV 0x8554 +#define GL_COMBINER5_NV 0x8555 +#define GL_COMBINER6_NV 0x8556 +#define GL_COMBINER7_NV 0x8557 +/* reuse GL_TEXTURE0_ARB */ +/* reuse GL_TEXTURE1_ARB */ +/* reuse GL_ZERO */ +/* reuse GL_NONE */ +/* reuse GL_FOG */ +#endif + +#ifndef GL_NV_fog_distance +#define GL_FOG_DISTANCE_MODE_NV 0x855A +#define GL_EYE_RADIAL_NV 0x855B +#define GL_EYE_PLANE_ABSOLUTE_NV 0x855C +/* reuse GL_EYE_PLANE */ +#endif + +#ifndef GL_NV_texgen_emboss +#define GL_EMBOSS_LIGHT_NV 0x855D +#define GL_EMBOSS_CONSTANT_NV 0x855E +#define GL_EMBOSS_MAP_NV 0x855F +#endif + +#ifndef GL_NV_blend_square +#endif + +#ifndef GL_NV_texture_env_combine4 +#define GL_COMBINE4_NV 0x8503 +#define GL_SOURCE3_RGB_NV 0x8583 +#define GL_SOURCE3_ALPHA_NV 0x858B +#define GL_OPERAND3_RGB_NV 0x8593 +#define GL_OPERAND3_ALPHA_NV 0x859B +#endif + +#ifndef GL_MESA_resize_buffers +#endif + +#ifndef GL_MESA_window_pos +#endif + +#ifndef GL_EXT_texture_compression_s3tc +#define GL_COMPRESSED_RGB_S3TC_DXT1_EXT 0x83F0 +#define GL_COMPRESSED_RGBA_S3TC_DXT1_EXT 0x83F1 +#define GL_COMPRESSED_RGBA_S3TC_DXT3_EXT 0x83F2 +#define GL_COMPRESSED_RGBA_S3TC_DXT5_EXT 0x83F3 +#endif + +#ifndef GL_IBM_cull_vertex +#define GL_CULL_VERTEX_IBM 103050 +#endif + +#ifndef GL_IBM_multimode_draw_arrays +#endif + +#ifndef GL_IBM_vertex_array_lists +#define GL_VERTEX_ARRAY_LIST_IBM 103070 +#define GL_NORMAL_ARRAY_LIST_IBM 103071 +#define GL_COLOR_ARRAY_LIST_IBM 103072 +#define GL_INDEX_ARRAY_LIST_IBM 103073 +#define GL_TEXTURE_COORD_ARRAY_LIST_IBM 103074 +#define GL_EDGE_FLAG_ARRAY_LIST_IBM 103075 +#define GL_FOG_COORDINATE_ARRAY_LIST_IBM 103076 +#define GL_SECONDARY_COLOR_ARRAY_LIST_IBM 103077 +#define GL_VERTEX_ARRAY_LIST_STRIDE_IBM 103080 +#define GL_NORMAL_ARRAY_LIST_STRIDE_IBM 103081 +#define GL_COLOR_ARRAY_LIST_STRIDE_IBM 103082 +#define GL_INDEX_ARRAY_LIST_STRIDE_IBM 103083 +#define GL_TEXTURE_COORD_ARRAY_LIST_STRIDE_IBM 103084 +#define GL_EDGE_FLAG_ARRAY_LIST_STRIDE_IBM 103085 +#define GL_FOG_COORDINATE_ARRAY_LIST_STRIDE_IBM 103086 +#define GL_SECONDARY_COLOR_ARRAY_LIST_STRIDE_IBM 103087 +#endif + +#ifndef GL_SGIX_subsample +#define GL_PACK_SUBSAMPLE_RATE_SGIX 0x85A0 +#define GL_UNPACK_SUBSAMPLE_RATE_SGIX 0x85A1 +#define GL_PIXEL_SUBSAMPLE_4444_SGIX 0x85A2 +#define GL_PIXEL_SUBSAMPLE_2424_SGIX 0x85A3 +#define GL_PIXEL_SUBSAMPLE_4242_SGIX 0x85A4 +#endif + +#ifndef GL_SGIX_ycrcb_subsample +#endif + +#ifndef GL_SGIX_ycrcba +#define GL_YCRCB_SGIX 0x8318 +#define GL_YCRCBA_SGIX 0x8319 +#endif + +#ifndef GL_SGI_depth_pass_instrument +#define GL_DEPTH_PASS_INSTRUMENT_SGIX 0x8310 +#define GL_DEPTH_PASS_INSTRUMENT_COUNTERS_SGIX 0x8311 +#define GL_DEPTH_PASS_INSTRUMENT_MAX_SGIX 0x8312 +#endif + +#ifndef GL_3DFX_texture_compression_FXT1 +#define GL_COMPRESSED_RGB_FXT1_3DFX 0x86B0 +#define GL_COMPRESSED_RGBA_FXT1_3DFX 0x86B1 +#endif + +#ifndef GL_3DFX_multisample +#define GL_MULTISAMPLE_3DFX 0x86B2 +#define GL_SAMPLE_BUFFERS_3DFX 0x86B3 +#define GL_SAMPLES_3DFX 0x86B4 +#define GL_MULTISAMPLE_BIT_3DFX 0x20000000 +#endif + +#ifndef GL_3DFX_tbuffer +#endif + +#ifndef GL_EXT_multisample +#define GL_MULTISAMPLE_EXT 0x809D +#define GL_SAMPLE_ALPHA_TO_MASK_EXT 0x809E +#define GL_SAMPLE_ALPHA_TO_ONE_EXT 0x809F +#define GL_SAMPLE_MASK_EXT 0x80A0 +#define GL_1PASS_EXT 0x80A1 +#define GL_2PASS_0_EXT 0x80A2 +#define GL_2PASS_1_EXT 0x80A3 +#define GL_4PASS_0_EXT 0x80A4 +#define GL_4PASS_1_EXT 0x80A5 +#define GL_4PASS_2_EXT 0x80A6 +#define GL_4PASS_3_EXT 0x80A7 +#define GL_SAMPLE_BUFFERS_EXT 0x80A8 +#define GL_SAMPLES_EXT 0x80A9 +#define GL_SAMPLE_MASK_VALUE_EXT 0x80AA +#define GL_SAMPLE_MASK_INVERT_EXT 0x80AB +#define GL_SAMPLE_PATTERN_EXT 0x80AC +#define GL_MULTISAMPLE_BIT_EXT 0x20000000 +#endif + +#ifndef GL_SGIX_vertex_preclip +#define GL_VERTEX_PRECLIP_SGIX 0x83EE +#define GL_VERTEX_PRECLIP_HINT_SGIX 0x83EF +#endif + +#ifndef GL_SGIX_convolution_accuracy +#define GL_CONVOLUTION_HINT_SGIX 0x8316 +#endif + +#ifndef GL_SGIX_resample +#define GL_PACK_RESAMPLE_SGIX 0x842C +#define GL_UNPACK_RESAMPLE_SGIX 0x842D +#define GL_RESAMPLE_REPLICATE_SGIX 0x842E +#define GL_RESAMPLE_ZERO_FILL_SGIX 0x842F +#define GL_RESAMPLE_DECIMATE_SGIX 0x8430 +#endif + +#ifndef GL_SGIS_point_line_texgen +#define GL_EYE_DISTANCE_TO_POINT_SGIS 0x81F0 +#define GL_OBJECT_DISTANCE_TO_POINT_SGIS 0x81F1 +#define GL_EYE_DISTANCE_TO_LINE_SGIS 0x81F2 +#define GL_OBJECT_DISTANCE_TO_LINE_SGIS 0x81F3 +#define GL_EYE_POINT_SGIS 0x81F4 +#define GL_OBJECT_POINT_SGIS 0x81F5 +#define GL_EYE_LINE_SGIS 0x81F6 +#define GL_OBJECT_LINE_SGIS 0x81F7 +#endif + +#ifndef GL_SGIS_texture_color_mask +#define GL_TEXTURE_COLOR_WRITEMASK_SGIS 0x81EF +#endif + +#ifndef GL_EXT_texture_env_dot3 +#define GL_DOT3_RGB_EXT 0x8740 +#define GL_DOT3_RGBA_EXT 0x8741 +#endif + +#ifndef GL_ATI_texture_mirror_once +#define GL_MIRROR_CLAMP_ATI 0x8742 +#define GL_MIRROR_CLAMP_TO_EDGE_ATI 0x8743 +#endif + +#ifndef GL_NV_fence +#define GL_ALL_COMPLETED_NV 0x84F2 +#define GL_FENCE_STATUS_NV 0x84F3 +#define GL_FENCE_CONDITION_NV 0x84F4 +#endif + +#ifndef GL_IBM_texture_mirrored_repeat +#define GL_MIRRORED_REPEAT_IBM 0x8370 +#endif + +#ifndef GL_NV_evaluators +#define GL_EVAL_2D_NV 0x86C0 +#define GL_EVAL_TRIANGULAR_2D_NV 0x86C1 +#define GL_MAP_TESSELLATION_NV 0x86C2 +#define GL_MAP_ATTRIB_U_ORDER_NV 0x86C3 +#define GL_MAP_ATTRIB_V_ORDER_NV 0x86C4 +#define GL_EVAL_FRACTIONAL_TESSELLATION_NV 0x86C5 +#define GL_EVAL_VERTEX_ATTRIB0_NV 0x86C6 +#define GL_EVAL_VERTEX_ATTRIB1_NV 0x86C7 +#define GL_EVAL_VERTEX_ATTRIB2_NV 0x86C8 +#define GL_EVAL_VERTEX_ATTRIB3_NV 0x86C9 +#define GL_EVAL_VERTEX_ATTRIB4_NV 0x86CA +#define GL_EVAL_VERTEX_ATTRIB5_NV 0x86CB +#define GL_EVAL_VERTEX_ATTRIB6_NV 0x86CC +#define GL_EVAL_VERTEX_ATTRIB7_NV 0x86CD +#define GL_EVAL_VERTEX_ATTRIB8_NV 0x86CE +#define GL_EVAL_VERTEX_ATTRIB9_NV 0x86CF +#define GL_EVAL_VERTEX_ATTRIB10_NV 0x86D0 +#define GL_EVAL_VERTEX_ATTRIB11_NV 0x86D1 +#define GL_EVAL_VERTEX_ATTRIB12_NV 0x86D2 +#define GL_EVAL_VERTEX_ATTRIB13_NV 0x86D3 +#define GL_EVAL_VERTEX_ATTRIB14_NV 0x86D4 +#define GL_EVAL_VERTEX_ATTRIB15_NV 0x86D5 +#define GL_MAX_MAP_TESSELLATION_NV 0x86D6 +#define GL_MAX_RATIONAL_EVAL_ORDER_NV 0x86D7 +#endif + +#ifndef GL_NV_packed_depth_stencil +#define GL_DEPTH_STENCIL_NV 0x84F9 +#define GL_UNSIGNED_INT_24_8_NV 0x84FA +#endif + +#ifndef GL_NV_register_combiners2 +#define GL_PER_STAGE_CONSTANTS_NV 0x8535 +#endif + +#ifndef GL_NV_texture_compression_vtc +#endif + +#ifndef GL_NV_texture_rectangle +#define GL_TEXTURE_RECTANGLE_NV 0x84F5 +#define GL_TEXTURE_BINDING_RECTANGLE_NV 0x84F6 +#define GL_PROXY_TEXTURE_RECTANGLE_NV 0x84F7 +#define GL_MAX_RECTANGLE_TEXTURE_SIZE_NV 0x84F8 +#endif + +#ifndef GL_NV_texture_shader +#define GL_OFFSET_TEXTURE_RECTANGLE_NV 0x864C +#define GL_OFFSET_TEXTURE_RECTANGLE_SCALE_NV 0x864D +#define GL_DOT_PRODUCT_TEXTURE_RECTANGLE_NV 0x864E +#define GL_RGBA_UNSIGNED_DOT_PRODUCT_MAPPING_NV 0x86D9 +#define GL_UNSIGNED_INT_S8_S8_8_8_NV 0x86DA +#define GL_UNSIGNED_INT_8_8_S8_S8_REV_NV 0x86DB +#define GL_DSDT_MAG_INTENSITY_NV 0x86DC +#define GL_SHADER_CONSISTENT_NV 0x86DD +#define GL_TEXTURE_SHADER_NV 0x86DE +#define GL_SHADER_OPERATION_NV 0x86DF +#define GL_CULL_MODES_NV 0x86E0 +#define GL_OFFSET_TEXTURE_MATRIX_NV 0x86E1 +#define GL_OFFSET_TEXTURE_SCALE_NV 0x86E2 +#define GL_OFFSET_TEXTURE_BIAS_NV 0x86E3 +#define GL_OFFSET_TEXTURE_2D_MATRIX_NV GL_OFFSET_TEXTURE_MATRIX_NV +#define GL_OFFSET_TEXTURE_2D_SCALE_NV GL_OFFSET_TEXTURE_SCALE_NV +#define GL_OFFSET_TEXTURE_2D_BIAS_NV GL_OFFSET_TEXTURE_BIAS_NV +#define GL_PREVIOUS_TEXTURE_INPUT_NV 0x86E4 +#define GL_CONST_EYE_NV 0x86E5 +#define GL_PASS_THROUGH_NV 0x86E6 +#define GL_CULL_FRAGMENT_NV 0x86E7 +#define GL_OFFSET_TEXTURE_2D_NV 0x86E8 +#define GL_DEPENDENT_AR_TEXTURE_2D_NV 0x86E9 +#define GL_DEPENDENT_GB_TEXTURE_2D_NV 0x86EA +#define GL_DOT_PRODUCT_NV 0x86EC +#define GL_DOT_PRODUCT_DEPTH_REPLACE_NV 0x86ED +#define GL_DOT_PRODUCT_TEXTURE_2D_NV 0x86EE +#define GL_DOT_PRODUCT_TEXTURE_CUBE_MAP_NV 0x86F0 +#define GL_DOT_PRODUCT_DIFFUSE_CUBE_MAP_NV 0x86F1 +#define GL_DOT_PRODUCT_REFLECT_CUBE_MAP_NV 0x86F2 +#define GL_DOT_PRODUCT_CONST_EYE_REFLECT_CUBE_MAP_NV 0x86F3 +#define GL_HILO_NV 0x86F4 +#define GL_DSDT_NV 0x86F5 +#define GL_DSDT_MAG_NV 0x86F6 +#define GL_DSDT_MAG_VIB_NV 0x86F7 +#define GL_HILO16_NV 0x86F8 +#define GL_SIGNED_HILO_NV 0x86F9 +#define GL_SIGNED_HILO16_NV 0x86FA +#define GL_SIGNED_RGBA_NV 0x86FB +#define GL_SIGNED_RGBA8_NV 0x86FC +#define GL_SIGNED_RGB_NV 0x86FE +#define GL_SIGNED_RGB8_NV 0x86FF +#define GL_SIGNED_LUMINANCE_NV 0x8701 +#define GL_SIGNED_LUMINANCE8_NV 0x8702 +#define GL_SIGNED_LUMINANCE_ALPHA_NV 0x8703 +#define GL_SIGNED_LUMINANCE8_ALPHA8_NV 0x8704 +#define GL_SIGNED_ALPHA_NV 0x8705 +#define GL_SIGNED_ALPHA8_NV 0x8706 +#define GL_SIGNED_INTENSITY_NV 0x8707 +#define GL_SIGNED_INTENSITY8_NV 0x8708 +#define GL_DSDT8_NV 0x8709 +#define GL_DSDT8_MAG8_NV 0x870A +#define GL_DSDT8_MAG8_INTENSITY8_NV 0x870B +#define GL_SIGNED_RGB_UNSIGNED_ALPHA_NV 0x870C +#define GL_SIGNED_RGB8_UNSIGNED_ALPHA8_NV 0x870D +#define GL_HI_SCALE_NV 0x870E +#define GL_LO_SCALE_NV 0x870F +#define GL_DS_SCALE_NV 0x8710 +#define GL_DT_SCALE_NV 0x8711 +#define GL_MAGNITUDE_SCALE_NV 0x8712 +#define GL_VIBRANCE_SCALE_NV 0x8713 +#define GL_HI_BIAS_NV 0x8714 +#define GL_LO_BIAS_NV 0x8715 +#define GL_DS_BIAS_NV 0x8716 +#define GL_DT_BIAS_NV 0x8717 +#define GL_MAGNITUDE_BIAS_NV 0x8718 +#define GL_VIBRANCE_BIAS_NV 0x8719 +#define GL_TEXTURE_BORDER_VALUES_NV 0x871A +#define GL_TEXTURE_HI_SIZE_NV 0x871B +#define GL_TEXTURE_LO_SIZE_NV 0x871C +#define GL_TEXTURE_DS_SIZE_NV 0x871D +#define GL_TEXTURE_DT_SIZE_NV 0x871E +#define GL_TEXTURE_MAG_SIZE_NV 0x871F +#endif + +#ifndef GL_NV_texture_shader2 +#define GL_DOT_PRODUCT_TEXTURE_3D_NV 0x86EF +#endif + +#ifndef GL_NV_vertex_array_range2 +#define GL_VERTEX_ARRAY_RANGE_WITHOUT_FLUSH_NV 0x8533 +#endif + +#ifndef GL_NV_vertex_program +#define GL_VERTEX_PROGRAM_NV 0x8620 +#define GL_VERTEX_STATE_PROGRAM_NV 0x8621 +#define GL_ATTRIB_ARRAY_SIZE_NV 0x8623 +#define GL_ATTRIB_ARRAY_STRIDE_NV 0x8624 +#define GL_ATTRIB_ARRAY_TYPE_NV 0x8625 +#define GL_CURRENT_ATTRIB_NV 0x8626 +#define GL_PROGRAM_LENGTH_NV 0x8627 +#define GL_PROGRAM_STRING_NV 0x8628 +#define GL_MODELVIEW_PROJECTION_NV 0x8629 +#define GL_IDENTITY_NV 0x862A +#define GL_INVERSE_NV 0x862B +#define GL_TRANSPOSE_NV 0x862C +#define GL_INVERSE_TRANSPOSE_NV 0x862D +#define GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV 0x862E +#define GL_MAX_TRACK_MATRICES_NV 0x862F +#define GL_MATRIX0_NV 0x8630 +#define GL_MATRIX1_NV 0x8631 +#define GL_MATRIX2_NV 0x8632 +#define GL_MATRIX3_NV 0x8633 +#define GL_MATRIX4_NV 0x8634 +#define GL_MATRIX5_NV 0x8635 +#define GL_MATRIX6_NV 0x8636 +#define GL_MATRIX7_NV 0x8637 +#define GL_CURRENT_MATRIX_STACK_DEPTH_NV 0x8640 +#define GL_CURRENT_MATRIX_NV 0x8641 +#define GL_VERTEX_PROGRAM_POINT_SIZE_NV 0x8642 +#define GL_VERTEX_PROGRAM_TWO_SIDE_NV 0x8643 +#define GL_PROGRAM_PARAMETER_NV 0x8644 +#define GL_ATTRIB_ARRAY_POINTER_NV 0x8645 +#define GL_PROGRAM_TARGET_NV 0x8646 +#define GL_PROGRAM_RESIDENT_NV 0x8647 +#define GL_TRACK_MATRIX_NV 0x8648 +#define GL_TRACK_MATRIX_TRANSFORM_NV 0x8649 +#define GL_VERTEX_PROGRAM_BINDING_NV 0x864A +#define GL_PROGRAM_ERROR_POSITION_NV 0x864B +#define GL_VERTEX_ATTRIB_ARRAY0_NV 0x8650 +#define GL_VERTEX_ATTRIB_ARRAY1_NV 0x8651 +#define GL_VERTEX_ATTRIB_ARRAY2_NV 0x8652 +#define GL_VERTEX_ATTRIB_ARRAY3_NV 0x8653 +#define GL_VERTEX_ATTRIB_ARRAY4_NV 0x8654 +#define GL_VERTEX_ATTRIB_ARRAY5_NV 0x8655 +#define GL_VERTEX_ATTRIB_ARRAY6_NV 0x8656 +#define GL_VERTEX_ATTRIB_ARRAY7_NV 0x8657 +#define GL_VERTEX_ATTRIB_ARRAY8_NV 0x8658 +#define GL_VERTEX_ATTRIB_ARRAY9_NV 0x8659 +#define GL_VERTEX_ATTRIB_ARRAY10_NV 0x865A +#define GL_VERTEX_ATTRIB_ARRAY11_NV 0x865B +#define GL_VERTEX_ATTRIB_ARRAY12_NV 0x865C +#define GL_VERTEX_ATTRIB_ARRAY13_NV 0x865D +#define GL_VERTEX_ATTRIB_ARRAY14_NV 0x865E +#define GL_VERTEX_ATTRIB_ARRAY15_NV 0x865F +#define GL_MAP1_VERTEX_ATTRIB0_4_NV 0x8660 +#define GL_MAP1_VERTEX_ATTRIB1_4_NV 0x8661 +#define GL_MAP1_VERTEX_ATTRIB2_4_NV 0x8662 +#define GL_MAP1_VERTEX_ATTRIB3_4_NV 0x8663 +#define GL_MAP1_VERTEX_ATTRIB4_4_NV 0x8664 +#define GL_MAP1_VERTEX_ATTRIB5_4_NV 0x8665 +#define GL_MAP1_VERTEX_ATTRIB6_4_NV 0x8666 +#define GL_MAP1_VERTEX_ATTRIB7_4_NV 0x8667 +#define GL_MAP1_VERTEX_ATTRIB8_4_NV 0x8668 +#define GL_MAP1_VERTEX_ATTRIB9_4_NV 0x8669 +#define GL_MAP1_VERTEX_ATTRIB10_4_NV 0x866A +#define GL_MAP1_VERTEX_ATTRIB11_4_NV 0x866B +#define GL_MAP1_VERTEX_ATTRIB12_4_NV 0x866C +#define GL_MAP1_VERTEX_ATTRIB13_4_NV 0x866D +#define GL_MAP1_VERTEX_ATTRIB14_4_NV 0x866E +#define GL_MAP1_VERTEX_ATTRIB15_4_NV 0x866F +#define GL_MAP2_VERTEX_ATTRIB0_4_NV 0x8670 +#define GL_MAP2_VERTEX_ATTRIB1_4_NV 0x8671 +#define GL_MAP2_VERTEX_ATTRIB2_4_NV 0x8672 +#define GL_MAP2_VERTEX_ATTRIB3_4_NV 0x8673 +#define GL_MAP2_VERTEX_ATTRIB4_4_NV 0x8674 +#define GL_MAP2_VERTEX_ATTRIB5_4_NV 0x8675 +#define GL_MAP2_VERTEX_ATTRIB6_4_NV 0x8676 +#define GL_MAP2_VERTEX_ATTRIB7_4_NV 0x8677 +#define GL_MAP2_VERTEX_ATTRIB8_4_NV 0x8678 +#define GL_MAP2_VERTEX_ATTRIB9_4_NV 0x8679 +#define GL_MAP2_VERTEX_ATTRIB10_4_NV 0x867A +#define GL_MAP2_VERTEX_ATTRIB11_4_NV 0x867B +#define GL_MAP2_VERTEX_ATTRIB12_4_NV 0x867C +#define GL_MAP2_VERTEX_ATTRIB13_4_NV 0x867D +#define GL_MAP2_VERTEX_ATTRIB14_4_NV 0x867E +#define GL_MAP2_VERTEX_ATTRIB15_4_NV 0x867F +#endif + +#ifndef GL_SGIX_texture_coordinate_clamp +#define GL_TEXTURE_MAX_CLAMP_S_SGIX 0x8369 +#define GL_TEXTURE_MAX_CLAMP_T_SGIX 0x836A +#define GL_TEXTURE_MAX_CLAMP_R_SGIX 0x836B +#endif + +#ifndef GL_SGIX_scalebias_hint +#define GL_SCALEBIAS_HINT_SGIX 0x8322 +#endif + +#ifndef GL_OML_interlace +#define GL_INTERLACE_OML 0x8980 +#define GL_INTERLACE_READ_OML 0x8981 +#endif + +#ifndef GL_OML_subsample +#define GL_FORMAT_SUBSAMPLE_24_24_OML 0x8982 +#define GL_FORMAT_SUBSAMPLE_244_244_OML 0x8983 +#endif + +#ifndef GL_OML_resample +#define GL_PACK_RESAMPLE_OML 0x8984 +#define GL_UNPACK_RESAMPLE_OML 0x8985 +#define GL_RESAMPLE_REPLICATE_OML 0x8986 +#define GL_RESAMPLE_ZERO_FILL_OML 0x8987 +#define GL_RESAMPLE_AVERAGE_OML 0x8988 +#define GL_RESAMPLE_DECIMATE_OML 0x8989 +#endif + +#ifndef GL_NV_copy_depth_to_color +#define GL_DEPTH_STENCIL_TO_RGBA_NV 0x886E +#define GL_DEPTH_STENCIL_TO_BGRA_NV 0x886F +#endif + +#ifndef GL_ATI_envmap_bumpmap +#define GL_BUMP_ROT_MATRIX_ATI 0x8775 +#define GL_BUMP_ROT_MATRIX_SIZE_ATI 0x8776 +#define GL_BUMP_NUM_TEX_UNITS_ATI 0x8777 +#define GL_BUMP_TEX_UNITS_ATI 0x8778 +#define GL_DUDV_ATI 0x8779 +#define GL_DU8DV8_ATI 0x877A +#define GL_BUMP_ENVMAP_ATI 0x877B +#define GL_BUMP_TARGET_ATI 0x877C +#endif + +#ifndef GL_ATI_fragment_shader +#define GL_FRAGMENT_SHADER_ATI 0x8920 +#define GL_REG_0_ATI 0x8921 +#define GL_REG_1_ATI 0x8922 +#define GL_REG_2_ATI 0x8923 +#define GL_REG_3_ATI 0x8924 +#define GL_REG_4_ATI 0x8925 +#define GL_REG_5_ATI 0x8926 +#define GL_REG_6_ATI 0x8927 +#define GL_REG_7_ATI 0x8928 +#define GL_REG_8_ATI 0x8929 +#define GL_REG_9_ATI 0x892A +#define GL_REG_10_ATI 0x892B +#define GL_REG_11_ATI 0x892C +#define GL_REG_12_ATI 0x892D +#define GL_REG_13_ATI 0x892E +#define GL_REG_14_ATI 0x892F +#define GL_REG_15_ATI 0x8930 +#define GL_REG_16_ATI 0x8931 +#define GL_REG_17_ATI 0x8932 +#define GL_REG_18_ATI 0x8933 +#define GL_REG_19_ATI 0x8934 +#define GL_REG_20_ATI 0x8935 +#define GL_REG_21_ATI 0x8936 +#define GL_REG_22_ATI 0x8937 +#define GL_REG_23_ATI 0x8938 +#define GL_REG_24_ATI 0x8939 +#define GL_REG_25_ATI 0x893A +#define GL_REG_26_ATI 0x893B +#define GL_REG_27_ATI 0x893C +#define GL_REG_28_ATI 0x893D +#define GL_REG_29_ATI 0x893E +#define GL_REG_30_ATI 0x893F +#define GL_REG_31_ATI 0x8940 +#define GL_CON_0_ATI 0x8941 +#define GL_CON_1_ATI 0x8942 +#define GL_CON_2_ATI 0x8943 +#define GL_CON_3_ATI 0x8944 +#define GL_CON_4_ATI 0x8945 +#define GL_CON_5_ATI 0x8946 +#define GL_CON_6_ATI 0x8947 +#define GL_CON_7_ATI 0x8948 +#define GL_CON_8_ATI 0x8949 +#define GL_CON_9_ATI 0x894A +#define GL_CON_10_ATI 0x894B +#define GL_CON_11_ATI 0x894C +#define GL_CON_12_ATI 0x894D +#define GL_CON_13_ATI 0x894E +#define GL_CON_14_ATI 0x894F +#define GL_CON_15_ATI 0x8950 +#define GL_CON_16_ATI 0x8951 +#define GL_CON_17_ATI 0x8952 +#define GL_CON_18_ATI 0x8953 +#define GL_CON_19_ATI 0x8954 +#define GL_CON_20_ATI 0x8955 +#define GL_CON_21_ATI 0x8956 +#define GL_CON_22_ATI 0x8957 +#define GL_CON_23_ATI 0x8958 +#define GL_CON_24_ATI 0x8959 +#define GL_CON_25_ATI 0x895A +#define GL_CON_26_ATI 0x895B +#define GL_CON_27_ATI 0x895C +#define GL_CON_28_ATI 0x895D +#define GL_CON_29_ATI 0x895E +#define GL_CON_30_ATI 0x895F +#define GL_CON_31_ATI 0x8960 +#define GL_MOV_ATI 0x8961 +#define GL_ADD_ATI 0x8963 +#define GL_MUL_ATI 0x8964 +#define GL_SUB_ATI 0x8965 +#define GL_DOT3_ATI 0x8966 +#define GL_DOT4_ATI 0x8967 +#define GL_MAD_ATI 0x8968 +#define GL_LERP_ATI 0x8969 +#define GL_CND_ATI 0x896A +#define GL_CND0_ATI 0x896B +#define GL_DOT2_ADD_ATI 0x896C +#define GL_SECONDARY_INTERPOLATOR_ATI 0x896D +#define GL_NUM_FRAGMENT_REGISTERS_ATI 0x896E +#define GL_NUM_FRAGMENT_CONSTANTS_ATI 0x896F +#define GL_NUM_PASSES_ATI 0x8970 +#define GL_NUM_INSTRUCTIONS_PER_PASS_ATI 0x8971 +#define GL_NUM_INSTRUCTIONS_TOTAL_ATI 0x8972 +#define GL_NUM_INPUT_INTERPOLATOR_COMPONENTS_ATI 0x8973 +#define GL_NUM_LOOPBACK_COMPONENTS_ATI 0x8974 +#define GL_COLOR_ALPHA_PAIRING_ATI 0x8975 +#define GL_SWIZZLE_STR_ATI 0x8976 +#define GL_SWIZZLE_STQ_ATI 0x8977 +#define GL_SWIZZLE_STR_DR_ATI 0x8978 +#define GL_SWIZZLE_STQ_DQ_ATI 0x8979 +#define GL_SWIZZLE_STRQ_ATI 0x897A +#define GL_SWIZZLE_STRQ_DQ_ATI 0x897B +#define GL_RED_BIT_ATI 0x00000001 +#define GL_GREEN_BIT_ATI 0x00000002 +#define GL_BLUE_BIT_ATI 0x00000004 +#define GL_2X_BIT_ATI 0x00000001 +#define GL_4X_BIT_ATI 0x00000002 +#define GL_8X_BIT_ATI 0x00000004 +#define GL_HALF_BIT_ATI 0x00000008 +#define GL_QUARTER_BIT_ATI 0x00000010 +#define GL_EIGHTH_BIT_ATI 0x00000020 +#define GL_SATURATE_BIT_ATI 0x00000040 +#define GL_COMP_BIT_ATI 0x00000002 +#define GL_NEGATE_BIT_ATI 0x00000004 +#define GL_BIAS_BIT_ATI 0x00000008 +#endif + +#ifndef GL_ATI_pn_triangles +#define GL_PN_TRIANGLES_ATI 0x87F0 +#define GL_MAX_PN_TRIANGLES_TESSELATION_LEVEL_ATI 0x87F1 +#define GL_PN_TRIANGLES_POINT_MODE_ATI 0x87F2 +#define GL_PN_TRIANGLES_NORMAL_MODE_ATI 0x87F3 +#define GL_PN_TRIANGLES_TESSELATION_LEVEL_ATI 0x87F4 +#define GL_PN_TRIANGLES_POINT_MODE_LINEAR_ATI 0x87F5 +#define GL_PN_TRIANGLES_POINT_MODE_CUBIC_ATI 0x87F6 +#define GL_PN_TRIANGLES_NORMAL_MODE_LINEAR_ATI 0x87F7 +#define GL_PN_TRIANGLES_NORMAL_MODE_QUADRATIC_ATI 0x87F8 +#endif + +#ifndef GL_ATI_vertex_array_object +#define GL_STATIC_ATI 0x8760 +#define GL_DYNAMIC_ATI 0x8761 +#define GL_PRESERVE_ATI 0x8762 +#define GL_DISCARD_ATI 0x8763 +#define GL_OBJECT_BUFFER_SIZE_ATI 0x8764 +#define GL_OBJECT_BUFFER_USAGE_ATI 0x8765 +#define GL_ARRAY_OBJECT_BUFFER_ATI 0x8766 +#define GL_ARRAY_OBJECT_OFFSET_ATI 0x8767 +#endif + +#ifndef GL_EXT_vertex_shader +#define GL_VERTEX_SHADER_EXT 0x8780 +#define GL_VERTEX_SHADER_BINDING_EXT 0x8781 +#define GL_OP_INDEX_EXT 0x8782 +#define GL_OP_NEGATE_EXT 0x8783 +#define GL_OP_DOT3_EXT 0x8784 +#define GL_OP_DOT4_EXT 0x8785 +#define GL_OP_MUL_EXT 0x8786 +#define GL_OP_ADD_EXT 0x8787 +#define GL_OP_MADD_EXT 0x8788 +#define GL_OP_FRAC_EXT 0x8789 +#define GL_OP_MAX_EXT 0x878A +#define GL_OP_MIN_EXT 0x878B +#define GL_OP_SET_GE_EXT 0x878C +#define GL_OP_SET_LT_EXT 0x878D +#define GL_OP_CLAMP_EXT 0x878E +#define GL_OP_FLOOR_EXT 0x878F +#define GL_OP_ROUND_EXT 0x8790 +#define GL_OP_EXP_BASE_2_EXT 0x8791 +#define GL_OP_LOG_BASE_2_EXT 0x8792 +#define GL_OP_POWER_EXT 0x8793 +#define GL_OP_RECIP_EXT 0x8794 +#define GL_OP_RECIP_SQRT_EXT 0x8795 +#define GL_OP_SUB_EXT 0x8796 +#define GL_OP_CROSS_PRODUCT_EXT 0x8797 +#define GL_OP_MULTIPLY_MATRIX_EXT 0x8798 +#define GL_OP_MOV_EXT 0x8799 +#define GL_OUTPUT_VERTEX_EXT 0x879A +#define GL_OUTPUT_COLOR0_EXT 0x879B +#define GL_OUTPUT_COLOR1_EXT 0x879C +#define GL_OUTPUT_TEXTURE_COORD0_EXT 0x879D +#define GL_OUTPUT_TEXTURE_COORD1_EXT 0x879E +#define GL_OUTPUT_TEXTURE_COORD2_EXT 0x879F +#define GL_OUTPUT_TEXTURE_COORD3_EXT 0x87A0 +#define GL_OUTPUT_TEXTURE_COORD4_EXT 0x87A1 +#define GL_OUTPUT_TEXTURE_COORD5_EXT 0x87A2 +#define GL_OUTPUT_TEXTURE_COORD6_EXT 0x87A3 +#define GL_OUTPUT_TEXTURE_COORD7_EXT 0x87A4 +#define GL_OUTPUT_TEXTURE_COORD8_EXT 0x87A5 +#define GL_OUTPUT_TEXTURE_COORD9_EXT 0x87A6 +#define GL_OUTPUT_TEXTURE_COORD10_EXT 0x87A7 +#define GL_OUTPUT_TEXTURE_COORD11_EXT 0x87A8 +#define GL_OUTPUT_TEXTURE_COORD12_EXT 0x87A9 +#define GL_OUTPUT_TEXTURE_COORD13_EXT 0x87AA +#define GL_OUTPUT_TEXTURE_COORD14_EXT 0x87AB +#define GL_OUTPUT_TEXTURE_COORD15_EXT 0x87AC +#define GL_OUTPUT_TEXTURE_COORD16_EXT 0x87AD +#define GL_OUTPUT_TEXTURE_COORD17_EXT 0x87AE +#define GL_OUTPUT_TEXTURE_COORD18_EXT 0x87AF +#define GL_OUTPUT_TEXTURE_COORD19_EXT 0x87B0 +#define GL_OUTPUT_TEXTURE_COORD20_EXT 0x87B1 +#define GL_OUTPUT_TEXTURE_COORD21_EXT 0x87B2 +#define GL_OUTPUT_TEXTURE_COORD22_EXT 0x87B3 +#define GL_OUTPUT_TEXTURE_COORD23_EXT 0x87B4 +#define GL_OUTPUT_TEXTURE_COORD24_EXT 0x87B5 +#define GL_OUTPUT_TEXTURE_COORD25_EXT 0x87B6 +#define GL_OUTPUT_TEXTURE_COORD26_EXT 0x87B7 +#define GL_OUTPUT_TEXTURE_COORD27_EXT 0x87B8 +#define GL_OUTPUT_TEXTURE_COORD28_EXT 0x87B9 +#define GL_OUTPUT_TEXTURE_COORD29_EXT 0x87BA +#define GL_OUTPUT_TEXTURE_COORD30_EXT 0x87BB +#define GL_OUTPUT_TEXTURE_COORD31_EXT 0x87BC +#define GL_OUTPUT_FOG_EXT 0x87BD +#define GL_SCALAR_EXT 0x87BE +#define GL_VECTOR_EXT 0x87BF +#define GL_MATRIX_EXT 0x87C0 +#define GL_VARIANT_EXT 0x87C1 +#define GL_INVARIANT_EXT 0x87C2 +#define GL_LOCAL_CONSTANT_EXT 0x87C3 +#define GL_LOCAL_EXT 0x87C4 +#define GL_MAX_VERTEX_SHADER_INSTRUCTIONS_EXT 0x87C5 +#define GL_MAX_VERTEX_SHADER_VARIANTS_EXT 0x87C6 +#define GL_MAX_VERTEX_SHADER_INVARIANTS_EXT 0x87C7 +#define GL_MAX_VERTEX_SHADER_LOCAL_CONSTANTS_EXT 0x87C8 +#define GL_MAX_VERTEX_SHADER_LOCALS_EXT 0x87C9 +#define GL_MAX_OPTIMIZED_VERTEX_SHADER_INSTRUCTIONS_EXT 0x87CA +#define GL_MAX_OPTIMIZED_VERTEX_SHADER_VARIANTS_EXT 0x87CB +#define GL_MAX_OPTIMIZED_VERTEX_SHADER_LOCAL_CONSTANTS_EXT 0x87CC +#define GL_MAX_OPTIMIZED_VERTEX_SHADER_INVARIANTS_EXT 0x87CD +#define GL_MAX_OPTIMIZED_VERTEX_SHADER_LOCALS_EXT 0x87CE +#define GL_VERTEX_SHADER_INSTRUCTIONS_EXT 0x87CF +#define GL_VERTEX_SHADER_VARIANTS_EXT 0x87D0 +#define GL_VERTEX_SHADER_INVARIANTS_EXT 0x87D1 +#define GL_VERTEX_SHADER_LOCAL_CONSTANTS_EXT 0x87D2 +#define GL_VERTEX_SHADER_LOCALS_EXT 0x87D3 +#define GL_VERTEX_SHADER_OPTIMIZED_EXT 0x87D4 +#define GL_X_EXT 0x87D5 +#define GL_Y_EXT 0x87D6 +#define GL_Z_EXT 0x87D7 +#define GL_W_EXT 0x87D8 +#define GL_NEGATIVE_X_EXT 0x87D9 +#define GL_NEGATIVE_Y_EXT 0x87DA +#define GL_NEGATIVE_Z_EXT 0x87DB +#define GL_NEGATIVE_W_EXT 0x87DC +#define GL_ZERO_EXT 0x87DD +#define GL_ONE_EXT 0x87DE +#define GL_NEGATIVE_ONE_EXT 0x87DF +#define GL_NORMALIZED_RANGE_EXT 0x87E0 +#define GL_FULL_RANGE_EXT 0x87E1 +#define GL_CURRENT_VERTEX_EXT 0x87E2 +#define GL_MVP_MATRIX_EXT 0x87E3 +#define GL_VARIANT_VALUE_EXT 0x87E4 +#define GL_VARIANT_DATATYPE_EXT 0x87E5 +#define GL_VARIANT_ARRAY_STRIDE_EXT 0x87E6 +#define GL_VARIANT_ARRAY_TYPE_EXT 0x87E7 +#define GL_VARIANT_ARRAY_EXT 0x87E8 +#define GL_VARIANT_ARRAY_POINTER_EXT 0x87E9 +#define GL_INVARIANT_VALUE_EXT 0x87EA +#define GL_INVARIANT_DATATYPE_EXT 0x87EB +#define GL_LOCAL_CONSTANT_VALUE_EXT 0x87EC +#define GL_LOCAL_CONSTANT_DATATYPE_EXT 0x87ED +#endif + +#ifndef GL_ATI_vertex_streams +#define GL_MAX_VERTEX_STREAMS_ATI 0x876B +#define GL_VERTEX_STREAM0_ATI 0x876C +#define GL_VERTEX_STREAM1_ATI 0x876D +#define GL_VERTEX_STREAM2_ATI 0x876E +#define GL_VERTEX_STREAM3_ATI 0x876F +#define GL_VERTEX_STREAM4_ATI 0x8770 +#define GL_VERTEX_STREAM5_ATI 0x8771 +#define GL_VERTEX_STREAM6_ATI 0x8772 +#define GL_VERTEX_STREAM7_ATI 0x8773 +#define GL_VERTEX_SOURCE_ATI 0x8774 +#endif + +#ifndef GL_ATI_element_array +#define GL_ELEMENT_ARRAY_ATI 0x8768 +#define GL_ELEMENT_ARRAY_TYPE_ATI 0x8769 +#define GL_ELEMENT_ARRAY_POINTER_ATI 0x876A +#endif + +#ifndef GL_SUN_mesh_array +#define GL_QUAD_MESH_SUN 0x8614 +#define GL_TRIANGLE_MESH_SUN 0x8615 +#endif + +#ifndef GL_SUN_slice_accum +#define GL_SLICE_ACCUM_SUN 0x85CC +#endif + +#ifndef GL_NV_multisample_filter_hint +#define GL_MULTISAMPLE_FILTER_HINT_NV 0x8534 +#endif + +#ifndef GL_NV_depth_clamp +#define GL_DEPTH_CLAMP_NV 0x864F +#endif + +#ifndef GL_NV_occlusion_query +#define GL_PIXEL_COUNTER_BITS_NV 0x8864 +#define GL_CURRENT_OCCLUSION_QUERY_ID_NV 0x8865 +#define GL_PIXEL_COUNT_NV 0x8866 +#define GL_PIXEL_COUNT_AVAILABLE_NV 0x8867 +#endif + +#ifndef GL_NV_point_sprite +#define GL_POINT_SPRITE_NV 0x8861 +#define GL_COORD_REPLACE_NV 0x8862 +#define GL_POINT_SPRITE_R_MODE_NV 0x8863 +#endif + +#ifndef GL_NV_texture_shader3 +#define GL_OFFSET_PROJECTIVE_TEXTURE_2D_NV 0x8850 +#define GL_OFFSET_PROJECTIVE_TEXTURE_2D_SCALE_NV 0x8851 +#define GL_OFFSET_PROJECTIVE_TEXTURE_RECTANGLE_NV 0x8852 +#define GL_OFFSET_PROJECTIVE_TEXTURE_RECTANGLE_SCALE_NV 0x8853 +#define GL_OFFSET_HILO_TEXTURE_2D_NV 0x8854 +#define GL_OFFSET_HILO_TEXTURE_RECTANGLE_NV 0x8855 +#define GL_OFFSET_HILO_PROJECTIVE_TEXTURE_2D_NV 0x8856 +#define GL_OFFSET_HILO_PROJECTIVE_TEXTURE_RECTANGLE_NV 0x8857 +#define GL_DEPENDENT_HILO_TEXTURE_2D_NV 0x8858 +#define GL_DEPENDENT_RGB_TEXTURE_3D_NV 0x8859 +#define GL_DEPENDENT_RGB_TEXTURE_CUBE_MAP_NV 0x885A +#define GL_DOT_PRODUCT_PASS_THROUGH_NV 0x885B +#define GL_DOT_PRODUCT_TEXTURE_1D_NV 0x885C +#define GL_DOT_PRODUCT_AFFINE_DEPTH_REPLACE_NV 0x885D +#define GL_HILO8_NV 0x885E +#define GL_SIGNED_HILO8_NV 0x885F +#define GL_FORCE_BLUE_TO_ONE_NV 0x8860 +#endif + +#ifndef GL_NV_vertex_program1_1 +#endif + +#ifndef GL_EXT_shadow_funcs +#endif + +#ifndef GL_EXT_stencil_two_side +#define GL_STENCIL_TEST_TWO_SIDE_EXT 0x8910 +#define GL_ACTIVE_STENCIL_FACE_EXT 0x8911 +#endif + +#ifndef GL_ATI_text_fragment_shader +#define GL_TEXT_FRAGMENT_SHADER_ATI 0x8200 +#endif + +#ifndef GL_APPLE_client_storage +#define GL_UNPACK_CLIENT_STORAGE_APPLE 0x85B2 +#endif + +#ifndef GL_APPLE_element_array +#define GL_ELEMENT_ARRAY_APPLE 0x8768 +#define GL_ELEMENT_ARRAY_TYPE_APPLE 0x8769 +#define GL_ELEMENT_ARRAY_POINTER_APPLE 0x876A +#endif + +#ifndef GL_APPLE_fence +#define GL_DRAW_PIXELS_APPLE 0x8A0A +#define GL_FENCE_APPLE 0x8A0B +#endif + +#ifndef GL_APPLE_vertex_array_object +#define GL_VERTEX_ARRAY_BINDING_APPLE 0x85B5 +#endif + +#ifndef GL_APPLE_vertex_array_range +#define GL_VERTEX_ARRAY_RANGE_APPLE 0x851D +#define GL_VERTEX_ARRAY_RANGE_LENGTH_APPLE 0x851E +#define GL_VERTEX_ARRAY_STORAGE_HINT_APPLE 0x851F +#define GL_VERTEX_ARRAY_RANGE_POINTER_APPLE 0x8521 +#define GL_STORAGE_CACHED_APPLE 0x85BE +#define GL_STORAGE_SHARED_APPLE 0x85BF +#endif + +#ifndef GL_APPLE_ycbcr_422 +#define GL_YCBCR_422_APPLE 0x85B9 +#define GL_UNSIGNED_SHORT_8_8_APPLE 0x85BA +#define GL_UNSIGNED_SHORT_8_8_REV_APPLE 0x85BB +#endif + +#ifndef GL_S3_s3tc +#define GL_RGB_S3TC 0x83A0 +#define GL_RGB4_S3TC 0x83A1 +#define GL_RGBA_S3TC 0x83A2 +#define GL_RGBA4_S3TC 0x83A3 +#endif + +#ifndef GL_ATI_draw_buffers +#define GL_MAX_DRAW_BUFFERS_ATI 0x8824 +#define GL_DRAW_BUFFER0_ATI 0x8825 +#define GL_DRAW_BUFFER1_ATI 0x8826 +#define GL_DRAW_BUFFER2_ATI 0x8827 +#define GL_DRAW_BUFFER3_ATI 0x8828 +#define GL_DRAW_BUFFER4_ATI 0x8829 +#define GL_DRAW_BUFFER5_ATI 0x882A +#define GL_DRAW_BUFFER6_ATI 0x882B +#define GL_DRAW_BUFFER7_ATI 0x882C +#define GL_DRAW_BUFFER8_ATI 0x882D +#define GL_DRAW_BUFFER9_ATI 0x882E +#define GL_DRAW_BUFFER10_ATI 0x882F +#define GL_DRAW_BUFFER11_ATI 0x8830 +#define GL_DRAW_BUFFER12_ATI 0x8831 +#define GL_DRAW_BUFFER13_ATI 0x8832 +#define GL_DRAW_BUFFER14_ATI 0x8833 +#define GL_DRAW_BUFFER15_ATI 0x8834 +#endif + +#ifndef GL_ATI_pixel_format_float +#define GL_TYPE_RGBA_FLOAT_ATI 0x8820 +#define GL_COLOR_CLEAR_UNCLAMPED_VALUE_ATI 0x8835 +#endif + +#ifndef GL_ATI_texture_env_combine3 +#define GL_MODULATE_ADD_ATI 0x8744 +#define GL_MODULATE_SIGNED_ADD_ATI 0x8745 +#define GL_MODULATE_SUBTRACT_ATI 0x8746 +#endif + +#ifndef GL_ATI_texture_float +#define GL_RGBA_FLOAT32_ATI 0x8814 +#define GL_RGB_FLOAT32_ATI 0x8815 +#define GL_ALPHA_FLOAT32_ATI 0x8816 +#define GL_INTENSITY_FLOAT32_ATI 0x8817 +#define GL_LUMINANCE_FLOAT32_ATI 0x8818 +#define GL_LUMINANCE_ALPHA_FLOAT32_ATI 0x8819 +#define GL_RGBA_FLOAT16_ATI 0x881A +#define GL_RGB_FLOAT16_ATI 0x881B +#define GL_ALPHA_FLOAT16_ATI 0x881C +#define GL_INTENSITY_FLOAT16_ATI 0x881D +#define GL_LUMINANCE_FLOAT16_ATI 0x881E +#define GL_LUMINANCE_ALPHA_FLOAT16_ATI 0x881F +#endif + +#ifndef GL_NV_float_buffer +#define GL_FLOAT_R_NV 0x8880 +#define GL_FLOAT_RG_NV 0x8881 +#define GL_FLOAT_RGB_NV 0x8882 +#define GL_FLOAT_RGBA_NV 0x8883 +#define GL_FLOAT_R16_NV 0x8884 +#define GL_FLOAT_R32_NV 0x8885 +#define GL_FLOAT_RG16_NV 0x8886 +#define GL_FLOAT_RG32_NV 0x8887 +#define GL_FLOAT_RGB16_NV 0x8888 +#define GL_FLOAT_RGB32_NV 0x8889 +#define GL_FLOAT_RGBA16_NV 0x888A +#define GL_FLOAT_RGBA32_NV 0x888B +#define GL_TEXTURE_FLOAT_COMPONENTS_NV 0x888C +#define GL_FLOAT_CLEAR_COLOR_VALUE_NV 0x888D +#define GL_FLOAT_RGBA_MODE_NV 0x888E +#endif + +#ifndef GL_NV_fragment_program +#define GL_MAX_FRAGMENT_PROGRAM_LOCAL_PARAMETERS_NV 0x8868 +#define GL_FRAGMENT_PROGRAM_NV 0x8870 +#define GL_MAX_TEXTURE_COORDS_NV 0x8871 +#define GL_MAX_TEXTURE_IMAGE_UNITS_NV 0x8872 +#define GL_FRAGMENT_PROGRAM_BINDING_NV 0x8873 +#define GL_PROGRAM_ERROR_STRING_NV 0x8874 +#endif + +#ifndef GL_NV_half_float +#define GL_HALF_FLOAT_NV 0x140B +#endif + +#ifndef GL_NV_pixel_data_range +#define GL_WRITE_PIXEL_DATA_RANGE_NV 0x8878 +#define GL_READ_PIXEL_DATA_RANGE_NV 0x8879 +#define GL_WRITE_PIXEL_DATA_RANGE_LENGTH_NV 0x887A +#define GL_READ_PIXEL_DATA_RANGE_LENGTH_NV 0x887B +#define GL_WRITE_PIXEL_DATA_RANGE_POINTER_NV 0x887C +#define GL_READ_PIXEL_DATA_RANGE_POINTER_NV 0x887D +#endif + +#ifndef GL_NV_primitive_restart +#define GL_PRIMITIVE_RESTART_NV 0x8558 +#define GL_PRIMITIVE_RESTART_INDEX_NV 0x8559 +#endif + +#ifndef GL_NV_texture_expand_normal +#define GL_TEXTURE_UNSIGNED_REMAP_MODE_NV 0x888F +#endif + +#ifndef GL_NV_vertex_program2 +#endif + +#ifndef GL_ATI_map_object_buffer +#endif + +#ifndef GL_ATI_separate_stencil +#define GL_STENCIL_BACK_FUNC_ATI 0x8800 +#define GL_STENCIL_BACK_FAIL_ATI 0x8801 +#define GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI 0x8802 +#define GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI 0x8803 +#endif + +#ifndef GL_ATI_vertex_attrib_array_object +#endif + +#ifndef GL_OES_read_format +#define GL_IMPLEMENTATION_COLOR_READ_TYPE_OES 0x8B9A +#define GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES 0x8B9B +#endif + +#ifndef GL_EXT_depth_bounds_test +#define GL_DEPTH_BOUNDS_TEST_EXT 0x8890 +#define GL_DEPTH_BOUNDS_EXT 0x8891 +#endif + +#ifndef GL_EXT_texture_mirror_clamp +#define GL_MIRROR_CLAMP_EXT 0x8742 +#define GL_MIRROR_CLAMP_TO_EDGE_EXT 0x8743 +#define GL_MIRROR_CLAMP_TO_BORDER_EXT 0x8912 +#endif + +#ifndef GL_EXT_blend_equation_separate +#define GL_BLEND_EQUATION_RGB_EXT 0x8009 +#define GL_BLEND_EQUATION_ALPHA_EXT 0x883D +#endif + +#ifndef GL_MESA_pack_invert +#define GL_PACK_INVERT_MESA 0x8758 +#endif + +#ifndef GL_MESA_ycbcr_texture +#define GL_UNSIGNED_SHORT_8_8_MESA 0x85BA +#define GL_UNSIGNED_SHORT_8_8_REV_MESA 0x85BB +#define GL_YCBCR_MESA 0x8757 +#endif + +#ifndef GL_EXT_pixel_buffer_object +#define GL_PIXEL_PACK_BUFFER_EXT 0x88EB +#define GL_PIXEL_UNPACK_BUFFER_EXT 0x88EC +#define GL_PIXEL_PACK_BUFFER_BINDING_EXT 0x88ED +#define GL_PIXEL_UNPACK_BUFFER_BINDING_EXT 0x88EF +#endif + +#ifndef GL_NV_fragment_program_option +#endif + +#ifndef GL_NV_fragment_program2 +#define GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV 0x88F4 +#define GL_MAX_PROGRAM_CALL_DEPTH_NV 0x88F5 +#define GL_MAX_PROGRAM_IF_DEPTH_NV 0x88F6 +#define GL_MAX_PROGRAM_LOOP_DEPTH_NV 0x88F7 +#define GL_MAX_PROGRAM_LOOP_COUNT_NV 0x88F8 +#endif + +#ifndef GL_NV_vertex_program2_option +/* reuse GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */ +/* reuse GL_MAX_PROGRAM_CALL_DEPTH_NV */ +#endif + +#ifndef GL_NV_vertex_program3 +/* reuse GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */ +#endif + +#ifndef GL_EXT_framebuffer_object +#define GL_INVALID_FRAMEBUFFER_OPERATION_EXT 0x0506 +#define GL_MAX_RENDERBUFFER_SIZE_EXT 0x84E8 +#define GL_FRAMEBUFFER_BINDING_EXT 0x8CA6 +#define GL_RENDERBUFFER_BINDING_EXT 0x8CA7 +#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT 0x8CD0 +#define GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT 0x8CD1 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT 0x8CD2 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT 0x8CD3 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT 0x8CD4 +#define GL_FRAMEBUFFER_COMPLETE_EXT 0x8CD5 +#define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT 0x8CD6 +#define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT 0x8CD7 +#define GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT 0x8CD9 +#define GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT 0x8CDA +#define GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT 0x8CDB +#define GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT 0x8CDC +#define GL_FRAMEBUFFER_UNSUPPORTED_EXT 0x8CDD +#define GL_MAX_COLOR_ATTACHMENTS_EXT 0x8CDF +#define GL_COLOR_ATTACHMENT0_EXT 0x8CE0 +#define GL_COLOR_ATTACHMENT1_EXT 0x8CE1 +#define GL_COLOR_ATTACHMENT2_EXT 0x8CE2 +#define GL_COLOR_ATTACHMENT3_EXT 0x8CE3 +#define GL_COLOR_ATTACHMENT4_EXT 0x8CE4 +#define GL_COLOR_ATTACHMENT5_EXT 0x8CE5 +#define GL_COLOR_ATTACHMENT6_EXT 0x8CE6 +#define GL_COLOR_ATTACHMENT7_EXT 0x8CE7 +#define GL_COLOR_ATTACHMENT8_EXT 0x8CE8 +#define GL_COLOR_ATTACHMENT9_EXT 0x8CE9 +#define GL_COLOR_ATTACHMENT10_EXT 0x8CEA +#define GL_COLOR_ATTACHMENT11_EXT 0x8CEB +#define GL_COLOR_ATTACHMENT12_EXT 0x8CEC +#define GL_COLOR_ATTACHMENT13_EXT 0x8CED +#define GL_COLOR_ATTACHMENT14_EXT 0x8CEE +#define GL_COLOR_ATTACHMENT15_EXT 0x8CEF +#define GL_DEPTH_ATTACHMENT_EXT 0x8D00 +#define GL_STENCIL_ATTACHMENT_EXT 0x8D20 +#define GL_FRAMEBUFFER_EXT 0x8D40 +#define GL_RENDERBUFFER_EXT 0x8D41 +#define GL_RENDERBUFFER_WIDTH_EXT 0x8D42 +#define GL_RENDERBUFFER_HEIGHT_EXT 0x8D43 +#define GL_RENDERBUFFER_INTERNAL_FORMAT_EXT 0x8D44 +#define GL_STENCIL_INDEX1_EXT 0x8D46 +#define GL_STENCIL_INDEX4_EXT 0x8D47 +#define GL_STENCIL_INDEX8_EXT 0x8D48 +#define GL_STENCIL_INDEX16_EXT 0x8D49 +#define GL_RENDERBUFFER_RED_SIZE_EXT 0x8D50 +#define GL_RENDERBUFFER_GREEN_SIZE_EXT 0x8D51 +#define GL_RENDERBUFFER_BLUE_SIZE_EXT 0x8D52 +#define GL_RENDERBUFFER_ALPHA_SIZE_EXT 0x8D53 +#define GL_RENDERBUFFER_DEPTH_SIZE_EXT 0x8D54 +#define GL_RENDERBUFFER_STENCIL_SIZE_EXT 0x8D55 +#endif + +#ifndef GL_GREMEDY_string_marker +#endif + +#ifndef GL_EXT_packed_depth_stencil +#define GL_DEPTH_STENCIL_EXT 0x84F9 +#define GL_UNSIGNED_INT_24_8_EXT 0x84FA +#define GL_DEPTH24_STENCIL8_EXT 0x88F0 +#define GL_TEXTURE_STENCIL_SIZE_EXT 0x88F1 +#endif + +#ifndef GL_EXT_stencil_clear_tag +#define GL_STENCIL_TAG_BITS_EXT 0x88F2 +#define GL_STENCIL_CLEAR_TAG_VALUE_EXT 0x88F3 +#endif + +#ifndef GL_EXT_texture_sRGB +#define GL_SRGB_EXT 0x8C40 +#define GL_SRGB8_EXT 0x8C41 +#define GL_SRGB_ALPHA_EXT 0x8C42 +#define GL_SRGB8_ALPHA8_EXT 0x8C43 +#define GL_SLUMINANCE_ALPHA_EXT 0x8C44 +#define GL_SLUMINANCE8_ALPHA8_EXT 0x8C45 +#define GL_SLUMINANCE_EXT 0x8C46 +#define GL_SLUMINANCE8_EXT 0x8C47 +#define GL_COMPRESSED_SRGB_EXT 0x8C48 +#define GL_COMPRESSED_SRGB_ALPHA_EXT 0x8C49 +#define GL_COMPRESSED_SLUMINANCE_EXT 0x8C4A +#define GL_COMPRESSED_SLUMINANCE_ALPHA_EXT 0x8C4B +#define GL_COMPRESSED_SRGB_S3TC_DXT1_EXT 0x8C4C +#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT 0x8C4D +#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT 0x8C4E +#define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT 0x8C4F +#endif + +#ifndef GL_EXT_framebuffer_blit +#define GL_READ_FRAMEBUFFER_EXT 0x8CA8 +#define GL_DRAW_FRAMEBUFFER_EXT 0x8CA9 +#define GL_DRAW_FRAMEBUFFER_BINDING_EXT GL_FRAMEBUFFER_BINDING_EXT +#define GL_READ_FRAMEBUFFER_BINDING_EXT 0x8CAA +#endif + +#ifndef GL_EXT_framebuffer_multisample +#define GL_RENDERBUFFER_SAMPLES_EXT 0x8CAB +#define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT 0x8D56 +#define GL_MAX_SAMPLES_EXT 0x8D57 +#endif + +#ifndef GL_MESAX_texture_stack +#define GL_TEXTURE_1D_STACK_MESAX 0x8759 +#define GL_TEXTURE_2D_STACK_MESAX 0x875A +#define GL_PROXY_TEXTURE_1D_STACK_MESAX 0x875B +#define GL_PROXY_TEXTURE_2D_STACK_MESAX 0x875C +#define GL_TEXTURE_1D_STACK_BINDING_MESAX 0x875D +#define GL_TEXTURE_2D_STACK_BINDING_MESAX 0x875E +#endif + +#ifndef GL_EXT_timer_query +#define GL_TIME_ELAPSED_EXT 0x88BF +#endif + +#ifndef GL_EXT_gpu_program_parameters +#endif + +#ifndef GL_APPLE_flush_buffer_range +#define GL_BUFFER_SERIALIZED_MODIFY_APPLE 0x8A12 +#define GL_BUFFER_FLUSHING_UNMAP_APPLE 0x8A13 +#endif + +#ifndef GL_NV_gpu_program4 +#define GL_MIN_PROGRAM_TEXEL_OFFSET_NV 0x8904 +#define GL_MAX_PROGRAM_TEXEL_OFFSET_NV 0x8905 +#define GL_PROGRAM_ATTRIB_COMPONENTS_NV 0x8906 +#define GL_PROGRAM_RESULT_COMPONENTS_NV 0x8907 +#define GL_MAX_PROGRAM_ATTRIB_COMPONENTS_NV 0x8908 +#define GL_MAX_PROGRAM_RESULT_COMPONENTS_NV 0x8909 +#define GL_MAX_PROGRAM_GENERIC_ATTRIBS_NV 0x8DA5 +#define GL_MAX_PROGRAM_GENERIC_RESULTS_NV 0x8DA6 +#endif + +#ifndef GL_NV_geometry_program4 +#define GL_LINES_ADJACENCY_EXT 0x000A +#define GL_LINE_STRIP_ADJACENCY_EXT 0x000B +#define GL_TRIANGLES_ADJACENCY_EXT 0x000C +#define GL_TRIANGLE_STRIP_ADJACENCY_EXT 0x000D +#define GL_GEOMETRY_PROGRAM_NV 0x8C26 +#define GL_MAX_PROGRAM_OUTPUT_VERTICES_NV 0x8C27 +#define GL_MAX_PROGRAM_TOTAL_OUTPUT_COMPONENTS_NV 0x8C28 +#define GL_GEOMETRY_VERTICES_OUT_EXT 0x8DDA +#define GL_GEOMETRY_INPUT_TYPE_EXT 0x8DDB +#define GL_GEOMETRY_OUTPUT_TYPE_EXT 0x8DDC +#define GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_EXT 0x8C29 +#define GL_FRAMEBUFFER_ATTACHMENT_LAYERED_EXT 0x8DA7 +#define GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_EXT 0x8DA8 +#define GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_EXT 0x8DA9 +#define GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT 0x8CD4 +#define GL_PROGRAM_POINT_SIZE_EXT 0x8642 +#endif + +#ifndef GL_EXT_geometry_shader4 +#define GL_GEOMETRY_SHADER_EXT 0x8DD9 +/* reuse GL_GEOMETRY_VERTICES_OUT_EXT */ +/* reuse GL_GEOMETRY_INPUT_TYPE_EXT */ +/* reuse GL_GEOMETRY_OUTPUT_TYPE_EXT */ +/* reuse GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_EXT */ +#define GL_MAX_GEOMETRY_VARYING_COMPONENTS_EXT 0x8DDD +#define GL_MAX_VERTEX_VARYING_COMPONENTS_EXT 0x8DDE +#define GL_MAX_VARYING_COMPONENTS_EXT 0x8B4B +#define GL_MAX_GEOMETRY_UNIFORM_COMPONENTS_EXT 0x8DDF +#define GL_MAX_GEOMETRY_OUTPUT_VERTICES_EXT 0x8DE0 +#define GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS_EXT 0x8DE1 +/* reuse GL_LINES_ADJACENCY_EXT */ +/* reuse GL_LINE_STRIP_ADJACENCY_EXT */ +/* reuse GL_TRIANGLES_ADJACENCY_EXT */ +/* reuse GL_TRIANGLE_STRIP_ADJACENCY_EXT */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_EXT */ +/* reuse GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_EXT */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_LAYERED_EXT */ +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */ +/* reuse GL_PROGRAM_POINT_SIZE_EXT */ +#endif + +#ifndef GL_NV_vertex_program4 +#define GL_VERTEX_ATTRIB_ARRAY_INTEGER_NV 0x88FD +#endif + +#ifndef GL_EXT_gpu_shader4 +#define GL_SAMPLER_1D_ARRAY_EXT 0x8DC0 +#define GL_SAMPLER_2D_ARRAY_EXT 0x8DC1 +#define GL_SAMPLER_BUFFER_EXT 0x8DC2 +#define GL_SAMPLER_1D_ARRAY_SHADOW_EXT 0x8DC3 +#define GL_SAMPLER_2D_ARRAY_SHADOW_EXT 0x8DC4 +#define GL_SAMPLER_CUBE_SHADOW_EXT 0x8DC5 +#define GL_UNSIGNED_INT_VEC2_EXT 0x8DC6 +#define GL_UNSIGNED_INT_VEC3_EXT 0x8DC7 +#define GL_UNSIGNED_INT_VEC4_EXT 0x8DC8 +#define GL_INT_SAMPLER_1D_EXT 0x8DC9 +#define GL_INT_SAMPLER_2D_EXT 0x8DCA +#define GL_INT_SAMPLER_3D_EXT 0x8DCB +#define GL_INT_SAMPLER_CUBE_EXT 0x8DCC +#define GL_INT_SAMPLER_2D_RECT_EXT 0x8DCD +#define GL_INT_SAMPLER_1D_ARRAY_EXT 0x8DCE +#define GL_INT_SAMPLER_2D_ARRAY_EXT 0x8DCF +#define GL_INT_SAMPLER_BUFFER_EXT 0x8DD0 +#define GL_UNSIGNED_INT_SAMPLER_1D_EXT 0x8DD1 +#define GL_UNSIGNED_INT_SAMPLER_2D_EXT 0x8DD2 +#define GL_UNSIGNED_INT_SAMPLER_3D_EXT 0x8DD3 +#define GL_UNSIGNED_INT_SAMPLER_CUBE_EXT 0x8DD4 +#define GL_UNSIGNED_INT_SAMPLER_2D_RECT_EXT 0x8DD5 +#define GL_UNSIGNED_INT_SAMPLER_1D_ARRAY_EXT 0x8DD6 +#define GL_UNSIGNED_INT_SAMPLER_2D_ARRAY_EXT 0x8DD7 +#define GL_UNSIGNED_INT_SAMPLER_BUFFER_EXT 0x8DD8 +#endif + +#ifndef GL_EXT_draw_instanced +#endif + +#ifndef GL_EXT_packed_float +#define GL_R11F_G11F_B10F_EXT 0x8C3A +#define GL_UNSIGNED_INT_10F_11F_11F_REV_EXT 0x8C3B +#define GL_RGBA_SIGNED_COMPONENTS_EXT 0x8C3C +#endif + +#ifndef GL_EXT_texture_array +#define GL_TEXTURE_1D_ARRAY_EXT 0x8C18 +#define GL_PROXY_TEXTURE_1D_ARRAY_EXT 0x8C19 +#define GL_TEXTURE_2D_ARRAY_EXT 0x8C1A +#define GL_PROXY_TEXTURE_2D_ARRAY_EXT 0x8C1B +#define GL_TEXTURE_BINDING_1D_ARRAY_EXT 0x8C1C +#define GL_TEXTURE_BINDING_2D_ARRAY_EXT 0x8C1D +#define GL_MAX_ARRAY_TEXTURE_LAYERS_EXT 0x88FF +#define GL_COMPARE_REF_DEPTH_TO_TEXTURE_EXT 0x884E +/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */ +#endif + +#ifndef GL_EXT_texture_buffer_object +#define GL_TEXTURE_BUFFER_EXT 0x8C2A +#define GL_MAX_TEXTURE_BUFFER_SIZE_EXT 0x8C2B +#define GL_TEXTURE_BINDING_BUFFER_EXT 0x8C2C +#define GL_TEXTURE_BUFFER_DATA_STORE_BINDING_EXT 0x8C2D +#define GL_TEXTURE_BUFFER_FORMAT_EXT 0x8C2E +#endif + +#ifndef GL_EXT_texture_compression_latc +#define GL_COMPRESSED_LUMINANCE_LATC1_EXT 0x8C70 +#define GL_COMPRESSED_SIGNED_LUMINANCE_LATC1_EXT 0x8C71 +#define GL_COMPRESSED_LUMINANCE_ALPHA_LATC2_EXT 0x8C72 +#define GL_COMPRESSED_SIGNED_LUMINANCE_ALPHA_LATC2_EXT 0x8C73 +#endif + +#ifndef GL_EXT_texture_compression_rgtc +#define GL_COMPRESSED_RED_RGTC1_EXT 0x8DBB +#define GL_COMPRESSED_SIGNED_RED_RGTC1_EXT 0x8DBC +#define GL_COMPRESSED_RED_GREEN_RGTC2_EXT 0x8DBD +#define GL_COMPRESSED_SIGNED_RED_GREEN_RGTC2_EXT 0x8DBE +#endif + +#ifndef GL_EXT_texture_shared_exponent +#define GL_RGB9_E5_EXT 0x8C3D +#define GL_UNSIGNED_INT_5_9_9_9_REV_EXT 0x8C3E +#define GL_TEXTURE_SHARED_SIZE_EXT 0x8C3F +#endif + +#ifndef GL_NV_depth_buffer_float +#define GL_DEPTH_COMPONENT32F_NV 0x8DAB +#define GL_DEPTH32F_STENCIL8_NV 0x8DAC +#define GL_FLOAT_32_UNSIGNED_INT_24_8_REV_NV 0x8DAD +#define GL_DEPTH_BUFFER_FLOAT_MODE_NV 0x8DAF +#endif + +#ifndef GL_NV_fragment_program4 +#endif + +#ifndef GL_NV_framebuffer_multisample_coverage +#define GL_RENDERBUFFER_COVERAGE_SAMPLES_NV 0x8CAB +#define GL_RENDERBUFFER_COLOR_SAMPLES_NV 0x8E10 +#define GL_MAX_MULTISAMPLE_COVERAGE_MODES_NV 0x8E11 +#define GL_MULTISAMPLE_COVERAGE_MODES_NV 0x8E12 +#endif + +#ifndef GL_EXT_framebuffer_sRGB +#define GL_FRAMEBUFFER_SRGB_EXT 0x8DB9 +#define GL_FRAMEBUFFER_SRGB_CAPABLE_EXT 0x8DBA +#endif + +#ifndef GL_NV_geometry_shader4 +#endif + +#ifndef GL_NV_parameter_buffer_object +#define GL_MAX_PROGRAM_PARAMETER_BUFFER_BINDINGS_NV 0x8DA0 +#define GL_MAX_PROGRAM_PARAMETER_BUFFER_SIZE_NV 0x8DA1 +#define GL_VERTEX_PROGRAM_PARAMETER_BUFFER_NV 0x8DA2 +#define GL_GEOMETRY_PROGRAM_PARAMETER_BUFFER_NV 0x8DA3 +#define GL_FRAGMENT_PROGRAM_PARAMETER_BUFFER_NV 0x8DA4 +#endif + +#ifndef GL_EXT_draw_buffers2 +#endif + +#ifndef GL_NV_transform_feedback +#define GL_BACK_PRIMARY_COLOR_NV 0x8C77 +#define GL_BACK_SECONDARY_COLOR_NV 0x8C78 +#define GL_TEXTURE_COORD_NV 0x8C79 +#define GL_CLIP_DISTANCE_NV 0x8C7A +#define GL_VERTEX_ID_NV 0x8C7B +#define GL_PRIMITIVE_ID_NV 0x8C7C +#define GL_GENERIC_ATTRIB_NV 0x8C7D +#define GL_TRANSFORM_FEEDBACK_ATTRIBS_NV 0x8C7E +#define GL_TRANSFORM_FEEDBACK_BUFFER_MODE_NV 0x8C7F +#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS_NV 0x8C80 +#define GL_ACTIVE_VARYINGS_NV 0x8C81 +#define GL_ACTIVE_VARYING_MAX_LENGTH_NV 0x8C82 +#define GL_TRANSFORM_FEEDBACK_VARYINGS_NV 0x8C83 +#define GL_TRANSFORM_FEEDBACK_BUFFER_START_NV 0x8C84 +#define GL_TRANSFORM_FEEDBACK_BUFFER_SIZE_NV 0x8C85 +#define GL_TRANSFORM_FEEDBACK_RECORD_NV 0x8C86 +#define GL_PRIMITIVES_GENERATED_NV 0x8C87 +#define GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN_NV 0x8C88 +#define GL_RASTERIZER_DISCARD_NV 0x8C89 +#define GL_MAX_TRANSFORM_FEEDBACK_INTERLEAVED_ATTRIBS_NV 0x8C8A +#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS_NV 0x8C8B +#define GL_INTERLEAVED_ATTRIBS_NV 0x8C8C +#define GL_SEPARATE_ATTRIBS_NV 0x8C8D +#define GL_TRANSFORM_FEEDBACK_BUFFER_NV 0x8C8E +#define GL_TRANSFORM_FEEDBACK_BUFFER_BINDING_NV 0x8C8F +#endif + +#ifndef GL_EXT_bindable_uniform +#define GL_MAX_VERTEX_BINDABLE_UNIFORMS_EXT 0x8DE2 +#define GL_MAX_FRAGMENT_BINDABLE_UNIFORMS_EXT 0x8DE3 +#define GL_MAX_GEOMETRY_BINDABLE_UNIFORMS_EXT 0x8DE4 +#define GL_MAX_BINDABLE_UNIFORM_SIZE_EXT 0x8DED +#define GL_UNIFORM_BUFFER_EXT 0x8DEE +#define GL_UNIFORM_BUFFER_BINDING_EXT 0x8DEF +#endif + +#ifndef GL_EXT_texture_integer +#define GL_RGBA32UI_EXT 0x8D70 +#define GL_RGB32UI_EXT 0x8D71 +#define GL_ALPHA32UI_EXT 0x8D72 +#define GL_INTENSITY32UI_EXT 0x8D73 +#define GL_LUMINANCE32UI_EXT 0x8D74 +#define GL_LUMINANCE_ALPHA32UI_EXT 0x8D75 +#define GL_RGBA16UI_EXT 0x8D76 +#define GL_RGB16UI_EXT 0x8D77 +#define GL_ALPHA16UI_EXT 0x8D78 +#define GL_INTENSITY16UI_EXT 0x8D79 +#define GL_LUMINANCE16UI_EXT 0x8D7A +#define GL_LUMINANCE_ALPHA16UI_EXT 0x8D7B +#define GL_RGBA8UI_EXT 0x8D7C +#define GL_RGB8UI_EXT 0x8D7D +#define GL_ALPHA8UI_EXT 0x8D7E +#define GL_INTENSITY8UI_EXT 0x8D7F +#define GL_LUMINANCE8UI_EXT 0x8D80 +#define GL_LUMINANCE_ALPHA8UI_EXT 0x8D81 +#define GL_RGBA32I_EXT 0x8D82 +#define GL_RGB32I_EXT 0x8D83 +#define GL_ALPHA32I_EXT 0x8D84 +#define GL_INTENSITY32I_EXT 0x8D85 +#define GL_LUMINANCE32I_EXT 0x8D86 +#define GL_LUMINANCE_ALPHA32I_EXT 0x8D87 +#define GL_RGBA16I_EXT 0x8D88 +#define GL_RGB16I_EXT 0x8D89 +#define GL_ALPHA16I_EXT 0x8D8A +#define GL_INTENSITY16I_EXT 0x8D8B +#define GL_LUMINANCE16I_EXT 0x8D8C +#define GL_LUMINANCE_ALPHA16I_EXT 0x8D8D +#define GL_RGBA8I_EXT 0x8D8E +#define GL_RGB8I_EXT 0x8D8F +#define GL_ALPHA8I_EXT 0x8D90 +#define GL_INTENSITY8I_EXT 0x8D91 +#define GL_LUMINANCE8I_EXT 0x8D92 +#define GL_LUMINANCE_ALPHA8I_EXT 0x8D93 +#define GL_RED_INTEGER_EXT 0x8D94 +#define GL_GREEN_INTEGER_EXT 0x8D95 +#define GL_BLUE_INTEGER_EXT 0x8D96 +#define GL_ALPHA_INTEGER_EXT 0x8D97 +#define GL_RGB_INTEGER_EXT 0x8D98 +#define GL_RGBA_INTEGER_EXT 0x8D99 +#define GL_BGR_INTEGER_EXT 0x8D9A +#define GL_BGRA_INTEGER_EXT 0x8D9B +#define GL_LUMINANCE_INTEGER_EXT 0x8D9C +#define GL_LUMINANCE_ALPHA_INTEGER_EXT 0x8D9D +#define GL_RGBA_INTEGER_MODE_EXT 0x8D9E +#endif + +#ifndef GL_GREMEDY_frame_terminator +#endif + +#ifndef GL_NV_conditional_render +#define GL_QUERY_WAIT_NV 0x8E13 +#define GL_QUERY_NO_WAIT_NV 0x8E14 +#define GL_QUERY_BY_REGION_WAIT_NV 0x8E15 +#define GL_QUERY_BY_REGION_NO_WAIT_NV 0x8E16 +#endif + +#ifndef GL_NV_present_video +#define GL_FRAME_NV 0x8E26 +#define GL_FIELDS_NV 0x8E27 +#define GL_CURRENT_TIME_NV 0x8E28 +#define GL_NUM_FILL_STREAMS_NV 0x8E29 +#define GL_PRESENT_TIME_NV 0x8E2A +#define GL_PRESENT_DURATION_NV 0x8E2B +#endif + +#ifndef GL_EXT_transform_feedback +#define GL_TRANSFORM_FEEDBACK_BUFFER_EXT 0x8C8E +#define GL_TRANSFORM_FEEDBACK_BUFFER_START_EXT 0x8C84 +#define GL_TRANSFORM_FEEDBACK_BUFFER_SIZE_EXT 0x8C85 +#define GL_TRANSFORM_FEEDBACK_BUFFER_BINDING_EXT 0x8C8F +#define GL_INTERLEAVED_ATTRIBS_EXT 0x8C8C +#define GL_SEPARATE_ATTRIBS_EXT 0x8C8D +#define GL_PRIMITIVES_GENERATED_EXT 0x8C87 +#define GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN_EXT 0x8C88 +#define GL_RASTERIZER_DISCARD_EXT 0x8C89 +#define GL_MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS_EXT 0x8C8A +#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS_EXT 0x8C8B +#define GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_COMPONENTS_EXT 0x8C80 +#define GL_TRANSFORM_FEEDBACK_VARYINGS_EXT 0x8C83 +#define GL_TRANSFORM_FEEDBACK_BUFFER_MODE_EXT 0x8C7F +#define GL_TRANSFORM_FEEDBACK_VARYING_MAX_LENGTH_EXT 0x8C76 +#endif + +#ifndef GL_EXT_direct_state_access +#define GL_PROGRAM_MATRIX_EXT 0x8E2D +#define GL_TRANSPOSE_PROGRAM_MATRIX_EXT 0x8E2E +#define GL_PROGRAM_MATRIX_STACK_DEPTH_EXT 0x8E2F +#endif + +#ifndef GL_EXT_vertex_array_bgra +/* reuse GL_BGRA */ +#endif + +#ifndef GL_EXT_texture_swizzle +#define GL_TEXTURE_SWIZZLE_R_EXT 0x8E42 +#define GL_TEXTURE_SWIZZLE_G_EXT 0x8E43 +#define GL_TEXTURE_SWIZZLE_B_EXT 0x8E44 +#define GL_TEXTURE_SWIZZLE_A_EXT 0x8E45 +#define GL_TEXTURE_SWIZZLE_RGBA_EXT 0x8E46 +#endif + +#ifndef GL_NV_explicit_multisample +#define GL_SAMPLE_POSITION_NV 0x8E50 +#define GL_SAMPLE_MASK_NV 0x8E51 +#define GL_SAMPLE_MASK_VALUE_NV 0x8E52 +#define GL_TEXTURE_BINDING_RENDERBUFFER_NV 0x8E53 +#define GL_TEXTURE_RENDERBUFFER_DATA_STORE_BINDING_NV 0x8E54 +#define GL_TEXTURE_RENDERBUFFER_NV 0x8E55 +#define GL_SAMPLER_RENDERBUFFER_NV 0x8E56 +#define GL_INT_SAMPLER_RENDERBUFFER_NV 0x8E57 +#define GL_UNSIGNED_INT_SAMPLER_RENDERBUFFER_NV 0x8E58 +#define GL_MAX_SAMPLE_MASK_WORDS_NV 0x8E59 +#endif + +#ifndef GL_NV_transform_feedback2 +#define GL_TRANSFORM_FEEDBACK_NV 0x8E22 +#define GL_TRANSFORM_FEEDBACK_BUFFER_PAUSED_NV 0x8E23 +#define GL_TRANSFORM_FEEDBACK_BUFFER_ACTIVE_NV 0x8E24 +#define GL_TRANSFORM_FEEDBACK_BINDING_NV 0x8E25 +#endif + +#ifndef GL_ATI_meminfo +#define GL_VBO_FREE_MEMORY_ATI 0x87FB +#define GL_TEXTURE_FREE_MEMORY_ATI 0x87FC +#define GL_RENDERBUFFER_FREE_MEMORY_ATI 0x87FD +#endif + +#ifndef GL_AMD_performance_monitor +#define GL_COUNTER_TYPE_AMD 0x8BC0 +#define GL_COUNTER_RANGE_AMD 0x8BC1 +#define GL_UNSIGNED_INT64_AMD 0x8BC2 +#define GL_PERCENTAGE_AMD 0x8BC3 +#define GL_PERFMON_RESULT_AVAILABLE_AMD 0x8BC4 +#define GL_PERFMON_RESULT_SIZE_AMD 0x8BC5 +#define GL_PERFMON_RESULT_AMD 0x8BC6 +#endif + +#ifndef GL_AMD_texture_texture4 +#endif + +#ifndef GL_AMD_vertex_shader_tesselator +#define GL_SAMPLER_BUFFER_AMD 0x9001 +#define GL_INT_SAMPLER_BUFFER_AMD 0x9002 +#define GL_UNSIGNED_INT_SAMPLER_BUFFER_AMD 0x9003 +#define GL_TESSELLATION_MODE_AMD 0x9004 +#define GL_TESSELLATION_FACTOR_AMD 0x9005 +#define GL_DISCRETE_AMD 0x9006 +#define GL_CONTINUOUS_AMD 0x9007 +#endif + +#ifndef GL_EXT_provoking_vertex +#define GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT 0x8E4C +#define GL_FIRST_VERTEX_CONVENTION_EXT 0x8E4D +#define GL_LAST_VERTEX_CONVENTION_EXT 0x8E4E +#define GL_PROVOKING_VERTEX_EXT 0x8E4F +#endif + +#ifndef GL_EXT_texture_snorm +#define GL_ALPHA_SNORM 0x9010 +#define GL_LUMINANCE_SNORM 0x9011 +#define GL_LUMINANCE_ALPHA_SNORM 0x9012 +#define GL_INTENSITY_SNORM 0x9013 +#define GL_ALPHA8_SNORM 0x9014 +#define GL_LUMINANCE8_SNORM 0x9015 +#define GL_LUMINANCE8_ALPHA8_SNORM 0x9016 +#define GL_INTENSITY8_SNORM 0x9017 +#define GL_ALPHA16_SNORM 0x9018 +#define GL_LUMINANCE16_SNORM 0x9019 +#define GL_LUMINANCE16_ALPHA16_SNORM 0x901A +#define GL_INTENSITY16_SNORM 0x901B +/* reuse GL_R_SNORM */ +/* reuse GL_RG_SNORM */ +/* reuse GL_RGB_SNORM */ +/* reuse GL_RGBA_SNORM */ +/* reuse GL_R8_SNORM */ +/* reuse GL_RG8_SNORM */ +/* reuse GL_RGB8_SNORM */ +/* reuse GL_RGBA8_SNORM */ +/* reuse GL_R16_SNORM */ +/* reuse GL_RG16_SNORM */ +/* reuse GL_RGB16_SNORM */ +/* reuse GL_RGBA16_SNORM */ +/* reuse GL_SIGNED_NORMALIZED */ +#endif + +#ifndef GL_AMD_draw_buffers_blend +#endif + +#ifndef GL_APPLE_texture_range +#define GL_TEXTURE_RANGE_LENGTH_APPLE 0x85B7 +#define GL_TEXTURE_RANGE_POINTER_APPLE 0x85B8 +#define GL_TEXTURE_STORAGE_HINT_APPLE 0x85BC +#define GL_STORAGE_PRIVATE_APPLE 0x85BD +/* reuse GL_STORAGE_CACHED_APPLE */ +/* reuse GL_STORAGE_SHARED_APPLE */ +#endif + +#ifndef GL_APPLE_float_pixels +#define GL_HALF_APPLE 0x140B +#define GL_RGBA_FLOAT32_APPLE 0x8814 +#define GL_RGB_FLOAT32_APPLE 0x8815 +#define GL_ALPHA_FLOAT32_APPLE 0x8816 +#define GL_INTENSITY_FLOAT32_APPLE 0x8817 +#define GL_LUMINANCE_FLOAT32_APPLE 0x8818 +#define GL_LUMINANCE_ALPHA_FLOAT32_APPLE 0x8819 +#define GL_RGBA_FLOAT16_APPLE 0x881A +#define GL_RGB_FLOAT16_APPLE 0x881B +#define GL_ALPHA_FLOAT16_APPLE 0x881C +#define GL_INTENSITY_FLOAT16_APPLE 0x881D +#define GL_LUMINANCE_FLOAT16_APPLE 0x881E +#define GL_LUMINANCE_ALPHA_FLOAT16_APPLE 0x881F +#define GL_COLOR_FLOAT_APPLE 0x8A0F +#endif + +#ifndef GL_APPLE_vertex_program_evaluators +#define GL_VERTEX_ATTRIB_MAP1_APPLE 0x8A00 +#define GL_VERTEX_ATTRIB_MAP2_APPLE 0x8A01 +#define GL_VERTEX_ATTRIB_MAP1_SIZE_APPLE 0x8A02 +#define GL_VERTEX_ATTRIB_MAP1_COEFF_APPLE 0x8A03 +#define GL_VERTEX_ATTRIB_MAP1_ORDER_APPLE 0x8A04 +#define GL_VERTEX_ATTRIB_MAP1_DOMAIN_APPLE 0x8A05 +#define GL_VERTEX_ATTRIB_MAP2_SIZE_APPLE 0x8A06 +#define GL_VERTEX_ATTRIB_MAP2_COEFF_APPLE 0x8A07 +#define GL_VERTEX_ATTRIB_MAP2_ORDER_APPLE 0x8A08 +#define GL_VERTEX_ATTRIB_MAP2_DOMAIN_APPLE 0x8A09 +#endif + +#ifndef GL_APPLE_aux_depth_stencil +#define GL_AUX_DEPTH_STENCIL_APPLE 0x8A14 +#endif + +#ifndef GL_APPLE_object_purgeable +#define GL_BUFFER_OBJECT_APPLE 0x85B3 +#define GL_RELEASED_APPLE 0x8A19 +#define GL_VOLATILE_APPLE 0x8A1A +#define GL_RETAINED_APPLE 0x8A1B +#define GL_UNDEFINED_APPLE 0x8A1C +#define GL_PURGEABLE_APPLE 0x8A1D +#endif + +#ifndef GL_APPLE_row_bytes +#define GL_PACK_ROW_BYTES_APPLE 0x8A15 +#define GL_UNPACK_ROW_BYTES_APPLE 0x8A16 +#endif + + +/*************************************************************/ + +#include +#ifndef GL_VERSION_2_0 +/* GL type for program/shader text */ +typedef char GLchar; +#endif + +#ifndef GL_VERSION_1_5 +/* GL types for handling large vertex buffer objects */ +typedef ptrdiff_t GLintptr; +typedef ptrdiff_t GLsizeiptr; +#endif + +#ifndef GL_ARB_vertex_buffer_object +/* GL types for handling large vertex buffer objects */ +typedef ptrdiff_t GLintptrARB; +typedef ptrdiff_t GLsizeiptrARB; +#endif + +#ifndef GL_ARB_shader_objects +/* GL types for program/shader text and shader object handles */ +typedef char GLcharARB; +typedef unsigned int GLhandleARB; +#endif + +/* GL type for "half" precision (s10e5) float data in host memory */ +#ifndef GL_ARB_half_float_pixel +typedef unsigned short GLhalfARB; +#endif + +#ifndef GL_NV_half_float +typedef unsigned short GLhalfNV; +#endif + +#ifndef GLEXT_64_TYPES_DEFINED +/* This code block is duplicated in glxext.h, so must be protected */ +#define GLEXT_64_TYPES_DEFINED +/* Define int32_t, int64_t, and uint64_t types for UST/MSC */ +/* (as used in the GL_EXT_timer_query extension). */ +#if defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L +#include +#elif defined(__sun__) || defined(__digital__) +#include +#if defined(__STDC__) +#if defined(__arch64__) || defined(_LP64) +typedef long int int64_t; +typedef unsigned long int uint64_t; +#else +typedef long long int int64_t; +typedef unsigned long long int uint64_t; +#endif /* __arch64__ */ +#endif /* __STDC__ */ +#elif defined( __VMS ) || defined(__sgi) +#include +#elif defined(__SCO__) || defined(__USLC__) +#include +#elif defined(__UNIXOS2__) || defined(__SOL64__) +typedef long int int32_t; +typedef long long int int64_t; +typedef unsigned long long int uint64_t; +#elif defined(_WIN32) && defined(__GNUC__) +#include +#elif defined(_WIN32) +typedef __int32 int32_t; +typedef __int64 int64_t; +typedef unsigned __int64 uint64_t; +#else +/* Fallback if nothing above works */ +#include +#endif +#endif + +#ifndef GL_EXT_timer_query +typedef int64_t GLint64EXT; +typedef uint64_t GLuint64EXT; +#endif + +#ifndef ARB_sync +typedef int64_t GLint64; +typedef uint64_t GLuint64; +typedef struct __GLsync *GLsync; +#endif + +#ifndef GL_VERSION_1_2 +#define GL_VERSION_1_2 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendColor (GLclampf, GLclampf, GLclampf, GLclampf); +GLAPI void APIENTRY glBlendEquation (GLenum); +GLAPI void APIENTRY glDrawRangeElements (GLenum, GLuint, GLuint, GLsizei, GLenum, const GLvoid *); +GLAPI void APIENTRY glTexImage3D (GLenum, GLint, GLint, GLsizei, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTexSubImage3D (GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyTexSubImage3D (GLenum, GLint, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDCOLORPROC) (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha); +typedef void (APIENTRYP PFNGLBLENDEQUATIONPROC) (GLenum mode); +typedef void (APIENTRYP PFNGLDRAWRANGEELEMENTSPROC) (GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid *indices); +typedef void (APIENTRYP PFNGLTEXIMAGE3DPROC) (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXSUBIMAGE3DPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLCOPYTEXSUBIMAGE3DPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height); +#endif + +#ifndef GL_VERSION_1_2_DEPRECATED +#define GL_VERSION_1_2_DEPRECATED 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorTable (GLenum, GLenum, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glColorTableParameterfv (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glColorTableParameteriv (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glCopyColorTable (GLenum, GLenum, GLint, GLint, GLsizei); +GLAPI void APIENTRY glGetColorTable (GLenum, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetColorTableParameterfv (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetColorTableParameteriv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glColorSubTable (GLenum, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyColorSubTable (GLenum, GLsizei, GLint, GLint, GLsizei); +GLAPI void APIENTRY glConvolutionFilter1D (GLenum, GLenum, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glConvolutionFilter2D (GLenum, GLenum, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glConvolutionParameterf (GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glConvolutionParameterfv (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glConvolutionParameteri (GLenum, GLenum, GLint); +GLAPI void APIENTRY glConvolutionParameteriv (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glCopyConvolutionFilter1D (GLenum, GLenum, GLint, GLint, GLsizei); +GLAPI void APIENTRY glCopyConvolutionFilter2D (GLenum, GLenum, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glGetConvolutionFilter (GLenum, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetConvolutionParameterfv (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetConvolutionParameteriv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetSeparableFilter (GLenum, GLenum, GLenum, GLvoid *, GLvoid *, GLvoid *); +GLAPI void APIENTRY glSeparableFilter2D (GLenum, GLenum, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *, const GLvoid *); +GLAPI void APIENTRY glGetHistogram (GLenum, GLboolean, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetHistogramParameterfv (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetHistogramParameteriv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetMinmax (GLenum, GLboolean, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetMinmaxParameterfv (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMinmaxParameteriv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glHistogram (GLenum, GLsizei, GLenum, GLboolean); +GLAPI void APIENTRY glMinmax (GLenum, GLenum, GLboolean); +GLAPI void APIENTRY glResetHistogram (GLenum); +GLAPI void APIENTRY glResetMinmax (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLORTABLEPROC) (GLenum target, GLenum internalformat, GLsizei width, GLenum format, GLenum type, const GLvoid *table); +typedef void (APIENTRYP PFNGLCOLORTABLEPARAMETERFVPROC) (GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLCOLORTABLEPARAMETERIVPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLCOPYCOLORTABLEPROC) (GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPROC) (GLenum target, GLenum format, GLenum type, GLvoid *table); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPARAMETERFVPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPARAMETERIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLCOLORSUBTABLEPROC) (GLenum target, GLsizei start, GLsizei count, GLenum format, GLenum type, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOPYCOLORSUBTABLEPROC) (GLenum target, GLsizei start, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLCONVOLUTIONFILTER1DPROC) (GLenum target, GLenum internalformat, GLsizei width, GLenum format, GLenum type, const GLvoid *image); +typedef void (APIENTRYP PFNGLCONVOLUTIONFILTER2DPROC) (GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *image); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERFPROC) (GLenum target, GLenum pname, GLfloat params); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERFVPROC) (GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERIPROC) (GLenum target, GLenum pname, GLint params); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERIVPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLCOPYCONVOLUTIONFILTER1DPROC) (GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLCOPYCONVOLUTIONFILTER2DPROC) (GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETCONVOLUTIONFILTERPROC) (GLenum target, GLenum format, GLenum type, GLvoid *image); +typedef void (APIENTRYP PFNGLGETCONVOLUTIONPARAMETERFVPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETCONVOLUTIONPARAMETERIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETSEPARABLEFILTERPROC) (GLenum target, GLenum format, GLenum type, GLvoid *row, GLvoid *column, GLvoid *span); +typedef void (APIENTRYP PFNGLSEPARABLEFILTER2DPROC) (GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *row, const GLvoid *column); +typedef void (APIENTRYP PFNGLGETHISTOGRAMPROC) (GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values); +typedef void (APIENTRYP PFNGLGETHISTOGRAMPARAMETERFVPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETHISTOGRAMPARAMETERIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMINMAXPROC) (GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values); +typedef void (APIENTRYP PFNGLGETMINMAXPARAMETERFVPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMINMAXPARAMETERIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLHISTOGRAMPROC) (GLenum target, GLsizei width, GLenum internalformat, GLboolean sink); +typedef void (APIENTRYP PFNGLMINMAXPROC) (GLenum target, GLenum internalformat, GLboolean sink); +typedef void (APIENTRYP PFNGLRESETHISTOGRAMPROC) (GLenum target); +typedef void (APIENTRYP PFNGLRESETMINMAXPROC) (GLenum target); +#endif + +#ifndef GL_VERSION_1_3 +#define GL_VERSION_1_3 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glActiveTexture (GLenum); +GLAPI void APIENTRY glSampleCoverage (GLclampf, GLboolean); +GLAPI void APIENTRY glCompressedTexImage3D (GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexImage2D (GLenum, GLint, GLenum, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexImage1D (GLenum, GLint, GLenum, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexSubImage3D (GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexSubImage2D (GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexSubImage1D (GLenum, GLint, GLint, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetCompressedTexImage (GLenum, GLint, GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLACTIVETEXTUREPROC) (GLenum texture); +typedef void (APIENTRYP PFNGLSAMPLECOVERAGEPROC) (GLclampf value, GLboolean invert); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE3DPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE2DPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE1DPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE3DPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE2DPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE1DPROC) (GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLGETCOMPRESSEDTEXIMAGEPROC) (GLenum target, GLint level, GLvoid *img); +#endif + +#ifndef GL_VERSION_1_3_DEPRECATED +#define GL_VERSION_1_3_DEPRECATED 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glClientActiveTexture (GLenum); +GLAPI void APIENTRY glMultiTexCoord1d (GLenum, GLdouble); +GLAPI void APIENTRY glMultiTexCoord1dv (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord1f (GLenum, GLfloat); +GLAPI void APIENTRY glMultiTexCoord1fv (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord1i (GLenum, GLint); +GLAPI void APIENTRY glMultiTexCoord1iv (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord1s (GLenum, GLshort); +GLAPI void APIENTRY glMultiTexCoord1sv (GLenum, const GLshort *); +GLAPI void APIENTRY glMultiTexCoord2d (GLenum, GLdouble, GLdouble); +GLAPI void APIENTRY glMultiTexCoord2dv (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord2f (GLenum, GLfloat, GLfloat); +GLAPI void APIENTRY glMultiTexCoord2fv (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord2i (GLenum, GLint, GLint); +GLAPI void APIENTRY glMultiTexCoord2iv (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord2s (GLenum, GLshort, GLshort); +GLAPI void APIENTRY glMultiTexCoord2sv (GLenum, const GLshort *); +GLAPI void APIENTRY glMultiTexCoord3d (GLenum, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMultiTexCoord3dv (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord3f (GLenum, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMultiTexCoord3fv (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord3i (GLenum, GLint, GLint, GLint); +GLAPI void APIENTRY glMultiTexCoord3iv (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord3s (GLenum, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glMultiTexCoord3sv (GLenum, const GLshort *); +GLAPI void APIENTRY glMultiTexCoord4d (GLenum, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMultiTexCoord4dv (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord4f (GLenum, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMultiTexCoord4fv (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord4i (GLenum, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glMultiTexCoord4iv (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord4s (GLenum, GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glMultiTexCoord4sv (GLenum, const GLshort *); +GLAPI void APIENTRY glLoadTransposeMatrixf (const GLfloat *); +GLAPI void APIENTRY glLoadTransposeMatrixd (const GLdouble *); +GLAPI void APIENTRY glMultTransposeMatrixf (const GLfloat *); +GLAPI void APIENTRY glMultTransposeMatrixd (const GLdouble *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCLIENTACTIVETEXTUREPROC) (GLenum texture); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1DPROC) (GLenum target, GLdouble s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1DVPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1FPROC) (GLenum target, GLfloat s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1FVPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1IPROC) (GLenum target, GLint s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1IVPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1SPROC) (GLenum target, GLshort s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1SVPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2DPROC) (GLenum target, GLdouble s, GLdouble t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2DVPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2FPROC) (GLenum target, GLfloat s, GLfloat t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2FVPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2IPROC) (GLenum target, GLint s, GLint t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2IVPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2SPROC) (GLenum target, GLshort s, GLshort t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2SVPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3DPROC) (GLenum target, GLdouble s, GLdouble t, GLdouble r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3DVPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3FPROC) (GLenum target, GLfloat s, GLfloat t, GLfloat r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3FVPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3IPROC) (GLenum target, GLint s, GLint t, GLint r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3IVPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3SPROC) (GLenum target, GLshort s, GLshort t, GLshort r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3SVPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4DPROC) (GLenum target, GLdouble s, GLdouble t, GLdouble r, GLdouble q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4DVPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4FPROC) (GLenum target, GLfloat s, GLfloat t, GLfloat r, GLfloat q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4FVPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4IPROC) (GLenum target, GLint s, GLint t, GLint r, GLint q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4IVPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4SPROC) (GLenum target, GLshort s, GLshort t, GLshort r, GLshort q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4SVPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLLOADTRANSPOSEMATRIXFPROC) (const GLfloat *m); +typedef void (APIENTRYP PFNGLLOADTRANSPOSEMATRIXDPROC) (const GLdouble *m); +typedef void (APIENTRYP PFNGLMULTTRANSPOSEMATRIXFPROC) (const GLfloat *m); +typedef void (APIENTRYP PFNGLMULTTRANSPOSEMATRIXDPROC) (const GLdouble *m); +#endif + +#ifndef GL_VERSION_1_4 +#define GL_VERSION_1_4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendFuncSeparate (GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glMultiDrawArrays (GLenum, GLint *, GLsizei *, GLsizei); +GLAPI void APIENTRY glMultiDrawElements (GLenum, const GLsizei *, GLenum, const GLvoid* *, GLsizei); +GLAPI void APIENTRY glPointParameterf (GLenum, GLfloat); +GLAPI void APIENTRY glPointParameterfv (GLenum, const GLfloat *); +GLAPI void APIENTRY glPointParameteri (GLenum, GLint); +GLAPI void APIENTRY glPointParameteriv (GLenum, const GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEPROC) (GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha); +typedef void (APIENTRYP PFNGLMULTIDRAWARRAYSPROC) (GLenum mode, GLint *first, GLsizei *count, GLsizei primcount); +typedef void (APIENTRYP PFNGLMULTIDRAWELEMENTSPROC) (GLenum mode, const GLsizei *count, GLenum type, const GLvoid* *indices, GLsizei primcount); +typedef void (APIENTRYP PFNGLPOINTPARAMETERFPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLPOINTPARAMETERFVPROC) (GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLPOINTPARAMETERIPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLPOINTPARAMETERIVPROC) (GLenum pname, const GLint *params); +#endif + +#ifndef GL_VERSION_1_4_DEPRECATED +#define GL_VERSION_1_4_DEPRECATED 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFogCoordf (GLfloat); +GLAPI void APIENTRY glFogCoordfv (const GLfloat *); +GLAPI void APIENTRY glFogCoordd (GLdouble); +GLAPI void APIENTRY glFogCoorddv (const GLdouble *); +GLAPI void APIENTRY glFogCoordPointer (GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glSecondaryColor3b (GLbyte, GLbyte, GLbyte); +GLAPI void APIENTRY glSecondaryColor3bv (const GLbyte *); +GLAPI void APIENTRY glSecondaryColor3d (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glSecondaryColor3dv (const GLdouble *); +GLAPI void APIENTRY glSecondaryColor3f (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glSecondaryColor3fv (const GLfloat *); +GLAPI void APIENTRY glSecondaryColor3i (GLint, GLint, GLint); +GLAPI void APIENTRY glSecondaryColor3iv (const GLint *); +GLAPI void APIENTRY glSecondaryColor3s (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glSecondaryColor3sv (const GLshort *); +GLAPI void APIENTRY glSecondaryColor3ub (GLubyte, GLubyte, GLubyte); +GLAPI void APIENTRY glSecondaryColor3ubv (const GLubyte *); +GLAPI void APIENTRY glSecondaryColor3ui (GLuint, GLuint, GLuint); +GLAPI void APIENTRY glSecondaryColor3uiv (const GLuint *); +GLAPI void APIENTRY glSecondaryColor3us (GLushort, GLushort, GLushort); +GLAPI void APIENTRY glSecondaryColor3usv (const GLushort *); +GLAPI void APIENTRY glSecondaryColorPointer (GLint, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glWindowPos2d (GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos2dv (const GLdouble *); +GLAPI void APIENTRY glWindowPos2f (GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos2fv (const GLfloat *); +GLAPI void APIENTRY glWindowPos2i (GLint, GLint); +GLAPI void APIENTRY glWindowPos2iv (const GLint *); +GLAPI void APIENTRY glWindowPos2s (GLshort, GLshort); +GLAPI void APIENTRY glWindowPos2sv (const GLshort *); +GLAPI void APIENTRY glWindowPos3d (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos3dv (const GLdouble *); +GLAPI void APIENTRY glWindowPos3f (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos3fv (const GLfloat *); +GLAPI void APIENTRY glWindowPos3i (GLint, GLint, GLint); +GLAPI void APIENTRY glWindowPos3iv (const GLint *); +GLAPI void APIENTRY glWindowPos3s (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glWindowPos3sv (const GLshort *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFOGCOORDFPROC) (GLfloat coord); +typedef void (APIENTRYP PFNGLFOGCOORDFVPROC) (const GLfloat *coord); +typedef void (APIENTRYP PFNGLFOGCOORDDPROC) (GLdouble coord); +typedef void (APIENTRYP PFNGLFOGCOORDDVPROC) (const GLdouble *coord); +typedef void (APIENTRYP PFNGLFOGCOORDPOINTERPROC) (GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3BPROC) (GLbyte red, GLbyte green, GLbyte blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3BVPROC) (const GLbyte *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3DPROC) (GLdouble red, GLdouble green, GLdouble blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3DVPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3FPROC) (GLfloat red, GLfloat green, GLfloat blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3FVPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3IPROC) (GLint red, GLint green, GLint blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3IVPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3SPROC) (GLshort red, GLshort green, GLshort blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3SVPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UBPROC) (GLubyte red, GLubyte green, GLubyte blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UBVPROC) (const GLubyte *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UIPROC) (GLuint red, GLuint green, GLuint blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UIVPROC) (const GLuint *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3USPROC) (GLushort red, GLushort green, GLushort blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3USVPROC) (const GLushort *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLORPOINTERPROC) (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLWINDOWPOS2DPROC) (GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLWINDOWPOS2DVPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2FPROC) (GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLWINDOWPOS2FVPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2IPROC) (GLint x, GLint y); +typedef void (APIENTRYP PFNGLWINDOWPOS2IVPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2SPROC) (GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLWINDOWPOS2SVPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3DPROC) (GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLWINDOWPOS3DVPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3FPROC) (GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLWINDOWPOS3FVPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3IPROC) (GLint x, GLint y, GLint z); +typedef void (APIENTRYP PFNGLWINDOWPOS3IVPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3SPROC) (GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLWINDOWPOS3SVPROC) (const GLshort *v); +#endif + +#ifndef GL_VERSION_1_5 +#define GL_VERSION_1_5 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGenQueries (GLsizei, GLuint *); +GLAPI void APIENTRY glDeleteQueries (GLsizei, const GLuint *); +GLAPI GLboolean APIENTRY glIsQuery (GLuint); +GLAPI void APIENTRY glBeginQuery (GLenum, GLuint); +GLAPI void APIENTRY glEndQuery (GLenum); +GLAPI void APIENTRY glGetQueryiv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetQueryObjectiv (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetQueryObjectuiv (GLuint, GLenum, GLuint *); +GLAPI void APIENTRY glBindBuffer (GLenum, GLuint); +GLAPI void APIENTRY glDeleteBuffers (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenBuffers (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsBuffer (GLuint); +GLAPI void APIENTRY glBufferData (GLenum, GLsizeiptr, const GLvoid *, GLenum); +GLAPI void APIENTRY glBufferSubData (GLenum, GLintptr, GLsizeiptr, const GLvoid *); +GLAPI void APIENTRY glGetBufferSubData (GLenum, GLintptr, GLsizeiptr, GLvoid *); +GLAPI GLvoid* APIENTRY glMapBuffer (GLenum, GLenum); +GLAPI GLboolean APIENTRY glUnmapBuffer (GLenum); +GLAPI void APIENTRY glGetBufferParameteriv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetBufferPointerv (GLenum, GLenum, GLvoid* *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGENQUERIESPROC) (GLsizei n, GLuint *ids); +typedef void (APIENTRYP PFNGLDELETEQUERIESPROC) (GLsizei n, const GLuint *ids); +typedef GLboolean (APIENTRYP PFNGLISQUERYPROC) (GLuint id); +typedef void (APIENTRYP PFNGLBEGINQUERYPROC) (GLenum target, GLuint id); +typedef void (APIENTRYP PFNGLENDQUERYPROC) (GLenum target); +typedef void (APIENTRYP PFNGLGETQUERYIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETQUERYOBJECTIVPROC) (GLuint id, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETQUERYOBJECTUIVPROC) (GLuint id, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLBINDBUFFERPROC) (GLenum target, GLuint buffer); +typedef void (APIENTRYP PFNGLDELETEBUFFERSPROC) (GLsizei n, const GLuint *buffers); +typedef void (APIENTRYP PFNGLGENBUFFERSPROC) (GLsizei n, GLuint *buffers); +typedef GLboolean (APIENTRYP PFNGLISBUFFERPROC) (GLuint buffer); +typedef void (APIENTRYP PFNGLBUFFERDATAPROC) (GLenum target, GLsizeiptr size, const GLvoid *data, GLenum usage); +typedef void (APIENTRYP PFNGLBUFFERSUBDATAPROC) (GLenum target, GLintptr offset, GLsizeiptr size, const GLvoid *data); +typedef void (APIENTRYP PFNGLGETBUFFERSUBDATAPROC) (GLenum target, GLintptr offset, GLsizeiptr size, GLvoid *data); +typedef GLvoid* (APIENTRYP PFNGLMAPBUFFERPROC) (GLenum target, GLenum access); +typedef GLboolean (APIENTRYP PFNGLUNMAPBUFFERPROC) (GLenum target); +typedef void (APIENTRYP PFNGLGETBUFFERPARAMETERIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETBUFFERPOINTERVPROC) (GLenum target, GLenum pname, GLvoid* *params); +#endif + +#ifndef GL_VERSION_2_0 +#define GL_VERSION_2_0 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendEquationSeparate (GLenum, GLenum); +GLAPI void APIENTRY glDrawBuffers (GLsizei, const GLenum *); +GLAPI void APIENTRY glStencilOpSeparate (GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glStencilFuncSeparate (GLenum, GLenum, GLint, GLuint); +GLAPI void APIENTRY glStencilMaskSeparate (GLenum, GLuint); +GLAPI void APIENTRY glAttachShader (GLuint, GLuint); +GLAPI void APIENTRY glBindAttribLocation (GLuint, GLuint, const GLchar *); +GLAPI void APIENTRY glCompileShader (GLuint); +GLAPI GLuint APIENTRY glCreateProgram (void); +GLAPI GLuint APIENTRY glCreateShader (GLenum); +GLAPI void APIENTRY glDeleteProgram (GLuint); +GLAPI void APIENTRY glDeleteShader (GLuint); +GLAPI void APIENTRY glDetachShader (GLuint, GLuint); +GLAPI void APIENTRY glDisableVertexAttribArray (GLuint); +GLAPI void APIENTRY glEnableVertexAttribArray (GLuint); +GLAPI void APIENTRY glGetActiveAttrib (GLuint, GLuint, GLsizei, GLsizei *, GLint *, GLenum *, GLchar *); +GLAPI void APIENTRY glGetActiveUniform (GLuint, GLuint, GLsizei, GLsizei *, GLint *, GLenum *, GLchar *); +GLAPI void APIENTRY glGetAttachedShaders (GLuint, GLsizei, GLsizei *, GLuint *); +GLAPI GLint APIENTRY glGetAttribLocation (GLuint, const GLchar *); +GLAPI void APIENTRY glGetProgramiv (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetProgramInfoLog (GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI void APIENTRY glGetShaderiv (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetShaderInfoLog (GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI void APIENTRY glGetShaderSource (GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI GLint APIENTRY glGetUniformLocation (GLuint, const GLchar *); +GLAPI void APIENTRY glGetUniformfv (GLuint, GLint, GLfloat *); +GLAPI void APIENTRY glGetUniformiv (GLuint, GLint, GLint *); +GLAPI void APIENTRY glGetVertexAttribdv (GLuint, GLenum, GLdouble *); +GLAPI void APIENTRY glGetVertexAttribfv (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetVertexAttribiv (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVertexAttribPointerv (GLuint, GLenum, GLvoid* *); +GLAPI GLboolean APIENTRY glIsProgram (GLuint); +GLAPI GLboolean APIENTRY glIsShader (GLuint); +GLAPI void APIENTRY glLinkProgram (GLuint); +GLAPI void APIENTRY glShaderSource (GLuint, GLsizei, const GLchar* *, const GLint *); +GLAPI void APIENTRY glUseProgram (GLuint); +GLAPI void APIENTRY glUniform1f (GLint, GLfloat); +GLAPI void APIENTRY glUniform2f (GLint, GLfloat, GLfloat); +GLAPI void APIENTRY glUniform3f (GLint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glUniform4f (GLint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glUniform1i (GLint, GLint); +GLAPI void APIENTRY glUniform2i (GLint, GLint, GLint); +GLAPI void APIENTRY glUniform3i (GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glUniform4i (GLint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glUniform1fv (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform2fv (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform3fv (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform4fv (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform1iv (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniform2iv (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniform3iv (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniform4iv (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniformMatrix2fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix3fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix4fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glValidateProgram (GLuint); +GLAPI void APIENTRY glVertexAttrib1d (GLuint, GLdouble); +GLAPI void APIENTRY glVertexAttrib1dv (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib1f (GLuint, GLfloat); +GLAPI void APIENTRY glVertexAttrib1fv (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib1s (GLuint, GLshort); +GLAPI void APIENTRY glVertexAttrib1sv (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib2d (GLuint, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib2dv (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib2f (GLuint, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib2fv (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib2s (GLuint, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib2sv (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib3d (GLuint, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib3dv (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib3f (GLuint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib3fv (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib3s (GLuint, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib3sv (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4Nbv (GLuint, const GLbyte *); +GLAPI void APIENTRY glVertexAttrib4Niv (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttrib4Nsv (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4Nub (GLuint, GLubyte, GLubyte, GLubyte, GLubyte); +GLAPI void APIENTRY glVertexAttrib4Nubv (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttrib4Nuiv (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttrib4Nusv (GLuint, const GLushort *); +GLAPI void APIENTRY glVertexAttrib4bv (GLuint, const GLbyte *); +GLAPI void APIENTRY glVertexAttrib4d (GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib4dv (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib4f (GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib4fv (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib4iv (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttrib4s (GLuint, GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib4sv (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4ubv (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttrib4uiv (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttrib4usv (GLuint, const GLushort *); +GLAPI void APIENTRY glVertexAttribPointer (GLuint, GLint, GLenum, GLboolean, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDEQUATIONSEPARATEPROC) (GLenum modeRGB, GLenum modeAlpha); +typedef void (APIENTRYP PFNGLDRAWBUFFERSPROC) (GLsizei n, const GLenum *bufs); +typedef void (APIENTRYP PFNGLSTENCILOPSEPARATEPROC) (GLenum face, GLenum sfail, GLenum dpfail, GLenum dppass); +typedef void (APIENTRYP PFNGLSTENCILFUNCSEPARATEPROC) (GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); +typedef void (APIENTRYP PFNGLSTENCILMASKSEPARATEPROC) (GLenum face, GLuint mask); +typedef void (APIENTRYP PFNGLATTACHSHADERPROC) (GLuint program, GLuint shader); +typedef void (APIENTRYP PFNGLBINDATTRIBLOCATIONPROC) (GLuint program, GLuint index, const GLchar *name); +typedef void (APIENTRYP PFNGLCOMPILESHADERPROC) (GLuint shader); +typedef GLuint (APIENTRYP PFNGLCREATEPROGRAMPROC) (void); +typedef GLuint (APIENTRYP PFNGLCREATESHADERPROC) (GLenum type); +typedef void (APIENTRYP PFNGLDELETEPROGRAMPROC) (GLuint program); +typedef void (APIENTRYP PFNGLDELETESHADERPROC) (GLuint shader); +typedef void (APIENTRYP PFNGLDETACHSHADERPROC) (GLuint program, GLuint shader); +typedef void (APIENTRYP PFNGLDISABLEVERTEXATTRIBARRAYPROC) (GLuint index); +typedef void (APIENTRYP PFNGLENABLEVERTEXATTRIBARRAYPROC) (GLuint index); +typedef void (APIENTRYP PFNGLGETACTIVEATTRIBPROC) (GLuint program, GLuint index, GLsizei bufSize, GLsizei *length, GLint *size, GLenum *type, GLchar *name); +typedef void (APIENTRYP PFNGLGETACTIVEUNIFORMPROC) (GLuint program, GLuint index, GLsizei bufSize, GLsizei *length, GLint *size, GLenum *type, GLchar *name); +typedef void (APIENTRYP PFNGLGETATTACHEDSHADERSPROC) (GLuint program, GLsizei maxCount, GLsizei *count, GLuint *obj); +typedef GLint (APIENTRYP PFNGLGETATTRIBLOCATIONPROC) (GLuint program, const GLchar *name); +typedef void (APIENTRYP PFNGLGETPROGRAMIVPROC) (GLuint program, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMINFOLOGPROC) (GLuint program, GLsizei bufSize, GLsizei *length, GLchar *infoLog); +typedef void (APIENTRYP PFNGLGETSHADERIVPROC) (GLuint shader, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETSHADERINFOLOGPROC) (GLuint shader, GLsizei bufSize, GLsizei *length, GLchar *infoLog); +typedef void (APIENTRYP PFNGLGETSHADERSOURCEPROC) (GLuint shader, GLsizei bufSize, GLsizei *length, GLchar *source); +typedef GLint (APIENTRYP PFNGLGETUNIFORMLOCATIONPROC) (GLuint program, const GLchar *name); +typedef void (APIENTRYP PFNGLGETUNIFORMFVPROC) (GLuint program, GLint location, GLfloat *params); +typedef void (APIENTRYP PFNGLGETUNIFORMIVPROC) (GLuint program, GLint location, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBDVPROC) (GLuint index, GLenum pname, GLdouble *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBFVPROC) (GLuint index, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIVPROC) (GLuint index, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBPOINTERVPROC) (GLuint index, GLenum pname, GLvoid* *pointer); +typedef GLboolean (APIENTRYP PFNGLISPROGRAMPROC) (GLuint program); +typedef GLboolean (APIENTRYP PFNGLISSHADERPROC) (GLuint shader); +typedef void (APIENTRYP PFNGLLINKPROGRAMPROC) (GLuint program); +typedef void (APIENTRYP PFNGLSHADERSOURCEPROC) (GLuint shader, GLsizei count, const GLchar* *string, const GLint *length); +typedef void (APIENTRYP PFNGLUSEPROGRAMPROC) (GLuint program); +typedef void (APIENTRYP PFNGLUNIFORM1FPROC) (GLint location, GLfloat v0); +typedef void (APIENTRYP PFNGLUNIFORM2FPROC) (GLint location, GLfloat v0, GLfloat v1); +typedef void (APIENTRYP PFNGLUNIFORM3FPROC) (GLint location, GLfloat v0, GLfloat v1, GLfloat v2); +typedef void (APIENTRYP PFNGLUNIFORM4FPROC) (GLint location, GLfloat v0, GLfloat v1, GLfloat v2, GLfloat v3); +typedef void (APIENTRYP PFNGLUNIFORM1IPROC) (GLint location, GLint v0); +typedef void (APIENTRYP PFNGLUNIFORM2IPROC) (GLint location, GLint v0, GLint v1); +typedef void (APIENTRYP PFNGLUNIFORM3IPROC) (GLint location, GLint v0, GLint v1, GLint v2); +typedef void (APIENTRYP PFNGLUNIFORM4IPROC) (GLint location, GLint v0, GLint v1, GLint v2, GLint v3); +typedef void (APIENTRYP PFNGLUNIFORM1FVPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM2FVPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM3FVPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM4FVPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM1IVPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORM2IVPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORM3IVPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORM4IVPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX2FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX3FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX4FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLVALIDATEPROGRAMPROC) (GLuint program); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1DPROC) (GLuint index, GLdouble x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1DVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1FPROC) (GLuint index, GLfloat x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1FVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1SPROC) (GLuint index, GLshort x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1SVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2DPROC) (GLuint index, GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2DVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2FPROC) (GLuint index, GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2FVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2SPROC) (GLuint index, GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2SVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3DPROC) (GLuint index, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3DVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3FPROC) (GLuint index, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3FVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3SPROC) (GLuint index, GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3SVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NBVPROC) (GLuint index, const GLbyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NIVPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NSVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUBPROC) (GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUBVPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUIVPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUSVPROC) (GLuint index, const GLushort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4BVPROC) (GLuint index, const GLbyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4DPROC) (GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4DVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4FPROC) (GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4FVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4IVPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4SPROC) (GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4SVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4UBVPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4UIVPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4USVPROC) (GLuint index, const GLushort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBPOINTERPROC) (GLuint index, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const GLvoid *pointer); +#endif + +#ifndef GL_VERSION_2_1 +#define GL_VERSION_2_1 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glUniformMatrix2x3fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix3x2fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix2x4fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix4x2fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix3x4fv (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix4x3fv (GLint, GLsizei, GLboolean, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLUNIFORMMATRIX2X3FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX3X2FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX2X4FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX4X2FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX3X4FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX4X3FVPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +#endif + +#ifndef GL_VERSION_3_0 +#define GL_VERSION_3_0 1 +/* OpenGL 3.0 also reuses entry points from these extensions: */ +/* ARB_framebuffer_object */ +/* ARB_map_buffer_range */ +/* ARB_vertex_array_object */ +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorMaski (GLuint, GLboolean, GLboolean, GLboolean, GLboolean); +GLAPI void APIENTRY glGetBooleani_v (GLenum, GLuint, GLboolean *); +GLAPI void APIENTRY glGetIntegeri_v (GLenum, GLuint, GLint *); +GLAPI void APIENTRY glEnablei (GLenum, GLuint); +GLAPI void APIENTRY glDisablei (GLenum, GLuint); +GLAPI GLboolean APIENTRY glIsEnabledi (GLenum, GLuint); +GLAPI void APIENTRY glBeginTransformFeedback (GLenum); +GLAPI void APIENTRY glEndTransformFeedback (void); +GLAPI void APIENTRY glBindBufferRange (GLenum, GLuint, GLuint, GLintptr, GLsizeiptr); +GLAPI void APIENTRY glBindBufferBase (GLenum, GLuint, GLuint); +GLAPI void APIENTRY glTransformFeedbackVaryings (GLuint, GLsizei, const GLchar* *, GLenum); +GLAPI void APIENTRY glGetTransformFeedbackVarying (GLuint, GLuint, GLsizei, GLsizei *, GLsizei *, GLenum *, GLchar *); +GLAPI void APIENTRY glClampColor (GLenum, GLenum); +GLAPI void APIENTRY glBeginConditionalRender (GLuint, GLenum); +GLAPI void APIENTRY glEndConditionalRender (void); +GLAPI void APIENTRY glVertexAttribIPointer (GLuint, GLint, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetVertexAttribIiv (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVertexAttribIuiv (GLuint, GLenum, GLuint *); +GLAPI void APIENTRY glVertexAttribI1i (GLuint, GLint); +GLAPI void APIENTRY glVertexAttribI2i (GLuint, GLint, GLint); +GLAPI void APIENTRY glVertexAttribI3i (GLuint, GLint, GLint, GLint); +GLAPI void APIENTRY glVertexAttribI4i (GLuint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glVertexAttribI1ui (GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI2ui (GLuint, GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI3ui (GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI4ui (GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI1iv (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI2iv (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI3iv (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI4iv (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI1uiv (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI2uiv (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI3uiv (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI4uiv (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI4bv (GLuint, const GLbyte *); +GLAPI void APIENTRY glVertexAttribI4sv (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttribI4ubv (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttribI4usv (GLuint, const GLushort *); +GLAPI void APIENTRY glGetUniformuiv (GLuint, GLint, GLuint *); +GLAPI void APIENTRY glBindFragDataLocation (GLuint, GLuint, const GLchar *); +GLAPI GLint APIENTRY glGetFragDataLocation (GLuint, const GLchar *); +GLAPI void APIENTRY glUniform1ui (GLint, GLuint); +GLAPI void APIENTRY glUniform2ui (GLint, GLuint, GLuint); +GLAPI void APIENTRY glUniform3ui (GLint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glUniform4ui (GLint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glUniform1uiv (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glUniform2uiv (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glUniform3uiv (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glUniform4uiv (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glTexParameterIiv (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glTexParameterIuiv (GLenum, GLenum, const GLuint *); +GLAPI void APIENTRY glGetTexParameterIiv (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetTexParameterIuiv (GLenum, GLenum, GLuint *); +GLAPI void APIENTRY glClearBufferiv (GLenum, GLint, const GLint *); +GLAPI void APIENTRY glClearBufferuiv (GLenum, GLint, const GLuint *); +GLAPI void APIENTRY glClearBufferfv (GLenum, GLint, const GLfloat *); +GLAPI void APIENTRY glClearBufferfi (GLenum, GLint, GLfloat, GLint); +GLAPI const GLubyte * APIENTRY glGetStringi (GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLORMASKIPROC) (GLuint index, GLboolean r, GLboolean g, GLboolean b, GLboolean a); +typedef void (APIENTRYP PFNGLGETBOOLEANI_VPROC) (GLenum target, GLuint index, GLboolean *data); +typedef void (APIENTRYP PFNGLGETINTEGERI_VPROC) (GLenum target, GLuint index, GLint *data); +typedef void (APIENTRYP PFNGLENABLEIPROC) (GLenum target, GLuint index); +typedef void (APIENTRYP PFNGLDISABLEIPROC) (GLenum target, GLuint index); +typedef GLboolean (APIENTRYP PFNGLISENABLEDIPROC) (GLenum target, GLuint index); +typedef void (APIENTRYP PFNGLBEGINTRANSFORMFEEDBACKPROC) (GLenum primitiveMode); +typedef void (APIENTRYP PFNGLENDTRANSFORMFEEDBACKPROC) (void); +typedef void (APIENTRYP PFNGLBINDBUFFERRANGEPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size); +typedef void (APIENTRYP PFNGLBINDBUFFERBASEPROC) (GLenum target, GLuint index, GLuint buffer); +typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKVARYINGSPROC) (GLuint program, GLsizei count, const GLchar* *varyings, GLenum bufferMode); +typedef void (APIENTRYP PFNGLGETTRANSFORMFEEDBACKVARYINGPROC) (GLuint program, GLuint index, GLsizei bufSize, GLsizei *length, GLsizei *size, GLenum *type, GLchar *name); +typedef void (APIENTRYP PFNGLCLAMPCOLORPROC) (GLenum target, GLenum clamp); +typedef void (APIENTRYP PFNGLBEGINCONDITIONALRENDERPROC) (GLuint id, GLenum mode); +typedef void (APIENTRYP PFNGLENDCONDITIONALRENDERPROC) (void); +typedef void (APIENTRYP PFNGLVERTEXATTRIBIPOINTERPROC) (GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIIVPROC) (GLuint index, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIUIVPROC) (GLuint index, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1IPROC) (GLuint index, GLint x); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2IPROC) (GLuint index, GLint x, GLint y); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3IPROC) (GLuint index, GLint x, GLint y, GLint z); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4IPROC) (GLuint index, GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1UIPROC) (GLuint index, GLuint x); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2UIPROC) (GLuint index, GLuint x, GLuint y); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3UIPROC) (GLuint index, GLuint x, GLuint y, GLuint z); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UIPROC) (GLuint index, GLuint x, GLuint y, GLuint z, GLuint w); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1IVPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2IVPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3IVPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4IVPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1UIVPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2UIVPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3UIVPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UIVPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4BVPROC) (GLuint index, const GLbyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4SVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UBVPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4USVPROC) (GLuint index, const GLushort *v); +typedef void (APIENTRYP PFNGLGETUNIFORMUIVPROC) (GLuint program, GLint location, GLuint *params); +typedef void (APIENTRYP PFNGLBINDFRAGDATALOCATIONPROC) (GLuint program, GLuint color, const GLchar *name); +typedef GLint (APIENTRYP PFNGLGETFRAGDATALOCATIONPROC) (GLuint program, const GLchar *name); +typedef void (APIENTRYP PFNGLUNIFORM1UIPROC) (GLint location, GLuint v0); +typedef void (APIENTRYP PFNGLUNIFORM2UIPROC) (GLint location, GLuint v0, GLuint v1); +typedef void (APIENTRYP PFNGLUNIFORM3UIPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2); +typedef void (APIENTRYP PFNGLUNIFORM4UIPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3); +typedef void (APIENTRYP PFNGLUNIFORM1UIVPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLUNIFORM2UIVPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLUNIFORM3UIVPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLUNIFORM4UIVPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLTEXPARAMETERIIVPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLTEXPARAMETERIUIVPROC) (GLenum target, GLenum pname, const GLuint *params); +typedef void (APIENTRYP PFNGLGETTEXPARAMETERIIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETTEXPARAMETERIUIVPROC) (GLenum target, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLCLEARBUFFERIVPROC) (GLenum buffer, GLint drawbuffer, const GLint *value); +typedef void (APIENTRYP PFNGLCLEARBUFFERUIVPROC) (GLenum buffer, GLint drawbuffer, const GLuint *value); +typedef void (APIENTRYP PFNGLCLEARBUFFERFVPROC) (GLenum buffer, GLint drawbuffer, const GLfloat *value); +typedef void (APIENTRYP PFNGLCLEARBUFFERFIPROC) (GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil); +typedef const GLubyte * (APIENTRYP PFNGLGETSTRINGIPROC) (GLenum name, GLuint index); +#endif + +#ifndef GL_VERSION_3_1 +#define GL_VERSION_3_1 1 +/* OpenGL 3.1 also reuses entry points from these extensions: */ +/* ARB_copy_buffer */ +/* ARB_uniform_buffer_object */ +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawArraysInstanced (GLenum, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glDrawElementsInstanced (GLenum, GLsizei, GLenum, const GLvoid *, GLsizei); +GLAPI void APIENTRY glTexBuffer (GLenum, GLenum, GLuint); +GLAPI void APIENTRY glPrimitiveRestartIndex (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWARRAYSINSTANCEDPROC) (GLenum mode, GLint first, GLsizei count, GLsizei primcount); +typedef void (APIENTRYP PFNGLDRAWELEMENTSINSTANCEDPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount); +typedef void (APIENTRYP PFNGLTEXBUFFERPROC) (GLenum target, GLenum internalformat, GLuint buffer); +typedef void (APIENTRYP PFNGLPRIMITIVERESTARTINDEXPROC) (GLuint index); +#endif + +#ifndef GL_VERSION_3_2 +#define GL_VERSION_3_2 1 +/* OpenGL 3.2 also reuses entry points from these extensions: */ +/* ARB_draw_elements_base_vertex */ +/* ARB_provoking_vertex */ +/* ARB_sync */ +/* ARB_texture_multisample */ +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetInteger64i_v (GLenum, GLuint, GLint64 *); +GLAPI void APIENTRY glGetBufferParameteri64v (GLenum, GLenum, GLint64 *); +GLAPI void APIENTRY glProgramParameteri (GLuint, GLenum, GLint); +GLAPI void APIENTRY glFramebufferTexture (GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTextureFace (GLenum, GLenum, GLuint, GLint, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETINTEGER64I_VPROC) (GLenum target, GLuint index, GLint64 *data); +typedef void (APIENTRYP PFNGLGETBUFFERPARAMETERI64VPROC) (GLenum target, GLenum pname, GLint64 *params); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETERIPROC) (GLuint program, GLenum pname, GLint value); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREFACEPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLenum face); +#endif + +#ifndef GL_ARB_multitexture +#define GL_ARB_multitexture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glActiveTextureARB (GLenum); +GLAPI void APIENTRY glClientActiveTextureARB (GLenum); +GLAPI void APIENTRY glMultiTexCoord1dARB (GLenum, GLdouble); +GLAPI void APIENTRY glMultiTexCoord1dvARB (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord1fARB (GLenum, GLfloat); +GLAPI void APIENTRY glMultiTexCoord1fvARB (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord1iARB (GLenum, GLint); +GLAPI void APIENTRY glMultiTexCoord1ivARB (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord1sARB (GLenum, GLshort); +GLAPI void APIENTRY glMultiTexCoord1svARB (GLenum, const GLshort *); +GLAPI void APIENTRY glMultiTexCoord2dARB (GLenum, GLdouble, GLdouble); +GLAPI void APIENTRY glMultiTexCoord2dvARB (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord2fARB (GLenum, GLfloat, GLfloat); +GLAPI void APIENTRY glMultiTexCoord2fvARB (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord2iARB (GLenum, GLint, GLint); +GLAPI void APIENTRY glMultiTexCoord2ivARB (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord2sARB (GLenum, GLshort, GLshort); +GLAPI void APIENTRY glMultiTexCoord2svARB (GLenum, const GLshort *); +GLAPI void APIENTRY glMultiTexCoord3dARB (GLenum, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMultiTexCoord3dvARB (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord3fARB (GLenum, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMultiTexCoord3fvARB (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord3iARB (GLenum, GLint, GLint, GLint); +GLAPI void APIENTRY glMultiTexCoord3ivARB (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord3sARB (GLenum, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glMultiTexCoord3svARB (GLenum, const GLshort *); +GLAPI void APIENTRY glMultiTexCoord4dARB (GLenum, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMultiTexCoord4dvARB (GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexCoord4fARB (GLenum, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMultiTexCoord4fvARB (GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexCoord4iARB (GLenum, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glMultiTexCoord4ivARB (GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexCoord4sARB (GLenum, GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glMultiTexCoord4svARB (GLenum, const GLshort *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLACTIVETEXTUREARBPROC) (GLenum texture); +typedef void (APIENTRYP PFNGLCLIENTACTIVETEXTUREARBPROC) (GLenum texture); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1DARBPROC) (GLenum target, GLdouble s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1DVARBPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1FARBPROC) (GLenum target, GLfloat s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1FVARBPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1IARBPROC) (GLenum target, GLint s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1IVARBPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1SARBPROC) (GLenum target, GLshort s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1SVARBPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2DARBPROC) (GLenum target, GLdouble s, GLdouble t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2DVARBPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2FARBPROC) (GLenum target, GLfloat s, GLfloat t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2FVARBPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2IARBPROC) (GLenum target, GLint s, GLint t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2IVARBPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2SARBPROC) (GLenum target, GLshort s, GLshort t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2SVARBPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3DARBPROC) (GLenum target, GLdouble s, GLdouble t, GLdouble r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3DVARBPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3FARBPROC) (GLenum target, GLfloat s, GLfloat t, GLfloat r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3FVARBPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3IARBPROC) (GLenum target, GLint s, GLint t, GLint r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3IVARBPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3SARBPROC) (GLenum target, GLshort s, GLshort t, GLshort r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3SVARBPROC) (GLenum target, const GLshort *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4DARBPROC) (GLenum target, GLdouble s, GLdouble t, GLdouble r, GLdouble q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4DVARBPROC) (GLenum target, const GLdouble *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4FARBPROC) (GLenum target, GLfloat s, GLfloat t, GLfloat r, GLfloat q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4FVARBPROC) (GLenum target, const GLfloat *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4IARBPROC) (GLenum target, GLint s, GLint t, GLint r, GLint q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4IVARBPROC) (GLenum target, const GLint *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4SARBPROC) (GLenum target, GLshort s, GLshort t, GLshort r, GLshort q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4SVARBPROC) (GLenum target, const GLshort *v); +#endif + +#ifndef GL_ARB_transpose_matrix +#define GL_ARB_transpose_matrix 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glLoadTransposeMatrixfARB (const GLfloat *); +GLAPI void APIENTRY glLoadTransposeMatrixdARB (const GLdouble *); +GLAPI void APIENTRY glMultTransposeMatrixfARB (const GLfloat *); +GLAPI void APIENTRY glMultTransposeMatrixdARB (const GLdouble *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLLOADTRANSPOSEMATRIXFARBPROC) (const GLfloat *m); +typedef void (APIENTRYP PFNGLLOADTRANSPOSEMATRIXDARBPROC) (const GLdouble *m); +typedef void (APIENTRYP PFNGLMULTTRANSPOSEMATRIXFARBPROC) (const GLfloat *m); +typedef void (APIENTRYP PFNGLMULTTRANSPOSEMATRIXDARBPROC) (const GLdouble *m); +#endif + +#ifndef GL_ARB_multisample +#define GL_ARB_multisample 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glSampleCoverageARB (GLclampf, GLboolean); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSAMPLECOVERAGEARBPROC) (GLclampf value, GLboolean invert); +#endif + +#ifndef GL_ARB_texture_env_add +#define GL_ARB_texture_env_add 1 +#endif + +#ifndef GL_ARB_texture_cube_map +#define GL_ARB_texture_cube_map 1 +#endif + +#ifndef GL_ARB_texture_compression +#define GL_ARB_texture_compression 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCompressedTexImage3DARB (GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexImage2DARB (GLenum, GLint, GLenum, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexImage1DARB (GLenum, GLint, GLenum, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexSubImage3DARB (GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexSubImage2DARB (GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTexSubImage1DARB (GLenum, GLint, GLint, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetCompressedTexImageARB (GLenum, GLint, GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE3DARBPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE2DARBPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXIMAGE1DARBPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE3DARBPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE2DARBPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXSUBIMAGE1DARBPROC) (GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const GLvoid *data); +typedef void (APIENTRYP PFNGLGETCOMPRESSEDTEXIMAGEARBPROC) (GLenum target, GLint level, GLvoid *img); +#endif + +#ifndef GL_ARB_texture_border_clamp +#define GL_ARB_texture_border_clamp 1 +#endif + +#ifndef GL_ARB_point_parameters +#define GL_ARB_point_parameters 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPointParameterfARB (GLenum, GLfloat); +GLAPI void APIENTRY glPointParameterfvARB (GLenum, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPOINTPARAMETERFARBPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLPOINTPARAMETERFVARBPROC) (GLenum pname, const GLfloat *params); +#endif + +#ifndef GL_ARB_vertex_blend +#define GL_ARB_vertex_blend 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glWeightbvARB (GLint, const GLbyte *); +GLAPI void APIENTRY glWeightsvARB (GLint, const GLshort *); +GLAPI void APIENTRY glWeightivARB (GLint, const GLint *); +GLAPI void APIENTRY glWeightfvARB (GLint, const GLfloat *); +GLAPI void APIENTRY glWeightdvARB (GLint, const GLdouble *); +GLAPI void APIENTRY glWeightubvARB (GLint, const GLubyte *); +GLAPI void APIENTRY glWeightusvARB (GLint, const GLushort *); +GLAPI void APIENTRY glWeightuivARB (GLint, const GLuint *); +GLAPI void APIENTRY glWeightPointerARB (GLint, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glVertexBlendARB (GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLWEIGHTBVARBPROC) (GLint size, const GLbyte *weights); +typedef void (APIENTRYP PFNGLWEIGHTSVARBPROC) (GLint size, const GLshort *weights); +typedef void (APIENTRYP PFNGLWEIGHTIVARBPROC) (GLint size, const GLint *weights); +typedef void (APIENTRYP PFNGLWEIGHTFVARBPROC) (GLint size, const GLfloat *weights); +typedef void (APIENTRYP PFNGLWEIGHTDVARBPROC) (GLint size, const GLdouble *weights); +typedef void (APIENTRYP PFNGLWEIGHTUBVARBPROC) (GLint size, const GLubyte *weights); +typedef void (APIENTRYP PFNGLWEIGHTUSVARBPROC) (GLint size, const GLushort *weights); +typedef void (APIENTRYP PFNGLWEIGHTUIVARBPROC) (GLint size, const GLuint *weights); +typedef void (APIENTRYP PFNGLWEIGHTPOINTERARBPROC) (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLVERTEXBLENDARBPROC) (GLint count); +#endif + +#ifndef GL_ARB_matrix_palette +#define GL_ARB_matrix_palette 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCurrentPaletteMatrixARB (GLint); +GLAPI void APIENTRY glMatrixIndexubvARB (GLint, const GLubyte *); +GLAPI void APIENTRY glMatrixIndexusvARB (GLint, const GLushort *); +GLAPI void APIENTRY glMatrixIndexuivARB (GLint, const GLuint *); +GLAPI void APIENTRY glMatrixIndexPointerARB (GLint, GLenum, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCURRENTPALETTEMATRIXARBPROC) (GLint index); +typedef void (APIENTRYP PFNGLMATRIXINDEXUBVARBPROC) (GLint size, const GLubyte *indices); +typedef void (APIENTRYP PFNGLMATRIXINDEXUSVARBPROC) (GLint size, const GLushort *indices); +typedef void (APIENTRYP PFNGLMATRIXINDEXUIVARBPROC) (GLint size, const GLuint *indices); +typedef void (APIENTRYP PFNGLMATRIXINDEXPOINTERARBPROC) (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +#endif + +#ifndef GL_ARB_texture_env_combine +#define GL_ARB_texture_env_combine 1 +#endif + +#ifndef GL_ARB_texture_env_crossbar +#define GL_ARB_texture_env_crossbar 1 +#endif + +#ifndef GL_ARB_texture_env_dot3 +#define GL_ARB_texture_env_dot3 1 +#endif + +#ifndef GL_ARB_texture_mirrored_repeat +#define GL_ARB_texture_mirrored_repeat 1 +#endif + +#ifndef GL_ARB_depth_texture +#define GL_ARB_depth_texture 1 +#endif + +#ifndef GL_ARB_shadow +#define GL_ARB_shadow 1 +#endif + +#ifndef GL_ARB_shadow_ambient +#define GL_ARB_shadow_ambient 1 +#endif + +#ifndef GL_ARB_window_pos +#define GL_ARB_window_pos 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glWindowPos2dARB (GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos2dvARB (const GLdouble *); +GLAPI void APIENTRY glWindowPos2fARB (GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos2fvARB (const GLfloat *); +GLAPI void APIENTRY glWindowPos2iARB (GLint, GLint); +GLAPI void APIENTRY glWindowPos2ivARB (const GLint *); +GLAPI void APIENTRY glWindowPos2sARB (GLshort, GLshort); +GLAPI void APIENTRY glWindowPos2svARB (const GLshort *); +GLAPI void APIENTRY glWindowPos3dARB (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos3dvARB (const GLdouble *); +GLAPI void APIENTRY glWindowPos3fARB (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos3fvARB (const GLfloat *); +GLAPI void APIENTRY glWindowPos3iARB (GLint, GLint, GLint); +GLAPI void APIENTRY glWindowPos3ivARB (const GLint *); +GLAPI void APIENTRY glWindowPos3sARB (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glWindowPos3svARB (const GLshort *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLWINDOWPOS2DARBPROC) (GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLWINDOWPOS2DVARBPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2FARBPROC) (GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLWINDOWPOS2FVARBPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2IARBPROC) (GLint x, GLint y); +typedef void (APIENTRYP PFNGLWINDOWPOS2IVARBPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2SARBPROC) (GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLWINDOWPOS2SVARBPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3DARBPROC) (GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLWINDOWPOS3DVARBPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3FARBPROC) (GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLWINDOWPOS3FVARBPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3IARBPROC) (GLint x, GLint y, GLint z); +typedef void (APIENTRYP PFNGLWINDOWPOS3IVARBPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3SARBPROC) (GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLWINDOWPOS3SVARBPROC) (const GLshort *v); +#endif + +#ifndef GL_ARB_vertex_program +#define GL_ARB_vertex_program 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexAttrib1dARB (GLuint, GLdouble); +GLAPI void APIENTRY glVertexAttrib1dvARB (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib1fARB (GLuint, GLfloat); +GLAPI void APIENTRY glVertexAttrib1fvARB (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib1sARB (GLuint, GLshort); +GLAPI void APIENTRY glVertexAttrib1svARB (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib2dARB (GLuint, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib2dvARB (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib2fARB (GLuint, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib2fvARB (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib2sARB (GLuint, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib2svARB (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib3dARB (GLuint, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib3dvARB (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib3fARB (GLuint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib3fvARB (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib3sARB (GLuint, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib3svARB (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4NbvARB (GLuint, const GLbyte *); +GLAPI void APIENTRY glVertexAttrib4NivARB (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttrib4NsvARB (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4NubARB (GLuint, GLubyte, GLubyte, GLubyte, GLubyte); +GLAPI void APIENTRY glVertexAttrib4NubvARB (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttrib4NuivARB (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttrib4NusvARB (GLuint, const GLushort *); +GLAPI void APIENTRY glVertexAttrib4bvARB (GLuint, const GLbyte *); +GLAPI void APIENTRY glVertexAttrib4dARB (GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib4dvARB (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib4fARB (GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib4fvARB (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib4ivARB (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttrib4sARB (GLuint, GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib4svARB (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4ubvARB (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttrib4uivARB (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttrib4usvARB (GLuint, const GLushort *); +GLAPI void APIENTRY glVertexAttribPointerARB (GLuint, GLint, GLenum, GLboolean, GLsizei, const GLvoid *); +GLAPI void APIENTRY glEnableVertexAttribArrayARB (GLuint); +GLAPI void APIENTRY glDisableVertexAttribArrayARB (GLuint); +GLAPI void APIENTRY glProgramStringARB (GLenum, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glBindProgramARB (GLenum, GLuint); +GLAPI void APIENTRY glDeleteProgramsARB (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenProgramsARB (GLsizei, GLuint *); +GLAPI void APIENTRY glProgramEnvParameter4dARB (GLenum, GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glProgramEnvParameter4dvARB (GLenum, GLuint, const GLdouble *); +GLAPI void APIENTRY glProgramEnvParameter4fARB (GLenum, GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramEnvParameter4fvARB (GLenum, GLuint, const GLfloat *); +GLAPI void APIENTRY glProgramLocalParameter4dARB (GLenum, GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glProgramLocalParameter4dvARB (GLenum, GLuint, const GLdouble *); +GLAPI void APIENTRY glProgramLocalParameter4fARB (GLenum, GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramLocalParameter4fvARB (GLenum, GLuint, const GLfloat *); +GLAPI void APIENTRY glGetProgramEnvParameterdvARB (GLenum, GLuint, GLdouble *); +GLAPI void APIENTRY glGetProgramEnvParameterfvARB (GLenum, GLuint, GLfloat *); +GLAPI void APIENTRY glGetProgramLocalParameterdvARB (GLenum, GLuint, GLdouble *); +GLAPI void APIENTRY glGetProgramLocalParameterfvARB (GLenum, GLuint, GLfloat *); +GLAPI void APIENTRY glGetProgramivARB (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetProgramStringARB (GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetVertexAttribdvARB (GLuint, GLenum, GLdouble *); +GLAPI void APIENTRY glGetVertexAttribfvARB (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetVertexAttribivARB (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVertexAttribPointervARB (GLuint, GLenum, GLvoid* *); +GLAPI GLboolean APIENTRY glIsProgramARB (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXATTRIB1DARBPROC) (GLuint index, GLdouble x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1DVARBPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1FARBPROC) (GLuint index, GLfloat x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1FVARBPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1SARBPROC) (GLuint index, GLshort x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1SVARBPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2DARBPROC) (GLuint index, GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2DVARBPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2FARBPROC) (GLuint index, GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2FVARBPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2SARBPROC) (GLuint index, GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2SVARBPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3DARBPROC) (GLuint index, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3DVARBPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3FARBPROC) (GLuint index, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3FVARBPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3SARBPROC) (GLuint index, GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3SVARBPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NBVARBPROC) (GLuint index, const GLbyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NIVARBPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NSVARBPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUBARBPROC) (GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUBVARBPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUIVARBPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4NUSVARBPROC) (GLuint index, const GLushort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4BVARBPROC) (GLuint index, const GLbyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4DARBPROC) (GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4DVARBPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4FARBPROC) (GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4FVARBPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4IVARBPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4SARBPROC) (GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4SVARBPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4UBVARBPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4UIVARBPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4USVARBPROC) (GLuint index, const GLushort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBPOINTERARBPROC) (GLuint index, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLENABLEVERTEXATTRIBARRAYARBPROC) (GLuint index); +typedef void (APIENTRYP PFNGLDISABLEVERTEXATTRIBARRAYARBPROC) (GLuint index); +typedef void (APIENTRYP PFNGLPROGRAMSTRINGARBPROC) (GLenum target, GLenum format, GLsizei len, const GLvoid *string); +typedef void (APIENTRYP PFNGLBINDPROGRAMARBPROC) (GLenum target, GLuint program); +typedef void (APIENTRYP PFNGLDELETEPROGRAMSARBPROC) (GLsizei n, const GLuint *programs); +typedef void (APIENTRYP PFNGLGENPROGRAMSARBPROC) (GLsizei n, GLuint *programs); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETER4DARBPROC) (GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETER4DVARBPROC) (GLenum target, GLuint index, const GLdouble *params); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETER4FARBPROC) (GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETER4FVARBPROC) (GLenum target, GLuint index, const GLfloat *params); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETER4DARBPROC) (GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETER4DVARBPROC) (GLenum target, GLuint index, const GLdouble *params); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETER4FARBPROC) (GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETER4FVARBPROC) (GLenum target, GLuint index, const GLfloat *params); +typedef void (APIENTRYP PFNGLGETPROGRAMENVPARAMETERDVARBPROC) (GLenum target, GLuint index, GLdouble *params); +typedef void (APIENTRYP PFNGLGETPROGRAMENVPARAMETERFVARBPROC) (GLenum target, GLuint index, GLfloat *params); +typedef void (APIENTRYP PFNGLGETPROGRAMLOCALPARAMETERDVARBPROC) (GLenum target, GLuint index, GLdouble *params); +typedef void (APIENTRYP PFNGLGETPROGRAMLOCALPARAMETERFVARBPROC) (GLenum target, GLuint index, GLfloat *params); +typedef void (APIENTRYP PFNGLGETPROGRAMIVARBPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMSTRINGARBPROC) (GLenum target, GLenum pname, GLvoid *string); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBDVARBPROC) (GLuint index, GLenum pname, GLdouble *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBFVARBPROC) (GLuint index, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIVARBPROC) (GLuint index, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBPOINTERVARBPROC) (GLuint index, GLenum pname, GLvoid* *pointer); +typedef GLboolean (APIENTRYP PFNGLISPROGRAMARBPROC) (GLuint program); +#endif + +#ifndef GL_ARB_fragment_program +#define GL_ARB_fragment_program 1 +/* All ARB_fragment_program entry points are shared with ARB_vertex_program. */ +#endif + +#ifndef GL_ARB_vertex_buffer_object +#define GL_ARB_vertex_buffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBindBufferARB (GLenum, GLuint); +GLAPI void APIENTRY glDeleteBuffersARB (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenBuffersARB (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsBufferARB (GLuint); +GLAPI void APIENTRY glBufferDataARB (GLenum, GLsizeiptrARB, const GLvoid *, GLenum); +GLAPI void APIENTRY glBufferSubDataARB (GLenum, GLintptrARB, GLsizeiptrARB, const GLvoid *); +GLAPI void APIENTRY glGetBufferSubDataARB (GLenum, GLintptrARB, GLsizeiptrARB, GLvoid *); +GLAPI GLvoid* APIENTRY glMapBufferARB (GLenum, GLenum); +GLAPI GLboolean APIENTRY glUnmapBufferARB (GLenum); +GLAPI void APIENTRY glGetBufferParameterivARB (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetBufferPointervARB (GLenum, GLenum, GLvoid* *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBINDBUFFERARBPROC) (GLenum target, GLuint buffer); +typedef void (APIENTRYP PFNGLDELETEBUFFERSARBPROC) (GLsizei n, const GLuint *buffers); +typedef void (APIENTRYP PFNGLGENBUFFERSARBPROC) (GLsizei n, GLuint *buffers); +typedef GLboolean (APIENTRYP PFNGLISBUFFERARBPROC) (GLuint buffer); +typedef void (APIENTRYP PFNGLBUFFERDATAARBPROC) (GLenum target, GLsizeiptrARB size, const GLvoid *data, GLenum usage); +typedef void (APIENTRYP PFNGLBUFFERSUBDATAARBPROC) (GLenum target, GLintptrARB offset, GLsizeiptrARB size, const GLvoid *data); +typedef void (APIENTRYP PFNGLGETBUFFERSUBDATAARBPROC) (GLenum target, GLintptrARB offset, GLsizeiptrARB size, GLvoid *data); +typedef GLvoid* (APIENTRYP PFNGLMAPBUFFERARBPROC) (GLenum target, GLenum access); +typedef GLboolean (APIENTRYP PFNGLUNMAPBUFFERARBPROC) (GLenum target); +typedef void (APIENTRYP PFNGLGETBUFFERPARAMETERIVARBPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETBUFFERPOINTERVARBPROC) (GLenum target, GLenum pname, GLvoid* *params); +#endif + +#ifndef GL_ARB_occlusion_query +#define GL_ARB_occlusion_query 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGenQueriesARB (GLsizei, GLuint *); +GLAPI void APIENTRY glDeleteQueriesARB (GLsizei, const GLuint *); +GLAPI GLboolean APIENTRY glIsQueryARB (GLuint); +GLAPI void APIENTRY glBeginQueryARB (GLenum, GLuint); +GLAPI void APIENTRY glEndQueryARB (GLenum); +GLAPI void APIENTRY glGetQueryivARB (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetQueryObjectivARB (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetQueryObjectuivARB (GLuint, GLenum, GLuint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGENQUERIESARBPROC) (GLsizei n, GLuint *ids); +typedef void (APIENTRYP PFNGLDELETEQUERIESARBPROC) (GLsizei n, const GLuint *ids); +typedef GLboolean (APIENTRYP PFNGLISQUERYARBPROC) (GLuint id); +typedef void (APIENTRYP PFNGLBEGINQUERYARBPROC) (GLenum target, GLuint id); +typedef void (APIENTRYP PFNGLENDQUERYARBPROC) (GLenum target); +typedef void (APIENTRYP PFNGLGETQUERYIVARBPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETQUERYOBJECTIVARBPROC) (GLuint id, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETQUERYOBJECTUIVARBPROC) (GLuint id, GLenum pname, GLuint *params); +#endif + +#ifndef GL_ARB_shader_objects +#define GL_ARB_shader_objects 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDeleteObjectARB (GLhandleARB); +GLAPI GLhandleARB APIENTRY glGetHandleARB (GLenum); +GLAPI void APIENTRY glDetachObjectARB (GLhandleARB, GLhandleARB); +GLAPI GLhandleARB APIENTRY glCreateShaderObjectARB (GLenum); +GLAPI void APIENTRY glShaderSourceARB (GLhandleARB, GLsizei, const GLcharARB* *, const GLint *); +GLAPI void APIENTRY glCompileShaderARB (GLhandleARB); +GLAPI GLhandleARB APIENTRY glCreateProgramObjectARB (void); +GLAPI void APIENTRY glAttachObjectARB (GLhandleARB, GLhandleARB); +GLAPI void APIENTRY glLinkProgramARB (GLhandleARB); +GLAPI void APIENTRY glUseProgramObjectARB (GLhandleARB); +GLAPI void APIENTRY glValidateProgramARB (GLhandleARB); +GLAPI void APIENTRY glUniform1fARB (GLint, GLfloat); +GLAPI void APIENTRY glUniform2fARB (GLint, GLfloat, GLfloat); +GLAPI void APIENTRY glUniform3fARB (GLint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glUniform4fARB (GLint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glUniform1iARB (GLint, GLint); +GLAPI void APIENTRY glUniform2iARB (GLint, GLint, GLint); +GLAPI void APIENTRY glUniform3iARB (GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glUniform4iARB (GLint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glUniform1fvARB (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform2fvARB (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform3fvARB (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform4fvARB (GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glUniform1ivARB (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniform2ivARB (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniform3ivARB (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniform4ivARB (GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glUniformMatrix2fvARB (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix3fvARB (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glUniformMatrix4fvARB (GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glGetObjectParameterfvARB (GLhandleARB, GLenum, GLfloat *); +GLAPI void APIENTRY glGetObjectParameterivARB (GLhandleARB, GLenum, GLint *); +GLAPI void APIENTRY glGetInfoLogARB (GLhandleARB, GLsizei, GLsizei *, GLcharARB *); +GLAPI void APIENTRY glGetAttachedObjectsARB (GLhandleARB, GLsizei, GLsizei *, GLhandleARB *); +GLAPI GLint APIENTRY glGetUniformLocationARB (GLhandleARB, const GLcharARB *); +GLAPI void APIENTRY glGetActiveUniformARB (GLhandleARB, GLuint, GLsizei, GLsizei *, GLint *, GLenum *, GLcharARB *); +GLAPI void APIENTRY glGetUniformfvARB (GLhandleARB, GLint, GLfloat *); +GLAPI void APIENTRY glGetUniformivARB (GLhandleARB, GLint, GLint *); +GLAPI void APIENTRY glGetShaderSourceARB (GLhandleARB, GLsizei, GLsizei *, GLcharARB *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDELETEOBJECTARBPROC) (GLhandleARB obj); +typedef GLhandleARB (APIENTRYP PFNGLGETHANDLEARBPROC) (GLenum pname); +typedef void (APIENTRYP PFNGLDETACHOBJECTARBPROC) (GLhandleARB containerObj, GLhandleARB attachedObj); +typedef GLhandleARB (APIENTRYP PFNGLCREATESHADEROBJECTARBPROC) (GLenum shaderType); +typedef void (APIENTRYP PFNGLSHADERSOURCEARBPROC) (GLhandleARB shaderObj, GLsizei count, const GLcharARB* *string, const GLint *length); +typedef void (APIENTRYP PFNGLCOMPILESHADERARBPROC) (GLhandleARB shaderObj); +typedef GLhandleARB (APIENTRYP PFNGLCREATEPROGRAMOBJECTARBPROC) (void); +typedef void (APIENTRYP PFNGLATTACHOBJECTARBPROC) (GLhandleARB containerObj, GLhandleARB obj); +typedef void (APIENTRYP PFNGLLINKPROGRAMARBPROC) (GLhandleARB programObj); +typedef void (APIENTRYP PFNGLUSEPROGRAMOBJECTARBPROC) (GLhandleARB programObj); +typedef void (APIENTRYP PFNGLVALIDATEPROGRAMARBPROC) (GLhandleARB programObj); +typedef void (APIENTRYP PFNGLUNIFORM1FARBPROC) (GLint location, GLfloat v0); +typedef void (APIENTRYP PFNGLUNIFORM2FARBPROC) (GLint location, GLfloat v0, GLfloat v1); +typedef void (APIENTRYP PFNGLUNIFORM3FARBPROC) (GLint location, GLfloat v0, GLfloat v1, GLfloat v2); +typedef void (APIENTRYP PFNGLUNIFORM4FARBPROC) (GLint location, GLfloat v0, GLfloat v1, GLfloat v2, GLfloat v3); +typedef void (APIENTRYP PFNGLUNIFORM1IARBPROC) (GLint location, GLint v0); +typedef void (APIENTRYP PFNGLUNIFORM2IARBPROC) (GLint location, GLint v0, GLint v1); +typedef void (APIENTRYP PFNGLUNIFORM3IARBPROC) (GLint location, GLint v0, GLint v1, GLint v2); +typedef void (APIENTRYP PFNGLUNIFORM4IARBPROC) (GLint location, GLint v0, GLint v1, GLint v2, GLint v3); +typedef void (APIENTRYP PFNGLUNIFORM1FVARBPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM2FVARBPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM3FVARBPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM4FVARBPROC) (GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORM1IVARBPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORM2IVARBPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORM3IVARBPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORM4IVARBPROC) (GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX2FVARBPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX3FVARBPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLUNIFORMMATRIX4FVARBPROC) (GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLGETOBJECTPARAMETERFVARBPROC) (GLhandleARB obj, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETOBJECTPARAMETERIVARBPROC) (GLhandleARB obj, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETINFOLOGARBPROC) (GLhandleARB obj, GLsizei maxLength, GLsizei *length, GLcharARB *infoLog); +typedef void (APIENTRYP PFNGLGETATTACHEDOBJECTSARBPROC) (GLhandleARB containerObj, GLsizei maxCount, GLsizei *count, GLhandleARB *obj); +typedef GLint (APIENTRYP PFNGLGETUNIFORMLOCATIONARBPROC) (GLhandleARB programObj, const GLcharARB *name); +typedef void (APIENTRYP PFNGLGETACTIVEUNIFORMARBPROC) (GLhandleARB programObj, GLuint index, GLsizei maxLength, GLsizei *length, GLint *size, GLenum *type, GLcharARB *name); +typedef void (APIENTRYP PFNGLGETUNIFORMFVARBPROC) (GLhandleARB programObj, GLint location, GLfloat *params); +typedef void (APIENTRYP PFNGLGETUNIFORMIVARBPROC) (GLhandleARB programObj, GLint location, GLint *params); +typedef void (APIENTRYP PFNGLGETSHADERSOURCEARBPROC) (GLhandleARB obj, GLsizei maxLength, GLsizei *length, GLcharARB *source); +#endif + +#ifndef GL_ARB_vertex_shader +#define GL_ARB_vertex_shader 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBindAttribLocationARB (GLhandleARB, GLuint, const GLcharARB *); +GLAPI void APIENTRY glGetActiveAttribARB (GLhandleARB, GLuint, GLsizei, GLsizei *, GLint *, GLenum *, GLcharARB *); +GLAPI GLint APIENTRY glGetAttribLocationARB (GLhandleARB, const GLcharARB *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBINDATTRIBLOCATIONARBPROC) (GLhandleARB programObj, GLuint index, const GLcharARB *name); +typedef void (APIENTRYP PFNGLGETACTIVEATTRIBARBPROC) (GLhandleARB programObj, GLuint index, GLsizei maxLength, GLsizei *length, GLint *size, GLenum *type, GLcharARB *name); +typedef GLint (APIENTRYP PFNGLGETATTRIBLOCATIONARBPROC) (GLhandleARB programObj, const GLcharARB *name); +#endif + +#ifndef GL_ARB_fragment_shader +#define GL_ARB_fragment_shader 1 +#endif + +#ifndef GL_ARB_shading_language_100 +#define GL_ARB_shading_language_100 1 +#endif + +#ifndef GL_ARB_texture_non_power_of_two +#define GL_ARB_texture_non_power_of_two 1 +#endif + +#ifndef GL_ARB_point_sprite +#define GL_ARB_point_sprite 1 +#endif + +#ifndef GL_ARB_fragment_program_shadow +#define GL_ARB_fragment_program_shadow 1 +#endif + +#ifndef GL_ARB_draw_buffers +#define GL_ARB_draw_buffers 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawBuffersARB (GLsizei, const GLenum *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWBUFFERSARBPROC) (GLsizei n, const GLenum *bufs); +#endif + +#ifndef GL_ARB_texture_rectangle +#define GL_ARB_texture_rectangle 1 +#endif + +#ifndef GL_ARB_color_buffer_float +#define GL_ARB_color_buffer_float 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glClampColorARB (GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCLAMPCOLORARBPROC) (GLenum target, GLenum clamp); +#endif + +#ifndef GL_ARB_half_float_pixel +#define GL_ARB_half_float_pixel 1 +#endif + +#ifndef GL_ARB_texture_float +#define GL_ARB_texture_float 1 +#endif + +#ifndef GL_ARB_pixel_buffer_object +#define GL_ARB_pixel_buffer_object 1 +#endif + +#ifndef GL_ARB_depth_buffer_float +#define GL_ARB_depth_buffer_float 1 +#endif + +#ifndef GL_ARB_draw_instanced +#define GL_ARB_draw_instanced 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawArraysInstancedARB (GLenum, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glDrawElementsInstancedARB (GLenum, GLsizei, GLenum, const GLvoid *, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWARRAYSINSTANCEDARBPROC) (GLenum mode, GLint first, GLsizei count, GLsizei primcount); +typedef void (APIENTRYP PFNGLDRAWELEMENTSINSTANCEDARBPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount); +#endif + +#ifndef GL_ARB_framebuffer_object +#define GL_ARB_framebuffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLboolean APIENTRY glIsRenderbuffer (GLuint); +GLAPI void APIENTRY glBindRenderbuffer (GLenum, GLuint); +GLAPI void APIENTRY glDeleteRenderbuffers (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenRenderbuffers (GLsizei, GLuint *); +GLAPI void APIENTRY glRenderbufferStorage (GLenum, GLenum, GLsizei, GLsizei); +GLAPI void APIENTRY glGetRenderbufferParameteriv (GLenum, GLenum, GLint *); +GLAPI GLboolean APIENTRY glIsFramebuffer (GLuint); +GLAPI void APIENTRY glBindFramebuffer (GLenum, GLuint); +GLAPI void APIENTRY glDeleteFramebuffers (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenFramebuffers (GLsizei, GLuint *); +GLAPI GLenum APIENTRY glCheckFramebufferStatus (GLenum); +GLAPI void APIENTRY glFramebufferTexture1D (GLenum, GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTexture2D (GLenum, GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTexture3D (GLenum, GLenum, GLenum, GLuint, GLint, GLint); +GLAPI void APIENTRY glFramebufferRenderbuffer (GLenum, GLenum, GLenum, GLuint); +GLAPI void APIENTRY glGetFramebufferAttachmentParameteriv (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGenerateMipmap (GLenum); +GLAPI void APIENTRY glBlitFramebuffer (GLint, GLint, GLint, GLint, GLint, GLint, GLint, GLint, GLbitfield, GLenum); +GLAPI void APIENTRY glRenderbufferStorageMultisample (GLenum, GLsizei, GLenum, GLsizei, GLsizei); +GLAPI void APIENTRY glFramebufferTextureLayer (GLenum, GLenum, GLuint, GLint, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLboolean (APIENTRYP PFNGLISRENDERBUFFERPROC) (GLuint renderbuffer); +typedef void (APIENTRYP PFNGLBINDRENDERBUFFERPROC) (GLenum target, GLuint renderbuffer); +typedef void (APIENTRYP PFNGLDELETERENDERBUFFERSPROC) (GLsizei n, const GLuint *renderbuffers); +typedef void (APIENTRYP PFNGLGENRENDERBUFFERSPROC) (GLsizei n, GLuint *renderbuffers); +typedef void (APIENTRYP PFNGLRENDERBUFFERSTORAGEPROC) (GLenum target, GLenum internalformat, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETRENDERBUFFERPARAMETERIVPROC) (GLenum target, GLenum pname, GLint *params); +typedef GLboolean (APIENTRYP PFNGLISFRAMEBUFFERPROC) (GLuint framebuffer); +typedef void (APIENTRYP PFNGLBINDFRAMEBUFFERPROC) (GLenum target, GLuint framebuffer); +typedef void (APIENTRYP PFNGLDELETEFRAMEBUFFERSPROC) (GLsizei n, const GLuint *framebuffers); +typedef void (APIENTRYP PFNGLGENFRAMEBUFFERSPROC) (GLsizei n, GLuint *framebuffers); +typedef GLenum (APIENTRYP PFNGLCHECKFRAMEBUFFERSTATUSPROC) (GLenum target); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURE1DPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURE2DPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURE3DPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); +typedef void (APIENTRYP PFNGLFRAMEBUFFERRENDERBUFFERPROC) (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); +typedef void (APIENTRYP PFNGLGETFRAMEBUFFERATTACHMENTPARAMETERIVPROC) (GLenum target, GLenum attachment, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGENERATEMIPMAPPROC) (GLenum target); +typedef void (APIENTRYP PFNGLBLITFRAMEBUFFERPROC) (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); +typedef void (APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURELAYERPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); +#endif + +#ifndef GL_ARB_framebuffer_sRGB +#define GL_ARB_framebuffer_sRGB 1 +#endif + +#ifndef GL_ARB_geometry_shader4 +#define GL_ARB_geometry_shader4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramParameteriARB (GLuint, GLenum, GLint); +GLAPI void APIENTRY glFramebufferTextureARB (GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTextureLayerARB (GLenum, GLenum, GLuint, GLint, GLint); +GLAPI void APIENTRY glFramebufferTextureFaceARB (GLenum, GLenum, GLuint, GLint, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMPARAMETERIARBPROC) (GLuint program, GLenum pname, GLint value); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREARBPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURELAYERARBPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREFACEARBPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLenum face); +#endif + +#ifndef GL_ARB_half_float_vertex +#define GL_ARB_half_float_vertex 1 +#endif + +#ifndef GL_ARB_instanced_arrays +#define GL_ARB_instanced_arrays 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexAttribDivisorARB (GLuint, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXATTRIBDIVISORARBPROC) (GLuint index, GLuint divisor); +#endif + +#ifndef GL_ARB_map_buffer_range +#define GL_ARB_map_buffer_range 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLvoid* APIENTRY glMapBufferRange (GLenum, GLintptr, GLsizeiptr, GLbitfield); +GLAPI void APIENTRY glFlushMappedBufferRange (GLenum, GLintptr, GLsizeiptr); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLvoid* (APIENTRYP PFNGLMAPBUFFERRANGEPROC) (GLenum target, GLintptr offset, GLsizeiptr length, GLbitfield access); +typedef void (APIENTRYP PFNGLFLUSHMAPPEDBUFFERRANGEPROC) (GLenum target, GLintptr offset, GLsizeiptr length); +#endif + +#ifndef GL_ARB_texture_buffer_object +#define GL_ARB_texture_buffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexBufferARB (GLenum, GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXBUFFERARBPROC) (GLenum target, GLenum internalformat, GLuint buffer); +#endif + +#ifndef GL_ARB_texture_compression_rgtc +#define GL_ARB_texture_compression_rgtc 1 +#endif + +#ifndef GL_ARB_texture_rg +#define GL_ARB_texture_rg 1 +#endif + +#ifndef GL_ARB_vertex_array_object +#define GL_ARB_vertex_array_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBindVertexArray (GLuint); +GLAPI void APIENTRY glDeleteVertexArrays (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenVertexArrays (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsVertexArray (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBINDVERTEXARRAYPROC) (GLuint array); +typedef void (APIENTRYP PFNGLDELETEVERTEXARRAYSPROC) (GLsizei n, const GLuint *arrays); +typedef void (APIENTRYP PFNGLGENVERTEXARRAYSPROC) (GLsizei n, GLuint *arrays); +typedef GLboolean (APIENTRYP PFNGLISVERTEXARRAYPROC) (GLuint array); +#endif + +#ifndef GL_ARB_uniform_buffer_object +#define GL_ARB_uniform_buffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetUniformIndices (GLuint, GLsizei, const GLchar* *, GLuint *); +GLAPI void APIENTRY glGetActiveUniformsiv (GLuint, GLsizei, const GLuint *, GLenum, GLint *); +GLAPI void APIENTRY glGetActiveUniformName (GLuint, GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI GLuint APIENTRY glGetUniformBlockIndex (GLuint, const GLchar *); +GLAPI void APIENTRY glGetActiveUniformBlockiv (GLuint, GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetActiveUniformBlockName (GLuint, GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI void APIENTRY glUniformBlockBinding (GLuint, GLuint, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETUNIFORMINDICESPROC) (GLuint program, GLsizei uniformCount, const GLchar* *uniformNames, GLuint *uniformIndices); +typedef void (APIENTRYP PFNGLGETACTIVEUNIFORMSIVPROC) (GLuint program, GLsizei uniformCount, const GLuint *uniformIndices, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETACTIVEUNIFORMNAMEPROC) (GLuint program, GLuint uniformIndex, GLsizei bufSize, GLsizei *length, GLchar *uniformName); +typedef GLuint (APIENTRYP PFNGLGETUNIFORMBLOCKINDEXPROC) (GLuint program, const GLchar *uniformBlockName); +typedef void (APIENTRYP PFNGLGETACTIVEUNIFORMBLOCKIVPROC) (GLuint program, GLuint uniformBlockIndex, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETACTIVEUNIFORMBLOCKNAMEPROC) (GLuint program, GLuint uniformBlockIndex, GLsizei bufSize, GLsizei *length, GLchar *uniformBlockName); +typedef void (APIENTRYP PFNGLUNIFORMBLOCKBINDINGPROC) (GLuint program, GLuint uniformBlockIndex, GLuint uniformBlockBinding); +#endif + +#ifndef GL_ARB_compatibility +#define GL_ARB_compatibility 1 +#endif + +#ifndef GL_ARB_copy_buffer +#define GL_ARB_copy_buffer 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCopyBufferSubData (GLenum, GLenum, GLintptr, GLintptr, GLsizeiptr); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOPYBUFFERSUBDATAPROC) (GLenum readTarget, GLenum writeTarget, GLintptr readOffset, GLintptr writeOffset, GLsizeiptr size); +#endif + +#ifndef GL_ARB_shader_texture_lod +#define GL_ARB_shader_texture_lod 1 +#endif + +#ifndef GL_ARB_depth_clamp +#define GL_ARB_depth_clamp 1 +#endif + +#ifndef GL_ARB_draw_elements_base_vertex +#define GL_ARB_draw_elements_base_vertex 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawElementsBaseVertex (GLenum, GLsizei, GLenum, const GLvoid *, GLint); +GLAPI void APIENTRY glDrawRangeElementsBaseVertex (GLenum, GLuint, GLuint, GLsizei, GLenum, const GLvoid *, GLint); +GLAPI void APIENTRY glDrawElementsInstancedBaseVertex (GLenum, GLsizei, GLenum, const GLvoid *, GLsizei, GLint); +GLAPI void APIENTRY glMultiDrawElementsBaseVertex (GLenum, const GLsizei *, GLenum, const GLvoid* *, GLsizei, const GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWELEMENTSBASEVERTEXPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLint basevertex); +typedef void (APIENTRYP PFNGLDRAWRANGEELEMENTSBASEVERTEXPROC) (GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid *indices, GLint basevertex); +typedef void (APIENTRYP PFNGLDRAWELEMENTSINSTANCEDBASEVERTEXPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount, GLint basevertex); +typedef void (APIENTRYP PFNGLMULTIDRAWELEMENTSBASEVERTEXPROC) (GLenum mode, const GLsizei *count, GLenum type, const GLvoid* *indices, GLsizei primcount, const GLint *basevertex); +#endif + +#ifndef GL_ARB_fragment_coord_conventions +#define GL_ARB_fragment_coord_conventions 1 +#endif + +#ifndef GL_ARB_provoking_vertex +#define GL_ARB_provoking_vertex 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProvokingVertex (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROVOKINGVERTEXPROC) (GLenum mode); +#endif + +#ifndef GL_ARB_seamless_cube_map +#define GL_ARB_seamless_cube_map 1 +#endif + +#ifndef GL_ARB_sync +#define GL_ARB_sync 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLsync APIENTRY glFenceSync (GLenum, GLbitfield); +GLAPI GLboolean APIENTRY glIsSync (GLsync); +GLAPI void APIENTRY glDeleteSync (GLsync); +GLAPI GLenum APIENTRY glClientWaitSync (GLsync, GLbitfield, GLuint64); +GLAPI void APIENTRY glWaitSync (GLsync, GLbitfield, GLuint64); +GLAPI void APIENTRY glGetInteger64v (GLenum, GLint64 *); +GLAPI void APIENTRY glGetSynciv (GLsync, GLenum, GLsizei, GLsizei *, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLsync (APIENTRYP PFNGLFENCESYNCPROC) (GLenum condition, GLbitfield flags); +typedef GLboolean (APIENTRYP PFNGLISSYNCPROC) (GLsync sync); +typedef void (APIENTRYP PFNGLDELETESYNCPROC) (GLsync sync); +typedef GLenum (APIENTRYP PFNGLCLIENTWAITSYNCPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout); +typedef void (APIENTRYP PFNGLWAITSYNCPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout); +typedef void (APIENTRYP PFNGLGETINTEGER64VPROC) (GLenum pname, GLint64 *params); +typedef void (APIENTRYP PFNGLGETSYNCIVPROC) (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length, GLint *values); +#endif + +#ifndef GL_ARB_texture_multisample +#define GL_ARB_texture_multisample 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexImage2DMultisample (GLenum, GLsizei, GLint, GLsizei, GLsizei, GLboolean); +GLAPI void APIENTRY glTexImage3DMultisample (GLenum, GLsizei, GLint, GLsizei, GLsizei, GLsizei, GLboolean); +GLAPI void APIENTRY glGetMultisamplefv (GLenum, GLuint, GLfloat *); +GLAPI void APIENTRY glSampleMaski (GLuint, GLbitfield); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXIMAGE2DMULTISAMPLEPROC) (GLenum target, GLsizei samples, GLint internalformat, GLsizei width, GLsizei height, GLboolean fixedsamplelocations); +typedef void (APIENTRYP PFNGLTEXIMAGE3DMULTISAMPLEPROC) (GLenum target, GLsizei samples, GLint internalformat, GLsizei width, GLsizei height, GLsizei depth, GLboolean fixedsamplelocations); +typedef void (APIENTRYP PFNGLGETMULTISAMPLEFVPROC) (GLenum pname, GLuint index, GLfloat *val); +typedef void (APIENTRYP PFNGLSAMPLEMASKIPROC) (GLuint index, GLbitfield mask); +#endif + +#ifndef GL_ARB_vertex_array_bgra +#define GL_ARB_vertex_array_bgra 1 +#endif + +#ifndef GL_ARB_draw_buffers_blend +#define GL_ARB_draw_buffers_blend 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendEquationi (GLuint, GLenum); +GLAPI void APIENTRY glBlendEquationSeparatei (GLuint, GLenum, GLenum); +GLAPI void APIENTRY glBlendFunci (GLuint, GLenum, GLenum); +GLAPI void APIENTRY glBlendFuncSeparatei (GLuint, GLenum, GLenum, GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDEQUATIONIPROC) (GLuint buf, GLenum mode); +typedef void (APIENTRYP PFNGLBLENDEQUATIONSEPARATEIPROC) (GLuint buf, GLenum modeRGB, GLenum modeAlpha); +typedef void (APIENTRYP PFNGLBLENDFUNCIPROC) (GLuint buf, GLenum src, GLenum dst); +typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEIPROC) (GLuint buf, GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha); +#endif + +#ifndef GL_ARB_sample_shading +#define GL_ARB_sample_shading 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glMinSampleShading (GLclampf); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLMINSAMPLESHADINGPROC) (GLclampf value); +#endif + +#ifndef GL_ARB_texture_cube_map_array +#define GL_ARB_texture_cube_map_array 1 +#endif + +#ifndef GL_ARB_texture_gather +#define GL_ARB_texture_gather 1 +#endif + +#ifndef GL_ARB_texture_query_lod +#define GL_ARB_texture_query_lod 1 +#endif + +#ifndef GL_EXT_abgr +#define GL_EXT_abgr 1 +#endif + +#ifndef GL_EXT_blend_color +#define GL_EXT_blend_color 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendColorEXT (GLclampf, GLclampf, GLclampf, GLclampf); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDCOLOREXTPROC) (GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha); +#endif + +#ifndef GL_EXT_polygon_offset +#define GL_EXT_polygon_offset 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPolygonOffsetEXT (GLfloat, GLfloat); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPOLYGONOFFSETEXTPROC) (GLfloat factor, GLfloat bias); +#endif + +#ifndef GL_EXT_texture +#define GL_EXT_texture 1 +#endif + +#ifndef GL_EXT_texture3D +#define GL_EXT_texture3D 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexImage3DEXT (GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTexSubImage3DEXT (GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXIMAGE3DEXTPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXSUBIMAGE3DEXTPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid *pixels); +#endif + +#ifndef GL_SGIS_texture_filter4 +#define GL_SGIS_texture_filter4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetTexFilterFuncSGIS (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glTexFilterFuncSGIS (GLenum, GLenum, GLsizei, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETTEXFILTERFUNCSGISPROC) (GLenum target, GLenum filter, GLfloat *weights); +typedef void (APIENTRYP PFNGLTEXFILTERFUNCSGISPROC) (GLenum target, GLenum filter, GLsizei n, const GLfloat *weights); +#endif + +#ifndef GL_EXT_subtexture +#define GL_EXT_subtexture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexSubImage1DEXT (GLenum, GLint, GLint, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTexSubImage2DEXT (GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXSUBIMAGE1DEXTPROC) (GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXSUBIMAGE2DEXTPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *pixels); +#endif + +#ifndef GL_EXT_copy_texture +#define GL_EXT_copy_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCopyTexImage1DEXT (GLenum, GLint, GLenum, GLint, GLint, GLsizei, GLint); +GLAPI void APIENTRY glCopyTexImage2DEXT (GLenum, GLint, GLenum, GLint, GLint, GLsizei, GLsizei, GLint); +GLAPI void APIENTRY glCopyTexSubImage1DEXT (GLenum, GLint, GLint, GLint, GLint, GLsizei); +GLAPI void APIENTRY glCopyTexSubImage2DEXT (GLenum, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glCopyTexSubImage3DEXT (GLenum, GLint, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOPYTEXIMAGE1DEXTPROC) (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLint border); +typedef void (APIENTRYP PFNGLCOPYTEXIMAGE2DEXTPROC) (GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border); +typedef void (APIENTRYP PFNGLCOPYTEXSUBIMAGE1DEXTPROC) (GLenum target, GLint level, GLint xoffset, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLCOPYTEXSUBIMAGE2DEXTPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLCOPYTEXSUBIMAGE3DEXTPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height); +#endif + +#ifndef GL_EXT_histogram +#define GL_EXT_histogram 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetHistogramEXT (GLenum, GLboolean, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetHistogramParameterfvEXT (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetHistogramParameterivEXT (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetMinmaxEXT (GLenum, GLboolean, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetMinmaxParameterfvEXT (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMinmaxParameterivEXT (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glHistogramEXT (GLenum, GLsizei, GLenum, GLboolean); +GLAPI void APIENTRY glMinmaxEXT (GLenum, GLenum, GLboolean); +GLAPI void APIENTRY glResetHistogramEXT (GLenum); +GLAPI void APIENTRY glResetMinmaxEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETHISTOGRAMEXTPROC) (GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values); +typedef void (APIENTRYP PFNGLGETHISTOGRAMPARAMETERFVEXTPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETHISTOGRAMPARAMETERIVEXTPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMINMAXEXTPROC) (GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values); +typedef void (APIENTRYP PFNGLGETMINMAXPARAMETERFVEXTPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMINMAXPARAMETERIVEXTPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLHISTOGRAMEXTPROC) (GLenum target, GLsizei width, GLenum internalformat, GLboolean sink); +typedef void (APIENTRYP PFNGLMINMAXEXTPROC) (GLenum target, GLenum internalformat, GLboolean sink); +typedef void (APIENTRYP PFNGLRESETHISTOGRAMEXTPROC) (GLenum target); +typedef void (APIENTRYP PFNGLRESETMINMAXEXTPROC) (GLenum target); +#endif + +#ifndef GL_EXT_convolution +#define GL_EXT_convolution 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glConvolutionFilter1DEXT (GLenum, GLenum, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glConvolutionFilter2DEXT (GLenum, GLenum, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glConvolutionParameterfEXT (GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glConvolutionParameterfvEXT (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glConvolutionParameteriEXT (GLenum, GLenum, GLint); +GLAPI void APIENTRY glConvolutionParameterivEXT (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glCopyConvolutionFilter1DEXT (GLenum, GLenum, GLint, GLint, GLsizei); +GLAPI void APIENTRY glCopyConvolutionFilter2DEXT (GLenum, GLenum, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glGetConvolutionFilterEXT (GLenum, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetConvolutionParameterfvEXT (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetConvolutionParameterivEXT (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetSeparableFilterEXT (GLenum, GLenum, GLenum, GLvoid *, GLvoid *, GLvoid *); +GLAPI void APIENTRY glSeparableFilter2DEXT (GLenum, GLenum, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCONVOLUTIONFILTER1DEXTPROC) (GLenum target, GLenum internalformat, GLsizei width, GLenum format, GLenum type, const GLvoid *image); +typedef void (APIENTRYP PFNGLCONVOLUTIONFILTER2DEXTPROC) (GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *image); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERFEXTPROC) (GLenum target, GLenum pname, GLfloat params); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERFVEXTPROC) (GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERIEXTPROC) (GLenum target, GLenum pname, GLint params); +typedef void (APIENTRYP PFNGLCONVOLUTIONPARAMETERIVEXTPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLCOPYCONVOLUTIONFILTER1DEXTPROC) (GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLCOPYCONVOLUTIONFILTER2DEXTPROC) (GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETCONVOLUTIONFILTEREXTPROC) (GLenum target, GLenum format, GLenum type, GLvoid *image); +typedef void (APIENTRYP PFNGLGETCONVOLUTIONPARAMETERFVEXTPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETCONVOLUTIONPARAMETERIVEXTPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETSEPARABLEFILTEREXTPROC) (GLenum target, GLenum format, GLenum type, GLvoid *row, GLvoid *column, GLvoid *span); +typedef void (APIENTRYP PFNGLSEPARABLEFILTER2DEXTPROC) (GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *row, const GLvoid *column); +#endif + +#ifndef GL_SGI_color_matrix +#define GL_SGI_color_matrix 1 +#endif + +#ifndef GL_SGI_color_table +#define GL_SGI_color_table 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorTableSGI (GLenum, GLenum, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glColorTableParameterfvSGI (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glColorTableParameterivSGI (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glCopyColorTableSGI (GLenum, GLenum, GLint, GLint, GLsizei); +GLAPI void APIENTRY glGetColorTableSGI (GLenum, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetColorTableParameterfvSGI (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetColorTableParameterivSGI (GLenum, GLenum, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLORTABLESGIPROC) (GLenum target, GLenum internalformat, GLsizei width, GLenum format, GLenum type, const GLvoid *table); +typedef void (APIENTRYP PFNGLCOLORTABLEPARAMETERFVSGIPROC) (GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLCOLORTABLEPARAMETERIVSGIPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLCOPYCOLORTABLESGIPROC) (GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLGETCOLORTABLESGIPROC) (GLenum target, GLenum format, GLenum type, GLvoid *table); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPARAMETERFVSGIPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPARAMETERIVSGIPROC) (GLenum target, GLenum pname, GLint *params); +#endif + +#ifndef GL_SGIX_pixel_texture +#define GL_SGIX_pixel_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPixelTexGenSGIX (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPIXELTEXGENSGIXPROC) (GLenum mode); +#endif + +#ifndef GL_SGIS_pixel_texture +#define GL_SGIS_pixel_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPixelTexGenParameteriSGIS (GLenum, GLint); +GLAPI void APIENTRY glPixelTexGenParameterivSGIS (GLenum, const GLint *); +GLAPI void APIENTRY glPixelTexGenParameterfSGIS (GLenum, GLfloat); +GLAPI void APIENTRY glPixelTexGenParameterfvSGIS (GLenum, const GLfloat *); +GLAPI void APIENTRY glGetPixelTexGenParameterivSGIS (GLenum, GLint *); +GLAPI void APIENTRY glGetPixelTexGenParameterfvSGIS (GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPIXELTEXGENPARAMETERISGISPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLPIXELTEXGENPARAMETERIVSGISPROC) (GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLPIXELTEXGENPARAMETERFSGISPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLPIXELTEXGENPARAMETERFVSGISPROC) (GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLGETPIXELTEXGENPARAMETERIVSGISPROC) (GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETPIXELTEXGENPARAMETERFVSGISPROC) (GLenum pname, GLfloat *params); +#endif + +#ifndef GL_SGIS_texture4D +#define GL_SGIS_texture4D 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexImage4DSGIS (GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTexSubImage4DSGIS (GLenum, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXIMAGE4DSGISPROC) (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLsizei size4d, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXSUBIMAGE4DSGISPROC) (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint woffset, GLsizei width, GLsizei height, GLsizei depth, GLsizei size4d, GLenum format, GLenum type, const GLvoid *pixels); +#endif + +#ifndef GL_SGI_texture_color_table +#define GL_SGI_texture_color_table 1 +#endif + +#ifndef GL_EXT_cmyka +#define GL_EXT_cmyka 1 +#endif + +#ifndef GL_EXT_texture_object +#define GL_EXT_texture_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLboolean APIENTRY glAreTexturesResidentEXT (GLsizei, const GLuint *, GLboolean *); +GLAPI void APIENTRY glBindTextureEXT (GLenum, GLuint); +GLAPI void APIENTRY glDeleteTexturesEXT (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenTexturesEXT (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsTextureEXT (GLuint); +GLAPI void APIENTRY glPrioritizeTexturesEXT (GLsizei, const GLuint *, const GLclampf *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLboolean (APIENTRYP PFNGLARETEXTURESRESIDENTEXTPROC) (GLsizei n, const GLuint *textures, GLboolean *residences); +typedef void (APIENTRYP PFNGLBINDTEXTUREEXTPROC) (GLenum target, GLuint texture); +typedef void (APIENTRYP PFNGLDELETETEXTURESEXTPROC) (GLsizei n, const GLuint *textures); +typedef void (APIENTRYP PFNGLGENTEXTURESEXTPROC) (GLsizei n, GLuint *textures); +typedef GLboolean (APIENTRYP PFNGLISTEXTUREEXTPROC) (GLuint texture); +typedef void (APIENTRYP PFNGLPRIORITIZETEXTURESEXTPROC) (GLsizei n, const GLuint *textures, const GLclampf *priorities); +#endif + +#ifndef GL_SGIS_detail_texture +#define GL_SGIS_detail_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDetailTexFuncSGIS (GLenum, GLsizei, const GLfloat *); +GLAPI void APIENTRY glGetDetailTexFuncSGIS (GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDETAILTEXFUNCSGISPROC) (GLenum target, GLsizei n, const GLfloat *points); +typedef void (APIENTRYP PFNGLGETDETAILTEXFUNCSGISPROC) (GLenum target, GLfloat *points); +#endif + +#ifndef GL_SGIS_sharpen_texture +#define GL_SGIS_sharpen_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glSharpenTexFuncSGIS (GLenum, GLsizei, const GLfloat *); +GLAPI void APIENTRY glGetSharpenTexFuncSGIS (GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSHARPENTEXFUNCSGISPROC) (GLenum target, GLsizei n, const GLfloat *points); +typedef void (APIENTRYP PFNGLGETSHARPENTEXFUNCSGISPROC) (GLenum target, GLfloat *points); +#endif + +#ifndef GL_EXT_packed_pixels +#define GL_EXT_packed_pixels 1 +#endif + +#ifndef GL_SGIS_texture_lod +#define GL_SGIS_texture_lod 1 +#endif + +#ifndef GL_SGIS_multisample +#define GL_SGIS_multisample 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glSampleMaskSGIS (GLclampf, GLboolean); +GLAPI void APIENTRY glSamplePatternSGIS (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSAMPLEMASKSGISPROC) (GLclampf value, GLboolean invert); +typedef void (APIENTRYP PFNGLSAMPLEPATTERNSGISPROC) (GLenum pattern); +#endif + +#ifndef GL_EXT_rescale_normal +#define GL_EXT_rescale_normal 1 +#endif + +#ifndef GL_EXT_vertex_array +#define GL_EXT_vertex_array 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glArrayElementEXT (GLint); +GLAPI void APIENTRY glColorPointerEXT (GLint, GLenum, GLsizei, GLsizei, const GLvoid *); +GLAPI void APIENTRY glDrawArraysEXT (GLenum, GLint, GLsizei); +GLAPI void APIENTRY glEdgeFlagPointerEXT (GLsizei, GLsizei, const GLboolean *); +GLAPI void APIENTRY glGetPointervEXT (GLenum, GLvoid* *); +GLAPI void APIENTRY glIndexPointerEXT (GLenum, GLsizei, GLsizei, const GLvoid *); +GLAPI void APIENTRY glNormalPointerEXT (GLenum, GLsizei, GLsizei, const GLvoid *); +GLAPI void APIENTRY glTexCoordPointerEXT (GLint, GLenum, GLsizei, GLsizei, const GLvoid *); +GLAPI void APIENTRY glVertexPointerEXT (GLint, GLenum, GLsizei, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLARRAYELEMENTEXTPROC) (GLint i); +typedef void (APIENTRYP PFNGLCOLORPOINTEREXTPROC) (GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLDRAWARRAYSEXTPROC) (GLenum mode, GLint first, GLsizei count); +typedef void (APIENTRYP PFNGLEDGEFLAGPOINTEREXTPROC) (GLsizei stride, GLsizei count, const GLboolean *pointer); +typedef void (APIENTRYP PFNGLGETPOINTERVEXTPROC) (GLenum pname, GLvoid* *params); +typedef void (APIENTRYP PFNGLINDEXPOINTEREXTPROC) (GLenum type, GLsizei stride, GLsizei count, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLNORMALPOINTEREXTPROC) (GLenum type, GLsizei stride, GLsizei count, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLTEXCOORDPOINTEREXTPROC) (GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLVERTEXPOINTEREXTPROC) (GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid *pointer); +#endif + +#ifndef GL_EXT_misc_attribute +#define GL_EXT_misc_attribute 1 +#endif + +#ifndef GL_SGIS_generate_mipmap +#define GL_SGIS_generate_mipmap 1 +#endif + +#ifndef GL_SGIX_clipmap +#define GL_SGIX_clipmap 1 +#endif + +#ifndef GL_SGIX_shadow +#define GL_SGIX_shadow 1 +#endif + +#ifndef GL_SGIS_texture_edge_clamp +#define GL_SGIS_texture_edge_clamp 1 +#endif + +#ifndef GL_SGIS_texture_border_clamp +#define GL_SGIS_texture_border_clamp 1 +#endif + +#ifndef GL_EXT_blend_minmax +#define GL_EXT_blend_minmax 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendEquationEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDEQUATIONEXTPROC) (GLenum mode); +#endif + +#ifndef GL_EXT_blend_subtract +#define GL_EXT_blend_subtract 1 +#endif + +#ifndef GL_EXT_blend_logic_op +#define GL_EXT_blend_logic_op 1 +#endif + +#ifndef GL_SGIX_interlace +#define GL_SGIX_interlace 1 +#endif + +#ifndef GL_SGIX_pixel_tiles +#define GL_SGIX_pixel_tiles 1 +#endif + +#ifndef GL_SGIX_texture_select +#define GL_SGIX_texture_select 1 +#endif + +#ifndef GL_SGIX_sprite +#define GL_SGIX_sprite 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glSpriteParameterfSGIX (GLenum, GLfloat); +GLAPI void APIENTRY glSpriteParameterfvSGIX (GLenum, const GLfloat *); +GLAPI void APIENTRY glSpriteParameteriSGIX (GLenum, GLint); +GLAPI void APIENTRY glSpriteParameterivSGIX (GLenum, const GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSPRITEPARAMETERFSGIXPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLSPRITEPARAMETERFVSGIXPROC) (GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLSPRITEPARAMETERISGIXPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLSPRITEPARAMETERIVSGIXPROC) (GLenum pname, const GLint *params); +#endif + +#ifndef GL_SGIX_texture_multi_buffer +#define GL_SGIX_texture_multi_buffer 1 +#endif + +#ifndef GL_EXT_point_parameters +#define GL_EXT_point_parameters 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPointParameterfEXT (GLenum, GLfloat); +GLAPI void APIENTRY glPointParameterfvEXT (GLenum, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPOINTPARAMETERFEXTPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLPOINTPARAMETERFVEXTPROC) (GLenum pname, const GLfloat *params); +#endif + +#ifndef GL_SGIS_point_parameters +#define GL_SGIS_point_parameters 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPointParameterfSGIS (GLenum, GLfloat); +GLAPI void APIENTRY glPointParameterfvSGIS (GLenum, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPOINTPARAMETERFSGISPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLPOINTPARAMETERFVSGISPROC) (GLenum pname, const GLfloat *params); +#endif + +#ifndef GL_SGIX_instruments +#define GL_SGIX_instruments 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLint APIENTRY glGetInstrumentsSGIX (void); +GLAPI void APIENTRY glInstrumentsBufferSGIX (GLsizei, GLint *); +GLAPI GLint APIENTRY glPollInstrumentsSGIX (GLint *); +GLAPI void APIENTRY glReadInstrumentsSGIX (GLint); +GLAPI void APIENTRY glStartInstrumentsSGIX (void); +GLAPI void APIENTRY glStopInstrumentsSGIX (GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLint (APIENTRYP PFNGLGETINSTRUMENTSSGIXPROC) (void); +typedef void (APIENTRYP PFNGLINSTRUMENTSBUFFERSGIXPROC) (GLsizei size, GLint *buffer); +typedef GLint (APIENTRYP PFNGLPOLLINSTRUMENTSSGIXPROC) (GLint *marker_p); +typedef void (APIENTRYP PFNGLREADINSTRUMENTSSGIXPROC) (GLint marker); +typedef void (APIENTRYP PFNGLSTARTINSTRUMENTSSGIXPROC) (void); +typedef void (APIENTRYP PFNGLSTOPINSTRUMENTSSGIXPROC) (GLint marker); +#endif + +#ifndef GL_SGIX_texture_scale_bias +#define GL_SGIX_texture_scale_bias 1 +#endif + +#ifndef GL_SGIX_framezoom +#define GL_SGIX_framezoom 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFrameZoomSGIX (GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFRAMEZOOMSGIXPROC) (GLint factor); +#endif + +#ifndef GL_SGIX_tag_sample_buffer +#define GL_SGIX_tag_sample_buffer 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTagSampleBufferSGIX (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTAGSAMPLEBUFFERSGIXPROC) (void); +#endif + +#ifndef GL_SGIX_polynomial_ffd +#define GL_SGIX_polynomial_ffd 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDeformationMap3dSGIX (GLenum, GLdouble, GLdouble, GLint, GLint, GLdouble, GLdouble, GLint, GLint, GLdouble, GLdouble, GLint, GLint, const GLdouble *); +GLAPI void APIENTRY glDeformationMap3fSGIX (GLenum, GLfloat, GLfloat, GLint, GLint, GLfloat, GLfloat, GLint, GLint, GLfloat, GLfloat, GLint, GLint, const GLfloat *); +GLAPI void APIENTRY glDeformSGIX (GLbitfield); +GLAPI void APIENTRY glLoadIdentityDeformationMapSGIX (GLbitfield); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDEFORMATIONMAP3DSGIXPROC) (GLenum target, GLdouble u1, GLdouble u2, GLint ustride, GLint uorder, GLdouble v1, GLdouble v2, GLint vstride, GLint vorder, GLdouble w1, GLdouble w2, GLint wstride, GLint worder, const GLdouble *points); +typedef void (APIENTRYP PFNGLDEFORMATIONMAP3FSGIXPROC) (GLenum target, GLfloat u1, GLfloat u2, GLint ustride, GLint uorder, GLfloat v1, GLfloat v2, GLint vstride, GLint vorder, GLfloat w1, GLfloat w2, GLint wstride, GLint worder, const GLfloat *points); +typedef void (APIENTRYP PFNGLDEFORMSGIXPROC) (GLbitfield mask); +typedef void (APIENTRYP PFNGLLOADIDENTITYDEFORMATIONMAPSGIXPROC) (GLbitfield mask); +#endif + +#ifndef GL_SGIX_reference_plane +#define GL_SGIX_reference_plane 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glReferencePlaneSGIX (const GLdouble *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLREFERENCEPLANESGIXPROC) (const GLdouble *equation); +#endif + +#ifndef GL_SGIX_flush_raster +#define GL_SGIX_flush_raster 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFlushRasterSGIX (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFLUSHRASTERSGIXPROC) (void); +#endif + +#ifndef GL_SGIX_depth_texture +#define GL_SGIX_depth_texture 1 +#endif + +#ifndef GL_SGIS_fog_function +#define GL_SGIS_fog_function 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFogFuncSGIS (GLsizei, const GLfloat *); +GLAPI void APIENTRY glGetFogFuncSGIS (GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFOGFUNCSGISPROC) (GLsizei n, const GLfloat *points); +typedef void (APIENTRYP PFNGLGETFOGFUNCSGISPROC) (GLfloat *points); +#endif + +#ifndef GL_SGIX_fog_offset +#define GL_SGIX_fog_offset 1 +#endif + +#ifndef GL_HP_image_transform +#define GL_HP_image_transform 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glImageTransformParameteriHP (GLenum, GLenum, GLint); +GLAPI void APIENTRY glImageTransformParameterfHP (GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glImageTransformParameterivHP (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glImageTransformParameterfvHP (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glGetImageTransformParameterivHP (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetImageTransformParameterfvHP (GLenum, GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLIMAGETRANSFORMPARAMETERIHPPROC) (GLenum target, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLIMAGETRANSFORMPARAMETERFHPPROC) (GLenum target, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLIMAGETRANSFORMPARAMETERIVHPPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLIMAGETRANSFORMPARAMETERFVHPPROC) (GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLGETIMAGETRANSFORMPARAMETERIVHPPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETIMAGETRANSFORMPARAMETERFVHPPROC) (GLenum target, GLenum pname, GLfloat *params); +#endif + +#ifndef GL_HP_convolution_border_modes +#define GL_HP_convolution_border_modes 1 +#endif + +#ifndef GL_SGIX_texture_add_env +#define GL_SGIX_texture_add_env 1 +#endif + +#ifndef GL_EXT_color_subtable +#define GL_EXT_color_subtable 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorSubTableEXT (GLenum, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyColorSubTableEXT (GLenum, GLsizei, GLint, GLint, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +//typedef void (APIENTRYP PFNGLCOLORSUBTABLEEXTPROC) (GLenum target, GLsizei start, GLsizei count, GLenum format, GLenum type, const GLvoid *data); +typedef void (APIENTRYP PFNGLCOPYCOLORSUBTABLEEXTPROC) (GLenum target, GLsizei start, GLint x, GLint y, GLsizei width); +#endif + +#ifndef GL_PGI_vertex_hints +#define GL_PGI_vertex_hints 1 +#endif + +#ifndef GL_PGI_misc_hints +#define GL_PGI_misc_hints 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glHintPGI (GLenum, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLHINTPGIPROC) (GLenum target, GLint mode); +#endif + +#ifndef GL_EXT_paletted_texture +#define GL_EXT_paletted_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorTableEXT (GLenum, GLenum, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glGetColorTableEXT (GLenum, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetColorTableParameterivEXT (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetColorTableParameterfvEXT (GLenum, GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLORTABLEEXTPROC) (GLenum target, GLenum internalFormat, GLsizei width, GLenum format, GLenum type, const GLvoid *table); +typedef void (APIENTRYP PFNGLGETCOLORTABLEEXTPROC) (GLenum target, GLenum format, GLenum type, GLvoid *data); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPARAMETERIVEXTPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETCOLORTABLEPARAMETERFVEXTPROC) (GLenum target, GLenum pname, GLfloat *params); +#endif + +#ifndef GL_EXT_clip_volume_hint +#define GL_EXT_clip_volume_hint 1 +#endif + +#ifndef GL_SGIX_list_priority +#define GL_SGIX_list_priority 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetListParameterfvSGIX (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetListParameterivSGIX (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glListParameterfSGIX (GLuint, GLenum, GLfloat); +GLAPI void APIENTRY glListParameterfvSGIX (GLuint, GLenum, const GLfloat *); +GLAPI void APIENTRY glListParameteriSGIX (GLuint, GLenum, GLint); +GLAPI void APIENTRY glListParameterivSGIX (GLuint, GLenum, const GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETLISTPARAMETERFVSGIXPROC) (GLuint list, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETLISTPARAMETERIVSGIXPROC) (GLuint list, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLLISTPARAMETERFSGIXPROC) (GLuint list, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLLISTPARAMETERFVSGIXPROC) (GLuint list, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLLISTPARAMETERISGIXPROC) (GLuint list, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLLISTPARAMETERIVSGIXPROC) (GLuint list, GLenum pname, const GLint *params); +#endif + +#ifndef GL_SGIX_ir_instrument1 +#define GL_SGIX_ir_instrument1 1 +#endif + +#ifndef GL_SGIX_calligraphic_fragment +#define GL_SGIX_calligraphic_fragment 1 +#endif + +#ifndef GL_SGIX_texture_lod_bias +#define GL_SGIX_texture_lod_bias 1 +#endif + +#ifndef GL_SGIX_shadow_ambient +#define GL_SGIX_shadow_ambient 1 +#endif + +#ifndef GL_EXT_index_texture +#define GL_EXT_index_texture 1 +#endif + +#ifndef GL_EXT_index_material +#define GL_EXT_index_material 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glIndexMaterialEXT (GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLINDEXMATERIALEXTPROC) (GLenum face, GLenum mode); +#endif + +#ifndef GL_EXT_index_func +#define GL_EXT_index_func 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glIndexFuncEXT (GLenum, GLclampf); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLINDEXFUNCEXTPROC) (GLenum func, GLclampf ref); +#endif + +#ifndef GL_EXT_index_array_formats +#define GL_EXT_index_array_formats 1 +#endif + +#ifndef GL_EXT_compiled_vertex_array +#define GL_EXT_compiled_vertex_array 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glLockArraysEXT (GLint, GLsizei); +GLAPI void APIENTRY glUnlockArraysEXT (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLLOCKARRAYSEXTPROC) (GLint first, GLsizei count); +typedef void (APIENTRYP PFNGLUNLOCKARRAYSEXTPROC) (void); +#endif + +#ifndef GL_EXT_cull_vertex +#define GL_EXT_cull_vertex 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCullParameterdvEXT (GLenum, GLdouble *); +GLAPI void APIENTRY glCullParameterfvEXT (GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCULLPARAMETERDVEXTPROC) (GLenum pname, GLdouble *params); +typedef void (APIENTRYP PFNGLCULLPARAMETERFVEXTPROC) (GLenum pname, GLfloat *params); +#endif + +#ifndef GL_SGIX_ycrcb +#define GL_SGIX_ycrcb 1 +#endif + +#ifndef GL_SGIX_fragment_lighting +#define GL_SGIX_fragment_lighting 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFragmentColorMaterialSGIX (GLenum, GLenum); +GLAPI void APIENTRY glFragmentLightfSGIX (GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glFragmentLightfvSGIX (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glFragmentLightiSGIX (GLenum, GLenum, GLint); +GLAPI void APIENTRY glFragmentLightivSGIX (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glFragmentLightModelfSGIX (GLenum, GLfloat); +GLAPI void APIENTRY glFragmentLightModelfvSGIX (GLenum, const GLfloat *); +GLAPI void APIENTRY glFragmentLightModeliSGIX (GLenum, GLint); +GLAPI void APIENTRY glFragmentLightModelivSGIX (GLenum, const GLint *); +GLAPI void APIENTRY glFragmentMaterialfSGIX (GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glFragmentMaterialfvSGIX (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glFragmentMaterialiSGIX (GLenum, GLenum, GLint); +GLAPI void APIENTRY glFragmentMaterialivSGIX (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glGetFragmentLightfvSGIX (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetFragmentLightivSGIX (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetFragmentMaterialfvSGIX (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetFragmentMaterialivSGIX (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glLightEnviSGIX (GLenum, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFRAGMENTCOLORMATERIALSGIXPROC) (GLenum face, GLenum mode); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTFSGIXPROC) (GLenum light, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTFVSGIXPROC) (GLenum light, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTISGIXPROC) (GLenum light, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTIVSGIXPROC) (GLenum light, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTMODELFSGIXPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTMODELFVSGIXPROC) (GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTMODELISGIXPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLFRAGMENTLIGHTMODELIVSGIXPROC) (GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLFRAGMENTMATERIALFSGIXPROC) (GLenum face, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLFRAGMENTMATERIALFVSGIXPROC) (GLenum face, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLFRAGMENTMATERIALISGIXPROC) (GLenum face, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLFRAGMENTMATERIALIVSGIXPROC) (GLenum face, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLGETFRAGMENTLIGHTFVSGIXPROC) (GLenum light, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETFRAGMENTLIGHTIVSGIXPROC) (GLenum light, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETFRAGMENTMATERIALFVSGIXPROC) (GLenum face, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETFRAGMENTMATERIALIVSGIXPROC) (GLenum face, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLLIGHTENVISGIXPROC) (GLenum pname, GLint param); +#endif + +#ifndef GL_IBM_rasterpos_clip +#define GL_IBM_rasterpos_clip 1 +#endif + +#ifndef GL_HP_texture_lighting +#define GL_HP_texture_lighting 1 +#endif + +#ifndef GL_EXT_draw_range_elements +#define GL_EXT_draw_range_elements 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawRangeElementsEXT (GLenum, GLuint, GLuint, GLsizei, GLenum, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWRANGEELEMENTSEXTPROC) (GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid *indices); +#endif + +#ifndef GL_WIN_phong_shading +#define GL_WIN_phong_shading 1 +#endif + +#ifndef GL_WIN_specular_fog +#define GL_WIN_specular_fog 1 +#endif + +#ifndef GL_EXT_light_texture +#define GL_EXT_light_texture 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glApplyTextureEXT (GLenum); +GLAPI void APIENTRY glTextureLightEXT (GLenum); +GLAPI void APIENTRY glTextureMaterialEXT (GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLAPPLYTEXTUREEXTPROC) (GLenum mode); +typedef void (APIENTRYP PFNGLTEXTURELIGHTEXTPROC) (GLenum pname); +typedef void (APIENTRYP PFNGLTEXTUREMATERIALEXTPROC) (GLenum face, GLenum mode); +#endif + +#ifndef GL_SGIX_blend_alpha_minmax +#define GL_SGIX_blend_alpha_minmax 1 +#endif + +#ifndef GL_EXT_bgra +#define GL_EXT_bgra 1 +#endif + +#ifndef GL_SGIX_async +#define GL_SGIX_async 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glAsyncMarkerSGIX (GLuint); +GLAPI GLint APIENTRY glFinishAsyncSGIX (GLuint *); +GLAPI GLint APIENTRY glPollAsyncSGIX (GLuint *); +GLAPI GLuint APIENTRY glGenAsyncMarkersSGIX (GLsizei); +GLAPI void APIENTRY glDeleteAsyncMarkersSGIX (GLuint, GLsizei); +GLAPI GLboolean APIENTRY glIsAsyncMarkerSGIX (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLASYNCMARKERSGIXPROC) (GLuint marker); +typedef GLint (APIENTRYP PFNGLFINISHASYNCSGIXPROC) (GLuint *markerp); +typedef GLint (APIENTRYP PFNGLPOLLASYNCSGIXPROC) (GLuint *markerp); +typedef GLuint (APIENTRYP PFNGLGENASYNCMARKERSSGIXPROC) (GLsizei range); +typedef void (APIENTRYP PFNGLDELETEASYNCMARKERSSGIXPROC) (GLuint marker, GLsizei range); +typedef GLboolean (APIENTRYP PFNGLISASYNCMARKERSGIXPROC) (GLuint marker); +#endif + +#ifndef GL_SGIX_async_pixel +#define GL_SGIX_async_pixel 1 +#endif + +#ifndef GL_SGIX_async_histogram +#define GL_SGIX_async_histogram 1 +#endif + +#ifndef GL_INTEL_parallel_arrays +#define GL_INTEL_parallel_arrays 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexPointervINTEL (GLint, GLenum, const GLvoid* *); +GLAPI void APIENTRY glNormalPointervINTEL (GLenum, const GLvoid* *); +GLAPI void APIENTRY glColorPointervINTEL (GLint, GLenum, const GLvoid* *); +GLAPI void APIENTRY glTexCoordPointervINTEL (GLint, GLenum, const GLvoid* *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXPOINTERVINTELPROC) (GLint size, GLenum type, const GLvoid* *pointer); +typedef void (APIENTRYP PFNGLNORMALPOINTERVINTELPROC) (GLenum type, const GLvoid* *pointer); +typedef void (APIENTRYP PFNGLCOLORPOINTERVINTELPROC) (GLint size, GLenum type, const GLvoid* *pointer); +typedef void (APIENTRYP PFNGLTEXCOORDPOINTERVINTELPROC) (GLint size, GLenum type, const GLvoid* *pointer); +#endif + +#ifndef GL_HP_occlusion_test +#define GL_HP_occlusion_test 1 +#endif + +#ifndef GL_EXT_pixel_transform +#define GL_EXT_pixel_transform 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPixelTransformParameteriEXT (GLenum, GLenum, GLint); +GLAPI void APIENTRY glPixelTransformParameterfEXT (GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glPixelTransformParameterivEXT (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glPixelTransformParameterfvEXT (GLenum, GLenum, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPIXELTRANSFORMPARAMETERIEXTPROC) (GLenum target, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLPIXELTRANSFORMPARAMETERFEXTPROC) (GLenum target, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLPIXELTRANSFORMPARAMETERIVEXTPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLPIXELTRANSFORMPARAMETERFVEXTPROC) (GLenum target, GLenum pname, const GLfloat *params); +#endif + +#ifndef GL_EXT_pixel_transform_color_table +#define GL_EXT_pixel_transform_color_table 1 +#endif + +#ifndef GL_EXT_shared_texture_palette +#define GL_EXT_shared_texture_palette 1 +#endif + +#ifndef GL_EXT_separate_specular_color +#define GL_EXT_separate_specular_color 1 +#endif + +#ifndef GL_EXT_secondary_color +#define GL_EXT_secondary_color 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glSecondaryColor3bEXT (GLbyte, GLbyte, GLbyte); +GLAPI void APIENTRY glSecondaryColor3bvEXT (const GLbyte *); +GLAPI void APIENTRY glSecondaryColor3dEXT (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glSecondaryColor3dvEXT (const GLdouble *); +GLAPI void APIENTRY glSecondaryColor3fEXT (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glSecondaryColor3fvEXT (const GLfloat *); +GLAPI void APIENTRY glSecondaryColor3iEXT (GLint, GLint, GLint); +GLAPI void APIENTRY glSecondaryColor3ivEXT (const GLint *); +GLAPI void APIENTRY glSecondaryColor3sEXT (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glSecondaryColor3svEXT (const GLshort *); +GLAPI void APIENTRY glSecondaryColor3ubEXT (GLubyte, GLubyte, GLubyte); +GLAPI void APIENTRY glSecondaryColor3ubvEXT (const GLubyte *); +GLAPI void APIENTRY glSecondaryColor3uiEXT (GLuint, GLuint, GLuint); +GLAPI void APIENTRY glSecondaryColor3uivEXT (const GLuint *); +GLAPI void APIENTRY glSecondaryColor3usEXT (GLushort, GLushort, GLushort); +GLAPI void APIENTRY glSecondaryColor3usvEXT (const GLushort *); +GLAPI void APIENTRY glSecondaryColorPointerEXT (GLint, GLenum, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3BEXTPROC) (GLbyte red, GLbyte green, GLbyte blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3BVEXTPROC) (const GLbyte *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3DEXTPROC) (GLdouble red, GLdouble green, GLdouble blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3DVEXTPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3FEXTPROC) (GLfloat red, GLfloat green, GLfloat blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3FVEXTPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3IEXTPROC) (GLint red, GLint green, GLint blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3IVEXTPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3SEXTPROC) (GLshort red, GLshort green, GLshort blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3SVEXTPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UBEXTPROC) (GLubyte red, GLubyte green, GLubyte blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UBVEXTPROC) (const GLubyte *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UIEXTPROC) (GLuint red, GLuint green, GLuint blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3UIVEXTPROC) (const GLuint *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3USEXTPROC) (GLushort red, GLushort green, GLushort blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3USVEXTPROC) (const GLushort *v); +typedef void (APIENTRYP PFNGLSECONDARYCOLORPOINTEREXTPROC) (GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +#endif + +#ifndef GL_EXT_texture_perturb_normal +#define GL_EXT_texture_perturb_normal 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTextureNormalEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXTURENORMALEXTPROC) (GLenum mode); +#endif + +#ifndef GL_EXT_multi_draw_arrays +#define GL_EXT_multi_draw_arrays 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glMultiDrawArraysEXT (GLenum, GLint *, GLsizei *, GLsizei); +GLAPI void APIENTRY glMultiDrawElementsEXT (GLenum, const GLsizei *, GLenum, const GLvoid* *, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLMULTIDRAWARRAYSEXTPROC) (GLenum mode, GLint *first, GLsizei *count, GLsizei primcount); +typedef void (APIENTRYP PFNGLMULTIDRAWELEMENTSEXTPROC) (GLenum mode, const GLsizei *count, GLenum type, const GLvoid* *indices, GLsizei primcount); +#endif + +#ifndef GL_EXT_fog_coord +#define GL_EXT_fog_coord 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFogCoordfEXT (GLfloat); +GLAPI void APIENTRY glFogCoordfvEXT (const GLfloat *); +GLAPI void APIENTRY glFogCoorddEXT (GLdouble); +GLAPI void APIENTRY glFogCoorddvEXT (const GLdouble *); +GLAPI void APIENTRY glFogCoordPointerEXT (GLenum, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFOGCOORDFEXTPROC) (GLfloat coord); +typedef void (APIENTRYP PFNGLFOGCOORDFVEXTPROC) (const GLfloat *coord); +typedef void (APIENTRYP PFNGLFOGCOORDDEXTPROC) (GLdouble coord); +typedef void (APIENTRYP PFNGLFOGCOORDDVEXTPROC) (const GLdouble *coord); +typedef void (APIENTRYP PFNGLFOGCOORDPOINTEREXTPROC) (GLenum type, GLsizei stride, const GLvoid *pointer); +#endif + +#ifndef GL_REND_screen_coordinates +#define GL_REND_screen_coordinates 1 +#endif + +#ifndef GL_EXT_coordinate_frame +#define GL_EXT_coordinate_frame 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTangent3bEXT (GLbyte, GLbyte, GLbyte); +GLAPI void APIENTRY glTangent3bvEXT (const GLbyte *); +GLAPI void APIENTRY glTangent3dEXT (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glTangent3dvEXT (const GLdouble *); +GLAPI void APIENTRY glTangent3fEXT (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTangent3fvEXT (const GLfloat *); +GLAPI void APIENTRY glTangent3iEXT (GLint, GLint, GLint); +GLAPI void APIENTRY glTangent3ivEXT (const GLint *); +GLAPI void APIENTRY glTangent3sEXT (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glTangent3svEXT (const GLshort *); +GLAPI void APIENTRY glBinormal3bEXT (GLbyte, GLbyte, GLbyte); +GLAPI void APIENTRY glBinormal3bvEXT (const GLbyte *); +GLAPI void APIENTRY glBinormal3dEXT (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glBinormal3dvEXT (const GLdouble *); +GLAPI void APIENTRY glBinormal3fEXT (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glBinormal3fvEXT (const GLfloat *); +GLAPI void APIENTRY glBinormal3iEXT (GLint, GLint, GLint); +GLAPI void APIENTRY glBinormal3ivEXT (const GLint *); +GLAPI void APIENTRY glBinormal3sEXT (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glBinormal3svEXT (const GLshort *); +GLAPI void APIENTRY glTangentPointerEXT (GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glBinormalPointerEXT (GLenum, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTANGENT3BEXTPROC) (GLbyte tx, GLbyte ty, GLbyte tz); +typedef void (APIENTRYP PFNGLTANGENT3BVEXTPROC) (const GLbyte *v); +typedef void (APIENTRYP PFNGLTANGENT3DEXTPROC) (GLdouble tx, GLdouble ty, GLdouble tz); +typedef void (APIENTRYP PFNGLTANGENT3DVEXTPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLTANGENT3FEXTPROC) (GLfloat tx, GLfloat ty, GLfloat tz); +typedef void (APIENTRYP PFNGLTANGENT3FVEXTPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLTANGENT3IEXTPROC) (GLint tx, GLint ty, GLint tz); +typedef void (APIENTRYP PFNGLTANGENT3IVEXTPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLTANGENT3SEXTPROC) (GLshort tx, GLshort ty, GLshort tz); +typedef void (APIENTRYP PFNGLTANGENT3SVEXTPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLBINORMAL3BEXTPROC) (GLbyte bx, GLbyte by, GLbyte bz); +typedef void (APIENTRYP PFNGLBINORMAL3BVEXTPROC) (const GLbyte *v); +typedef void (APIENTRYP PFNGLBINORMAL3DEXTPROC) (GLdouble bx, GLdouble by, GLdouble bz); +typedef void (APIENTRYP PFNGLBINORMAL3DVEXTPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLBINORMAL3FEXTPROC) (GLfloat bx, GLfloat by, GLfloat bz); +typedef void (APIENTRYP PFNGLBINORMAL3FVEXTPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLBINORMAL3IEXTPROC) (GLint bx, GLint by, GLint bz); +typedef void (APIENTRYP PFNGLBINORMAL3IVEXTPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLBINORMAL3SEXTPROC) (GLshort bx, GLshort by, GLshort bz); +typedef void (APIENTRYP PFNGLBINORMAL3SVEXTPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLTANGENTPOINTEREXTPROC) (GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLBINORMALPOINTEREXTPROC) (GLenum type, GLsizei stride, const GLvoid *pointer); +#endif + +#ifndef GL_EXT_texture_env_combine +#define GL_EXT_texture_env_combine 1 +#endif + +#ifndef GL_APPLE_specular_vector +#define GL_APPLE_specular_vector 1 +#endif + +#ifndef GL_APPLE_transform_hint +#define GL_APPLE_transform_hint 1 +#endif + +#ifndef GL_SGIX_fog_scale +#define GL_SGIX_fog_scale 1 +#endif + +#ifndef GL_SUNX_constant_data +#define GL_SUNX_constant_data 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFinishTextureSUNX (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFINISHTEXTURESUNXPROC) (void); +#endif + +#ifndef GL_SUN_global_alpha +#define GL_SUN_global_alpha 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGlobalAlphaFactorbSUN (GLbyte); +GLAPI void APIENTRY glGlobalAlphaFactorsSUN (GLshort); +GLAPI void APIENTRY glGlobalAlphaFactoriSUN (GLint); +GLAPI void APIENTRY glGlobalAlphaFactorfSUN (GLfloat); +GLAPI void APIENTRY glGlobalAlphaFactordSUN (GLdouble); +GLAPI void APIENTRY glGlobalAlphaFactorubSUN (GLubyte); +GLAPI void APIENTRY glGlobalAlphaFactorusSUN (GLushort); +GLAPI void APIENTRY glGlobalAlphaFactoruiSUN (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORBSUNPROC) (GLbyte factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORSSUNPROC) (GLshort factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORISUNPROC) (GLint factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORFSUNPROC) (GLfloat factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORDSUNPROC) (GLdouble factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORUBSUNPROC) (GLubyte factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORUSSUNPROC) (GLushort factor); +typedef void (APIENTRYP PFNGLGLOBALALPHAFACTORUISUNPROC) (GLuint factor); +#endif + +#ifndef GL_SUN_triangle_list +#define GL_SUN_triangle_list 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glReplacementCodeuiSUN (GLuint); +GLAPI void APIENTRY glReplacementCodeusSUN (GLushort); +GLAPI void APIENTRY glReplacementCodeubSUN (GLubyte); +GLAPI void APIENTRY glReplacementCodeuivSUN (const GLuint *); +GLAPI void APIENTRY glReplacementCodeusvSUN (const GLushort *); +GLAPI void APIENTRY glReplacementCodeubvSUN (const GLubyte *); +GLAPI void APIENTRY glReplacementCodePointerSUN (GLenum, GLsizei, const GLvoid* *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUISUNPROC) (GLuint code); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUSSUNPROC) (GLushort code); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUBSUNPROC) (GLubyte code); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUIVSUNPROC) (const GLuint *code); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUSVSUNPROC) (const GLushort *code); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUBVSUNPROC) (const GLubyte *code); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEPOINTERSUNPROC) (GLenum type, GLsizei stride, const GLvoid* *pointer); +#endif + +#ifndef GL_SUN_vertex +#define GL_SUN_vertex 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColor4ubVertex2fSUN (GLubyte, GLubyte, GLubyte, GLubyte, GLfloat, GLfloat); +GLAPI void APIENTRY glColor4ubVertex2fvSUN (const GLubyte *, const GLfloat *); +GLAPI void APIENTRY glColor4ubVertex3fSUN (GLubyte, GLubyte, GLubyte, GLubyte, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glColor4ubVertex3fvSUN (const GLubyte *, const GLfloat *); +GLAPI void APIENTRY glColor3fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glColor3fVertex3fvSUN (const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glNormal3fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glNormal3fVertex3fvSUN (const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glColor4fNormal3fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glColor4fNormal3fVertex3fvSUN (const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glTexCoord2fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord2fVertex3fvSUN (const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glTexCoord4fVertex4fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord4fVertex4fvSUN (const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glTexCoord2fColor4ubVertex3fSUN (GLfloat, GLfloat, GLubyte, GLubyte, GLubyte, GLubyte, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord2fColor4ubVertex3fvSUN (const GLfloat *, const GLubyte *, const GLfloat *); +GLAPI void APIENTRY glTexCoord2fColor3fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord2fColor3fVertex3fvSUN (const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glTexCoord2fNormal3fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord2fNormal3fVertex3fvSUN (const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glTexCoord2fColor4fNormal3fVertex3fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord2fColor4fNormal3fVertex3fvSUN (const GLfloat *, const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glTexCoord4fColor4fNormal3fVertex4fSUN (GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glTexCoord4fColor4fNormal3fVertex4fvSUN (const GLfloat *, const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiVertex3fvSUN (const GLuint *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiColor4ubVertex3fSUN (GLuint, GLubyte, GLubyte, GLubyte, GLubyte, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiColor4ubVertex3fvSUN (const GLuint *, const GLubyte *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiColor3fVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiColor3fVertex3fvSUN (const GLuint *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiNormal3fVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiNormal3fVertex3fvSUN (const GLuint *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiColor4fNormal3fVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiColor4fNormal3fVertex3fvSUN (const GLuint *, const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiTexCoord2fVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiTexCoord2fVertex3fvSUN (const GLuint *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiTexCoord2fNormal3fVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN (const GLuint *, const GLfloat *, const GLfloat *, const GLfloat *); +GLAPI void APIENTRY glReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN (GLuint, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fvSUN (const GLuint *, const GLfloat *, const GLfloat *, const GLfloat *, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLOR4UBVERTEX2FSUNPROC) (GLubyte r, GLubyte g, GLubyte b, GLubyte a, GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLCOLOR4UBVERTEX2FVSUNPROC) (const GLubyte *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLCOLOR4UBVERTEX3FSUNPROC) (GLubyte r, GLubyte g, GLubyte b, GLubyte a, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLCOLOR4UBVERTEX3FVSUNPROC) (const GLubyte *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLCOLOR3FVERTEX3FSUNPROC) (GLfloat r, GLfloat g, GLfloat b, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLCOLOR3FVERTEX3FVSUNPROC) (const GLfloat *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLNORMAL3FVERTEX3FSUNPROC) (GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLNORMAL3FVERTEX3FVSUNPROC) (const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLCOLOR4FNORMAL3FVERTEX3FSUNPROC) (GLfloat r, GLfloat g, GLfloat b, GLfloat a, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLCOLOR4FNORMAL3FVERTEX3FVSUNPROC) (const GLfloat *c, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD2FVERTEX3FSUNPROC) (GLfloat s, GLfloat t, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLTEXCOORD2FVERTEX3FVSUNPROC) (const GLfloat *tc, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD4FVERTEX4FSUNPROC) (GLfloat s, GLfloat t, GLfloat p, GLfloat q, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLTEXCOORD4FVERTEX4FVSUNPROC) (const GLfloat *tc, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD2FCOLOR4UBVERTEX3FSUNPROC) (GLfloat s, GLfloat t, GLubyte r, GLubyte g, GLubyte b, GLubyte a, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLTEXCOORD2FCOLOR4UBVERTEX3FVSUNPROC) (const GLfloat *tc, const GLubyte *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD2FCOLOR3FVERTEX3FSUNPROC) (GLfloat s, GLfloat t, GLfloat r, GLfloat g, GLfloat b, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLTEXCOORD2FCOLOR3FVERTEX3FVSUNPROC) (const GLfloat *tc, const GLfloat *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD2FNORMAL3FVERTEX3FSUNPROC) (GLfloat s, GLfloat t, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLTEXCOORD2FNORMAL3FVERTEX3FVSUNPROC) (const GLfloat *tc, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD2FCOLOR4FNORMAL3FVERTEX3FSUNPROC) (GLfloat s, GLfloat t, GLfloat r, GLfloat g, GLfloat b, GLfloat a, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLTEXCOORD2FCOLOR4FNORMAL3FVERTEX3FVSUNPROC) (const GLfloat *tc, const GLfloat *c, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLTEXCOORD4FCOLOR4FNORMAL3FVERTEX4FSUNPROC) (GLfloat s, GLfloat t, GLfloat p, GLfloat q, GLfloat r, GLfloat g, GLfloat b, GLfloat a, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLTEXCOORD4FCOLOR4FNORMAL3FVERTEX4FVSUNPROC) (const GLfloat *tc, const GLfloat *c, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUIVERTEX3FSUNPROC) (GLuint rc, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUIVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUICOLOR4UBVERTEX3FSUNPROC) (GLuint rc, GLubyte r, GLubyte g, GLubyte b, GLubyte a, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUICOLOR4UBVERTEX3FVSUNPROC) (const GLuint *rc, const GLubyte *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUICOLOR3FVERTEX3FSUNPROC) (GLuint rc, GLfloat r, GLfloat g, GLfloat b, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUICOLOR3FVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *c, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUINORMAL3FVERTEX3FSUNPROC) (GLuint rc, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUINORMAL3FVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUICOLOR4FNORMAL3FVERTEX3FSUNPROC) (GLuint rc, GLfloat r, GLfloat g, GLfloat b, GLfloat a, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUICOLOR4FNORMAL3FVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *c, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUITEXCOORD2FVERTEX3FSUNPROC) (GLuint rc, GLfloat s, GLfloat t, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUITEXCOORD2FVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *tc, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUITEXCOORD2FNORMAL3FVERTEX3FSUNPROC) (GLuint rc, GLfloat s, GLfloat t, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUITEXCOORD2FNORMAL3FVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *tc, const GLfloat *n, const GLfloat *v); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUITEXCOORD2FCOLOR4FNORMAL3FVERTEX3FSUNPROC) (GLuint rc, GLfloat s, GLfloat t, GLfloat r, GLfloat g, GLfloat b, GLfloat a, GLfloat nx, GLfloat ny, GLfloat nz, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLREPLACEMENTCODEUITEXCOORD2FCOLOR4FNORMAL3FVERTEX3FVSUNPROC) (const GLuint *rc, const GLfloat *tc, const GLfloat *c, const GLfloat *n, const GLfloat *v); +#endif + +#ifndef GL_EXT_blend_func_separate +#define GL_EXT_blend_func_separate 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendFuncSeparateEXT (GLenum, GLenum, GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEEXTPROC) (GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha); +#endif + +#ifndef GL_INGR_blend_func_separate +#define GL_INGR_blend_func_separate 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendFuncSeparateINGR (GLenum, GLenum, GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEINGRPROC) (GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha); +#endif + +#ifndef GL_INGR_color_clamp +#define GL_INGR_color_clamp 1 +#endif + +#ifndef GL_INGR_interlace_read +#define GL_INGR_interlace_read 1 +#endif + +#ifndef GL_EXT_stencil_wrap +#define GL_EXT_stencil_wrap 1 +#endif + +#ifndef GL_EXT_422_pixels +#define GL_EXT_422_pixels 1 +#endif + +#ifndef GL_NV_texgen_reflection +#define GL_NV_texgen_reflection 1 +#endif + +#ifndef GL_SUN_convolution_border_modes +#define GL_SUN_convolution_border_modes 1 +#endif + +#ifndef GL_EXT_texture_env_add +#define GL_EXT_texture_env_add 1 +#endif + +#ifndef GL_EXT_texture_lod_bias +#define GL_EXT_texture_lod_bias 1 +#endif + +#ifndef GL_EXT_texture_filter_anisotropic +#define GL_EXT_texture_filter_anisotropic 1 +#endif + +#ifndef GL_EXT_vertex_weighting +#define GL_EXT_vertex_weighting 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexWeightfEXT (GLfloat); +GLAPI void APIENTRY glVertexWeightfvEXT (const GLfloat *); +GLAPI void APIENTRY glVertexWeightPointerEXT (GLsizei, GLenum, GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXWEIGHTFEXTPROC) (GLfloat weight); +typedef void (APIENTRYP PFNGLVERTEXWEIGHTFVEXTPROC) (const GLfloat *weight); +typedef void (APIENTRYP PFNGLVERTEXWEIGHTPOINTEREXTPROC) (GLsizei size, GLenum type, GLsizei stride, const GLvoid *pointer); +#endif + +#ifndef GL_NV_light_max_exponent +#define GL_NV_light_max_exponent 1 +#endif + +#ifndef GL_NV_vertex_array_range +#define GL_NV_vertex_array_range 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFlushVertexArrayRangeNV (void); +GLAPI void APIENTRY glVertexArrayRangeNV (GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFLUSHVERTEXARRAYRANGENVPROC) (void); +typedef void (APIENTRYP PFNGLVERTEXARRAYRANGENVPROC) (GLsizei length, const GLvoid *pointer); +#endif + +#ifndef GL_NV_register_combiners +#define GL_NV_register_combiners 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCombinerParameterfvNV (GLenum, const GLfloat *); +GLAPI void APIENTRY glCombinerParameterfNV (GLenum, GLfloat); +GLAPI void APIENTRY glCombinerParameterivNV (GLenum, const GLint *); +GLAPI void APIENTRY glCombinerParameteriNV (GLenum, GLint); +GLAPI void APIENTRY glCombinerInputNV (GLenum, GLenum, GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glCombinerOutputNV (GLenum, GLenum, GLenum, GLenum, GLenum, GLenum, GLenum, GLboolean, GLboolean, GLboolean); +GLAPI void APIENTRY glFinalCombinerInputNV (GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glGetCombinerInputParameterfvNV (GLenum, GLenum, GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetCombinerInputParameterivNV (GLenum, GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetCombinerOutputParameterfvNV (GLenum, GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetCombinerOutputParameterivNV (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetFinalCombinerInputParameterfvNV (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetFinalCombinerInputParameterivNV (GLenum, GLenum, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOMBINERPARAMETERFVNVPROC) (GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLCOMBINERPARAMETERFNVPROC) (GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLCOMBINERPARAMETERIVNVPROC) (GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLCOMBINERPARAMETERINVPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLCOMBINERINPUTNVPROC) (GLenum stage, GLenum portion, GLenum variable, GLenum input, GLenum mapping, GLenum componentUsage); +typedef void (APIENTRYP PFNGLCOMBINEROUTPUTNVPROC) (GLenum stage, GLenum portion, GLenum abOutput, GLenum cdOutput, GLenum sumOutput, GLenum scale, GLenum bias, GLboolean abDotProduct, GLboolean cdDotProduct, GLboolean muxSum); +typedef void (APIENTRYP PFNGLFINALCOMBINERINPUTNVPROC) (GLenum variable, GLenum input, GLenum mapping, GLenum componentUsage); +typedef void (APIENTRYP PFNGLGETCOMBINERINPUTPARAMETERFVNVPROC) (GLenum stage, GLenum portion, GLenum variable, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETCOMBINERINPUTPARAMETERIVNVPROC) (GLenum stage, GLenum portion, GLenum variable, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETCOMBINEROUTPUTPARAMETERFVNVPROC) (GLenum stage, GLenum portion, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETCOMBINEROUTPUTPARAMETERIVNVPROC) (GLenum stage, GLenum portion, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETFINALCOMBINERINPUTPARAMETERFVNVPROC) (GLenum variable, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETFINALCOMBINERINPUTPARAMETERIVNVPROC) (GLenum variable, GLenum pname, GLint *params); +#endif + +#ifndef GL_NV_fog_distance +#define GL_NV_fog_distance 1 +#endif + +#ifndef GL_NV_texgen_emboss +#define GL_NV_texgen_emboss 1 +#endif + +#ifndef GL_NV_blend_square +#define GL_NV_blend_square 1 +#endif + +#ifndef GL_NV_texture_env_combine4 +#define GL_NV_texture_env_combine4 1 +#endif + +#ifndef GL_MESA_resize_buffers +#define GL_MESA_resize_buffers 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glResizeBuffersMESA (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLRESIZEBUFFERSMESAPROC) (void); +#endif + +#ifndef GL_MESA_window_pos +#define GL_MESA_window_pos 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glWindowPos2dMESA (GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos2dvMESA (const GLdouble *); +GLAPI void APIENTRY glWindowPos2fMESA (GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos2fvMESA (const GLfloat *); +GLAPI void APIENTRY glWindowPos2iMESA (GLint, GLint); +GLAPI void APIENTRY glWindowPos2ivMESA (const GLint *); +GLAPI void APIENTRY glWindowPos2sMESA (GLshort, GLshort); +GLAPI void APIENTRY glWindowPos2svMESA (const GLshort *); +GLAPI void APIENTRY glWindowPos3dMESA (GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos3dvMESA (const GLdouble *); +GLAPI void APIENTRY glWindowPos3fMESA (GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos3fvMESA (const GLfloat *); +GLAPI void APIENTRY glWindowPos3iMESA (GLint, GLint, GLint); +GLAPI void APIENTRY glWindowPos3ivMESA (const GLint *); +GLAPI void APIENTRY glWindowPos3sMESA (GLshort, GLshort, GLshort); +GLAPI void APIENTRY glWindowPos3svMESA (const GLshort *); +GLAPI void APIENTRY glWindowPos4dMESA (GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glWindowPos4dvMESA (const GLdouble *); +GLAPI void APIENTRY glWindowPos4fMESA (GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glWindowPos4fvMESA (const GLfloat *); +GLAPI void APIENTRY glWindowPos4iMESA (GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glWindowPos4ivMESA (const GLint *); +GLAPI void APIENTRY glWindowPos4sMESA (GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glWindowPos4svMESA (const GLshort *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLWINDOWPOS2DMESAPROC) (GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLWINDOWPOS2DVMESAPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2FMESAPROC) (GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLWINDOWPOS2FVMESAPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2IMESAPROC) (GLint x, GLint y); +typedef void (APIENTRYP PFNGLWINDOWPOS2IVMESAPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS2SMESAPROC) (GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLWINDOWPOS2SVMESAPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3DMESAPROC) (GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLWINDOWPOS3DVMESAPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3FMESAPROC) (GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLWINDOWPOS3FVMESAPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3IMESAPROC) (GLint x, GLint y, GLint z); +typedef void (APIENTRYP PFNGLWINDOWPOS3IVMESAPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS3SMESAPROC) (GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLWINDOWPOS3SVMESAPROC) (const GLshort *v); +typedef void (APIENTRYP PFNGLWINDOWPOS4DMESAPROC) (GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLWINDOWPOS4DVMESAPROC) (const GLdouble *v); +typedef void (APIENTRYP PFNGLWINDOWPOS4FMESAPROC) (GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLWINDOWPOS4FVMESAPROC) (const GLfloat *v); +typedef void (APIENTRYP PFNGLWINDOWPOS4IMESAPROC) (GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLWINDOWPOS4IVMESAPROC) (const GLint *v); +typedef void (APIENTRYP PFNGLWINDOWPOS4SMESAPROC) (GLshort x, GLshort y, GLshort z, GLshort w); +typedef void (APIENTRYP PFNGLWINDOWPOS4SVMESAPROC) (const GLshort *v); +#endif + +#ifndef GL_IBM_cull_vertex +#define GL_IBM_cull_vertex 1 +#endif + +#ifndef GL_IBM_multimode_draw_arrays +#define GL_IBM_multimode_draw_arrays 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glMultiModeDrawArraysIBM (const GLenum *, const GLint *, const GLsizei *, GLsizei, GLint); +GLAPI void APIENTRY glMultiModeDrawElementsIBM (const GLenum *, const GLsizei *, GLenum, const GLvoid* const *, GLsizei, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLMULTIMODEDRAWARRAYSIBMPROC) (const GLenum *mode, const GLint *first, const GLsizei *count, GLsizei primcount, GLint modestride); +typedef void (APIENTRYP PFNGLMULTIMODEDRAWELEMENTSIBMPROC) (const GLenum *mode, const GLsizei *count, GLenum type, const GLvoid* const *indices, GLsizei primcount, GLint modestride); +#endif + +#ifndef GL_IBM_vertex_array_lists +#define GL_IBM_vertex_array_lists 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorPointerListIBM (GLint, GLenum, GLint, const GLvoid* *, GLint); +GLAPI void APIENTRY glSecondaryColorPointerListIBM (GLint, GLenum, GLint, const GLvoid* *, GLint); +GLAPI void APIENTRY glEdgeFlagPointerListIBM (GLint, const GLboolean* *, GLint); +GLAPI void APIENTRY glFogCoordPointerListIBM (GLenum, GLint, const GLvoid* *, GLint); +GLAPI void APIENTRY glIndexPointerListIBM (GLenum, GLint, const GLvoid* *, GLint); +GLAPI void APIENTRY glNormalPointerListIBM (GLenum, GLint, const GLvoid* *, GLint); +GLAPI void APIENTRY glTexCoordPointerListIBM (GLint, GLenum, GLint, const GLvoid* *, GLint); +GLAPI void APIENTRY glVertexPointerListIBM (GLint, GLenum, GLint, const GLvoid* *, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLORPOINTERLISTIBMPROC) (GLint size, GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLSECONDARYCOLORPOINTERLISTIBMPROC) (GLint size, GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLEDGEFLAGPOINTERLISTIBMPROC) (GLint stride, const GLboolean* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLFOGCOORDPOINTERLISTIBMPROC) (GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLINDEXPOINTERLISTIBMPROC) (GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLNORMALPOINTERLISTIBMPROC) (GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLTEXCOORDPOINTERLISTIBMPROC) (GLint size, GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +typedef void (APIENTRYP PFNGLVERTEXPOINTERLISTIBMPROC) (GLint size, GLenum type, GLint stride, const GLvoid* *pointer, GLint ptrstride); +#endif + +#ifndef GL_SGIX_subsample +#define GL_SGIX_subsample 1 +#endif + +#ifndef GL_SGIX_ycrcba +#define GL_SGIX_ycrcba 1 +#endif + +#ifndef GL_SGIX_ycrcb_subsample +#define GL_SGIX_ycrcb_subsample 1 +#endif + +#ifndef GL_SGIX_depth_pass_instrument +#define GL_SGIX_depth_pass_instrument 1 +#endif + +#ifndef GL_3DFX_texture_compression_FXT1 +#define GL_3DFX_texture_compression_FXT1 1 +#endif + +#ifndef GL_3DFX_multisample +#define GL_3DFX_multisample 1 +#endif + +#ifndef GL_3DFX_tbuffer +#define GL_3DFX_tbuffer 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTbufferMask3DFX (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTBUFFERMASK3DFXPROC) (GLuint mask); +#endif + +#ifndef GL_EXT_multisample +#define GL_EXT_multisample 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glSampleMaskEXT (GLclampf, GLboolean); +GLAPI void APIENTRY glSamplePatternEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSAMPLEMASKEXTPROC) (GLclampf value, GLboolean invert); +typedef void (APIENTRYP PFNGLSAMPLEPATTERNEXTPROC) (GLenum pattern); +#endif + +#ifndef GL_SGIX_vertex_preclip +#define GL_SGIX_vertex_preclip 1 +#endif + +#ifndef GL_SGIX_convolution_accuracy +#define GL_SGIX_convolution_accuracy 1 +#endif + +#ifndef GL_SGIX_resample +#define GL_SGIX_resample 1 +#endif + +#ifndef GL_SGIS_point_line_texgen +#define GL_SGIS_point_line_texgen 1 +#endif + +#ifndef GL_SGIS_texture_color_mask +#define GL_SGIS_texture_color_mask 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTextureColorMaskSGIS (GLboolean, GLboolean, GLboolean, GLboolean); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXTURECOLORMASKSGISPROC) (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha); +#endif + +#ifndef GL_SGIX_igloo_interface +#define GL_SGIX_igloo_interface 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glIglooInterfaceSGIX (GLenum, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLIGLOOINTERFACESGIXPROC) (GLenum pname, const GLvoid *params); +#endif + +#ifndef GL_EXT_texture_env_dot3 +#define GL_EXT_texture_env_dot3 1 +#endif + +#ifndef GL_ATI_texture_mirror_once +#define GL_ATI_texture_mirror_once 1 +#endif + +#ifndef GL_NV_fence +#define GL_NV_fence 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDeleteFencesNV (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenFencesNV (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsFenceNV (GLuint); +GLAPI GLboolean APIENTRY glTestFenceNV (GLuint); +GLAPI void APIENTRY glGetFenceivNV (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glFinishFenceNV (GLuint); +GLAPI void APIENTRY glSetFenceNV (GLuint, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDELETEFENCESNVPROC) (GLsizei n, const GLuint *fences); +typedef void (APIENTRYP PFNGLGENFENCESNVPROC) (GLsizei n, GLuint *fences); +typedef GLboolean (APIENTRYP PFNGLISFENCENVPROC) (GLuint fence); +typedef GLboolean (APIENTRYP PFNGLTESTFENCENVPROC) (GLuint fence); +typedef void (APIENTRYP PFNGLGETFENCEIVNVPROC) (GLuint fence, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLFINISHFENCENVPROC) (GLuint fence); +typedef void (APIENTRYP PFNGLSETFENCENVPROC) (GLuint fence, GLenum condition); +#endif + +#ifndef GL_NV_evaluators +#define GL_NV_evaluators 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glMapControlPointsNV (GLenum, GLuint, GLenum, GLsizei, GLsizei, GLint, GLint, GLboolean, const GLvoid *); +GLAPI void APIENTRY glMapParameterivNV (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glMapParameterfvNV (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glGetMapControlPointsNV (GLenum, GLuint, GLenum, GLsizei, GLsizei, GLboolean, GLvoid *); +GLAPI void APIENTRY glGetMapParameterivNV (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetMapParameterfvNV (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMapAttribParameterivNV (GLenum, GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetMapAttribParameterfvNV (GLenum, GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glEvalMapsNV (GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLMAPCONTROLPOINTSNVPROC) (GLenum target, GLuint index, GLenum type, GLsizei ustride, GLsizei vstride, GLint uorder, GLint vorder, GLboolean packed, const GLvoid *points); +typedef void (APIENTRYP PFNGLMAPPARAMETERIVNVPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLMAPPARAMETERFVNVPROC) (GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLGETMAPCONTROLPOINTSNVPROC) (GLenum target, GLuint index, GLenum type, GLsizei ustride, GLsizei vstride, GLboolean packed, GLvoid *points); +typedef void (APIENTRYP PFNGLGETMAPPARAMETERIVNVPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMAPPARAMETERFVNVPROC) (GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMAPATTRIBPARAMETERIVNVPROC) (GLenum target, GLuint index, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMAPATTRIBPARAMETERFVNVPROC) (GLenum target, GLuint index, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLEVALMAPSNVPROC) (GLenum target, GLenum mode); +#endif + +#ifndef GL_NV_packed_depth_stencil +#define GL_NV_packed_depth_stencil 1 +#endif + +#ifndef GL_NV_register_combiners2 +#define GL_NV_register_combiners2 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glCombinerStageParameterfvNV (GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glGetCombinerStageParameterfvNV (GLenum, GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOMBINERSTAGEPARAMETERFVNVPROC) (GLenum stage, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLGETCOMBINERSTAGEPARAMETERFVNVPROC) (GLenum stage, GLenum pname, GLfloat *params); +#endif + +#ifndef GL_NV_texture_compression_vtc +#define GL_NV_texture_compression_vtc 1 +#endif + +#ifndef GL_NV_texture_rectangle +#define GL_NV_texture_rectangle 1 +#endif + +#ifndef GL_NV_texture_shader +#define GL_NV_texture_shader 1 +#endif + +#ifndef GL_NV_texture_shader2 +#define GL_NV_texture_shader2 1 +#endif + +#ifndef GL_NV_vertex_array_range2 +#define GL_NV_vertex_array_range2 1 +#endif + +#ifndef GL_NV_vertex_program +#define GL_NV_vertex_program 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLboolean APIENTRY glAreProgramsResidentNV (GLsizei, const GLuint *, GLboolean *); +GLAPI void APIENTRY glBindProgramNV (GLenum, GLuint); +GLAPI void APIENTRY glDeleteProgramsNV (GLsizei, const GLuint *); +GLAPI void APIENTRY glExecuteProgramNV (GLenum, GLuint, const GLfloat *); +GLAPI void APIENTRY glGenProgramsNV (GLsizei, GLuint *); +GLAPI void APIENTRY glGetProgramParameterdvNV (GLenum, GLuint, GLenum, GLdouble *); +GLAPI void APIENTRY glGetProgramParameterfvNV (GLenum, GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetProgramivNV (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetProgramStringNV (GLuint, GLenum, GLubyte *); +GLAPI void APIENTRY glGetTrackMatrixivNV (GLenum, GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVertexAttribdvNV (GLuint, GLenum, GLdouble *); +GLAPI void APIENTRY glGetVertexAttribfvNV (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetVertexAttribivNV (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVertexAttribPointervNV (GLuint, GLenum, GLvoid* *); +GLAPI GLboolean APIENTRY glIsProgramNV (GLuint); +GLAPI void APIENTRY glLoadProgramNV (GLenum, GLuint, GLsizei, const GLubyte *); +GLAPI void APIENTRY glProgramParameter4dNV (GLenum, GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glProgramParameter4dvNV (GLenum, GLuint, const GLdouble *); +GLAPI void APIENTRY glProgramParameter4fNV (GLenum, GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramParameter4fvNV (GLenum, GLuint, const GLfloat *); +GLAPI void APIENTRY glProgramParameters4dvNV (GLenum, GLuint, GLuint, const GLdouble *); +GLAPI void APIENTRY glProgramParameters4fvNV (GLenum, GLuint, GLuint, const GLfloat *); +GLAPI void APIENTRY glRequestResidentProgramsNV (GLsizei, const GLuint *); +GLAPI void APIENTRY glTrackMatrixNV (GLenum, GLuint, GLenum, GLenum); +GLAPI void APIENTRY glVertexAttribPointerNV (GLuint, GLint, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glVertexAttrib1dNV (GLuint, GLdouble); +GLAPI void APIENTRY glVertexAttrib1dvNV (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib1fNV (GLuint, GLfloat); +GLAPI void APIENTRY glVertexAttrib1fvNV (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib1sNV (GLuint, GLshort); +GLAPI void APIENTRY glVertexAttrib1svNV (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib2dNV (GLuint, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib2dvNV (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib2fNV (GLuint, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib2fvNV (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib2sNV (GLuint, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib2svNV (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib3dNV (GLuint, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib3dvNV (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib3fNV (GLuint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib3fvNV (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib3sNV (GLuint, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib3svNV (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4dNV (GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexAttrib4dvNV (GLuint, const GLdouble *); +GLAPI void APIENTRY glVertexAttrib4fNV (GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexAttrib4fvNV (GLuint, const GLfloat *); +GLAPI void APIENTRY glVertexAttrib4sNV (GLuint, GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexAttrib4svNV (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttrib4ubNV (GLuint, GLubyte, GLubyte, GLubyte, GLubyte); +GLAPI void APIENTRY glVertexAttrib4ubvNV (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttribs1dvNV (GLuint, GLsizei, const GLdouble *); +GLAPI void APIENTRY glVertexAttribs1fvNV (GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glVertexAttribs1svNV (GLuint, GLsizei, const GLshort *); +GLAPI void APIENTRY glVertexAttribs2dvNV (GLuint, GLsizei, const GLdouble *); +GLAPI void APIENTRY glVertexAttribs2fvNV (GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glVertexAttribs2svNV (GLuint, GLsizei, const GLshort *); +GLAPI void APIENTRY glVertexAttribs3dvNV (GLuint, GLsizei, const GLdouble *); +GLAPI void APIENTRY glVertexAttribs3fvNV (GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glVertexAttribs3svNV (GLuint, GLsizei, const GLshort *); +GLAPI void APIENTRY glVertexAttribs4dvNV (GLuint, GLsizei, const GLdouble *); +GLAPI void APIENTRY glVertexAttribs4fvNV (GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glVertexAttribs4svNV (GLuint, GLsizei, const GLshort *); +GLAPI void APIENTRY glVertexAttribs4ubvNV (GLuint, GLsizei, const GLubyte *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLboolean (APIENTRYP PFNGLAREPROGRAMSRESIDENTNVPROC) (GLsizei n, const GLuint *programs, GLboolean *residences); +typedef void (APIENTRYP PFNGLBINDPROGRAMNVPROC) (GLenum target, GLuint id); +typedef void (APIENTRYP PFNGLDELETEPROGRAMSNVPROC) (GLsizei n, const GLuint *programs); +typedef void (APIENTRYP PFNGLEXECUTEPROGRAMNVPROC) (GLenum target, GLuint id, const GLfloat *params); +typedef void (APIENTRYP PFNGLGENPROGRAMSNVPROC) (GLsizei n, GLuint *programs); +typedef void (APIENTRYP PFNGLGETPROGRAMPARAMETERDVNVPROC) (GLenum target, GLuint index, GLenum pname, GLdouble *params); +typedef void (APIENTRYP PFNGLGETPROGRAMPARAMETERFVNVPROC) (GLenum target, GLuint index, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETPROGRAMIVNVPROC) (GLuint id, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMSTRINGNVPROC) (GLuint id, GLenum pname, GLubyte *program); +typedef void (APIENTRYP PFNGLGETTRACKMATRIXIVNVPROC) (GLenum target, GLuint address, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBDVNVPROC) (GLuint index, GLenum pname, GLdouble *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBFVNVPROC) (GLuint index, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIVNVPROC) (GLuint index, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBPOINTERVNVPROC) (GLuint index, GLenum pname, GLvoid* *pointer); +typedef GLboolean (APIENTRYP PFNGLISPROGRAMNVPROC) (GLuint id); +typedef void (APIENTRYP PFNGLLOADPROGRAMNVPROC) (GLenum target, GLuint id, GLsizei len, const GLubyte *program); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETER4DNVPROC) (GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETER4DVNVPROC) (GLenum target, GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETER4FNVPROC) (GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETER4FVNVPROC) (GLenum target, GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETERS4DVNVPROC) (GLenum target, GLuint index, GLuint count, const GLdouble *v); +typedef void (APIENTRYP PFNGLPROGRAMPARAMETERS4FVNVPROC) (GLenum target, GLuint index, GLuint count, const GLfloat *v); +typedef void (APIENTRYP PFNGLREQUESTRESIDENTPROGRAMSNVPROC) (GLsizei n, const GLuint *programs); +typedef void (APIENTRYP PFNGLTRACKMATRIXNVPROC) (GLenum target, GLuint address, GLenum matrix, GLenum transform); +typedef void (APIENTRYP PFNGLVERTEXATTRIBPOINTERNVPROC) (GLuint index, GLint fsize, GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1DNVPROC) (GLuint index, GLdouble x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1DVNVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1FNVPROC) (GLuint index, GLfloat x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1FVNVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1SNVPROC) (GLuint index, GLshort x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1SVNVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2DNVPROC) (GLuint index, GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2DVNVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2FNVPROC) (GLuint index, GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2FVNVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2SNVPROC) (GLuint index, GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2SVNVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3DNVPROC) (GLuint index, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3DVNVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3FNVPROC) (GLuint index, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3FVNVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3SNVPROC) (GLuint index, GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3SVNVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4DNVPROC) (GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4DVNVPROC) (GLuint index, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4FNVPROC) (GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4FVNVPROC) (GLuint index, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4SNVPROC) (GLuint index, GLshort x, GLshort y, GLshort z, GLshort w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4SVNVPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4UBNVPROC) (GLuint index, GLubyte x, GLubyte y, GLubyte z, GLubyte w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4UBVNVPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS1DVNVPROC) (GLuint index, GLsizei count, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS1FVNVPROC) (GLuint index, GLsizei count, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS1SVNVPROC) (GLuint index, GLsizei count, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS2DVNVPROC) (GLuint index, GLsizei count, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS2FVNVPROC) (GLuint index, GLsizei count, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS2SVNVPROC) (GLuint index, GLsizei count, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS3DVNVPROC) (GLuint index, GLsizei count, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS3FVNVPROC) (GLuint index, GLsizei count, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS3SVNVPROC) (GLuint index, GLsizei count, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS4DVNVPROC) (GLuint index, GLsizei count, const GLdouble *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS4FVNVPROC) (GLuint index, GLsizei count, const GLfloat *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS4SVNVPROC) (GLuint index, GLsizei count, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS4UBVNVPROC) (GLuint index, GLsizei count, const GLubyte *v); +#endif + +#ifndef GL_SGIX_texture_coordinate_clamp +#define GL_SGIX_texture_coordinate_clamp 1 +#endif + +#ifndef GL_SGIX_scalebias_hint +#define GL_SGIX_scalebias_hint 1 +#endif + +#ifndef GL_OML_interlace +#define GL_OML_interlace 1 +#endif + +#ifndef GL_OML_subsample +#define GL_OML_subsample 1 +#endif + +#ifndef GL_OML_resample +#define GL_OML_resample 1 +#endif + +#ifndef GL_NV_copy_depth_to_color +#define GL_NV_copy_depth_to_color 1 +#endif + +#ifndef GL_ATI_envmap_bumpmap +#define GL_ATI_envmap_bumpmap 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexBumpParameterivATI (GLenum, const GLint *); +GLAPI void APIENTRY glTexBumpParameterfvATI (GLenum, const GLfloat *); +GLAPI void APIENTRY glGetTexBumpParameterivATI (GLenum, GLint *); +GLAPI void APIENTRY glGetTexBumpParameterfvATI (GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXBUMPPARAMETERIVATIPROC) (GLenum pname, const GLint *param); +typedef void (APIENTRYP PFNGLTEXBUMPPARAMETERFVATIPROC) (GLenum pname, const GLfloat *param); +typedef void (APIENTRYP PFNGLGETTEXBUMPPARAMETERIVATIPROC) (GLenum pname, GLint *param); +typedef void (APIENTRYP PFNGLGETTEXBUMPPARAMETERFVATIPROC) (GLenum pname, GLfloat *param); +#endif + +#ifndef GL_ATI_fragment_shader +#define GL_ATI_fragment_shader 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLuint APIENTRY glGenFragmentShadersATI (GLuint); +GLAPI void APIENTRY glBindFragmentShaderATI (GLuint); +GLAPI void APIENTRY glDeleteFragmentShaderATI (GLuint); +GLAPI void APIENTRY glBeginFragmentShaderATI (void); +GLAPI void APIENTRY glEndFragmentShaderATI (void); +GLAPI void APIENTRY glPassTexCoordATI (GLuint, GLuint, GLenum); +GLAPI void APIENTRY glSampleMapATI (GLuint, GLuint, GLenum); +GLAPI void APIENTRY glColorFragmentOp1ATI (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glColorFragmentOp2ATI (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glColorFragmentOp3ATI (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glAlphaFragmentOp1ATI (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glAlphaFragmentOp2ATI (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glAlphaFragmentOp3ATI (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glSetFragmentShaderConstantATI (GLuint, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLuint (APIENTRYP PFNGLGENFRAGMENTSHADERSATIPROC) (GLuint range); +typedef void (APIENTRYP PFNGLBINDFRAGMENTSHADERATIPROC) (GLuint id); +typedef void (APIENTRYP PFNGLDELETEFRAGMENTSHADERATIPROC) (GLuint id); +typedef void (APIENTRYP PFNGLBEGINFRAGMENTSHADERATIPROC) (void); +typedef void (APIENTRYP PFNGLENDFRAGMENTSHADERATIPROC) (void); +typedef void (APIENTRYP PFNGLPASSTEXCOORDATIPROC) (GLuint dst, GLuint coord, GLenum swizzle); +typedef void (APIENTRYP PFNGLSAMPLEMAPATIPROC) (GLuint dst, GLuint interp, GLenum swizzle); +typedef void (APIENTRYP PFNGLCOLORFRAGMENTOP1ATIPROC) (GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); +typedef void (APIENTRYP PFNGLCOLORFRAGMENTOP2ATIPROC) (GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); +typedef void (APIENTRYP PFNGLCOLORFRAGMENTOP3ATIPROC) (GLenum op, GLuint dst, GLuint dstMask, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); +typedef void (APIENTRYP PFNGLALPHAFRAGMENTOP1ATIPROC) (GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod); +typedef void (APIENTRYP PFNGLALPHAFRAGMENTOP2ATIPROC) (GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod); +typedef void (APIENTRYP PFNGLALPHAFRAGMENTOP3ATIPROC) (GLenum op, GLuint dst, GLuint dstMod, GLuint arg1, GLuint arg1Rep, GLuint arg1Mod, GLuint arg2, GLuint arg2Rep, GLuint arg2Mod, GLuint arg3, GLuint arg3Rep, GLuint arg3Mod); +typedef void (APIENTRYP PFNGLSETFRAGMENTSHADERCONSTANTATIPROC) (GLuint dst, const GLfloat *value); +#endif + +#ifndef GL_ATI_pn_triangles +#define GL_ATI_pn_triangles 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPNTrianglesiATI (GLenum, GLint); +GLAPI void APIENTRY glPNTrianglesfATI (GLenum, GLfloat); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPNTRIANGLESIATIPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLPNTRIANGLESFATIPROC) (GLenum pname, GLfloat param); +#endif + +#ifndef GL_ATI_vertex_array_object +#define GL_ATI_vertex_array_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLuint APIENTRY glNewObjectBufferATI (GLsizei, const GLvoid *, GLenum); +GLAPI GLboolean APIENTRY glIsObjectBufferATI (GLuint); +GLAPI void APIENTRY glUpdateObjectBufferATI (GLuint, GLuint, GLsizei, const GLvoid *, GLenum); +GLAPI void APIENTRY glGetObjectBufferfvATI (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetObjectBufferivATI (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glFreeObjectBufferATI (GLuint); +GLAPI void APIENTRY glArrayObjectATI (GLenum, GLint, GLenum, GLsizei, GLuint, GLuint); +GLAPI void APIENTRY glGetArrayObjectfvATI (GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetArrayObjectivATI (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glVariantArrayObjectATI (GLuint, GLenum, GLsizei, GLuint, GLuint); +GLAPI void APIENTRY glGetVariantArrayObjectfvATI (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetVariantArrayObjectivATI (GLuint, GLenum, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLuint (APIENTRYP PFNGLNEWOBJECTBUFFERATIPROC) (GLsizei size, const GLvoid *pointer, GLenum usage); +typedef GLboolean (APIENTRYP PFNGLISOBJECTBUFFERATIPROC) (GLuint buffer); +typedef void (APIENTRYP PFNGLUPDATEOBJECTBUFFERATIPROC) (GLuint buffer, GLuint offset, GLsizei size, const GLvoid *pointer, GLenum preserve); +typedef void (APIENTRYP PFNGLGETOBJECTBUFFERFVATIPROC) (GLuint buffer, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETOBJECTBUFFERIVATIPROC) (GLuint buffer, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLFREEOBJECTBUFFERATIPROC) (GLuint buffer); +typedef void (APIENTRYP PFNGLARRAYOBJECTATIPROC) (GLenum array, GLint size, GLenum type, GLsizei stride, GLuint buffer, GLuint offset); +typedef void (APIENTRYP PFNGLGETARRAYOBJECTFVATIPROC) (GLenum array, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETARRAYOBJECTIVATIPROC) (GLenum array, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLVARIANTARRAYOBJECTATIPROC) (GLuint id, GLenum type, GLsizei stride, GLuint buffer, GLuint offset); +typedef void (APIENTRYP PFNGLGETVARIANTARRAYOBJECTFVATIPROC) (GLuint id, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETVARIANTARRAYOBJECTIVATIPROC) (GLuint id, GLenum pname, GLint *params); +#endif + +#ifndef GL_EXT_vertex_shader +#define GL_EXT_vertex_shader 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBeginVertexShaderEXT (void); +GLAPI void APIENTRY glEndVertexShaderEXT (void); +GLAPI void APIENTRY glBindVertexShaderEXT (GLuint); +GLAPI GLuint APIENTRY glGenVertexShadersEXT (GLuint); +GLAPI void APIENTRY glDeleteVertexShaderEXT (GLuint); +GLAPI void APIENTRY glShaderOp1EXT (GLenum, GLuint, GLuint); +GLAPI void APIENTRY glShaderOp2EXT (GLenum, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glShaderOp3EXT (GLenum, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glSwizzleEXT (GLuint, GLuint, GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glWriteMaskEXT (GLuint, GLuint, GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glInsertComponentEXT (GLuint, GLuint, GLuint); +GLAPI void APIENTRY glExtractComponentEXT (GLuint, GLuint, GLuint); +GLAPI GLuint APIENTRY glGenSymbolsEXT (GLenum, GLenum, GLenum, GLuint); +GLAPI void APIENTRY glSetInvariantEXT (GLuint, GLenum, const GLvoid *); +GLAPI void APIENTRY glSetLocalConstantEXT (GLuint, GLenum, const GLvoid *); +GLAPI void APIENTRY glVariantbvEXT (GLuint, const GLbyte *); +GLAPI void APIENTRY glVariantsvEXT (GLuint, const GLshort *); +GLAPI void APIENTRY glVariantivEXT (GLuint, const GLint *); +GLAPI void APIENTRY glVariantfvEXT (GLuint, const GLfloat *); +GLAPI void APIENTRY glVariantdvEXT (GLuint, const GLdouble *); +GLAPI void APIENTRY glVariantubvEXT (GLuint, const GLubyte *); +GLAPI void APIENTRY glVariantusvEXT (GLuint, const GLushort *); +GLAPI void APIENTRY glVariantuivEXT (GLuint, const GLuint *); +GLAPI void APIENTRY glVariantPointerEXT (GLuint, GLenum, GLuint, const GLvoid *); +GLAPI void APIENTRY glEnableVariantClientStateEXT (GLuint); +GLAPI void APIENTRY glDisableVariantClientStateEXT (GLuint); +GLAPI GLuint APIENTRY glBindLightParameterEXT (GLenum, GLenum); +GLAPI GLuint APIENTRY glBindMaterialParameterEXT (GLenum, GLenum); +GLAPI GLuint APIENTRY glBindTexGenParameterEXT (GLenum, GLenum, GLenum); +GLAPI GLuint APIENTRY glBindTextureUnitParameterEXT (GLenum, GLenum); +GLAPI GLuint APIENTRY glBindParameterEXT (GLenum); +GLAPI GLboolean APIENTRY glIsVariantEnabledEXT (GLuint, GLenum); +GLAPI void APIENTRY glGetVariantBooleanvEXT (GLuint, GLenum, GLboolean *); +GLAPI void APIENTRY glGetVariantIntegervEXT (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVariantFloatvEXT (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetVariantPointervEXT (GLuint, GLenum, GLvoid* *); +GLAPI void APIENTRY glGetInvariantBooleanvEXT (GLuint, GLenum, GLboolean *); +GLAPI void APIENTRY glGetInvariantIntegervEXT (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetInvariantFloatvEXT (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetLocalConstantBooleanvEXT (GLuint, GLenum, GLboolean *); +GLAPI void APIENTRY glGetLocalConstantIntegervEXT (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetLocalConstantFloatvEXT (GLuint, GLenum, GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBEGINVERTEXSHADEREXTPROC) (void); +typedef void (APIENTRYP PFNGLENDVERTEXSHADEREXTPROC) (void); +typedef void (APIENTRYP PFNGLBINDVERTEXSHADEREXTPROC) (GLuint id); +typedef GLuint (APIENTRYP PFNGLGENVERTEXSHADERSEXTPROC) (GLuint range); +typedef void (APIENTRYP PFNGLDELETEVERTEXSHADEREXTPROC) (GLuint id); +typedef void (APIENTRYP PFNGLSHADEROP1EXTPROC) (GLenum op, GLuint res, GLuint arg1); +typedef void (APIENTRYP PFNGLSHADEROP2EXTPROC) (GLenum op, GLuint res, GLuint arg1, GLuint arg2); +typedef void (APIENTRYP PFNGLSHADEROP3EXTPROC) (GLenum op, GLuint res, GLuint arg1, GLuint arg2, GLuint arg3); +typedef void (APIENTRYP PFNGLSWIZZLEEXTPROC) (GLuint res, GLuint in, GLenum outX, GLenum outY, GLenum outZ, GLenum outW); +typedef void (APIENTRYP PFNGLWRITEMASKEXTPROC) (GLuint res, GLuint in, GLenum outX, GLenum outY, GLenum outZ, GLenum outW); +typedef void (APIENTRYP PFNGLINSERTCOMPONENTEXTPROC) (GLuint res, GLuint src, GLuint num); +typedef void (APIENTRYP PFNGLEXTRACTCOMPONENTEXTPROC) (GLuint res, GLuint src, GLuint num); +typedef GLuint (APIENTRYP PFNGLGENSYMBOLSEXTPROC) (GLenum datatype, GLenum storagetype, GLenum range, GLuint components); +typedef void (APIENTRYP PFNGLSETINVARIANTEXTPROC) (GLuint id, GLenum type, const GLvoid *addr); +typedef void (APIENTRYP PFNGLSETLOCALCONSTANTEXTPROC) (GLuint id, GLenum type, const GLvoid *addr); +typedef void (APIENTRYP PFNGLVARIANTBVEXTPROC) (GLuint id, const GLbyte *addr); +typedef void (APIENTRYP PFNGLVARIANTSVEXTPROC) (GLuint id, const GLshort *addr); +typedef void (APIENTRYP PFNGLVARIANTIVEXTPROC) (GLuint id, const GLint *addr); +typedef void (APIENTRYP PFNGLVARIANTFVEXTPROC) (GLuint id, const GLfloat *addr); +typedef void (APIENTRYP PFNGLVARIANTDVEXTPROC) (GLuint id, const GLdouble *addr); +typedef void (APIENTRYP PFNGLVARIANTUBVEXTPROC) (GLuint id, const GLubyte *addr); +typedef void (APIENTRYP PFNGLVARIANTUSVEXTPROC) (GLuint id, const GLushort *addr); +typedef void (APIENTRYP PFNGLVARIANTUIVEXTPROC) (GLuint id, const GLuint *addr); +typedef void (APIENTRYP PFNGLVARIANTPOINTEREXTPROC) (GLuint id, GLenum type, GLuint stride, const GLvoid *addr); +typedef void (APIENTRYP PFNGLENABLEVARIANTCLIENTSTATEEXTPROC) (GLuint id); +typedef void (APIENTRYP PFNGLDISABLEVARIANTCLIENTSTATEEXTPROC) (GLuint id); +typedef GLuint (APIENTRYP PFNGLBINDLIGHTPARAMETEREXTPROC) (GLenum light, GLenum value); +typedef GLuint (APIENTRYP PFNGLBINDMATERIALPARAMETEREXTPROC) (GLenum face, GLenum value); +typedef GLuint (APIENTRYP PFNGLBINDTEXGENPARAMETEREXTPROC) (GLenum unit, GLenum coord, GLenum value); +typedef GLuint (APIENTRYP PFNGLBINDTEXTUREUNITPARAMETEREXTPROC) (GLenum unit, GLenum value); +typedef GLuint (APIENTRYP PFNGLBINDPARAMETEREXTPROC) (GLenum value); +typedef GLboolean (APIENTRYP PFNGLISVARIANTENABLEDEXTPROC) (GLuint id, GLenum cap); +typedef void (APIENTRYP PFNGLGETVARIANTBOOLEANVEXTPROC) (GLuint id, GLenum value, GLboolean *data); +typedef void (APIENTRYP PFNGLGETVARIANTINTEGERVEXTPROC) (GLuint id, GLenum value, GLint *data); +typedef void (APIENTRYP PFNGLGETVARIANTFLOATVEXTPROC) (GLuint id, GLenum value, GLfloat *data); +typedef void (APIENTRYP PFNGLGETVARIANTPOINTERVEXTPROC) (GLuint id, GLenum value, GLvoid* *data); +typedef void (APIENTRYP PFNGLGETINVARIANTBOOLEANVEXTPROC) (GLuint id, GLenum value, GLboolean *data); +typedef void (APIENTRYP PFNGLGETINVARIANTINTEGERVEXTPROC) (GLuint id, GLenum value, GLint *data); +typedef void (APIENTRYP PFNGLGETINVARIANTFLOATVEXTPROC) (GLuint id, GLenum value, GLfloat *data); +typedef void (APIENTRYP PFNGLGETLOCALCONSTANTBOOLEANVEXTPROC) (GLuint id, GLenum value, GLboolean *data); +typedef void (APIENTRYP PFNGLGETLOCALCONSTANTINTEGERVEXTPROC) (GLuint id, GLenum value, GLint *data); +typedef void (APIENTRYP PFNGLGETLOCALCONSTANTFLOATVEXTPROC) (GLuint id, GLenum value, GLfloat *data); +#endif + +#ifndef GL_ATI_vertex_streams +#define GL_ATI_vertex_streams 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexStream1sATI (GLenum, GLshort); +GLAPI void APIENTRY glVertexStream1svATI (GLenum, const GLshort *); +GLAPI void APIENTRY glVertexStream1iATI (GLenum, GLint); +GLAPI void APIENTRY glVertexStream1ivATI (GLenum, const GLint *); +GLAPI void APIENTRY glVertexStream1fATI (GLenum, GLfloat); +GLAPI void APIENTRY glVertexStream1fvATI (GLenum, const GLfloat *); +GLAPI void APIENTRY glVertexStream1dATI (GLenum, GLdouble); +GLAPI void APIENTRY glVertexStream1dvATI (GLenum, const GLdouble *); +GLAPI void APIENTRY glVertexStream2sATI (GLenum, GLshort, GLshort); +GLAPI void APIENTRY glVertexStream2svATI (GLenum, const GLshort *); +GLAPI void APIENTRY glVertexStream2iATI (GLenum, GLint, GLint); +GLAPI void APIENTRY glVertexStream2ivATI (GLenum, const GLint *); +GLAPI void APIENTRY glVertexStream2fATI (GLenum, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexStream2fvATI (GLenum, const GLfloat *); +GLAPI void APIENTRY glVertexStream2dATI (GLenum, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexStream2dvATI (GLenum, const GLdouble *); +GLAPI void APIENTRY glVertexStream3sATI (GLenum, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexStream3svATI (GLenum, const GLshort *); +GLAPI void APIENTRY glVertexStream3iATI (GLenum, GLint, GLint, GLint); +GLAPI void APIENTRY glVertexStream3ivATI (GLenum, const GLint *); +GLAPI void APIENTRY glVertexStream3fATI (GLenum, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexStream3fvATI (GLenum, const GLfloat *); +GLAPI void APIENTRY glVertexStream3dATI (GLenum, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexStream3dvATI (GLenum, const GLdouble *); +GLAPI void APIENTRY glVertexStream4sATI (GLenum, GLshort, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glVertexStream4svATI (GLenum, const GLshort *); +GLAPI void APIENTRY glVertexStream4iATI (GLenum, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glVertexStream4ivATI (GLenum, const GLint *); +GLAPI void APIENTRY glVertexStream4fATI (GLenum, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glVertexStream4fvATI (GLenum, const GLfloat *); +GLAPI void APIENTRY glVertexStream4dATI (GLenum, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glVertexStream4dvATI (GLenum, const GLdouble *); +GLAPI void APIENTRY glNormalStream3bATI (GLenum, GLbyte, GLbyte, GLbyte); +GLAPI void APIENTRY glNormalStream3bvATI (GLenum, const GLbyte *); +GLAPI void APIENTRY glNormalStream3sATI (GLenum, GLshort, GLshort, GLshort); +GLAPI void APIENTRY glNormalStream3svATI (GLenum, const GLshort *); +GLAPI void APIENTRY glNormalStream3iATI (GLenum, GLint, GLint, GLint); +GLAPI void APIENTRY glNormalStream3ivATI (GLenum, const GLint *); +GLAPI void APIENTRY glNormalStream3fATI (GLenum, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glNormalStream3fvATI (GLenum, const GLfloat *); +GLAPI void APIENTRY glNormalStream3dATI (GLenum, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glNormalStream3dvATI (GLenum, const GLdouble *); +GLAPI void APIENTRY glClientActiveVertexStreamATI (GLenum); +GLAPI void APIENTRY glVertexBlendEnviATI (GLenum, GLint); +GLAPI void APIENTRY glVertexBlendEnvfATI (GLenum, GLfloat); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXSTREAM1SATIPROC) (GLenum stream, GLshort x); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1SVATIPROC) (GLenum stream, const GLshort *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1IATIPROC) (GLenum stream, GLint x); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1IVATIPROC) (GLenum stream, const GLint *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1FATIPROC) (GLenum stream, GLfloat x); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1FVATIPROC) (GLenum stream, const GLfloat *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1DATIPROC) (GLenum stream, GLdouble x); +typedef void (APIENTRYP PFNGLVERTEXSTREAM1DVATIPROC) (GLenum stream, const GLdouble *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2SATIPROC) (GLenum stream, GLshort x, GLshort y); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2SVATIPROC) (GLenum stream, const GLshort *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2IATIPROC) (GLenum stream, GLint x, GLint y); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2IVATIPROC) (GLenum stream, const GLint *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2FATIPROC) (GLenum stream, GLfloat x, GLfloat y); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2FVATIPROC) (GLenum stream, const GLfloat *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2DATIPROC) (GLenum stream, GLdouble x, GLdouble y); +typedef void (APIENTRYP PFNGLVERTEXSTREAM2DVATIPROC) (GLenum stream, const GLdouble *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3SATIPROC) (GLenum stream, GLshort x, GLshort y, GLshort z); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3SVATIPROC) (GLenum stream, const GLshort *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3IATIPROC) (GLenum stream, GLint x, GLint y, GLint z); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3IVATIPROC) (GLenum stream, const GLint *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3FATIPROC) (GLenum stream, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3FVATIPROC) (GLenum stream, const GLfloat *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3DATIPROC) (GLenum stream, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLVERTEXSTREAM3DVATIPROC) (GLenum stream, const GLdouble *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4SATIPROC) (GLenum stream, GLshort x, GLshort y, GLshort z, GLshort w); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4SVATIPROC) (GLenum stream, const GLshort *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4IATIPROC) (GLenum stream, GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4IVATIPROC) (GLenum stream, const GLint *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4FATIPROC) (GLenum stream, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4FVATIPROC) (GLenum stream, const GLfloat *coords); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4DATIPROC) (GLenum stream, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLVERTEXSTREAM4DVATIPROC) (GLenum stream, const GLdouble *coords); +typedef void (APIENTRYP PFNGLNORMALSTREAM3BATIPROC) (GLenum stream, GLbyte nx, GLbyte ny, GLbyte nz); +typedef void (APIENTRYP PFNGLNORMALSTREAM3BVATIPROC) (GLenum stream, const GLbyte *coords); +typedef void (APIENTRYP PFNGLNORMALSTREAM3SATIPROC) (GLenum stream, GLshort nx, GLshort ny, GLshort nz); +typedef void (APIENTRYP PFNGLNORMALSTREAM3SVATIPROC) (GLenum stream, const GLshort *coords); +typedef void (APIENTRYP PFNGLNORMALSTREAM3IATIPROC) (GLenum stream, GLint nx, GLint ny, GLint nz); +typedef void (APIENTRYP PFNGLNORMALSTREAM3IVATIPROC) (GLenum stream, const GLint *coords); +typedef void (APIENTRYP PFNGLNORMALSTREAM3FATIPROC) (GLenum stream, GLfloat nx, GLfloat ny, GLfloat nz); +typedef void (APIENTRYP PFNGLNORMALSTREAM3FVATIPROC) (GLenum stream, const GLfloat *coords); +typedef void (APIENTRYP PFNGLNORMALSTREAM3DATIPROC) (GLenum stream, GLdouble nx, GLdouble ny, GLdouble nz); +typedef void (APIENTRYP PFNGLNORMALSTREAM3DVATIPROC) (GLenum stream, const GLdouble *coords); +typedef void (APIENTRYP PFNGLCLIENTACTIVEVERTEXSTREAMATIPROC) (GLenum stream); +typedef void (APIENTRYP PFNGLVERTEXBLENDENVIATIPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLVERTEXBLENDENVFATIPROC) (GLenum pname, GLfloat param); +#endif + +#ifndef GL_ATI_element_array +#define GL_ATI_element_array 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glElementPointerATI (GLenum, const GLvoid *); +GLAPI void APIENTRY glDrawElementArrayATI (GLenum, GLsizei); +GLAPI void APIENTRY glDrawRangeElementArrayATI (GLenum, GLuint, GLuint, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLELEMENTPOINTERATIPROC) (GLenum type, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLDRAWELEMENTARRAYATIPROC) (GLenum mode, GLsizei count); +typedef void (APIENTRYP PFNGLDRAWRANGEELEMENTARRAYATIPROC) (GLenum mode, GLuint start, GLuint end, GLsizei count); +#endif + +#ifndef GL_SUN_mesh_array +#define GL_SUN_mesh_array 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawMeshArraysSUN (GLenum, GLint, GLsizei, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWMESHARRAYSSUNPROC) (GLenum mode, GLint first, GLsizei count, GLsizei width); +#endif + +#ifndef GL_SUN_slice_accum +#define GL_SUN_slice_accum 1 +#endif + +#ifndef GL_NV_multisample_filter_hint +#define GL_NV_multisample_filter_hint 1 +#endif + +#ifndef GL_NV_depth_clamp +#define GL_NV_depth_clamp 1 +#endif + +#ifndef GL_NV_occlusion_query +#define GL_NV_occlusion_query 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGenOcclusionQueriesNV (GLsizei, GLuint *); +GLAPI void APIENTRY glDeleteOcclusionQueriesNV (GLsizei, const GLuint *); +GLAPI GLboolean APIENTRY glIsOcclusionQueryNV (GLuint); +GLAPI void APIENTRY glBeginOcclusionQueryNV (GLuint); +GLAPI void APIENTRY glEndOcclusionQueryNV (void); +GLAPI void APIENTRY glGetOcclusionQueryivNV (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetOcclusionQueryuivNV (GLuint, GLenum, GLuint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGENOCCLUSIONQUERIESNVPROC) (GLsizei n, GLuint *ids); +typedef void (APIENTRYP PFNGLDELETEOCCLUSIONQUERIESNVPROC) (GLsizei n, const GLuint *ids); +typedef GLboolean (APIENTRYP PFNGLISOCCLUSIONQUERYNVPROC) (GLuint id); +typedef void (APIENTRYP PFNGLBEGINOCCLUSIONQUERYNVPROC) (GLuint id); +typedef void (APIENTRYP PFNGLENDOCCLUSIONQUERYNVPROC) (void); +typedef void (APIENTRYP PFNGLGETOCCLUSIONQUERYIVNVPROC) (GLuint id, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETOCCLUSIONQUERYUIVNVPROC) (GLuint id, GLenum pname, GLuint *params); +#endif + +#ifndef GL_NV_point_sprite +#define GL_NV_point_sprite 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPointParameteriNV (GLenum, GLint); +GLAPI void APIENTRY glPointParameterivNV (GLenum, const GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPOINTPARAMETERINVPROC) (GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLPOINTPARAMETERIVNVPROC) (GLenum pname, const GLint *params); +#endif + +#ifndef GL_NV_texture_shader3 +#define GL_NV_texture_shader3 1 +#endif + +#ifndef GL_NV_vertex_program1_1 +#define GL_NV_vertex_program1_1 1 +#endif + +#ifndef GL_EXT_shadow_funcs +#define GL_EXT_shadow_funcs 1 +#endif + +#ifndef GL_EXT_stencil_two_side +#define GL_EXT_stencil_two_side 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glActiveStencilFaceEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLACTIVESTENCILFACEEXTPROC) (GLenum face); +#endif + +#ifndef GL_ATI_text_fragment_shader +#define GL_ATI_text_fragment_shader 1 +#endif + +#ifndef GL_APPLE_client_storage +#define GL_APPLE_client_storage 1 +#endif + +#ifndef GL_APPLE_element_array +#define GL_APPLE_element_array 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glElementPointerAPPLE (GLenum, const GLvoid *); +GLAPI void APIENTRY glDrawElementArrayAPPLE (GLenum, GLint, GLsizei); +GLAPI void APIENTRY glDrawRangeElementArrayAPPLE (GLenum, GLuint, GLuint, GLint, GLsizei); +GLAPI void APIENTRY glMultiDrawElementArrayAPPLE (GLenum, const GLint *, const GLsizei *, GLsizei); +GLAPI void APIENTRY glMultiDrawRangeElementArrayAPPLE (GLenum, GLuint, GLuint, const GLint *, const GLsizei *, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLELEMENTPOINTERAPPLEPROC) (GLenum type, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLDRAWELEMENTARRAYAPPLEPROC) (GLenum mode, GLint first, GLsizei count); +typedef void (APIENTRYP PFNGLDRAWRANGEELEMENTARRAYAPPLEPROC) (GLenum mode, GLuint start, GLuint end, GLint first, GLsizei count); +typedef void (APIENTRYP PFNGLMULTIDRAWELEMENTARRAYAPPLEPROC) (GLenum mode, const GLint *first, const GLsizei *count, GLsizei primcount); +typedef void (APIENTRYP PFNGLMULTIDRAWRANGEELEMENTARRAYAPPLEPROC) (GLenum mode, GLuint start, GLuint end, const GLint *first, const GLsizei *count, GLsizei primcount); +#endif + +#ifndef GL_APPLE_fence +#define GL_APPLE_fence 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGenFencesAPPLE (GLsizei, GLuint *); +GLAPI void APIENTRY glDeleteFencesAPPLE (GLsizei, const GLuint *); +GLAPI void APIENTRY glSetFenceAPPLE (GLuint); +GLAPI GLboolean APIENTRY glIsFenceAPPLE (GLuint); +GLAPI GLboolean APIENTRY glTestFenceAPPLE (GLuint); +GLAPI void APIENTRY glFinishFenceAPPLE (GLuint); +GLAPI GLboolean APIENTRY glTestObjectAPPLE (GLenum, GLuint); +GLAPI void APIENTRY glFinishObjectAPPLE (GLenum, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGENFENCESAPPLEPROC) (GLsizei n, GLuint *fences); +typedef void (APIENTRYP PFNGLDELETEFENCESAPPLEPROC) (GLsizei n, const GLuint *fences); +typedef void (APIENTRYP PFNGLSETFENCEAPPLEPROC) (GLuint fence); +typedef GLboolean (APIENTRYP PFNGLISFENCEAPPLEPROC) (GLuint fence); +typedef GLboolean (APIENTRYP PFNGLTESTFENCEAPPLEPROC) (GLuint fence); +typedef void (APIENTRYP PFNGLFINISHFENCEAPPLEPROC) (GLuint fence); +typedef GLboolean (APIENTRYP PFNGLTESTOBJECTAPPLEPROC) (GLenum object, GLuint name); +typedef void (APIENTRYP PFNGLFINISHOBJECTAPPLEPROC) (GLenum object, GLint name); +#endif + +#ifndef GL_APPLE_vertex_array_object +#define GL_APPLE_vertex_array_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBindVertexArrayAPPLE (GLuint); +GLAPI void APIENTRY glDeleteVertexArraysAPPLE (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenVertexArraysAPPLE (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsVertexArrayAPPLE (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBINDVERTEXARRAYAPPLEPROC) (GLuint array); +typedef void (APIENTRYP PFNGLDELETEVERTEXARRAYSAPPLEPROC) (GLsizei n, const GLuint *arrays); +typedef void (APIENTRYP PFNGLGENVERTEXARRAYSAPPLEPROC) (GLsizei n, GLuint *arrays); +typedef GLboolean (APIENTRYP PFNGLISVERTEXARRAYAPPLEPROC) (GLuint array); +#endif + +#ifndef GL_APPLE_vertex_array_range +#define GL_APPLE_vertex_array_range 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexArrayRangeAPPLE (GLsizei, GLvoid *); +GLAPI void APIENTRY glFlushVertexArrayRangeAPPLE (GLsizei, GLvoid *); +GLAPI void APIENTRY glVertexArrayParameteriAPPLE (GLenum, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXARRAYRANGEAPPLEPROC) (GLsizei length, GLvoid *pointer); +typedef void (APIENTRYP PFNGLFLUSHVERTEXARRAYRANGEAPPLEPROC) (GLsizei length, GLvoid *pointer); +typedef void (APIENTRYP PFNGLVERTEXARRAYPARAMETERIAPPLEPROC) (GLenum pname, GLint param); +#endif + +#ifndef GL_APPLE_ycbcr_422 +#define GL_APPLE_ycbcr_422 1 +#endif + +#ifndef GL_S3_s3tc +#define GL_S3_s3tc 1 +#endif + +#ifndef GL_ATI_draw_buffers +#define GL_ATI_draw_buffers 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawBuffersATI (GLsizei, const GLenum *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWBUFFERSATIPROC) (GLsizei n, const GLenum *bufs); +#endif + +#ifndef GL_ATI_pixel_format_float +#define GL_ATI_pixel_format_float 1 +/* This is really a WGL extension, but defines some associated GL enums. + * ATI does not export "GL_ATI_pixel_format_float" in the GL_EXTENSIONS string. + */ +#endif + +#ifndef GL_ATI_texture_env_combine3 +#define GL_ATI_texture_env_combine3 1 +#endif + +#ifndef GL_ATI_texture_float +#define GL_ATI_texture_float 1 +#endif + +#ifndef GL_NV_float_buffer +#define GL_NV_float_buffer 1 +#endif + +#ifndef GL_NV_fragment_program +#define GL_NV_fragment_program 1 +/* Some NV_fragment_program entry points are shared with ARB_vertex_program. */ +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramNamedParameter4fNV (GLuint, GLsizei, const GLubyte *, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramNamedParameter4dNV (GLuint, GLsizei, const GLubyte *, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glProgramNamedParameter4fvNV (GLuint, GLsizei, const GLubyte *, const GLfloat *); +GLAPI void APIENTRY glProgramNamedParameter4dvNV (GLuint, GLsizei, const GLubyte *, const GLdouble *); +GLAPI void APIENTRY glGetProgramNamedParameterfvNV (GLuint, GLsizei, const GLubyte *, GLfloat *); +GLAPI void APIENTRY glGetProgramNamedParameterdvNV (GLuint, GLsizei, const GLubyte *, GLdouble *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMNAMEDPARAMETER4FNVPROC) (GLuint id, GLsizei len, const GLubyte *name, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLPROGRAMNAMEDPARAMETER4DNVPROC) (GLuint id, GLsizei len, const GLubyte *name, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLPROGRAMNAMEDPARAMETER4FVNVPROC) (GLuint id, GLsizei len, const GLubyte *name, const GLfloat *v); +typedef void (APIENTRYP PFNGLPROGRAMNAMEDPARAMETER4DVNVPROC) (GLuint id, GLsizei len, const GLubyte *name, const GLdouble *v); +typedef void (APIENTRYP PFNGLGETPROGRAMNAMEDPARAMETERFVNVPROC) (GLuint id, GLsizei len, const GLubyte *name, GLfloat *params); +typedef void (APIENTRYP PFNGLGETPROGRAMNAMEDPARAMETERDVNVPROC) (GLuint id, GLsizei len, const GLubyte *name, GLdouble *params); +#endif + +#ifndef GL_NV_half_float +#define GL_NV_half_float 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertex2hNV (GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glVertex2hvNV (const GLhalfNV *); +GLAPI void APIENTRY glVertex3hNV (GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glVertex3hvNV (const GLhalfNV *); +GLAPI void APIENTRY glVertex4hNV (GLhalfNV, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glVertex4hvNV (const GLhalfNV *); +GLAPI void APIENTRY glNormal3hNV (GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glNormal3hvNV (const GLhalfNV *); +GLAPI void APIENTRY glColor3hNV (GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glColor3hvNV (const GLhalfNV *); +GLAPI void APIENTRY glColor4hNV (GLhalfNV, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glColor4hvNV (const GLhalfNV *); +GLAPI void APIENTRY glTexCoord1hNV (GLhalfNV); +GLAPI void APIENTRY glTexCoord1hvNV (const GLhalfNV *); +GLAPI void APIENTRY glTexCoord2hNV (GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glTexCoord2hvNV (const GLhalfNV *); +GLAPI void APIENTRY glTexCoord3hNV (GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glTexCoord3hvNV (const GLhalfNV *); +GLAPI void APIENTRY glTexCoord4hNV (GLhalfNV, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glTexCoord4hvNV (const GLhalfNV *); +GLAPI void APIENTRY glMultiTexCoord1hNV (GLenum, GLhalfNV); +GLAPI void APIENTRY glMultiTexCoord1hvNV (GLenum, const GLhalfNV *); +GLAPI void APIENTRY glMultiTexCoord2hNV (GLenum, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glMultiTexCoord2hvNV (GLenum, const GLhalfNV *); +GLAPI void APIENTRY glMultiTexCoord3hNV (GLenum, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glMultiTexCoord3hvNV (GLenum, const GLhalfNV *); +GLAPI void APIENTRY glMultiTexCoord4hNV (GLenum, GLhalfNV, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glMultiTexCoord4hvNV (GLenum, const GLhalfNV *); +GLAPI void APIENTRY glFogCoordhNV (GLhalfNV); +GLAPI void APIENTRY glFogCoordhvNV (const GLhalfNV *); +GLAPI void APIENTRY glSecondaryColor3hNV (GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glSecondaryColor3hvNV (const GLhalfNV *); +GLAPI void APIENTRY glVertexWeighthNV (GLhalfNV); +GLAPI void APIENTRY glVertexWeighthvNV (const GLhalfNV *); +GLAPI void APIENTRY glVertexAttrib1hNV (GLuint, GLhalfNV); +GLAPI void APIENTRY glVertexAttrib1hvNV (GLuint, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttrib2hNV (GLuint, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glVertexAttrib2hvNV (GLuint, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttrib3hNV (GLuint, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glVertexAttrib3hvNV (GLuint, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttrib4hNV (GLuint, GLhalfNV, GLhalfNV, GLhalfNV, GLhalfNV); +GLAPI void APIENTRY glVertexAttrib4hvNV (GLuint, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttribs1hvNV (GLuint, GLsizei, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttribs2hvNV (GLuint, GLsizei, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttribs3hvNV (GLuint, GLsizei, const GLhalfNV *); +GLAPI void APIENTRY glVertexAttribs4hvNV (GLuint, GLsizei, const GLhalfNV *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEX2HNVPROC) (GLhalfNV x, GLhalfNV y); +typedef void (APIENTRYP PFNGLVERTEX2HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEX3HNVPROC) (GLhalfNV x, GLhalfNV y, GLhalfNV z); +typedef void (APIENTRYP PFNGLVERTEX3HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEX4HNVPROC) (GLhalfNV x, GLhalfNV y, GLhalfNV z, GLhalfNV w); +typedef void (APIENTRYP PFNGLVERTEX4HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLNORMAL3HNVPROC) (GLhalfNV nx, GLhalfNV ny, GLhalfNV nz); +typedef void (APIENTRYP PFNGLNORMAL3HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLCOLOR3HNVPROC) (GLhalfNV red, GLhalfNV green, GLhalfNV blue); +typedef void (APIENTRYP PFNGLCOLOR3HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLCOLOR4HNVPROC) (GLhalfNV red, GLhalfNV green, GLhalfNV blue, GLhalfNV alpha); +typedef void (APIENTRYP PFNGLCOLOR4HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLTEXCOORD1HNVPROC) (GLhalfNV s); +typedef void (APIENTRYP PFNGLTEXCOORD1HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLTEXCOORD2HNVPROC) (GLhalfNV s, GLhalfNV t); +typedef void (APIENTRYP PFNGLTEXCOORD2HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLTEXCOORD3HNVPROC) (GLhalfNV s, GLhalfNV t, GLhalfNV r); +typedef void (APIENTRYP PFNGLTEXCOORD3HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLTEXCOORD4HNVPROC) (GLhalfNV s, GLhalfNV t, GLhalfNV r, GLhalfNV q); +typedef void (APIENTRYP PFNGLTEXCOORD4HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1HNVPROC) (GLenum target, GLhalfNV s); +typedef void (APIENTRYP PFNGLMULTITEXCOORD1HVNVPROC) (GLenum target, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2HNVPROC) (GLenum target, GLhalfNV s, GLhalfNV t); +typedef void (APIENTRYP PFNGLMULTITEXCOORD2HVNVPROC) (GLenum target, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3HNVPROC) (GLenum target, GLhalfNV s, GLhalfNV t, GLhalfNV r); +typedef void (APIENTRYP PFNGLMULTITEXCOORD3HVNVPROC) (GLenum target, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4HNVPROC) (GLenum target, GLhalfNV s, GLhalfNV t, GLhalfNV r, GLhalfNV q); +typedef void (APIENTRYP PFNGLMULTITEXCOORD4HVNVPROC) (GLenum target, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLFOGCOORDHNVPROC) (GLhalfNV fog); +typedef void (APIENTRYP PFNGLFOGCOORDHVNVPROC) (const GLhalfNV *fog); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3HNVPROC) (GLhalfNV red, GLhalfNV green, GLhalfNV blue); +typedef void (APIENTRYP PFNGLSECONDARYCOLOR3HVNVPROC) (const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXWEIGHTHNVPROC) (GLhalfNV weight); +typedef void (APIENTRYP PFNGLVERTEXWEIGHTHVNVPROC) (const GLhalfNV *weight); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1HNVPROC) (GLuint index, GLhalfNV x); +typedef void (APIENTRYP PFNGLVERTEXATTRIB1HVNVPROC) (GLuint index, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2HNVPROC) (GLuint index, GLhalfNV x, GLhalfNV y); +typedef void (APIENTRYP PFNGLVERTEXATTRIB2HVNVPROC) (GLuint index, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3HNVPROC) (GLuint index, GLhalfNV x, GLhalfNV y, GLhalfNV z); +typedef void (APIENTRYP PFNGLVERTEXATTRIB3HVNVPROC) (GLuint index, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4HNVPROC) (GLuint index, GLhalfNV x, GLhalfNV y, GLhalfNV z, GLhalfNV w); +typedef void (APIENTRYP PFNGLVERTEXATTRIB4HVNVPROC) (GLuint index, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS1HVNVPROC) (GLuint index, GLsizei n, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS2HVNVPROC) (GLuint index, GLsizei n, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS3HVNVPROC) (GLuint index, GLsizei n, const GLhalfNV *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBS4HVNVPROC) (GLuint index, GLsizei n, const GLhalfNV *v); +#endif + +#ifndef GL_NV_pixel_data_range +#define GL_NV_pixel_data_range 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPixelDataRangeNV (GLenum, GLsizei, GLvoid *); +GLAPI void APIENTRY glFlushPixelDataRangeNV (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPIXELDATARANGENVPROC) (GLenum target, GLsizei length, GLvoid *pointer); +typedef void (APIENTRYP PFNGLFLUSHPIXELDATARANGENVPROC) (GLenum target); +#endif + +#ifndef GL_NV_primitive_restart +#define GL_NV_primitive_restart 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPrimitiveRestartNV (void); +GLAPI void APIENTRY glPrimitiveRestartIndexNV (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPRIMITIVERESTARTNVPROC) (void); +typedef void (APIENTRYP PFNGLPRIMITIVERESTARTINDEXNVPROC) (GLuint index); +#endif + +#ifndef GL_NV_texture_expand_normal +#define GL_NV_texture_expand_normal 1 +#endif + +#ifndef GL_NV_vertex_program2 +#define GL_NV_vertex_program2 1 +#endif + +#ifndef GL_ATI_map_object_buffer +#define GL_ATI_map_object_buffer 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLvoid* APIENTRY glMapObjectBufferATI (GLuint); +GLAPI void APIENTRY glUnmapObjectBufferATI (GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLvoid* (APIENTRYP PFNGLMAPOBJECTBUFFERATIPROC) (GLuint buffer); +typedef void (APIENTRYP PFNGLUNMAPOBJECTBUFFERATIPROC) (GLuint buffer); +#endif + +#ifndef GL_ATI_separate_stencil +#define GL_ATI_separate_stencil 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glStencilOpSeparateATI (GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glStencilFuncSeparateATI (GLenum, GLenum, GLint, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSTENCILOPSEPARATEATIPROC) (GLenum face, GLenum sfail, GLenum dpfail, GLenum dppass); +typedef void (APIENTRYP PFNGLSTENCILFUNCSEPARATEATIPROC) (GLenum frontfunc, GLenum backfunc, GLint ref, GLuint mask); +#endif + +#ifndef GL_ATI_vertex_attrib_array_object +#define GL_ATI_vertex_attrib_array_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexAttribArrayObjectATI (GLuint, GLint, GLenum, GLboolean, GLsizei, GLuint, GLuint); +GLAPI void APIENTRY glGetVertexAttribArrayObjectfvATI (GLuint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetVertexAttribArrayObjectivATI (GLuint, GLenum, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXATTRIBARRAYOBJECTATIPROC) (GLuint index, GLint size, GLenum type, GLboolean normalized, GLsizei stride, GLuint buffer, GLuint offset); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBARRAYOBJECTFVATIPROC) (GLuint index, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBARRAYOBJECTIVATIPROC) (GLuint index, GLenum pname, GLint *params); +#endif + +#ifndef GL_OES_read_format +#define GL_OES_read_format 1 +#endif + +#ifndef GL_EXT_depth_bounds_test +#define GL_EXT_depth_bounds_test 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDepthBoundsEXT (GLclampd, GLclampd); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDEPTHBOUNDSEXTPROC) (GLclampd zmin, GLclampd zmax); +#endif + +#ifndef GL_EXT_texture_mirror_clamp +#define GL_EXT_texture_mirror_clamp 1 +#endif + +#ifndef GL_EXT_blend_equation_separate +#define GL_EXT_blend_equation_separate 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendEquationSeparateEXT (GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDEQUATIONSEPARATEEXTPROC) (GLenum modeRGB, GLenum modeAlpha); +#endif + +#ifndef GL_MESA_pack_invert +#define GL_MESA_pack_invert 1 +#endif + +#ifndef GL_MESA_ycbcr_texture +#define GL_MESA_ycbcr_texture 1 +#endif + +#ifndef GL_EXT_pixel_buffer_object +#define GL_EXT_pixel_buffer_object 1 +#endif + +#ifndef GL_NV_fragment_program_option +#define GL_NV_fragment_program_option 1 +#endif + +#ifndef GL_NV_fragment_program2 +#define GL_NV_fragment_program2 1 +#endif + +#ifndef GL_NV_vertex_program2_option +#define GL_NV_vertex_program2_option 1 +#endif + +#ifndef GL_NV_vertex_program3 +#define GL_NV_vertex_program3 1 +#endif + +#ifndef GL_EXT_framebuffer_object +#define GL_EXT_framebuffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLboolean APIENTRY glIsRenderbufferEXT (GLuint); +GLAPI void APIENTRY glBindRenderbufferEXT (GLenum, GLuint); +GLAPI void APIENTRY glDeleteRenderbuffersEXT (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenRenderbuffersEXT (GLsizei, GLuint *); +GLAPI void APIENTRY glRenderbufferStorageEXT (GLenum, GLenum, GLsizei, GLsizei); +GLAPI void APIENTRY glGetRenderbufferParameterivEXT (GLenum, GLenum, GLint *); +GLAPI GLboolean APIENTRY glIsFramebufferEXT (GLuint); +GLAPI void APIENTRY glBindFramebufferEXT (GLenum, GLuint); +GLAPI void APIENTRY glDeleteFramebuffersEXT (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenFramebuffersEXT (GLsizei, GLuint *); +GLAPI GLenum APIENTRY glCheckFramebufferStatusEXT (GLenum); +GLAPI void APIENTRY glFramebufferTexture1DEXT (GLenum, GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTexture2DEXT (GLenum, GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTexture3DEXT (GLenum, GLenum, GLenum, GLuint, GLint, GLint); +GLAPI void APIENTRY glFramebufferRenderbufferEXT (GLenum, GLenum, GLenum, GLuint); +GLAPI void APIENTRY glGetFramebufferAttachmentParameterivEXT (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGenerateMipmapEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLboolean (APIENTRYP PFNGLISRENDERBUFFEREXTPROC) (GLuint renderbuffer); +typedef void (APIENTRYP PFNGLBINDRENDERBUFFEREXTPROC) (GLenum target, GLuint renderbuffer); +typedef void (APIENTRYP PFNGLDELETERENDERBUFFERSEXTPROC) (GLsizei n, const GLuint *renderbuffers); +typedef void (APIENTRYP PFNGLGENRENDERBUFFERSEXTPROC) (GLsizei n, GLuint *renderbuffers); +typedef void (APIENTRYP PFNGLRENDERBUFFERSTORAGEEXTPROC) (GLenum target, GLenum internalformat, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETRENDERBUFFERPARAMETERIVEXTPROC) (GLenum target, GLenum pname, GLint *params); +typedef GLboolean (APIENTRYP PFNGLISFRAMEBUFFEREXTPROC) (GLuint framebuffer); +typedef void (APIENTRYP PFNGLBINDFRAMEBUFFEREXTPROC) (GLenum target, GLuint framebuffer); +typedef void (APIENTRYP PFNGLDELETEFRAMEBUFFERSEXTPROC) (GLsizei n, const GLuint *framebuffers); +typedef void (APIENTRYP PFNGLGENFRAMEBUFFERSEXTPROC) (GLsizei n, GLuint *framebuffers); +typedef GLenum (APIENTRYP PFNGLCHECKFRAMEBUFFERSTATUSEXTPROC) (GLenum target); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURE1DEXTPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURE2DEXTPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURE3DEXTPROC) (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); +typedef void (APIENTRYP PFNGLFRAMEBUFFERRENDERBUFFEREXTPROC) (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); +typedef void (APIENTRYP PFNGLGETFRAMEBUFFERATTACHMENTPARAMETERIVEXTPROC) (GLenum target, GLenum attachment, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGENERATEMIPMAPEXTPROC) (GLenum target); +#endif + +#ifndef GL_GREMEDY_string_marker +#define GL_GREMEDY_string_marker 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glStringMarkerGREMEDY (GLsizei, const GLvoid *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSTRINGMARKERGREMEDYPROC) (GLsizei len, const GLvoid *string); +#endif + +#ifndef GL_EXT_packed_depth_stencil +#define GL_EXT_packed_depth_stencil 1 +#endif + +#ifndef GL_EXT_stencil_clear_tag +#define GL_EXT_stencil_clear_tag 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glStencilClearTagEXT (GLsizei, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLSTENCILCLEARTAGEXTPROC) (GLsizei stencilTagBits, GLuint stencilClearTag); +#endif + +#ifndef GL_EXT_texture_sRGB +#define GL_EXT_texture_sRGB 1 +#endif + +#ifndef GL_EXT_framebuffer_blit +#define GL_EXT_framebuffer_blit 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlitFramebufferEXT (GLint, GLint, GLint, GLint, GLint, GLint, GLint, GLint, GLbitfield, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLITFRAMEBUFFEREXTPROC) (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter); +#endif + +#ifndef GL_EXT_framebuffer_multisample +#define GL_EXT_framebuffer_multisample 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glRenderbufferStorageMultisampleEXT (GLenum, GLsizei, GLenum, GLsizei, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLEEXTPROC) (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height); +#endif + +#ifndef GL_MESAX_texture_stack +#define GL_MESAX_texture_stack 1 +#endif + +#ifndef GL_EXT_timer_query +#define GL_EXT_timer_query 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetQueryObjecti64vEXT (GLuint, GLenum, GLint64EXT *); +GLAPI void APIENTRY glGetQueryObjectui64vEXT (GLuint, GLenum, GLuint64EXT *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETQUERYOBJECTI64VEXTPROC) (GLuint id, GLenum pname, GLint64EXT *params); +typedef void (APIENTRYP PFNGLGETQUERYOBJECTUI64VEXTPROC) (GLuint id, GLenum pname, GLuint64EXT *params); +#endif + +#ifndef GL_EXT_gpu_program_parameters +#define GL_EXT_gpu_program_parameters 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramEnvParameters4fvEXT (GLenum, GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glProgramLocalParameters4fvEXT (GLenum, GLuint, GLsizei, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERS4FVEXTPROC) (GLenum target, GLuint index, GLsizei count, const GLfloat *params); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERS4FVEXTPROC) (GLenum target, GLuint index, GLsizei count, const GLfloat *params); +#endif + +#ifndef GL_APPLE_flush_buffer_range +#define GL_APPLE_flush_buffer_range 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBufferParameteriAPPLE (GLenum, GLenum, GLint); +GLAPI void APIENTRY glFlushMappedBufferRangeAPPLE (GLenum, GLintptr, GLsizeiptr); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBUFFERPARAMETERIAPPLEPROC) (GLenum target, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLFLUSHMAPPEDBUFFERRANGEAPPLEPROC) (GLenum target, GLintptr offset, GLsizeiptr size); +#endif + +#ifndef GL_NV_gpu_program4 +#define GL_NV_gpu_program4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramLocalParameterI4iNV (GLenum, GLuint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glProgramLocalParameterI4ivNV (GLenum, GLuint, const GLint *); +GLAPI void APIENTRY glProgramLocalParametersI4ivNV (GLenum, GLuint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramLocalParameterI4uiNV (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glProgramLocalParameterI4uivNV (GLenum, GLuint, const GLuint *); +GLAPI void APIENTRY glProgramLocalParametersI4uivNV (GLenum, GLuint, GLsizei, const GLuint *); +GLAPI void APIENTRY glProgramEnvParameterI4iNV (GLenum, GLuint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glProgramEnvParameterI4ivNV (GLenum, GLuint, const GLint *); +GLAPI void APIENTRY glProgramEnvParametersI4ivNV (GLenum, GLuint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramEnvParameterI4uiNV (GLenum, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glProgramEnvParameterI4uivNV (GLenum, GLuint, const GLuint *); +GLAPI void APIENTRY glProgramEnvParametersI4uivNV (GLenum, GLuint, GLsizei, const GLuint *); +GLAPI void APIENTRY glGetProgramLocalParameterIivNV (GLenum, GLuint, GLint *); +GLAPI void APIENTRY glGetProgramLocalParameterIuivNV (GLenum, GLuint, GLuint *); +GLAPI void APIENTRY glGetProgramEnvParameterIivNV (GLenum, GLuint, GLint *); +GLAPI void APIENTRY glGetProgramEnvParameterIuivNV (GLenum, GLuint, GLuint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERI4INVPROC) (GLenum target, GLuint index, GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERI4IVNVPROC) (GLenum target, GLuint index, const GLint *params); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERSI4IVNVPROC) (GLenum target, GLuint index, GLsizei count, const GLint *params); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERI4UINVPROC) (GLenum target, GLuint index, GLuint x, GLuint y, GLuint z, GLuint w); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERI4UIVNVPROC) (GLenum target, GLuint index, const GLuint *params); +typedef void (APIENTRYP PFNGLPROGRAMLOCALPARAMETERSI4UIVNVPROC) (GLenum target, GLuint index, GLsizei count, const GLuint *params); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERI4INVPROC) (GLenum target, GLuint index, GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERI4IVNVPROC) (GLenum target, GLuint index, const GLint *params); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERSI4IVNVPROC) (GLenum target, GLuint index, GLsizei count, const GLint *params); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERI4UINVPROC) (GLenum target, GLuint index, GLuint x, GLuint y, GLuint z, GLuint w); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERI4UIVNVPROC) (GLenum target, GLuint index, const GLuint *params); +typedef void (APIENTRYP PFNGLPROGRAMENVPARAMETERSI4UIVNVPROC) (GLenum target, GLuint index, GLsizei count, const GLuint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMLOCALPARAMETERIIVNVPROC) (GLenum target, GLuint index, GLint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMLOCALPARAMETERIUIVNVPROC) (GLenum target, GLuint index, GLuint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMENVPARAMETERIIVNVPROC) (GLenum target, GLuint index, GLint *params); +typedef void (APIENTRYP PFNGLGETPROGRAMENVPARAMETERIUIVNVPROC) (GLenum target, GLuint index, GLuint *params); +#endif + +#ifndef GL_NV_geometry_program4 +#define GL_NV_geometry_program4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramVertexLimitNV (GLenum, GLint); +GLAPI void APIENTRY glFramebufferTextureEXT (GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glFramebufferTextureLayerEXT (GLenum, GLenum, GLuint, GLint, GLint); +GLAPI void APIENTRY glFramebufferTextureFaceEXT (GLenum, GLenum, GLuint, GLint, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMVERTEXLIMITNVPROC) (GLenum target, GLint limit); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREEXTPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTURELAYEREXTPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer); +typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREFACEEXTPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLenum face); +#endif + +#ifndef GL_EXT_geometry_shader4 +#define GL_EXT_geometry_shader4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramParameteriEXT (GLuint, GLenum, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMPARAMETERIEXTPROC) (GLuint program, GLenum pname, GLint value); +#endif + +#ifndef GL_NV_vertex_program4 +#define GL_NV_vertex_program4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glVertexAttribI1iEXT (GLuint, GLint); +GLAPI void APIENTRY glVertexAttribI2iEXT (GLuint, GLint, GLint); +GLAPI void APIENTRY glVertexAttribI3iEXT (GLuint, GLint, GLint, GLint); +GLAPI void APIENTRY glVertexAttribI4iEXT (GLuint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glVertexAttribI1uiEXT (GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI2uiEXT (GLuint, GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI3uiEXT (GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI4uiEXT (GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glVertexAttribI1ivEXT (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI2ivEXT (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI3ivEXT (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI4ivEXT (GLuint, const GLint *); +GLAPI void APIENTRY glVertexAttribI1uivEXT (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI2uivEXT (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI3uivEXT (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI4uivEXT (GLuint, const GLuint *); +GLAPI void APIENTRY glVertexAttribI4bvEXT (GLuint, const GLbyte *); +GLAPI void APIENTRY glVertexAttribI4svEXT (GLuint, const GLshort *); +GLAPI void APIENTRY glVertexAttribI4ubvEXT (GLuint, const GLubyte *); +GLAPI void APIENTRY glVertexAttribI4usvEXT (GLuint, const GLushort *); +GLAPI void APIENTRY glVertexAttribIPointerEXT (GLuint, GLint, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetVertexAttribIivEXT (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVertexAttribIuivEXT (GLuint, GLenum, GLuint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1IEXTPROC) (GLuint index, GLint x); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2IEXTPROC) (GLuint index, GLint x, GLint y); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3IEXTPROC) (GLuint index, GLint x, GLint y, GLint z); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4IEXTPROC) (GLuint index, GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1UIEXTPROC) (GLuint index, GLuint x); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2UIEXTPROC) (GLuint index, GLuint x, GLuint y); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3UIEXTPROC) (GLuint index, GLuint x, GLuint y, GLuint z); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UIEXTPROC) (GLuint index, GLuint x, GLuint y, GLuint z, GLuint w); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1IVEXTPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2IVEXTPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3IVEXTPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4IVEXTPROC) (GLuint index, const GLint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI1UIVEXTPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI2UIVEXTPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI3UIVEXTPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UIVEXTPROC) (GLuint index, const GLuint *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4BVEXTPROC) (GLuint index, const GLbyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4SVEXTPROC) (GLuint index, const GLshort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UBVEXTPROC) (GLuint index, const GLubyte *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBI4USVEXTPROC) (GLuint index, const GLushort *v); +typedef void (APIENTRYP PFNGLVERTEXATTRIBIPOINTEREXTPROC) (GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIIVEXTPROC) (GLuint index, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIUIVEXTPROC) (GLuint index, GLenum pname, GLuint *params); +#endif + +#ifndef GL_EXT_gpu_shader4 +#define GL_EXT_gpu_shader4 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetUniformuivEXT (GLuint, GLint, GLuint *); +GLAPI void APIENTRY glBindFragDataLocationEXT (GLuint, GLuint, const GLchar *); +GLAPI GLint APIENTRY glGetFragDataLocationEXT (GLuint, const GLchar *); +GLAPI void APIENTRY glUniform1uiEXT (GLint, GLuint); +GLAPI void APIENTRY glUniform2uiEXT (GLint, GLuint, GLuint); +GLAPI void APIENTRY glUniform3uiEXT (GLint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glUniform4uiEXT (GLint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glUniform1uivEXT (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glUniform2uivEXT (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glUniform3uivEXT (GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glUniform4uivEXT (GLint, GLsizei, const GLuint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETUNIFORMUIVEXTPROC) (GLuint program, GLint location, GLuint *params); +typedef void (APIENTRYP PFNGLBINDFRAGDATALOCATIONEXTPROC) (GLuint program, GLuint color, const GLchar *name); +typedef GLint (APIENTRYP PFNGLGETFRAGDATALOCATIONEXTPROC) (GLuint program, const GLchar *name); +typedef void (APIENTRYP PFNGLUNIFORM1UIEXTPROC) (GLint location, GLuint v0); +typedef void (APIENTRYP PFNGLUNIFORM2UIEXTPROC) (GLint location, GLuint v0, GLuint v1); +typedef void (APIENTRYP PFNGLUNIFORM3UIEXTPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2); +typedef void (APIENTRYP PFNGLUNIFORM4UIEXTPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3); +typedef void (APIENTRYP PFNGLUNIFORM1UIVEXTPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLUNIFORM2UIVEXTPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLUNIFORM3UIVEXTPROC) (GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLUNIFORM4UIVEXTPROC) (GLint location, GLsizei count, const GLuint *value); +#endif + +#ifndef GL_EXT_draw_instanced +#define GL_EXT_draw_instanced 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDrawArraysInstancedEXT (GLenum, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glDrawElementsInstancedEXT (GLenum, GLsizei, GLenum, const GLvoid *, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDRAWARRAYSINSTANCEDEXTPROC) (GLenum mode, GLint start, GLsizei count, GLsizei primcount); +typedef void (APIENTRYP PFNGLDRAWELEMENTSINSTANCEDEXTPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount); +#endif + +#ifndef GL_EXT_packed_float +#define GL_EXT_packed_float 1 +#endif + +#ifndef GL_EXT_texture_array +#define GL_EXT_texture_array 1 +#endif + +#ifndef GL_EXT_texture_buffer_object +#define GL_EXT_texture_buffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexBufferEXT (GLenum, GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXBUFFEREXTPROC) (GLenum target, GLenum internalformat, GLuint buffer); +#endif + +#ifndef GL_EXT_texture_compression_latc +#define GL_EXT_texture_compression_latc 1 +#endif + +#ifndef GL_EXT_texture_compression_rgtc +#define GL_EXT_texture_compression_rgtc 1 +#endif + +#ifndef GL_EXT_texture_shared_exponent +#define GL_EXT_texture_shared_exponent 1 +#endif + +#ifndef GL_NV_depth_buffer_float +#define GL_NV_depth_buffer_float 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glDepthRangedNV (GLdouble, GLdouble); +GLAPI void APIENTRY glClearDepthdNV (GLdouble); +GLAPI void APIENTRY glDepthBoundsdNV (GLdouble, GLdouble); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLDEPTHRANGEDNVPROC) (GLdouble zNear, GLdouble zFar); +typedef void (APIENTRYP PFNGLCLEARDEPTHDNVPROC) (GLdouble depth); +typedef void (APIENTRYP PFNGLDEPTHBOUNDSDNVPROC) (GLdouble zmin, GLdouble zmax); +#endif + +#ifndef GL_NV_fragment_program4 +#define GL_NV_fragment_program4 1 +#endif + +#ifndef GL_NV_framebuffer_multisample_coverage +#define GL_NV_framebuffer_multisample_coverage 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glRenderbufferStorageMultisampleCoverageNV (GLenum, GLsizei, GLsizei, GLenum, GLsizei, GLsizei); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLRENDERBUFFERSTORAGEMULTISAMPLECOVERAGENVPROC) (GLenum target, GLsizei coverageSamples, GLsizei colorSamples, GLenum internalformat, GLsizei width, GLsizei height); +#endif + +#ifndef GL_EXT_framebuffer_sRGB +#define GL_EXT_framebuffer_sRGB 1 +#endif + +#ifndef GL_NV_geometry_shader4 +#define GL_NV_geometry_shader4 1 +#endif + +#ifndef GL_NV_parameter_buffer_object +#define GL_NV_parameter_buffer_object 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProgramBufferParametersfvNV (GLenum, GLuint, GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glProgramBufferParametersIivNV (GLenum, GLuint, GLuint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramBufferParametersIuivNV (GLenum, GLuint, GLuint, GLsizei, const GLuint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROGRAMBUFFERPARAMETERSFVNVPROC) (GLenum target, GLuint buffer, GLuint index, GLsizei count, const GLfloat *params); +typedef void (APIENTRYP PFNGLPROGRAMBUFFERPARAMETERSIIVNVPROC) (GLenum target, GLuint buffer, GLuint index, GLsizei count, const GLint *params); +typedef void (APIENTRYP PFNGLPROGRAMBUFFERPARAMETERSIUIVNVPROC) (GLenum target, GLuint buffer, GLuint index, GLsizei count, const GLuint *params); +#endif + +#ifndef GL_EXT_draw_buffers2 +#define GL_EXT_draw_buffers2 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glColorMaskIndexedEXT (GLuint, GLboolean, GLboolean, GLboolean, GLboolean); +GLAPI void APIENTRY glGetBooleanIndexedvEXT (GLenum, GLuint, GLboolean *); +GLAPI void APIENTRY glGetIntegerIndexedvEXT (GLenum, GLuint, GLint *); +GLAPI void APIENTRY glEnableIndexedEXT (GLenum, GLuint); +GLAPI void APIENTRY glDisableIndexedEXT (GLenum, GLuint); +GLAPI GLboolean APIENTRY glIsEnabledIndexedEXT (GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCOLORMASKINDEXEDEXTPROC) (GLuint index, GLboolean r, GLboolean g, GLboolean b, GLboolean a); +typedef void (APIENTRYP PFNGLGETBOOLEANINDEXEDVEXTPROC) (GLenum target, GLuint index, GLboolean *data); +typedef void (APIENTRYP PFNGLGETINTEGERINDEXEDVEXTPROC) (GLenum target, GLuint index, GLint *data); +typedef void (APIENTRYP PFNGLENABLEINDEXEDEXTPROC) (GLenum target, GLuint index); +typedef void (APIENTRYP PFNGLDISABLEINDEXEDEXTPROC) (GLenum target, GLuint index); +typedef GLboolean (APIENTRYP PFNGLISENABLEDINDEXEDEXTPROC) (GLenum target, GLuint index); +#endif + +#ifndef GL_NV_transform_feedback +#define GL_NV_transform_feedback 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBeginTransformFeedbackNV (GLenum); +GLAPI void APIENTRY glEndTransformFeedbackNV (void); +GLAPI void APIENTRY glTransformFeedbackAttribsNV (GLuint, const GLint *, GLenum); +GLAPI void APIENTRY glBindBufferRangeNV (GLenum, GLuint, GLuint, GLintptr, GLsizeiptr); +GLAPI void APIENTRY glBindBufferOffsetNV (GLenum, GLuint, GLuint, GLintptr); +GLAPI void APIENTRY glBindBufferBaseNV (GLenum, GLuint, GLuint); +GLAPI void APIENTRY glTransformFeedbackVaryingsNV (GLuint, GLsizei, const GLchar* *, GLenum); +GLAPI void APIENTRY glActiveVaryingNV (GLuint, const GLchar *); +GLAPI GLint APIENTRY glGetVaryingLocationNV (GLuint, const GLchar *); +GLAPI void APIENTRY glGetActiveVaryingNV (GLuint, GLuint, GLsizei, GLsizei *, GLsizei *, GLenum *, GLchar *); +GLAPI void APIENTRY glGetTransformFeedbackVaryingNV (GLuint, GLuint, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBEGINTRANSFORMFEEDBACKNVPROC) (GLenum primitiveMode); +typedef void (APIENTRYP PFNGLENDTRANSFORMFEEDBACKNVPROC) (void); +typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKATTRIBSNVPROC) (GLuint count, const GLint *attribs, GLenum bufferMode); +typedef void (APIENTRYP PFNGLBINDBUFFERRANGENVPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size); +typedef void (APIENTRYP PFNGLBINDBUFFEROFFSETNVPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset); +typedef void (APIENTRYP PFNGLBINDBUFFERBASENVPROC) (GLenum target, GLuint index, GLuint buffer); +typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKVARYINGSNVPROC) (GLuint program, GLsizei count, const GLchar* *varyings, GLenum bufferMode); +typedef void (APIENTRYP PFNGLACTIVEVARYINGNVPROC) (GLuint program, const GLchar *name); +typedef GLint (APIENTRYP PFNGLGETVARYINGLOCATIONNVPROC) (GLuint program, const GLchar *name); +typedef void (APIENTRYP PFNGLGETACTIVEVARYINGNVPROC) (GLuint program, GLuint index, GLsizei bufSize, GLsizei *length, GLsizei *size, GLenum *type, GLchar *name); +typedef void (APIENTRYP PFNGLGETTRANSFORMFEEDBACKVARYINGNVPROC) (GLuint program, GLuint index, GLint *location); +#endif + +#ifndef GL_EXT_bindable_uniform +#define GL_EXT_bindable_uniform 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glUniformBufferEXT (GLuint, GLint, GLuint); +GLAPI GLint APIENTRY glGetUniformBufferSizeEXT (GLuint, GLint); +GLAPI GLintptr APIENTRY glGetUniformOffsetEXT (GLuint, GLint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLUNIFORMBUFFEREXTPROC) (GLuint program, GLint location, GLuint buffer); +typedef GLint (APIENTRYP PFNGLGETUNIFORMBUFFERSIZEEXTPROC) (GLuint program, GLint location); +typedef GLintptr (APIENTRYP PFNGLGETUNIFORMOFFSETEXTPROC) (GLuint program, GLint location); +#endif + +#ifndef GL_EXT_texture_integer +#define GL_EXT_texture_integer 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTexParameterIivEXT (GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glTexParameterIuivEXT (GLenum, GLenum, const GLuint *); +GLAPI void APIENTRY glGetTexParameterIivEXT (GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetTexParameterIuivEXT (GLenum, GLenum, GLuint *); +GLAPI void APIENTRY glClearColorIiEXT (GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glClearColorIuiEXT (GLuint, GLuint, GLuint, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXPARAMETERIIVEXTPROC) (GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLTEXPARAMETERIUIVEXTPROC) (GLenum target, GLenum pname, const GLuint *params); +typedef void (APIENTRYP PFNGLGETTEXPARAMETERIIVEXTPROC) (GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETTEXPARAMETERIUIVEXTPROC) (GLenum target, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLCLEARCOLORIIEXTPROC) (GLint red, GLint green, GLint blue, GLint alpha); +typedef void (APIENTRYP PFNGLCLEARCOLORIUIEXTPROC) (GLuint red, GLuint green, GLuint blue, GLuint alpha); +#endif + +#ifndef GL_GREMEDY_frame_terminator +#define GL_GREMEDY_frame_terminator 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glFrameTerminatorGREMEDY (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLFRAMETERMINATORGREMEDYPROC) (void); +#endif + +#ifndef GL_NV_conditional_render +#define GL_NV_conditional_render 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBeginConditionalRenderNV (GLuint, GLenum); +GLAPI void APIENTRY glEndConditionalRenderNV (void); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBEGINCONDITIONALRENDERNVPROC) (GLuint id, GLenum mode); +typedef void (APIENTRYP PFNGLENDCONDITIONALRENDERNVPROC) (void); +#endif + +#ifndef GL_NV_present_video +#define GL_NV_present_video 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glPresentFrameKeyedNV (GLuint, GLuint64EXT, GLuint, GLuint, GLenum, GLenum, GLuint, GLuint, GLenum, GLuint, GLuint); +GLAPI void APIENTRY glPresentFrameDualFillNV (GLuint, GLuint64EXT, GLuint, GLuint, GLenum, GLenum, GLuint, GLenum, GLuint, GLenum, GLuint, GLenum, GLuint); +GLAPI void APIENTRY glGetVideoivNV (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetVideouivNV (GLuint, GLenum, GLuint *); +GLAPI void APIENTRY glGetVideoi64vNV (GLuint, GLenum, GLint64EXT *); +GLAPI void APIENTRY glGetVideoui64vNV (GLuint, GLenum, GLuint64EXT *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPRESENTFRAMEKEYEDNVPROC) (GLuint video_slot, GLuint64EXT minPresentTime, GLuint beginPresentTimeId, GLuint presentDurationId, GLenum type, GLenum target0, GLuint fill0, GLuint key0, GLenum target1, GLuint fill1, GLuint key1); +typedef void (APIENTRYP PFNGLPRESENTFRAMEDUALFILLNVPROC) (GLuint video_slot, GLuint64EXT minPresentTime, GLuint beginPresentTimeId, GLuint presentDurationId, GLenum type, GLenum target0, GLuint fill0, GLenum target1, GLuint fill1, GLenum target2, GLuint fill2, GLenum target3, GLuint fill3); +typedef void (APIENTRYP PFNGLGETVIDEOIVNVPROC) (GLuint video_slot, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETVIDEOUIVNVPROC) (GLuint video_slot, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLGETVIDEOI64VNVPROC) (GLuint video_slot, GLenum pname, GLint64EXT *params); +typedef void (APIENTRYP PFNGLGETVIDEOUI64VNVPROC) (GLuint video_slot, GLenum pname, GLuint64EXT *params); +#endif + +#ifndef GL_EXT_transform_feedback +#define GL_EXT_transform_feedback 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBeginTransformFeedbackEXT (GLenum); +GLAPI void APIENTRY glEndTransformFeedbackEXT (void); +GLAPI void APIENTRY glBindBufferRangeEXT (GLenum, GLuint, GLuint, GLintptr, GLsizeiptr); +GLAPI void APIENTRY glBindBufferOffsetEXT (GLenum, GLuint, GLuint, GLintptr); +GLAPI void APIENTRY glBindBufferBaseEXT (GLenum, GLuint, GLuint); +GLAPI void APIENTRY glTransformFeedbackVaryingsEXT (GLuint, GLsizei, const GLchar* *, GLenum); +GLAPI void APIENTRY glGetTransformFeedbackVaryingEXT (GLuint, GLuint, GLsizei, GLsizei *, GLsizei *, GLenum *, GLchar *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBEGINTRANSFORMFEEDBACKEXTPROC) (GLenum primitiveMode); +typedef void (APIENTRYP PFNGLENDTRANSFORMFEEDBACKEXTPROC) (void); +typedef void (APIENTRYP PFNGLBINDBUFFERRANGEEXTPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size); +typedef void (APIENTRYP PFNGLBINDBUFFEROFFSETEXTPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset); +typedef void (APIENTRYP PFNGLBINDBUFFERBASEEXTPROC) (GLenum target, GLuint index, GLuint buffer); +typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKVARYINGSEXTPROC) (GLuint program, GLsizei count, const GLchar* *varyings, GLenum bufferMode); +typedef void (APIENTRYP PFNGLGETTRANSFORMFEEDBACKVARYINGEXTPROC) (GLuint program, GLuint index, GLsizei bufSize, GLsizei *length, GLsizei *size, GLenum *type, GLchar *name); +#endif + +#ifndef GL_EXT_direct_state_access +#define GL_EXT_direct_state_access 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glClientAttribDefaultEXT (GLbitfield); +GLAPI void APIENTRY glPushClientAttribDefaultEXT (GLbitfield); +GLAPI void APIENTRY glMatrixLoadfEXT (GLenum, const GLfloat *); +GLAPI void APIENTRY glMatrixLoaddEXT (GLenum, const GLdouble *); +GLAPI void APIENTRY glMatrixMultfEXT (GLenum, const GLfloat *); +GLAPI void APIENTRY glMatrixMultdEXT (GLenum, const GLdouble *); +GLAPI void APIENTRY glMatrixLoadIdentityEXT (GLenum); +GLAPI void APIENTRY glMatrixRotatefEXT (GLenum, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMatrixRotatedEXT (GLenum, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMatrixScalefEXT (GLenum, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMatrixScaledEXT (GLenum, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMatrixTranslatefEXT (GLenum, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glMatrixTranslatedEXT (GLenum, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMatrixFrustumEXT (GLenum, GLdouble, GLdouble, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMatrixOrthoEXT (GLenum, GLdouble, GLdouble, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glMatrixPopEXT (GLenum); +GLAPI void APIENTRY glMatrixPushEXT (GLenum); +GLAPI void APIENTRY glMatrixLoadTransposefEXT (GLenum, const GLfloat *); +GLAPI void APIENTRY glMatrixLoadTransposedEXT (GLenum, const GLdouble *); +GLAPI void APIENTRY glMatrixMultTransposefEXT (GLenum, const GLfloat *); +GLAPI void APIENTRY glMatrixMultTransposedEXT (GLenum, const GLdouble *); +GLAPI void APIENTRY glTextureParameterfEXT (GLuint, GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glTextureParameterfvEXT (GLuint, GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glTextureParameteriEXT (GLuint, GLenum, GLenum, GLint); +GLAPI void APIENTRY glTextureParameterivEXT (GLuint, GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glTextureImage1DEXT (GLuint, GLenum, GLint, GLenum, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTextureImage2DEXT (GLuint, GLenum, GLint, GLenum, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTextureSubImage1DEXT (GLuint, GLenum, GLint, GLint, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTextureSubImage2DEXT (GLuint, GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyTextureImage1DEXT (GLuint, GLenum, GLint, GLenum, GLint, GLint, GLsizei, GLint); +GLAPI void APIENTRY glCopyTextureImage2DEXT (GLuint, GLenum, GLint, GLenum, GLint, GLint, GLsizei, GLsizei, GLint); +GLAPI void APIENTRY glCopyTextureSubImage1DEXT (GLuint, GLenum, GLint, GLint, GLint, GLint, GLsizei); +GLAPI void APIENTRY glCopyTextureSubImage2DEXT (GLuint, GLenum, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glGetTextureImageEXT (GLuint, GLenum, GLint, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetTextureParameterfvEXT (GLuint, GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetTextureParameterivEXT (GLuint, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetTextureLevelParameterfvEXT (GLuint, GLenum, GLint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetTextureLevelParameterivEXT (GLuint, GLenum, GLint, GLenum, GLint *); +GLAPI void APIENTRY glTextureImage3DEXT (GLuint, GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glTextureSubImage3DEXT (GLuint, GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyTextureSubImage3DEXT (GLuint, GLenum, GLint, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glMultiTexParameterfEXT (GLenum, GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glMultiTexParameterfvEXT (GLenum, GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexParameteriEXT (GLenum, GLenum, GLenum, GLint); +GLAPI void APIENTRY glMultiTexParameterivEXT (GLenum, GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexImage1DEXT (GLenum, GLenum, GLint, GLenum, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glMultiTexImage2DEXT (GLenum, GLenum, GLint, GLenum, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glMultiTexSubImage1DEXT (GLenum, GLenum, GLint, GLint, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glMultiTexSubImage2DEXT (GLenum, GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyMultiTexImage1DEXT (GLenum, GLenum, GLint, GLenum, GLint, GLint, GLsizei, GLint); +GLAPI void APIENTRY glCopyMultiTexImage2DEXT (GLenum, GLenum, GLint, GLenum, GLint, GLint, GLsizei, GLsizei, GLint); +GLAPI void APIENTRY glCopyMultiTexSubImage1DEXT (GLenum, GLenum, GLint, GLint, GLint, GLint, GLsizei); +GLAPI void APIENTRY glCopyMultiTexSubImage2DEXT (GLenum, GLenum, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glGetMultiTexImageEXT (GLenum, GLenum, GLint, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glGetMultiTexParameterfvEXT (GLenum, GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMultiTexParameterivEXT (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetMultiTexLevelParameterfvEXT (GLenum, GLenum, GLint, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMultiTexLevelParameterivEXT (GLenum, GLenum, GLint, GLenum, GLint *); +GLAPI void APIENTRY glMultiTexImage3DEXT (GLenum, GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glMultiTexSubImage3DEXT (GLenum, GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLenum, const GLvoid *); +GLAPI void APIENTRY glCopyMultiTexSubImage3DEXT (GLenum, GLenum, GLint, GLint, GLint, GLint, GLint, GLint, GLsizei, GLsizei); +GLAPI void APIENTRY glBindMultiTextureEXT (GLenum, GLenum, GLuint); +GLAPI void APIENTRY glEnableClientStateIndexedEXT (GLenum, GLuint); +GLAPI void APIENTRY glDisableClientStateIndexedEXT (GLenum, GLuint); +GLAPI void APIENTRY glMultiTexCoordPointerEXT (GLenum, GLint, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glMultiTexEnvfEXT (GLenum, GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glMultiTexEnvfvEXT (GLenum, GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexEnviEXT (GLenum, GLenum, GLenum, GLint); +GLAPI void APIENTRY glMultiTexEnvivEXT (GLenum, GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexGendEXT (GLenum, GLenum, GLenum, GLdouble); +GLAPI void APIENTRY glMultiTexGendvEXT (GLenum, GLenum, GLenum, const GLdouble *); +GLAPI void APIENTRY glMultiTexGenfEXT (GLenum, GLenum, GLenum, GLfloat); +GLAPI void APIENTRY glMultiTexGenfvEXT (GLenum, GLenum, GLenum, const GLfloat *); +GLAPI void APIENTRY glMultiTexGeniEXT (GLenum, GLenum, GLenum, GLint); +GLAPI void APIENTRY glMultiTexGenivEXT (GLenum, GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glGetMultiTexEnvfvEXT (GLenum, GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMultiTexEnvivEXT (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetMultiTexGendvEXT (GLenum, GLenum, GLenum, GLdouble *); +GLAPI void APIENTRY glGetMultiTexGenfvEXT (GLenum, GLenum, GLenum, GLfloat *); +GLAPI void APIENTRY glGetMultiTexGenivEXT (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetFloatIndexedvEXT (GLenum, GLuint, GLfloat *); +GLAPI void APIENTRY glGetDoubleIndexedvEXT (GLenum, GLuint, GLdouble *); +GLAPI void APIENTRY glGetPointerIndexedvEXT (GLenum, GLuint, GLvoid* *); +GLAPI void APIENTRY glCompressedTextureImage3DEXT (GLuint, GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTextureImage2DEXT (GLuint, GLenum, GLint, GLenum, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTextureImage1DEXT (GLuint, GLenum, GLint, GLenum, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTextureSubImage3DEXT (GLuint, GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTextureSubImage2DEXT (GLuint, GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedTextureSubImage1DEXT (GLuint, GLenum, GLint, GLint, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetCompressedTextureImageEXT (GLuint, GLenum, GLint, GLvoid *); +GLAPI void APIENTRY glCompressedMultiTexImage3DEXT (GLenum, GLenum, GLint, GLenum, GLsizei, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedMultiTexImage2DEXT (GLenum, GLenum, GLint, GLenum, GLsizei, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedMultiTexImage1DEXT (GLenum, GLenum, GLint, GLenum, GLsizei, GLint, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedMultiTexSubImage3DEXT (GLenum, GLenum, GLint, GLint, GLint, GLint, GLsizei, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedMultiTexSubImage2DEXT (GLenum, GLenum, GLint, GLint, GLint, GLsizei, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glCompressedMultiTexSubImage1DEXT (GLenum, GLenum, GLint, GLint, GLsizei, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetCompressedMultiTexImageEXT (GLenum, GLenum, GLint, GLvoid *); +GLAPI void APIENTRY glNamedProgramStringEXT (GLuint, GLenum, GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glNamedProgramLocalParameter4dEXT (GLuint, GLenum, GLuint, GLdouble, GLdouble, GLdouble, GLdouble); +GLAPI void APIENTRY glNamedProgramLocalParameter4dvEXT (GLuint, GLenum, GLuint, const GLdouble *); +GLAPI void APIENTRY glNamedProgramLocalParameter4fEXT (GLuint, GLenum, GLuint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glNamedProgramLocalParameter4fvEXT (GLuint, GLenum, GLuint, const GLfloat *); +GLAPI void APIENTRY glGetNamedProgramLocalParameterdvEXT (GLuint, GLenum, GLuint, GLdouble *); +GLAPI void APIENTRY glGetNamedProgramLocalParameterfvEXT (GLuint, GLenum, GLuint, GLfloat *); +GLAPI void APIENTRY glGetNamedProgramivEXT (GLuint, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetNamedProgramStringEXT (GLuint, GLenum, GLenum, GLvoid *); +GLAPI void APIENTRY glNamedProgramLocalParameters4fvEXT (GLuint, GLenum, GLuint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glNamedProgramLocalParameterI4iEXT (GLuint, GLenum, GLuint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glNamedProgramLocalParameterI4ivEXT (GLuint, GLenum, GLuint, const GLint *); +GLAPI void APIENTRY glNamedProgramLocalParametersI4ivEXT (GLuint, GLenum, GLuint, GLsizei, const GLint *); +GLAPI void APIENTRY glNamedProgramLocalParameterI4uiEXT (GLuint, GLenum, GLuint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glNamedProgramLocalParameterI4uivEXT (GLuint, GLenum, GLuint, const GLuint *); +GLAPI void APIENTRY glNamedProgramLocalParametersI4uivEXT (GLuint, GLenum, GLuint, GLsizei, const GLuint *); +GLAPI void APIENTRY glGetNamedProgramLocalParameterIivEXT (GLuint, GLenum, GLuint, GLint *); +GLAPI void APIENTRY glGetNamedProgramLocalParameterIuivEXT (GLuint, GLenum, GLuint, GLuint *); +GLAPI void APIENTRY glTextureParameterIivEXT (GLuint, GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glTextureParameterIuivEXT (GLuint, GLenum, GLenum, const GLuint *); +GLAPI void APIENTRY glGetTextureParameterIivEXT (GLuint, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetTextureParameterIuivEXT (GLuint, GLenum, GLenum, GLuint *); +GLAPI void APIENTRY glMultiTexParameterIivEXT (GLenum, GLenum, GLenum, const GLint *); +GLAPI void APIENTRY glMultiTexParameterIuivEXT (GLenum, GLenum, GLenum, const GLuint *); +GLAPI void APIENTRY glGetMultiTexParameterIivEXT (GLenum, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGetMultiTexParameterIuivEXT (GLenum, GLenum, GLenum, GLuint *); +GLAPI void APIENTRY glProgramUniform1fEXT (GLuint, GLint, GLfloat); +GLAPI void APIENTRY glProgramUniform2fEXT (GLuint, GLint, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramUniform3fEXT (GLuint, GLint, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramUniform4fEXT (GLuint, GLint, GLfloat, GLfloat, GLfloat, GLfloat); +GLAPI void APIENTRY glProgramUniform1iEXT (GLuint, GLint, GLint); +GLAPI void APIENTRY glProgramUniform2iEXT (GLuint, GLint, GLint, GLint); +GLAPI void APIENTRY glProgramUniform3iEXT (GLuint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glProgramUniform4iEXT (GLuint, GLint, GLint, GLint, GLint, GLint); +GLAPI void APIENTRY glProgramUniform1fvEXT (GLuint, GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glProgramUniform2fvEXT (GLuint, GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glProgramUniform3fvEXT (GLuint, GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glProgramUniform4fvEXT (GLuint, GLint, GLsizei, const GLfloat *); +GLAPI void APIENTRY glProgramUniform1ivEXT (GLuint, GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramUniform2ivEXT (GLuint, GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramUniform3ivEXT (GLuint, GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramUniform4ivEXT (GLuint, GLint, GLsizei, const GLint *); +GLAPI void APIENTRY glProgramUniformMatrix2fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix3fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix4fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix2x3fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix3x2fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix2x4fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix4x2fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix3x4fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniformMatrix4x3fvEXT (GLuint, GLint, GLsizei, GLboolean, const GLfloat *); +GLAPI void APIENTRY glProgramUniform1uiEXT (GLuint, GLint, GLuint); +GLAPI void APIENTRY glProgramUniform2uiEXT (GLuint, GLint, GLuint, GLuint); +GLAPI void APIENTRY glProgramUniform3uiEXT (GLuint, GLint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glProgramUniform4uiEXT (GLuint, GLint, GLuint, GLuint, GLuint, GLuint); +GLAPI void APIENTRY glProgramUniform1uivEXT (GLuint, GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glProgramUniform2uivEXT (GLuint, GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glProgramUniform3uivEXT (GLuint, GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glProgramUniform4uivEXT (GLuint, GLint, GLsizei, const GLuint *); +GLAPI void APIENTRY glNamedBufferDataEXT (GLuint, GLsizeiptr, const GLvoid *, GLenum); +GLAPI void APIENTRY glNamedBufferSubDataEXT (GLuint, GLintptr, GLsizeiptr, const GLvoid *); +GLAPI GLvoid* APIENTRY glMapNamedBufferEXT (GLuint, GLenum); +GLAPI GLboolean APIENTRY glUnmapNamedBufferEXT (GLuint); +GLAPI void APIENTRY glGetNamedBufferParameterivEXT (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glGetNamedBufferPointervEXT (GLuint, GLenum, GLvoid* *); +GLAPI void APIENTRY glGetNamedBufferSubDataEXT (GLuint, GLintptr, GLsizeiptr, GLvoid *); +GLAPI void APIENTRY glTextureBufferEXT (GLuint, GLenum, GLenum, GLuint); +GLAPI void APIENTRY glMultiTexBufferEXT (GLenum, GLenum, GLenum, GLuint); +GLAPI void APIENTRY glNamedRenderbufferStorageEXT (GLuint, GLenum, GLsizei, GLsizei); +GLAPI void APIENTRY glGetNamedRenderbufferParameterivEXT (GLuint, GLenum, GLint *); +GLAPI GLenum APIENTRY glCheckNamedFramebufferStatusEXT (GLuint, GLenum); +GLAPI void APIENTRY glNamedFramebufferTexture1DEXT (GLuint, GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glNamedFramebufferTexture2DEXT (GLuint, GLenum, GLenum, GLuint, GLint); +GLAPI void APIENTRY glNamedFramebufferTexture3DEXT (GLuint, GLenum, GLenum, GLuint, GLint, GLint); +GLAPI void APIENTRY glNamedFramebufferRenderbufferEXT (GLuint, GLenum, GLenum, GLuint); +GLAPI void APIENTRY glGetNamedFramebufferAttachmentParameterivEXT (GLuint, GLenum, GLenum, GLint *); +GLAPI void APIENTRY glGenerateTextureMipmapEXT (GLuint, GLenum); +GLAPI void APIENTRY glGenerateMultiTexMipmapEXT (GLenum, GLenum); +GLAPI void APIENTRY glFramebufferDrawBufferEXT (GLuint, GLenum); +GLAPI void APIENTRY glFramebufferDrawBuffersEXT (GLuint, GLsizei, const GLenum *); +GLAPI void APIENTRY glFramebufferReadBufferEXT (GLuint, GLenum); +GLAPI void APIENTRY glGetFramebufferParameterivEXT (GLuint, GLenum, GLint *); +GLAPI void APIENTRY glNamedRenderbufferStorageMultisampleEXT (GLuint, GLsizei, GLenum, GLsizei, GLsizei); +GLAPI void APIENTRY glNamedRenderbufferStorageMultisampleCoverageEXT (GLuint, GLsizei, GLsizei, GLenum, GLsizei, GLsizei); +GLAPI void APIENTRY glNamedFramebufferTextureEXT (GLuint, GLenum, GLuint, GLint); +GLAPI void APIENTRY glNamedFramebufferTextureLayerEXT (GLuint, GLenum, GLuint, GLint, GLint); +GLAPI void APIENTRY glNamedFramebufferTextureFaceEXT (GLuint, GLenum, GLuint, GLint, GLenum); +GLAPI void APIENTRY glTextureRenderbufferEXT (GLuint, GLenum, GLuint); +GLAPI void APIENTRY glMultiTexRenderbufferEXT (GLenum, GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLCLIENTATTRIBDEFAULTEXTPROC) (GLbitfield mask); +typedef void (APIENTRYP PFNGLPUSHCLIENTATTRIBDEFAULTEXTPROC) (GLbitfield mask); +typedef void (APIENTRYP PFNGLMATRIXLOADFEXTPROC) (GLenum mode, const GLfloat *m); +typedef void (APIENTRYP PFNGLMATRIXLOADDEXTPROC) (GLenum mode, const GLdouble *m); +typedef void (APIENTRYP PFNGLMATRIXMULTFEXTPROC) (GLenum mode, const GLfloat *m); +typedef void (APIENTRYP PFNGLMATRIXMULTDEXTPROC) (GLenum mode, const GLdouble *m); +typedef void (APIENTRYP PFNGLMATRIXLOADIDENTITYEXTPROC) (GLenum mode); +typedef void (APIENTRYP PFNGLMATRIXROTATEFEXTPROC) (GLenum mode, GLfloat angle, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLMATRIXROTATEDEXTPROC) (GLenum mode, GLdouble angle, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLMATRIXSCALEFEXTPROC) (GLenum mode, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLMATRIXSCALEDEXTPROC) (GLenum mode, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLMATRIXTRANSLATEFEXTPROC) (GLenum mode, GLfloat x, GLfloat y, GLfloat z); +typedef void (APIENTRYP PFNGLMATRIXTRANSLATEDEXTPROC) (GLenum mode, GLdouble x, GLdouble y, GLdouble z); +typedef void (APIENTRYP PFNGLMATRIXFRUSTUMEXTPROC) (GLenum mode, GLdouble left, GLdouble right, GLdouble bottom, GLdouble top, GLdouble zNear, GLdouble zFar); +typedef void (APIENTRYP PFNGLMATRIXORTHOEXTPROC) (GLenum mode, GLdouble left, GLdouble right, GLdouble bottom, GLdouble top, GLdouble zNear, GLdouble zFar); +typedef void (APIENTRYP PFNGLMATRIXPOPEXTPROC) (GLenum mode); +typedef void (APIENTRYP PFNGLMATRIXPUSHEXTPROC) (GLenum mode); +typedef void (APIENTRYP PFNGLMATRIXLOADTRANSPOSEFEXTPROC) (GLenum mode, const GLfloat *m); +typedef void (APIENTRYP PFNGLMATRIXLOADTRANSPOSEDEXTPROC) (GLenum mode, const GLdouble *m); +typedef void (APIENTRYP PFNGLMATRIXMULTTRANSPOSEFEXTPROC) (GLenum mode, const GLfloat *m); +typedef void (APIENTRYP PFNGLMATRIXMULTTRANSPOSEDEXTPROC) (GLenum mode, const GLdouble *m); +typedef void (APIENTRYP PFNGLTEXTUREPARAMETERFEXTPROC) (GLuint texture, GLenum target, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLTEXTUREPARAMETERFVEXTPROC) (GLuint texture, GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLTEXTUREPARAMETERIEXTPROC) (GLuint texture, GLenum target, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLTEXTUREPARAMETERIVEXTPROC) (GLuint texture, GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLTEXTUREIMAGE1DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXTUREIMAGE2DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXTURESUBIMAGE1DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXTURESUBIMAGE2DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLCOPYTEXTUREIMAGE1DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLint border); +typedef void (APIENTRYP PFNGLCOPYTEXTUREIMAGE2DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border); +typedef void (APIENTRYP PFNGLCOPYTEXTURESUBIMAGE1DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLCOPYTEXTURESUBIMAGE2DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETTEXTUREIMAGEEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum format, GLenum type, GLvoid *pixels); +typedef void (APIENTRYP PFNGLGETTEXTUREPARAMETERFVEXTPROC) (GLuint texture, GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETTEXTUREPARAMETERIVEXTPROC) (GLuint texture, GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETTEXTURELEVELPARAMETERFVEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETTEXTURELEVELPARAMETERIVEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLTEXTUREIMAGE3DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLTEXTURESUBIMAGE3DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLCOPYTEXTURESUBIMAGE3DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLMULTITEXPARAMETERFEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLMULTITEXPARAMETERFVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLMULTITEXPARAMETERIEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLMULTITEXPARAMETERIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLMULTITEXIMAGE1DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLMULTITEXIMAGE2DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLMULTITEXSUBIMAGE1DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLMULTITEXSUBIMAGE2DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLCOPYMULTITEXIMAGE1DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLint border); +typedef void (APIENTRYP PFNGLCOPYMULTITEXIMAGE2DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width, GLsizei height, GLint border); +typedef void (APIENTRYP PFNGLCOPYMULTITEXSUBIMAGE1DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint x, GLint y, GLsizei width); +typedef void (APIENTRYP PFNGLCOPYMULTITEXSUBIMAGE2DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETMULTITEXIMAGEEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum format, GLenum type, GLvoid *pixels); +typedef void (APIENTRYP PFNGLGETMULTITEXPARAMETERFVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMULTITEXPARAMETERIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMULTITEXLEVELPARAMETERFVEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMULTITEXLEVELPARAMETERIVEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLMULTITEXIMAGE3DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLMULTITEXSUBIMAGE3DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid *pixels); +typedef void (APIENTRYP PFNGLCOPYMULTITEXSUBIMAGE3DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLBINDMULTITEXTUREEXTPROC) (GLenum texunit, GLenum target, GLuint texture); +typedef void (APIENTRYP PFNGLENABLECLIENTSTATEINDEXEDEXTPROC) (GLenum array, GLuint index); +typedef void (APIENTRYP PFNGLDISABLECLIENTSTATEINDEXEDEXTPROC) (GLenum array, GLuint index); +typedef void (APIENTRYP PFNGLMULTITEXCOORDPOINTEREXTPROC) (GLenum texunit, GLint size, GLenum type, GLsizei stride, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLMULTITEXENVFEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLMULTITEXENVFVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLMULTITEXENVIEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLMULTITEXENVIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLMULTITEXGENDEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, GLdouble param); +typedef void (APIENTRYP PFNGLMULTITEXGENDVEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, const GLdouble *params); +typedef void (APIENTRYP PFNGLMULTITEXGENFEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, GLfloat param); +typedef void (APIENTRYP PFNGLMULTITEXGENFVEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, const GLfloat *params); +typedef void (APIENTRYP PFNGLMULTITEXGENIEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, GLint param); +typedef void (APIENTRYP PFNGLMULTITEXGENIVEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLGETMULTITEXENVFVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMULTITEXENVIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMULTITEXGENDVEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, GLdouble *params); +typedef void (APIENTRYP PFNGLGETMULTITEXGENFVEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, GLfloat *params); +typedef void (APIENTRYP PFNGLGETMULTITEXGENIVEXTPROC) (GLenum texunit, GLenum coord, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETFLOATINDEXEDVEXTPROC) (GLenum target, GLuint index, GLfloat *data); +typedef void (APIENTRYP PFNGLGETDOUBLEINDEXEDVEXTPROC) (GLenum target, GLuint index, GLdouble *data); +typedef void (APIENTRYP PFNGLGETPOINTERINDEXEDVEXTPROC) (GLenum target, GLuint index, GLvoid* *data); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXTUREIMAGE3DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXTUREIMAGE2DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXTUREIMAGE1DEXTPROC) (GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXTURESUBIMAGE3DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXTURESUBIMAGE2DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDTEXTURESUBIMAGE1DEXTPROC) (GLuint texture, GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLGETCOMPRESSEDTEXTUREIMAGEEXTPROC) (GLuint texture, GLenum target, GLint lod, GLvoid *img); +typedef void (APIENTRYP PFNGLCOMPRESSEDMULTITEXIMAGE3DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDMULTITEXIMAGE2DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDMULTITEXIMAGE1DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDMULTITEXSUBIMAGE3DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDMULTITEXSUBIMAGE2DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLCOMPRESSEDMULTITEXSUBIMAGE1DEXTPROC) (GLenum texunit, GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const GLvoid *bits); +typedef void (APIENTRYP PFNGLGETCOMPRESSEDMULTITEXIMAGEEXTPROC) (GLenum texunit, GLenum target, GLint lod, GLvoid *img); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMSTRINGEXTPROC) (GLuint program, GLenum target, GLenum format, GLsizei len, const GLvoid *string); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETER4DEXTPROC) (GLuint program, GLenum target, GLuint index, GLdouble x, GLdouble y, GLdouble z, GLdouble w); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETER4DVEXTPROC) (GLuint program, GLenum target, GLuint index, const GLdouble *params); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETER4FEXTPROC) (GLuint program, GLenum target, GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETER4FVEXTPROC) (GLuint program, GLenum target, GLuint index, const GLfloat *params); +typedef void (APIENTRYP PFNGLGETNAMEDPROGRAMLOCALPARAMETERDVEXTPROC) (GLuint program, GLenum target, GLuint index, GLdouble *params); +typedef void (APIENTRYP PFNGLGETNAMEDPROGRAMLOCALPARAMETERFVEXTPROC) (GLuint program, GLenum target, GLuint index, GLfloat *params); +typedef void (APIENTRYP PFNGLGETNAMEDPROGRAMIVEXTPROC) (GLuint program, GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETNAMEDPROGRAMSTRINGEXTPROC) (GLuint program, GLenum target, GLenum pname, GLvoid *string); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERS4FVEXTPROC) (GLuint program, GLenum target, GLuint index, GLsizei count, const GLfloat *params); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERI4IEXTPROC) (GLuint program, GLenum target, GLuint index, GLint x, GLint y, GLint z, GLint w); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERI4IVEXTPROC) (GLuint program, GLenum target, GLuint index, const GLint *params); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERSI4IVEXTPROC) (GLuint program, GLenum target, GLuint index, GLsizei count, const GLint *params); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERI4UIEXTPROC) (GLuint program, GLenum target, GLuint index, GLuint x, GLuint y, GLuint z, GLuint w); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERI4UIVEXTPROC) (GLuint program, GLenum target, GLuint index, const GLuint *params); +typedef void (APIENTRYP PFNGLNAMEDPROGRAMLOCALPARAMETERSI4UIVEXTPROC) (GLuint program, GLenum target, GLuint index, GLsizei count, const GLuint *params); +typedef void (APIENTRYP PFNGLGETNAMEDPROGRAMLOCALPARAMETERIIVEXTPROC) (GLuint program, GLenum target, GLuint index, GLint *params); +typedef void (APIENTRYP PFNGLGETNAMEDPROGRAMLOCALPARAMETERIUIVEXTPROC) (GLuint program, GLenum target, GLuint index, GLuint *params); +typedef void (APIENTRYP PFNGLTEXTUREPARAMETERIIVEXTPROC) (GLuint texture, GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLTEXTUREPARAMETERIUIVEXTPROC) (GLuint texture, GLenum target, GLenum pname, const GLuint *params); +typedef void (APIENTRYP PFNGLGETTEXTUREPARAMETERIIVEXTPROC) (GLuint texture, GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETTEXTUREPARAMETERIUIVEXTPROC) (GLuint texture, GLenum target, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLMULTITEXPARAMETERIIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, const GLint *params); +typedef void (APIENTRYP PFNGLMULTITEXPARAMETERIUIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, const GLuint *params); +typedef void (APIENTRYP PFNGLGETMULTITEXPARAMETERIIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETMULTITEXPARAMETERIUIVEXTPROC) (GLenum texunit, GLenum target, GLenum pname, GLuint *params); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM1FEXTPROC) (GLuint program, GLint location, GLfloat v0); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM2FEXTPROC) (GLuint program, GLint location, GLfloat v0, GLfloat v1); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM3FEXTPROC) (GLuint program, GLint location, GLfloat v0, GLfloat v1, GLfloat v2); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM4FEXTPROC) (GLuint program, GLint location, GLfloat v0, GLfloat v1, GLfloat v2, GLfloat v3); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM1IEXTPROC) (GLuint program, GLint location, GLint v0); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM2IEXTPROC) (GLuint program, GLint location, GLint v0, GLint v1); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM3IEXTPROC) (GLuint program, GLint location, GLint v0, GLint v1, GLint v2); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM4IEXTPROC) (GLuint program, GLint location, GLint v0, GLint v1, GLint v2, GLint v3); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM1FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM2FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM3FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM4FVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM1IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM2IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM3IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM4IVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX2FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX3FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX4FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX2X3FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX3X2FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX2X4FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX4X2FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX3X4FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORMMATRIX4X3FVEXTPROC) (GLuint program, GLint location, GLsizei count, GLboolean transpose, const GLfloat *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM1UIEXTPROC) (GLuint program, GLint location, GLuint v0); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM2UIEXTPROC) (GLuint program, GLint location, GLuint v0, GLuint v1); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM3UIEXTPROC) (GLuint program, GLint location, GLuint v0, GLuint v1, GLuint v2); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM4UIEXTPROC) (GLuint program, GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM1UIVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM2UIVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM3UIVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLPROGRAMUNIFORM4UIVEXTPROC) (GLuint program, GLint location, GLsizei count, const GLuint *value); +typedef void (APIENTRYP PFNGLNAMEDBUFFERDATAEXTPROC) (GLuint buffer, GLsizeiptr size, const GLvoid *data, GLenum usage); +typedef void (APIENTRYP PFNGLNAMEDBUFFERSUBDATAEXTPROC) (GLuint buffer, GLintptr offset, GLsizeiptr size, const GLvoid *data); +typedef GLvoid* (APIENTRYP PFNGLMAPNAMEDBUFFEREXTPROC) (GLuint buffer, GLenum access); +typedef GLboolean (APIENTRYP PFNGLUNMAPNAMEDBUFFEREXTPROC) (GLuint buffer); +typedef void (APIENTRYP PFNGLGETNAMEDBUFFERPARAMETERIVEXTPROC) (GLuint buffer, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGETNAMEDBUFFERPOINTERVEXTPROC) (GLuint buffer, GLenum pname, GLvoid* *params); +typedef void (APIENTRYP PFNGLGETNAMEDBUFFERSUBDATAEXTPROC) (GLuint buffer, GLintptr offset, GLsizeiptr size, GLvoid *data); +typedef void (APIENTRYP PFNGLTEXTUREBUFFEREXTPROC) (GLuint texture, GLenum target, GLenum internalformat, GLuint buffer); +typedef void (APIENTRYP PFNGLMULTITEXBUFFEREXTPROC) (GLenum texunit, GLenum target, GLenum internalformat, GLuint buffer); +typedef void (APIENTRYP PFNGLNAMEDRENDERBUFFERSTORAGEEXTPROC) (GLuint renderbuffer, GLenum internalformat, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLGETNAMEDRENDERBUFFERPARAMETERIVEXTPROC) (GLuint renderbuffer, GLenum pname, GLint *params); +typedef GLenum (APIENTRYP PFNGLCHECKNAMEDFRAMEBUFFERSTATUSEXTPROC) (GLuint framebuffer, GLenum target); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERTEXTURE1DEXTPROC) (GLuint framebuffer, GLenum attachment, GLenum textarget, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERTEXTURE2DEXTPROC) (GLuint framebuffer, GLenum attachment, GLenum textarget, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERTEXTURE3DEXTPROC) (GLuint framebuffer, GLenum attachment, GLenum textarget, GLuint texture, GLint level, GLint zoffset); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERRENDERBUFFEREXTPROC) (GLuint framebuffer, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer); +typedef void (APIENTRYP PFNGLGETNAMEDFRAMEBUFFERATTACHMENTPARAMETERIVEXTPROC) (GLuint framebuffer, GLenum attachment, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLGENERATETEXTUREMIPMAPEXTPROC) (GLuint texture, GLenum target); +typedef void (APIENTRYP PFNGLGENERATEMULTITEXMIPMAPEXTPROC) (GLenum texunit, GLenum target); +typedef void (APIENTRYP PFNGLFRAMEBUFFERDRAWBUFFEREXTPROC) (GLuint framebuffer, GLenum mode); +typedef void (APIENTRYP PFNGLFRAMEBUFFERDRAWBUFFERSEXTPROC) (GLuint framebuffer, GLsizei n, const GLenum *bufs); +typedef void (APIENTRYP PFNGLFRAMEBUFFERREADBUFFEREXTPROC) (GLuint framebuffer, GLenum mode); +typedef void (APIENTRYP PFNGLGETFRAMEBUFFERPARAMETERIVEXTPROC) (GLuint framebuffer, GLenum pname, GLint *params); +typedef void (APIENTRYP PFNGLNAMEDRENDERBUFFERSTORAGEMULTISAMPLEEXTPROC) (GLuint renderbuffer, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLNAMEDRENDERBUFFERSTORAGEMULTISAMPLECOVERAGEEXTPROC) (GLuint renderbuffer, GLsizei coverageSamples, GLsizei colorSamples, GLenum internalformat, GLsizei width, GLsizei height); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERTEXTUREEXTPROC) (GLuint framebuffer, GLenum attachment, GLuint texture, GLint level); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERTEXTURELAYEREXTPROC) (GLuint framebuffer, GLenum attachment, GLuint texture, GLint level, GLint layer); +typedef void (APIENTRYP PFNGLNAMEDFRAMEBUFFERTEXTUREFACEEXTPROC) (GLuint framebuffer, GLenum attachment, GLuint texture, GLint level, GLenum face); +typedef void (APIENTRYP PFNGLTEXTURERENDERBUFFEREXTPROC) (GLuint texture, GLenum target, GLuint renderbuffer); +typedef void (APIENTRYP PFNGLMULTITEXRENDERBUFFEREXTPROC) (GLenum texunit, GLenum target, GLuint renderbuffer); +#endif + +#ifndef GL_EXT_vertex_array_bgra +#define GL_EXT_vertex_array_bgra 1 +#endif + +#ifndef GL_EXT_texture_swizzle +#define GL_EXT_texture_swizzle 1 +#endif + +#ifndef GL_NV_explicit_multisample +#define GL_NV_explicit_multisample 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetMultisamplefvNV (GLenum, GLuint, GLfloat *); +GLAPI void APIENTRY glSampleMaskIndexedNV (GLuint, GLbitfield); +GLAPI void APIENTRY glTexRenderbufferNV (GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETMULTISAMPLEFVNVPROC) (GLenum pname, GLuint index, GLfloat *val); +typedef void (APIENTRYP PFNGLSAMPLEMASKINDEXEDNVPROC) (GLuint index, GLbitfield mask); +typedef void (APIENTRYP PFNGLTEXRENDERBUFFERNVPROC) (GLenum target, GLuint renderbuffer); +#endif + +#ifndef GL_NV_transform_feedback2 +#define GL_NV_transform_feedback2 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBindTransformFeedbackNV (GLenum, GLuint); +GLAPI void APIENTRY glDeleteTransformFeedbacksNV (GLsizei, const GLuint *); +GLAPI void APIENTRY glGenTransformFeedbacksNV (GLsizei, GLuint *); +GLAPI GLboolean APIENTRY glIsTransformFeedbackNV (GLuint); +GLAPI void APIENTRY glPauseTransformFeedbackNV (void); +GLAPI void APIENTRY glResumeTransformFeedbackNV (void); +GLAPI void APIENTRY glDrawTransformFeedbackNV (GLenum, GLuint); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBINDTRANSFORMFEEDBACKNVPROC) (GLenum target, GLuint id); +typedef void (APIENTRYP PFNGLDELETETRANSFORMFEEDBACKSNVPROC) (GLsizei n, const GLuint *ids); +typedef void (APIENTRYP PFNGLGENTRANSFORMFEEDBACKSNVPROC) (GLsizei n, GLuint *ids); +typedef GLboolean (APIENTRYP PFNGLISTRANSFORMFEEDBACKNVPROC) (GLuint id); +typedef void (APIENTRYP PFNGLPAUSETRANSFORMFEEDBACKNVPROC) (void); +typedef void (APIENTRYP PFNGLRESUMETRANSFORMFEEDBACKNVPROC) (void); +typedef void (APIENTRYP PFNGLDRAWTRANSFORMFEEDBACKNVPROC) (GLenum mode, GLuint id); +#endif + +#ifndef GL_ATI_meminfo +#define GL_ATI_meminfo 1 +#endif + +#ifndef GL_AMD_performance_monitor +#define GL_AMD_performance_monitor 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glGetPerfMonitorGroupsAMD (GLint *, GLsizei, GLuint *); +GLAPI void APIENTRY glGetPerfMonitorCountersAMD (GLuint, GLint *, GLint *, GLsizei, GLuint *); +GLAPI void APIENTRY glGetPerfMonitorGroupStringAMD (GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI void APIENTRY glGetPerfMonitorCounterStringAMD (GLuint, GLuint, GLsizei, GLsizei *, GLchar *); +GLAPI void APIENTRY glGetPerfMonitorCounterInfoAMD (GLuint, GLuint, GLenum, void *); +GLAPI void APIENTRY glGenPerfMonitorsAMD (GLsizei, GLuint *); +GLAPI void APIENTRY glDeletePerfMonitorsAMD (GLsizei, GLuint *); +GLAPI void APIENTRY glSelectPerfMonitorCountersAMD (GLuint, GLboolean, GLuint, GLint, GLuint *); +GLAPI void APIENTRY glBeginPerfMonitorAMD (GLuint); +GLAPI void APIENTRY glEndPerfMonitorAMD (GLuint); +GLAPI void APIENTRY glGetPerfMonitorCounterDataAMD (GLuint, GLenum, GLsizei, GLuint *, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLGETPERFMONITORGROUPSAMDPROC) (GLint *numGroups, GLsizei groupsSize, GLuint *groups); +typedef void (APIENTRYP PFNGLGETPERFMONITORCOUNTERSAMDPROC) (GLuint group, GLint *numCounters, GLint *maxActiveCounters, GLsizei counterSize, GLuint *counters); +typedef void (APIENTRYP PFNGLGETPERFMONITORGROUPSTRINGAMDPROC) (GLuint group, GLsizei bufSize, GLsizei *length, GLchar *groupString); +typedef void (APIENTRYP PFNGLGETPERFMONITORCOUNTERSTRINGAMDPROC) (GLuint group, GLuint counter, GLsizei bufSize, GLsizei *length, GLchar *counterString); +typedef void (APIENTRYP PFNGLGETPERFMONITORCOUNTERINFOAMDPROC) (GLuint group, GLuint counter, GLenum pname, void *data); +typedef void (APIENTRYP PFNGLGENPERFMONITORSAMDPROC) (GLsizei n, GLuint *monitors); +typedef void (APIENTRYP PFNGLDELETEPERFMONITORSAMDPROC) (GLsizei n, GLuint *monitors); +typedef void (APIENTRYP PFNGLSELECTPERFMONITORCOUNTERSAMDPROC) (GLuint monitor, GLboolean enable, GLuint group, GLint numCounters, GLuint *counterList); +typedef void (APIENTRYP PFNGLBEGINPERFMONITORAMDPROC) (GLuint monitor); +typedef void (APIENTRYP PFNGLENDPERFMONITORAMDPROC) (GLuint monitor); +typedef void (APIENTRYP PFNGLGETPERFMONITORCOUNTERDATAAMDPROC) (GLuint monitor, GLenum pname, GLsizei dataSize, GLuint *data, GLint *bytesWritten); +#endif + +#ifndef GL_AMD_texture_texture4 +#define GL_AMD_texture_texture4 1 +#endif + +#ifndef GL_AMD_vertex_shader_tesselator +#define GL_AMD_vertex_shader_tesselator 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTessellationFactorAMD (GLfloat); +GLAPI void APIENTRY glTessellationModeAMD (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTESSELLATIONFACTORAMDPROC) (GLfloat factor); +typedef void (APIENTRYP PFNGLTESSELLATIONMODEAMDPROC) (GLenum mode); +#endif + +#ifndef GL_EXT_provoking_vertex +#define GL_EXT_provoking_vertex 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glProvokingVertexEXT (GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLPROVOKINGVERTEXEXTPROC) (GLenum mode); +#endif + +#ifndef GL_EXT_texture_snorm +#define GL_EXT_texture_snorm 1 +#endif + +#ifndef GL_AMD_draw_buffers_blend +#define GL_AMD_draw_buffers_blend 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glBlendFuncIndexedAMD (GLuint, GLenum, GLenum); +GLAPI void APIENTRY glBlendFuncSeparateIndexedAMD (GLuint, GLenum, GLenum, GLenum, GLenum); +GLAPI void APIENTRY glBlendEquationIndexedAMD (GLuint, GLenum); +GLAPI void APIENTRY glBlendEquationSeparateIndexedAMD (GLuint, GLenum, GLenum); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLBLENDFUNCINDEXEDAMDPROC) (GLuint buf, GLenum src, GLenum dst); +typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEINDEXEDAMDPROC) (GLuint buf, GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha); +typedef void (APIENTRYP PFNGLBLENDEQUATIONINDEXEDAMDPROC) (GLuint buf, GLenum mode); +typedef void (APIENTRYP PFNGLBLENDEQUATIONSEPARATEINDEXEDAMDPROC) (GLuint buf, GLenum modeRGB, GLenum modeAlpha); +#endif + +#ifndef GL_APPLE_texture_range +#define GL_APPLE_texture_range 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glTextureRangeAPPLE (GLenum, GLsizei, const GLvoid *); +GLAPI void APIENTRY glGetTexParameterPointervAPPLE (GLenum, GLenum, GLvoid* *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLTEXTURERANGEAPPLEPROC) (GLenum target, GLsizei length, const GLvoid *pointer); +typedef void (APIENTRYP PFNGLGETTEXPARAMETERPOINTERVAPPLEPROC) (GLenum target, GLenum pname, GLvoid* *params); +#endif + +#ifndef GL_APPLE_float_pixels +#define GL_APPLE_float_pixels 1 +#endif + +#ifndef GL_APPLE_vertex_program_evaluators +#define GL_APPLE_vertex_program_evaluators 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI void APIENTRY glEnableVertexAttribAPPLE (GLuint, GLenum); +GLAPI void APIENTRY glDisableVertexAttribAPPLE (GLuint, GLenum); +GLAPI GLboolean APIENTRY glIsVertexAttribEnabledAPPLE (GLuint, GLenum); +GLAPI void APIENTRY glMapVertexAttrib1dAPPLE (GLuint, GLuint, GLdouble, GLdouble, GLint, GLint, const GLdouble *); +GLAPI void APIENTRY glMapVertexAttrib1fAPPLE (GLuint, GLuint, GLfloat, GLfloat, GLint, GLint, const GLfloat *); +GLAPI void APIENTRY glMapVertexAttrib2dAPPLE (GLuint, GLuint, GLdouble, GLdouble, GLint, GLint, GLdouble, GLdouble, GLint, GLint, const GLdouble *); +GLAPI void APIENTRY glMapVertexAttrib2fAPPLE (GLuint, GLuint, GLfloat, GLfloat, GLint, GLint, GLfloat, GLfloat, GLint, GLint, const GLfloat *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef void (APIENTRYP PFNGLENABLEVERTEXATTRIBAPPLEPROC) (GLuint index, GLenum pname); +typedef void (APIENTRYP PFNGLDISABLEVERTEXATTRIBAPPLEPROC) (GLuint index, GLenum pname); +typedef GLboolean (APIENTRYP PFNGLISVERTEXATTRIBENABLEDAPPLEPROC) (GLuint index, GLenum pname); +typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB1DAPPLEPROC) (GLuint index, GLuint size, GLdouble u1, GLdouble u2, GLint stride, GLint order, const GLdouble *points); +typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB1FAPPLEPROC) (GLuint index, GLuint size, GLfloat u1, GLfloat u2, GLint stride, GLint order, const GLfloat *points); +typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB2DAPPLEPROC) (GLuint index, GLuint size, GLdouble u1, GLdouble u2, GLint ustride, GLint uorder, GLdouble v1, GLdouble v2, GLint vstride, GLint vorder, const GLdouble *points); +typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB2FAPPLEPROC) (GLuint index, GLuint size, GLfloat u1, GLfloat u2, GLint ustride, GLint uorder, GLfloat v1, GLfloat v2, GLint vstride, GLint vorder, const GLfloat *points); +#endif + +#ifndef GL_APPLE_aux_depth_stencil +#define GL_APPLE_aux_depth_stencil 1 +#endif + +#ifndef GL_APPLE_object_purgeable +#define GL_APPLE_object_purgeable 1 +#ifdef GL_GLEXT_PROTOTYPES +GLAPI GLenum APIENTRY glObjectPurgeableAPPLE (GLenum, GLuint, GLenum); +GLAPI GLenum APIENTRY glObjectUnpurgeableAPPLE (GLenum, GLuint, GLenum); +GLAPI void APIENTRY glGetObjectParameterivAPPLE (GLenum, GLuint, GLenum, GLint *); +#endif /* GL_GLEXT_PROTOTYPES */ +typedef GLenum (APIENTRYP PFNGLOBJECTPURGEABLEAPPLEPROC) (GLenum objectType, GLuint name, GLenum option); +typedef GLenum (APIENTRYP PFNGLOBJECTUNPURGEABLEAPPLEPROC) (GLenum objectType, GLuint name, GLenum option); +typedef void (APIENTRYP PFNGLGETOBJECTPARAMETERIVAPPLEPROC) (GLenum objectType, GLuint name, GLenum pname, GLint *params); +#endif + +#ifndef GL_APPLE_row_bytes +#define GL_APPLE_row_bytes 1 +#endif + + +#ifdef __cplusplus +} +#endif + +#endif diff --git a/opencl/api/opencl/khronos/headers/KHR/khrplatform.h b/opencl/api/opencl/khronos/headers/KHR/khrplatform.h new file mode 100644 index 0000000000..c9e6f17d34 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/KHR/khrplatform.h @@ -0,0 +1,282 @@ +#ifndef __khrplatform_h_ +#define __khrplatform_h_ + +/* +** Copyright (c) 2008-2009 The Khronos Group Inc. +** +** Permission is hereby granted, free of charge, to any person obtaining a +** copy of this software and/or associated documentation files (the +** "Materials"), to deal in the Materials without restriction, including +** without limitation the rights to use, copy, modify, merge, publish, +** distribute, sublicense, and/or sell copies of the Materials, and to +** permit persons to whom the Materials are furnished to do so, subject to +** the following conditions: +** +** The above copyright notice and this permission notice shall be included +** in all copies or substantial portions of the Materials. +** +** THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY +** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. +*/ + +/* Khronos platform-specific types and definitions. + * + * $Revision: 23298 $ on $Date: 2013-09-30 17:07:13 -0700 (Mon, 30 Sep 2013) $ + * + * Adopters may modify this file to suit their platform. Adopters are + * encouraged to submit platform specific modifications to the Khronos + * group so that they can be included in future versions of this file. + * Please submit changes by sending them to the public Khronos Bugzilla + * (http://khronos.org/bugzilla) by filing a bug against product + * "Khronos (general)" component "Registry". + * + * A predefined template which fills in some of the bug fields can be + * reached using http://tinyurl.com/khrplatform-h-bugreport, but you + * must create a Bugzilla login first. + * + * + * See the Implementer's Guidelines for information about where this file + * should be located on your system and for more details of its use: + * http://www.khronos.org/registry/implementers_guide.pdf + * + * This file should be included as + * #include + * by Khronos client API header files that use its types and defines. + * + * The types in khrplatform.h should only be used to define API-specific types. + * + * Types defined in khrplatform.h: + * khronos_int8_t signed 8 bit + * khronos_uint8_t unsigned 8 bit + * khronos_int16_t signed 16 bit + * khronos_uint16_t unsigned 16 bit + * khronos_int32_t signed 32 bit + * khronos_uint32_t unsigned 32 bit + * khronos_int64_t signed 64 bit + * khronos_uint64_t unsigned 64 bit + * khronos_intptr_t signed same number of bits as a pointer + * khronos_uintptr_t unsigned same number of bits as a pointer + * khronos_ssize_t signed size + * khronos_usize_t unsigned size + * khronos_float_t signed 32 bit floating point + * khronos_time_ns_t unsigned 64 bit time in nanoseconds + * khronos_utime_nanoseconds_t unsigned time interval or absolute time in + * nanoseconds + * khronos_stime_nanoseconds_t signed time interval in nanoseconds + * khronos_boolean_enum_t enumerated boolean type. This should + * only be used as a base type when a client API's boolean type is + * an enum. Client APIs which use an integer or other type for + * booleans cannot use this as the base type for their boolean. + * + * Tokens defined in khrplatform.h: + * + * KHRONOS_FALSE, KHRONOS_TRUE Enumerated boolean false/true values. + * + * KHRONOS_SUPPORT_INT64 is 1 if 64 bit integers are supported; otherwise 0. + * KHRONOS_SUPPORT_FLOAT is 1 if floats are supported; otherwise 0. + * + * Calling convention macros defined in this file: + * KHRONOS_APICALL + * KHRONOS_APIENTRY + * KHRONOS_APIATTRIBUTES + * + * These may be used in function prototypes as: + * + * KHRONOS_APICALL void KHRONOS_APIENTRY funcname( + * int arg1, + * int arg2) KHRONOS_APIATTRIBUTES; + */ + +/*------------------------------------------------------------------------- + * Definition of KHRONOS_APICALL + *------------------------------------------------------------------------- + * This precedes the return type of the function in the function prototype. + */ +#if defined(_WIN32) && !defined(__SCITECH_SNAP__) +# define KHRONOS_APICALL __declspec(dllimport) +#elif defined (__SYMBIAN32__) +# define KHRONOS_APICALL IMPORT_C +#else +# define KHRONOS_APICALL +#endif + +/*------------------------------------------------------------------------- + * Definition of KHRONOS_APIENTRY + *------------------------------------------------------------------------- + * This follows the return type of the function and precedes the function + * name in the function prototype. + */ +#if defined(_WIN32) && !defined(_WIN32_WCE) && !defined(__SCITECH_SNAP__) + /* Win32 but not WinCE */ +# define KHRONOS_APIENTRY __stdcall +#else +# define KHRONOS_APIENTRY +#endif + +/*------------------------------------------------------------------------- + * Definition of KHRONOS_APIATTRIBUTES + *------------------------------------------------------------------------- + * This follows the closing parenthesis of the function prototype arguments. + */ +#if defined (__ARMCC_2__) +#define KHRONOS_APIATTRIBUTES __softfp +#else +#define KHRONOS_APIATTRIBUTES +#endif + +/*------------------------------------------------------------------------- + * basic type definitions + *-----------------------------------------------------------------------*/ +#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || defined(__GNUC__) || defined(__SCO__) || defined(__USLC__) + + +/* + * Using + */ +#include +typedef int32_t khronos_int32_t; +typedef uint32_t khronos_uint32_t; +typedef int64_t khronos_int64_t; +typedef uint64_t khronos_uint64_t; +#define KHRONOS_SUPPORT_INT64 1 +#define KHRONOS_SUPPORT_FLOAT 1 + +#elif defined(__VMS ) || defined(__sgi) + +/* + * Using + */ +#include +typedef int32_t khronos_int32_t; +typedef uint32_t khronos_uint32_t; +typedef int64_t khronos_int64_t; +typedef uint64_t khronos_uint64_t; +#define KHRONOS_SUPPORT_INT64 1 +#define KHRONOS_SUPPORT_FLOAT 1 + +#elif defined(_WIN32) && !defined(__SCITECH_SNAP__) + +/* + * Win32 + */ +typedef __int32 khronos_int32_t; +typedef unsigned __int32 khronos_uint32_t; +typedef __int64 khronos_int64_t; +typedef unsigned __int64 khronos_uint64_t; +#define KHRONOS_SUPPORT_INT64 1 +#define KHRONOS_SUPPORT_FLOAT 1 + +#elif defined(__sun__) || defined(__digital__) + +/* + * Sun or Digital + */ +typedef int khronos_int32_t; +typedef unsigned int khronos_uint32_t; +#if defined(__arch64__) || defined(_LP64) +typedef long int khronos_int64_t; +typedef unsigned long int khronos_uint64_t; +#else +typedef long long int khronos_int64_t; +typedef unsigned long long int khronos_uint64_t; +#endif /* __arch64__ */ +#define KHRONOS_SUPPORT_INT64 1 +#define KHRONOS_SUPPORT_FLOAT 1 + +#elif 0 + +/* + * Hypothetical platform with no float or int64 support + */ +typedef int khronos_int32_t; +typedef unsigned int khronos_uint32_t; +#define KHRONOS_SUPPORT_INT64 0 +#define KHRONOS_SUPPORT_FLOAT 0 + +#else + +/* + * Generic fallback + */ +#include +typedef int32_t khronos_int32_t; +typedef uint32_t khronos_uint32_t; +typedef int64_t khronos_int64_t; +typedef uint64_t khronos_uint64_t; +#define KHRONOS_SUPPORT_INT64 1 +#define KHRONOS_SUPPORT_FLOAT 1 + +#endif + + +/* + * Types that are (so far) the same on all platforms + */ +typedef signed char khronos_int8_t; +typedef unsigned char khronos_uint8_t; +typedef signed short int khronos_int16_t; +typedef unsigned short int khronos_uint16_t; + +/* + * Types that differ between LLP64 and LP64 architectures - in LLP64, + * pointers are 64 bits, but 'long' is still 32 bits. Win64 appears + * to be the only LLP64 architecture in current use. + */ +#ifdef _WIN64 +typedef signed long long int khronos_intptr_t; +typedef unsigned long long int khronos_uintptr_t; +typedef signed long long int khronos_ssize_t; +typedef unsigned long long int khronos_usize_t; +#else +typedef signed long int khronos_intptr_t; +typedef unsigned long int khronos_uintptr_t; +typedef signed long int khronos_ssize_t; +typedef unsigned long int khronos_usize_t; +#endif + +#if KHRONOS_SUPPORT_FLOAT +/* + * Float type + */ +typedef float khronos_float_t; +#endif + +#if KHRONOS_SUPPORT_INT64 +/* Time types + * + * These types can be used to represent a time interval in nanoseconds or + * an absolute Unadjusted System Time. Unadjusted System Time is the number + * of nanoseconds since some arbitrary system event (e.g. since the last + * time the system booted). The Unadjusted System Time is an unsigned + * 64 bit value that wraps back to 0 every 584 years. Time intervals + * may be either signed or unsigned. + */ +typedef khronos_uint64_t khronos_utime_nanoseconds_t; +typedef khronos_int64_t khronos_stime_nanoseconds_t; +#endif + +/* + * Dummy value used to pad enum types to 32 bits. + */ +#ifndef KHRONOS_MAX_ENUM +#define KHRONOS_MAX_ENUM 0x7FFFFFFF +#endif + +/* + * Enumerated boolean type + * + * Values other than zero should be considered to be true. Therefore + * comparisons should not be made against KHRONOS_TRUE. + */ +typedef enum { + KHRONOS_FALSE = 0, + KHRONOS_TRUE = 1, + KHRONOS_BOOLEAN_ENUM_FORCE_SIZE = KHRONOS_MAX_ENUM +} khronos_boolean_enum_t; + +#endif /* __khrplatform_h_ */ diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.h new file mode 100644 index 0000000000..1ade724fe3 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.h @@ -0,0 +1,1216 @@ +/******************************************************************************* + * Copyright (c) 2008 - 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +#ifndef __OPENCL_CL_H +#define __OPENCL_CL_H + +#ifdef __APPLE__ +#include +#else +#include +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +/******************************************************************************/ + +typedef struct _cl_platform_id * cl_platform_id; +typedef struct _cl_device_id * cl_device_id; +typedef struct _cl_context * cl_context; +typedef struct _cl_command_queue * cl_command_queue; +typedef struct _cl_mem * cl_mem; +typedef struct _cl_program * cl_program; +typedef struct _cl_kernel * cl_kernel; +typedef struct _cl_event * cl_event; +typedef struct _cl_sampler * cl_sampler; + +typedef cl_uint cl_bool; /* WARNING! Unlike cl_ types in cl_platform.h, cl_bool is not guaranteed to be the same size as the bool in kernels. */ +typedef cl_ulong cl_bitfield; +typedef cl_bitfield cl_device_type; +typedef cl_uint cl_platform_info; +typedef cl_uint cl_device_info; +typedef cl_bitfield cl_device_fp_config; +typedef cl_uint cl_device_mem_cache_type; +typedef cl_uint cl_device_local_mem_type; +typedef cl_bitfield cl_device_exec_capabilities; +typedef cl_bitfield cl_command_queue_properties; +typedef intptr_t cl_device_partition_property; +typedef cl_bitfield cl_device_affinity_domain; + +typedef intptr_t cl_context_properties; +typedef cl_uint cl_context_info; +typedef cl_uint cl_command_queue_info; +typedef cl_uint cl_channel_order; +typedef cl_uint cl_channel_type; +typedef cl_bitfield cl_mem_flags; +typedef cl_uint cl_mem_object_type; +typedef cl_uint cl_mem_info; +typedef cl_bitfield cl_mem_migration_flags; +typedef cl_uint cl_image_info; +typedef cl_uint cl_buffer_create_type; +typedef cl_uint cl_addressing_mode; +typedef cl_uint cl_filter_mode; +typedef cl_uint cl_sampler_info; +typedef cl_bitfield cl_map_flags; +typedef cl_uint cl_program_info; +typedef cl_uint cl_program_build_info; +typedef cl_uint cl_program_binary_type; +typedef cl_int cl_build_status; +typedef cl_uint cl_kernel_info; +typedef cl_uint cl_kernel_arg_info; +typedef cl_uint cl_kernel_arg_address_qualifier; +typedef cl_uint cl_kernel_arg_access_qualifier; +typedef cl_bitfield cl_kernel_arg_type_qualifier; +typedef cl_uint cl_kernel_work_group_info; +typedef cl_uint cl_event_info; +typedef cl_uint cl_command_type; +typedef cl_uint cl_profiling_info; + + +typedef struct _cl_image_format { + cl_channel_order image_channel_order; + cl_channel_type image_channel_data_type; +} cl_image_format; + +typedef struct _cl_image_desc { + cl_mem_object_type image_type; + size_t image_width; + size_t image_height; + size_t image_depth; + size_t image_array_size; + size_t image_row_pitch; + size_t image_slice_pitch; + cl_uint num_mip_levels; + cl_uint num_samples; + cl_mem buffer; +} cl_image_desc; + +typedef struct _cl_buffer_region { + size_t origin; + size_t size; +} cl_buffer_region; + + +/******************************************************************************/ + +/* Error Codes */ +#define CL_SUCCESS 0 +#define CL_DEVICE_NOT_FOUND -1 +#define CL_DEVICE_NOT_AVAILABLE -2 +#define CL_COMPILER_NOT_AVAILABLE -3 +#define CL_MEM_OBJECT_ALLOCATION_FAILURE -4 +#define CL_OUT_OF_RESOURCES -5 +#define CL_OUT_OF_HOST_MEMORY -6 +#define CL_PROFILING_INFO_NOT_AVAILABLE -7 +#define CL_MEM_COPY_OVERLAP -8 +#define CL_IMAGE_FORMAT_MISMATCH -9 +#define CL_IMAGE_FORMAT_NOT_SUPPORTED -10 +#define CL_BUILD_PROGRAM_FAILURE -11 +#define CL_MAP_FAILURE -12 +#define CL_MISALIGNED_SUB_BUFFER_OFFSET -13 +#define CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST -14 +#define CL_COMPILE_PROGRAM_FAILURE -15 +#define CL_LINKER_NOT_AVAILABLE -16 +#define CL_LINK_PROGRAM_FAILURE -17 +#define CL_DEVICE_PARTITION_FAILED -18 +#define CL_KERNEL_ARG_INFO_NOT_AVAILABLE -19 + +#define CL_INVALID_VALUE -30 +#define CL_INVALID_DEVICE_TYPE -31 +#define CL_INVALID_PLATFORM -32 +#define CL_INVALID_DEVICE -33 +#define CL_INVALID_CONTEXT -34 +#define CL_INVALID_QUEUE_PROPERTIES -35 +#define CL_INVALID_COMMAND_QUEUE -36 +#define CL_INVALID_HOST_PTR -37 +#define CL_INVALID_MEM_OBJECT -38 +#define CL_INVALID_IMAGE_FORMAT_DESCRIPTOR -39 +#define CL_INVALID_IMAGE_SIZE -40 +#define CL_INVALID_SAMPLER -41 +#define CL_INVALID_BINARY -42 +#define CL_INVALID_BUILD_OPTIONS -43 +#define CL_INVALID_PROGRAM -44 +#define CL_INVALID_PROGRAM_EXECUTABLE -45 +#define CL_INVALID_KERNEL_NAME -46 +#define CL_INVALID_KERNEL_DEFINITION -47 +#define CL_INVALID_KERNEL -48 +#define CL_INVALID_ARG_INDEX -49 +#define CL_INVALID_ARG_VALUE -50 +#define CL_INVALID_ARG_SIZE -51 +#define CL_INVALID_KERNEL_ARGS -52 +#define CL_INVALID_WORK_DIMENSION -53 +#define CL_INVALID_WORK_GROUP_SIZE -54 +#define CL_INVALID_WORK_ITEM_SIZE -55 +#define CL_INVALID_GLOBAL_OFFSET -56 +#define CL_INVALID_EVENT_WAIT_LIST -57 +#define CL_INVALID_EVENT -58 +#define CL_INVALID_OPERATION -59 +#define CL_INVALID_GL_OBJECT -60 +#define CL_INVALID_BUFFER_SIZE -61 +#define CL_INVALID_MIP_LEVEL -62 +#define CL_INVALID_GLOBAL_WORK_SIZE -63 +#define CL_INVALID_PROPERTY -64 +#define CL_INVALID_IMAGE_DESCRIPTOR -65 +#define CL_INVALID_COMPILER_OPTIONS -66 +#define CL_INVALID_LINKER_OPTIONS -67 +#define CL_INVALID_DEVICE_PARTITION_COUNT -68 + +/* OpenCL Version */ +#define CL_VERSION_1_0 1 +#define CL_VERSION_1_1 1 +#define CL_VERSION_1_2 1 + +/* cl_bool */ +#define CL_FALSE 0 +#define CL_TRUE 1 +#define CL_BLOCKING CL_TRUE +#define CL_NON_BLOCKING CL_FALSE + +/* cl_platform_info */ +#define CL_PLATFORM_PROFILE 0x0900 +#define CL_PLATFORM_VERSION 0x0901 +#define CL_PLATFORM_NAME 0x0902 +#define CL_PLATFORM_VENDOR 0x0903 +#define CL_PLATFORM_EXTENSIONS 0x0904 + +/* cl_device_type - bitfield */ +#define CL_DEVICE_TYPE_DEFAULT (1 << 0) +#define CL_DEVICE_TYPE_CPU (1 << 1) +#define CL_DEVICE_TYPE_GPU (1 << 2) +#define CL_DEVICE_TYPE_ACCELERATOR (1 << 3) +#define CL_DEVICE_TYPE_CUSTOM (1 << 4) +#define CL_DEVICE_TYPE_ALL 0xFFFFFFFF + +/* cl_device_info */ +#define CL_DEVICE_TYPE 0x1000 +#define CL_DEVICE_VENDOR_ID 0x1001 +#define CL_DEVICE_MAX_COMPUTE_UNITS 0x1002 +#define CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 0x1003 +#define CL_DEVICE_MAX_WORK_GROUP_SIZE 0x1004 +#define CL_DEVICE_MAX_WORK_ITEM_SIZES 0x1005 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 0x1006 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT 0x1007 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT 0x1008 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG 0x1009 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT 0x100A +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE 0x100B +#define CL_DEVICE_MAX_CLOCK_FREQUENCY 0x100C +#define CL_DEVICE_ADDRESS_BITS 0x100D +#define CL_DEVICE_MAX_READ_IMAGE_ARGS 0x100E +#define CL_DEVICE_MAX_WRITE_IMAGE_ARGS 0x100F +#define CL_DEVICE_MAX_MEM_ALLOC_SIZE 0x1010 +#define CL_DEVICE_IMAGE2D_MAX_WIDTH 0x1011 +#define CL_DEVICE_IMAGE2D_MAX_HEIGHT 0x1012 +#define CL_DEVICE_IMAGE3D_MAX_WIDTH 0x1013 +#define CL_DEVICE_IMAGE3D_MAX_HEIGHT 0x1014 +#define CL_DEVICE_IMAGE3D_MAX_DEPTH 0x1015 +#define CL_DEVICE_IMAGE_SUPPORT 0x1016 +#define CL_DEVICE_MAX_PARAMETER_SIZE 0x1017 +#define CL_DEVICE_MAX_SAMPLERS 0x1018 +#define CL_DEVICE_MEM_BASE_ADDR_ALIGN 0x1019 +#define CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE 0x101A +#define CL_DEVICE_SINGLE_FP_CONFIG 0x101B +#define CL_DEVICE_GLOBAL_MEM_CACHE_TYPE 0x101C +#define CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE 0x101D +#define CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 0x101E +#define CL_DEVICE_GLOBAL_MEM_SIZE 0x101F +#define CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 0x1020 +#define CL_DEVICE_MAX_CONSTANT_ARGS 0x1021 +#define CL_DEVICE_LOCAL_MEM_TYPE 0x1022 +#define CL_DEVICE_LOCAL_MEM_SIZE 0x1023 +#define CL_DEVICE_ERROR_CORRECTION_SUPPORT 0x1024 +#define CL_DEVICE_PROFILING_TIMER_RESOLUTION 0x1025 +#define CL_DEVICE_ENDIAN_LITTLE 0x1026 +#define CL_DEVICE_AVAILABLE 0x1027 +#define CL_DEVICE_COMPILER_AVAILABLE 0x1028 +#define CL_DEVICE_EXECUTION_CAPABILITIES 0x1029 +#define CL_DEVICE_QUEUE_PROPERTIES 0x102A +#define CL_DEVICE_NAME 0x102B +#define CL_DEVICE_VENDOR 0x102C +#define CL_DRIVER_VERSION 0x102D +#define CL_DEVICE_PROFILE 0x102E +#define CL_DEVICE_VERSION 0x102F +#define CL_DEVICE_EXTENSIONS 0x1030 +#define CL_DEVICE_PLATFORM 0x1031 +#define CL_DEVICE_DOUBLE_FP_CONFIG 0x1032 +/* 0x1033 reserved for CL_DEVICE_HALF_FP_CONFIG */ +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF 0x1034 +#define CL_DEVICE_HOST_UNIFIED_MEMORY 0x1035 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR 0x1036 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT 0x1037 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 0x1038 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 0x1039 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 0x103A +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 0x103B +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 0x103C +#define CL_DEVICE_OPENCL_C_VERSION 0x103D +#define CL_DEVICE_LINKER_AVAILABLE 0x103E +#define CL_DEVICE_BUILT_IN_KERNELS 0x103F +#define CL_DEVICE_IMAGE_MAX_BUFFER_SIZE 0x1040 +#define CL_DEVICE_IMAGE_MAX_ARRAY_SIZE 0x1041 +#define CL_DEVICE_PARENT_DEVICE 0x1042 +#define CL_DEVICE_PARTITION_MAX_SUB_DEVICES 0x1043 +#define CL_DEVICE_PARTITION_PROPERTIES 0x1044 +#define CL_DEVICE_PARTITION_AFFINITY_DOMAIN 0x1045 +#define CL_DEVICE_PARTITION_TYPE 0x1046 +#define CL_DEVICE_REFERENCE_COUNT 0x1047 +#define CL_DEVICE_PREFERRED_INTEROP_USER_SYNC 0x1048 +#define CL_DEVICE_PRINTF_BUFFER_SIZE 0x1049 +#define CL_DEVICE_IMAGE_PITCH_ALIGNMENT 0x104A +#define CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT 0x104B +/* 0x104C reserved for CL_DEVICE_SVM_CAPABILITIES */ + +/* cl_device_fp_config - bitfield */ +#define CL_FP_DENORM (1 << 0) +#define CL_FP_INF_NAN (1 << 1) +#define CL_FP_ROUND_TO_NEAREST (1 << 2) +#define CL_FP_ROUND_TO_ZERO (1 << 3) +#define CL_FP_ROUND_TO_INF (1 << 4) +#define CL_FP_FMA (1 << 5) +#define CL_FP_SOFT_FLOAT (1 << 6) +#define CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT (1 << 7) + +/* cl_device_mem_cache_type */ +#define CL_NONE 0x0 +#define CL_READ_ONLY_CACHE 0x1 +#define CL_READ_WRITE_CACHE 0x2 + +/* cl_device_local_mem_type */ +#define CL_LOCAL 0x1 +#define CL_GLOBAL 0x2 + +/* cl_device_exec_capabilities - bitfield */ +#define CL_EXEC_KERNEL (1 << 0) +#define CL_EXEC_NATIVE_KERNEL (1 << 1) + +/* cl_command_queue_properties - bitfield */ +#define CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE (1 << 0) +#define CL_QUEUE_PROFILING_ENABLE (1 << 1) + +/* cl_context_info */ +#define CL_CONTEXT_REFERENCE_COUNT 0x1080 +#define CL_CONTEXT_DEVICES 0x1081 +#define CL_CONTEXT_PROPERTIES 0x1082 +#define CL_CONTEXT_NUM_DEVICES 0x1083 + +/* cl_context_properties */ +#define CL_CONTEXT_PLATFORM 0x1084 +#define CL_CONTEXT_INTEROP_USER_SYNC 0x1085 + +/* cl_device_partition_property */ +#define CL_DEVICE_PARTITION_EQUALLY 0x1086 +#define CL_DEVICE_PARTITION_BY_COUNTS 0x1087 +#define CL_DEVICE_PARTITION_BY_COUNTS_LIST_END 0x0 +#define CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN 0x1088 + +/* cl_device_affinity_domain */ +#define CL_DEVICE_AFFINITY_DOMAIN_NUMA (1 << 0) +#define CL_DEVICE_AFFINITY_DOMAIN_L4_CACHE (1 << 1) +#define CL_DEVICE_AFFINITY_DOMAIN_L3_CACHE (1 << 2) +#define CL_DEVICE_AFFINITY_DOMAIN_L2_CACHE (1 << 3) +#define CL_DEVICE_AFFINITY_DOMAIN_L1_CACHE (1 << 4) +#define CL_DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE (1 << 5) + +/* cl_command_queue_info */ +#define CL_QUEUE_CONTEXT 0x1090 +#define CL_QUEUE_DEVICE 0x1091 +#define CL_QUEUE_REFERENCE_COUNT 0x1092 +#define CL_QUEUE_PROPERTIES 0x1093 + +/* cl_mem_flags - bitfield */ +#define CL_MEM_READ_WRITE (1 << 0) +#define CL_MEM_WRITE_ONLY (1 << 1) +#define CL_MEM_READ_ONLY (1 << 2) +#define CL_MEM_USE_HOST_PTR (1 << 3) +#define CL_MEM_ALLOC_HOST_PTR (1 << 4) +#define CL_MEM_COPY_HOST_PTR (1 << 5) +// reserved (1 << 6) +#define CL_MEM_HOST_WRITE_ONLY (1 << 7) +#define CL_MEM_HOST_READ_ONLY (1 << 8) +#define CL_MEM_HOST_NO_ACCESS (1 << 9) + +/* cl_mem_migration_flags - bitfield */ +#define CL_MIGRATE_MEM_OBJECT_HOST (1 << 0) +#define CL_MIGRATE_MEM_OBJECT_CONTENT_UNDEFINED (1 << 1) + +/* cl_channel_order */ +#define CL_R 0x10B0 +#define CL_A 0x10B1 +#define CL_RG 0x10B2 +#define CL_RA 0x10B3 +#define CL_RGB 0x10B4 +#define CL_RGBA 0x10B5 +#define CL_BGRA 0x10B6 +#define CL_ARGB 0x10B7 +#define CL_INTENSITY 0x10B8 +#define CL_LUMINANCE 0x10B9 +#define CL_Rx 0x10BA +#define CL_RGx 0x10BB +#define CL_RGBx 0x10BC +#define CL_DEPTH 0x10BD +#define CL_DEPTH_STENCIL 0x10BE + +/* cl_channel_type */ +#define CL_SNORM_INT8 0x10D0 +#define CL_SNORM_INT16 0x10D1 +#define CL_UNORM_INT8 0x10D2 +#define CL_UNORM_INT16 0x10D3 +#define CL_UNORM_SHORT_565 0x10D4 +#define CL_UNORM_SHORT_555 0x10D5 +#define CL_UNORM_INT_101010 0x10D6 +#define CL_SIGNED_INT8 0x10D7 +#define CL_SIGNED_INT16 0x10D8 +#define CL_SIGNED_INT32 0x10D9 +#define CL_UNSIGNED_INT8 0x10DA +#define CL_UNSIGNED_INT16 0x10DB +#define CL_UNSIGNED_INT32 0x10DC +#define CL_HALF_FLOAT 0x10DD +#define CL_FLOAT 0x10DE +#define CL_UNORM_INT24 0x10DF + +/* cl_mem_object_type */ +#define CL_MEM_OBJECT_BUFFER 0x10F0 +#define CL_MEM_OBJECT_IMAGE2D 0x10F1 +#define CL_MEM_OBJECT_IMAGE3D 0x10F2 +#define CL_MEM_OBJECT_IMAGE2D_ARRAY 0x10F3 +#define CL_MEM_OBJECT_IMAGE1D 0x10F4 +#define CL_MEM_OBJECT_IMAGE1D_ARRAY 0x10F5 +#define CL_MEM_OBJECT_IMAGE1D_BUFFER 0x10F6 + +/* cl_mem_info */ +#define CL_MEM_TYPE 0x1100 +#define CL_MEM_FLAGS 0x1101 +#define CL_MEM_SIZE 0x1102 +#define CL_MEM_HOST_PTR 0x1103 +#define CL_MEM_MAP_COUNT 0x1104 +#define CL_MEM_REFERENCE_COUNT 0x1105 +#define CL_MEM_CONTEXT 0x1106 +#define CL_MEM_ASSOCIATED_MEMOBJECT 0x1107 +#define CL_MEM_OFFSET 0x1108 + +/* cl_image_info */ +#define CL_IMAGE_FORMAT 0x1110 +#define CL_IMAGE_ELEMENT_SIZE 0x1111 +#define CL_IMAGE_ROW_PITCH 0x1112 +#define CL_IMAGE_SLICE_PITCH 0x1113 +#define CL_IMAGE_WIDTH 0x1114 +#define CL_IMAGE_HEIGHT 0x1115 +#define CL_IMAGE_DEPTH 0x1116 +#define CL_IMAGE_ARRAY_SIZE 0x1117 +#define CL_IMAGE_BUFFER 0x1118 +#define CL_IMAGE_NUM_MIP_LEVELS 0x1119 +#define CL_IMAGE_NUM_SAMPLES 0x111A + +/* cl_addressing_mode */ +#define CL_ADDRESS_NONE 0x1130 +#define CL_ADDRESS_CLAMP_TO_EDGE 0x1131 +#define CL_ADDRESS_CLAMP 0x1132 +#define CL_ADDRESS_REPEAT 0x1133 +#define CL_ADDRESS_MIRRORED_REPEAT 0x1134 + +/* cl_filter_mode */ +#define CL_FILTER_NEAREST 0x1140 +#define CL_FILTER_LINEAR 0x1141 + +/* cl_sampler_info */ +#define CL_SAMPLER_REFERENCE_COUNT 0x1150 +#define CL_SAMPLER_CONTEXT 0x1151 +#define CL_SAMPLER_NORMALIZED_COORDS 0x1152 +#define CL_SAMPLER_ADDRESSING_MODE 0x1153 +#define CL_SAMPLER_FILTER_MODE 0x1154 + +/* cl_map_flags - bitfield */ +#define CL_MAP_READ (1 << 0) +#define CL_MAP_WRITE (1 << 1) +#define CL_MAP_WRITE_INVALIDATE_REGION (1 << 2) + +/* cl_program_info */ +#define CL_PROGRAM_REFERENCE_COUNT 0x1160 +#define CL_PROGRAM_CONTEXT 0x1161 +#define CL_PROGRAM_NUM_DEVICES 0x1162 +#define CL_PROGRAM_DEVICES 0x1163 +#define CL_PROGRAM_SOURCE 0x1164 +#define CL_PROGRAM_BINARY_SIZES 0x1165 +#define CL_PROGRAM_BINARIES 0x1166 +#define CL_PROGRAM_NUM_KERNELS 0x1167 +#define CL_PROGRAM_KERNEL_NAMES 0x1168 + +/* cl_program_build_info */ +#define CL_PROGRAM_BUILD_STATUS 0x1181 +#define CL_PROGRAM_BUILD_OPTIONS 0x1182 +#define CL_PROGRAM_BUILD_LOG 0x1183 +#define CL_PROGRAM_BINARY_TYPE 0x1184 + +/* cl_program_binary_type */ +#define CL_PROGRAM_BINARY_TYPE_NONE 0x0 +#define CL_PROGRAM_BINARY_TYPE_COMPILED_OBJECT 0x1 +#define CL_PROGRAM_BINARY_TYPE_LIBRARY 0x2 +#define CL_PROGRAM_BINARY_TYPE_EXECUTABLE 0x4 + +/* cl_build_status */ +#define CL_BUILD_SUCCESS 0 +#define CL_BUILD_NONE -1 +#define CL_BUILD_ERROR -2 +#define CL_BUILD_IN_PROGRESS -3 + +/* cl_kernel_info */ +#define CL_KERNEL_FUNCTION_NAME 0x1190 +#define CL_KERNEL_NUM_ARGS 0x1191 +#define CL_KERNEL_REFERENCE_COUNT 0x1192 +#define CL_KERNEL_CONTEXT 0x1193 +#define CL_KERNEL_PROGRAM 0x1194 +#define CL_KERNEL_ATTRIBUTES 0x1195 + +/* cl_kernel_arg_info */ +#define CL_KERNEL_ARG_ADDRESS_QUALIFIER 0x1196 +#define CL_KERNEL_ARG_ACCESS_QUALIFIER 0x1197 +#define CL_KERNEL_ARG_TYPE_NAME 0x1198 +#define CL_KERNEL_ARG_TYPE_QUALIFIER 0x1199 +#define CL_KERNEL_ARG_NAME 0x119A + +/* cl_kernel_arg_address_qualifier */ +#define CL_KERNEL_ARG_ADDRESS_GLOBAL 0x119B +#define CL_KERNEL_ARG_ADDRESS_LOCAL 0x119C +#define CL_KERNEL_ARG_ADDRESS_CONSTANT 0x119D +#define CL_KERNEL_ARG_ADDRESS_PRIVATE 0x119E + +/* cl_kernel_arg_access_qualifier */ +#define CL_KERNEL_ARG_ACCESS_READ_ONLY 0x11A0 +#define CL_KERNEL_ARG_ACCESS_WRITE_ONLY 0x11A1 +#define CL_KERNEL_ARG_ACCESS_READ_WRITE 0x11A2 +#define CL_KERNEL_ARG_ACCESS_NONE 0x11A3 + +/* cl_kernel_arg_type_qualifer */ +#define CL_KERNEL_ARG_TYPE_NONE 0 +#define CL_KERNEL_ARG_TYPE_CONST (1 << 0) +#define CL_KERNEL_ARG_TYPE_RESTRICT (1 << 1) +#define CL_KERNEL_ARG_TYPE_VOLATILE (1 << 2) + +/* cl_kernel_work_group_info */ +#define CL_KERNEL_WORK_GROUP_SIZE 0x11B0 +#define CL_KERNEL_COMPILE_WORK_GROUP_SIZE 0x11B1 +#define CL_KERNEL_LOCAL_MEM_SIZE 0x11B2 +#define CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 0x11B3 +#define CL_KERNEL_PRIVATE_MEM_SIZE 0x11B4 +#define CL_KERNEL_GLOBAL_WORK_SIZE 0x11B5 + +/* cl_event_info */ +#define CL_EVENT_COMMAND_QUEUE 0x11D0 +#define CL_EVENT_COMMAND_TYPE 0x11D1 +#define CL_EVENT_REFERENCE_COUNT 0x11D2 +#define CL_EVENT_COMMAND_EXECUTION_STATUS 0x11D3 +#define CL_EVENT_CONTEXT 0x11D4 + +/* cl_command_type */ +#define CL_COMMAND_NDRANGE_KERNEL 0x11F0 +#define CL_COMMAND_TASK 0x11F1 +#define CL_COMMAND_NATIVE_KERNEL 0x11F2 +#define CL_COMMAND_READ_BUFFER 0x11F3 +#define CL_COMMAND_WRITE_BUFFER 0x11F4 +#define CL_COMMAND_COPY_BUFFER 0x11F5 +#define CL_COMMAND_READ_IMAGE 0x11F6 +#define CL_COMMAND_WRITE_IMAGE 0x11F7 +#define CL_COMMAND_COPY_IMAGE 0x11F8 +#define CL_COMMAND_COPY_IMAGE_TO_BUFFER 0x11F9 +#define CL_COMMAND_COPY_BUFFER_TO_IMAGE 0x11FA +#define CL_COMMAND_MAP_BUFFER 0x11FB +#define CL_COMMAND_MAP_IMAGE 0x11FC +#define CL_COMMAND_UNMAP_MEM_OBJECT 0x11FD +#define CL_COMMAND_MARKER 0x11FE +#define CL_COMMAND_ACQUIRE_GL_OBJECTS 0x11FF +#define CL_COMMAND_RELEASE_GL_OBJECTS 0x1200 +#define CL_COMMAND_READ_BUFFER_RECT 0x1201 +#define CL_COMMAND_WRITE_BUFFER_RECT 0x1202 +#define CL_COMMAND_COPY_BUFFER_RECT 0x1203 +#define CL_COMMAND_USER 0x1204 +#define CL_COMMAND_BARRIER 0x1205 +#define CL_COMMAND_MIGRATE_MEM_OBJECTS 0x1206 +#define CL_COMMAND_FILL_BUFFER 0x1207 +#define CL_COMMAND_FILL_IMAGE 0x1208 +/* 0x1209-0x120F reserved for SVM commands */ + +/* command execution status */ +#define CL_COMPLETE 0x0 +#define CL_RUNNING 0x1 +#define CL_SUBMITTED 0x2 +#define CL_QUEUED 0x3 + +/* cl_buffer_create_type */ +#define CL_BUFFER_CREATE_TYPE_REGION 0x1220 + +/* cl_profiling_info */ +#define CL_PROFILING_COMMAND_QUEUED 0x1280 +#define CL_PROFILING_COMMAND_SUBMIT 0x1281 +#define CL_PROFILING_COMMAND_START 0x1282 +#define CL_PROFILING_COMMAND_END 0x1283 + +/********************************************************************************************************/ + +/* Platform API */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetPlatformIDs(cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetPlatformInfo(cl_platform_id /* platform */, + cl_platform_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Device APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDs(cl_platform_id /* platform */, + cl_device_type /* device_type */, + cl_uint /* num_entries */, + cl_device_id * /* devices */, + cl_uint * /* num_devices */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceInfo(cl_device_id /* device */, + cl_device_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clCreateSubDevices(cl_device_id /* in_device */, + const cl_device_partition_property * /* properties */, + cl_uint /* num_devices */, + cl_device_id * /* out_devices */, + cl_uint * /* num_devices_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; + +/* Context APIs */ +extern CL_API_ENTRY cl_context CL_API_CALL +clCreateContext(const cl_context_properties * /* properties */, + cl_uint /* num_devices */, + const cl_device_id * /* devices */, + void (CL_CALLBACK * /* pfn_notify */)(const char *, const void *, size_t, void *), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_context CL_API_CALL +clCreateContextFromType(const cl_context_properties * /* properties */, + cl_device_type /* device_type */, + void (CL_CALLBACK * /* pfn_notify*/ )(const char *, const void *, size_t, void *), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainContext(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseContext(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetContextInfo(cl_context /* context */, + cl_context_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Command Queue APIs */ +extern CL_API_ENTRY cl_command_queue CL_API_CALL +clCreateCommandQueue(cl_context /* context */, + cl_device_id /* device */, + cl_command_queue_properties /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetCommandQueueInfo(cl_command_queue /* command_queue */, + cl_command_queue_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Memory Object APIs */ +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateBuffer(cl_context /* context */, + cl_mem_flags /* flags */, + size_t /* size */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateSubBuffer(cl_mem /* buffer */, + cl_mem_flags /* flags */, + cl_buffer_create_type /* buffer_create_type */, + const void * /* buffer_create_info */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateImage(cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + const cl_image_desc * /* image_desc */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetSupportedImageFormats(cl_context /* context */, + cl_mem_flags /* flags */, + cl_mem_object_type /* image_type */, + cl_uint /* num_entries */, + cl_image_format * /* image_formats */, + cl_uint * /* num_image_formats */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetMemObjectInfo(cl_mem /* memobj */, + cl_mem_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetImageInfo(cl_mem /* image */, + cl_image_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetMemObjectDestructorCallback( cl_mem /* memobj */, + void (CL_CALLBACK * /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), + void * /*user_data */ ) CL_API_SUFFIX__VERSION_1_1; + +/* Sampler APIs */ +extern CL_API_ENTRY cl_sampler CL_API_CALL +clCreateSampler(cl_context /* context */, + cl_bool /* normalized_coords */, + cl_addressing_mode /* addressing_mode */, + cl_filter_mode /* filter_mode */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetSamplerInfo(cl_sampler /* sampler */, + cl_sampler_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Program Object APIs */ +extern CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithSource(cl_context /* context */, + cl_uint /* count */, + const char ** /* strings */, + const size_t * /* lengths */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithBinary(cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const size_t * /* lengths */, + const unsigned char ** /* binaries */, + cl_int * /* binary_status */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithBuiltInKernels(cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* kernel_names */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clBuildProgram(cl_program /* program */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clCompileProgram(cl_program /* program */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + cl_uint /* num_input_headers */, + const cl_program * /* input_headers */, + const char ** /* header_include_names */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_program CL_API_CALL +clLinkProgram(cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + cl_uint /* num_input_programs */, + const cl_program * /* input_programs */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */, + cl_int * /* errcode_ret */ ) CL_API_SUFFIX__VERSION_1_2; + + +extern CL_API_ENTRY cl_int CL_API_CALL +clUnloadPlatformCompiler(cl_platform_id /* platform */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetProgramInfo(cl_program /* program */, + cl_program_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetProgramBuildInfo(cl_program /* program */, + cl_device_id /* device */, + cl_program_build_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Kernel Object APIs */ +extern CL_API_ENTRY cl_kernel CL_API_CALL +clCreateKernel(cl_program /* program */, + const char * /* kernel_name */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clCreateKernelsInProgram(cl_program /* program */, + cl_uint /* num_kernels */, + cl_kernel * /* kernels */, + cl_uint * /* num_kernels_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetKernelArg(cl_kernel /* kernel */, + cl_uint /* arg_index */, + size_t /* arg_size */, + const void * /* arg_value */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelInfo(cl_kernel /* kernel */, + cl_kernel_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelArgInfo(cl_kernel /* kernel */, + cl_uint /* arg_indx */, + cl_kernel_arg_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelWorkGroupInfo(cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_work_group_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Event Object APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clWaitForEvents(cl_uint /* num_events */, + const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetEventInfo(cl_event /* event */, + cl_event_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_event CL_API_CALL +clCreateUserEvent(cl_context /* context */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetUserEventStatus(cl_event /* event */, + cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetEventCallback( cl_event /* event */, + cl_int /* command_exec_callback_type */, + void (CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_1; + +/* Profiling APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetEventProfilingInfo(cl_event /* event */, + cl_profiling_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Flush and Finish APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clFlush(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clFinish(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +/* Enqueued Commands APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + size_t /* offset */, + size_t /* size */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadBufferRect(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + const size_t * /* buffer_offset */, + const size_t * /* host_offset */, + const size_t * /* region */, + size_t /* buffer_row_pitch */, + size_t /* buffer_slice_pitch */, + size_t /* host_row_pitch */, + size_t /* host_slice_pitch */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_write */, + size_t /* offset */, + size_t /* size */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteBufferRect(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_write */, + const size_t * /* buffer_offset */, + const size_t * /* host_offset */, + const size_t * /* region */, + size_t /* buffer_row_pitch */, + size_t /* buffer_slice_pitch */, + size_t /* host_row_pitch */, + size_t /* host_slice_pitch */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueFillBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + const void * /* pattern */, + size_t /* pattern_size */, + size_t /* offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBuffer(cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_buffer */, + size_t /* src_offset */, + size_t /* dst_offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBufferRect(cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_buffer */, + const size_t * /* src_origin */, + const size_t * /* dst_origin */, + const size_t * /* region */, + size_t /* src_row_pitch */, + size_t /* src_slice_pitch */, + size_t /* dst_row_pitch */, + size_t /* dst_slice_pitch */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_read */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t /* row_pitch */, + size_t /* slice_pitch */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_write */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t /* input_row_pitch */, + size_t /* input_slice_pitch */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueFillImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + const void * /* fill_color */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyImage(cl_command_queue /* command_queue */, + cl_mem /* src_image */, + cl_mem /* dst_image */, + const size_t * /* src_origin[3] */, + const size_t * /* dst_origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyImageToBuffer(cl_command_queue /* command_queue */, + cl_mem /* src_image */, + cl_mem /* dst_buffer */, + const size_t * /* src_origin[3] */, + const size_t * /* region[3] */, + size_t /* dst_offset */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBufferToImage(cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_image */, + size_t /* src_offset */, + const size_t * /* dst_origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY void * CL_API_CALL +clEnqueueMapBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + size_t /* offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY void * CL_API_CALL +clEnqueueMapImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t * /* image_row_pitch */, + size_t * /* image_slice_pitch */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueUnmapMemObject(cl_command_queue /* command_queue */, + cl_mem /* memobj */, + void * /* mapped_ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMigrateMemObjects(cl_command_queue /* command_queue */, + cl_uint /* num_mem_objects */, + const cl_mem * /* mem_objects */, + cl_mem_migration_flags /* flags */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueNDRangeKernel(cl_command_queue /* command_queue */, + cl_kernel /* kernel */, + cl_uint /* work_dim */, + const size_t * /* global_work_offset */, + const size_t * /* global_work_size */, + const size_t * /* local_work_size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueTask(cl_command_queue /* command_queue */, + cl_kernel /* kernel */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueNativeKernel(cl_command_queue /* command_queue */, + void (CL_CALLBACK * /*user_func*/)(void *), + void * /* args */, + size_t /* cb_args */, + cl_uint /* num_mem_objects */, + const cl_mem * /* mem_list */, + const void ** /* args_mem_loc */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMarkerWithWaitList(cl_command_queue /* command_queue */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueBarrierWithWaitList(cl_command_queue /* command_queue */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + + +/* Extension function access + * + * Returns the extension function address for the given function name, + * or NULL if a valid function can not be found. The client must + * check to make sure the address is not NULL, before using or + * calling the returned function address. + */ +extern CL_API_ENTRY void * CL_API_CALL +clGetExtensionFunctionAddressForPlatform(cl_platform_id /* platform */, + const char * /* func_name */) CL_API_SUFFIX__VERSION_1_2; + + +// Deprecated OpenCL 1.1 APIs +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateImage2D(cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + size_t /* image_width */, + size_t /* image_height */, + size_t /* image_row_pitch */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateImage3D(cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + size_t /* image_width */, + size_t /* image_height */, + size_t /* image_depth */, + size_t /* image_row_pitch */, + size_t /* image_slice_pitch */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clEnqueueMarker(cl_command_queue /* command_queue */, + cl_event * /* event */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clEnqueueWaitForEvents(cl_command_queue /* command_queue */, + cl_uint /* num_events */, + const cl_event * /* event_list */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clEnqueueBarrier(cl_command_queue /* command_queue */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clUnloadCompiler(void) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED void * CL_API_CALL +clGetExtensionFunctionAddress(const char * /* func_name */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.hpp b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.hpp new file mode 100644 index 0000000000..667526e341 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl.hpp @@ -0,0 +1,12305 @@ +/******************************************************************************* + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/*! \file + * + * \brief C++ bindings for OpenCL 1.0 (rev 48), OpenCL 1.1 (rev 33) and + * OpenCL 1.2 (rev 15) + * \author Benedict R. Gaster, Laurent Morichetti and Lee Howes + * + * Additions and fixes from: + * Brian Cole, March 3rd 2010 and April 2012 + * Matt Gruenke, April 2012. + * Bruce Merry, February 2013. + * + * \version 1.2.5 + * \date June 2013 + * + * Optional extension support + * + * cl + * cl_ext_device_fission + * #define USE_CL_DEVICE_FISSION + */ + +/*! \mainpage + * \section intro Introduction + * For many large applications C++ is the language of choice and so it seems + * reasonable to define C++ bindings for OpenCL. + * + * + * The interface is contained with a single C++ header file \em cl.hpp and all + * definitions are contained within the namespace \em cl. There is no additional + * requirement to include \em cl.h and to use either the C++ or original C + * bindings it is enough to simply include \em cl.hpp. + * + * The bindings themselves are lightweight and correspond closely to the + * underlying C API. Using the C++ bindings introduces no additional execution + * overhead. + * + * For detail documentation on the bindings see: + * + * The OpenCL C++ Wrapper API 1.2 (revision 09) + * http://www.khronos.org/registry/cl/specs/opencl-cplusplus-1.2.pdf + * + * \section example Example + * + * The following example shows a general use case for the C++ + * bindings, including support for the optional exception feature and + * also the supplied vector and string classes, see following sections for + * decriptions of these features. + * + * \code + * #define __CL_ENABLE_EXCEPTIONS + * + * #if defined(__APPLE__) || defined(__MACOSX) + * #include + * #else + * #include + * #endif + * #include + * #include + * #include + * + * const char * helloStr = "__kernel void " + * "hello(void) " + * "{ " + * " " + * "} "; + * + * int + * main(void) + * { + * cl_int err = CL_SUCCESS; + * try { + * + * std::vector platforms; + * cl::Platform::get(&platforms); + * if (platforms.size() == 0) { + * std::cout << "Platform size 0\n"; + * return -1; + * } + * + * cl_context_properties properties[] = + * { CL_CONTEXT_PLATFORM, (cl_context_properties)(platforms[0])(), 0}; + * cl::Context context(CL_DEVICE_TYPE_CPU, properties); + * + * std::vector devices = context.getInfo(); + * + * cl::Program::Sources source(1, + * std::make_pair(helloStr,strlen(helloStr))); + * cl::Program program_ = cl::Program(context, source); + * program_.build(devices); + * + * cl::Kernel kernel(program_, "hello", &err); + * + * cl::Event event; + * cl::CommandQueue queue(context, devices[0], 0, &err); + * queue.enqueueNDRangeKernel( + * kernel, + * cl::NullRange, + * cl::NDRange(4,4), + * cl::NullRange, + * NULL, + * &event); + * + * event.wait(); + * } + * catch (cl::Error err) { + * std::cerr + * << "ERROR: " + * << err.what() + * << "(" + * << err.err() + * << ")" + * << std::endl; + * } + * + * return EXIT_SUCCESS; + * } + * + * \endcode + * + */ +#ifndef CL_HPP_ +#define CL_HPP_ + +#ifdef _WIN32 + +#include +#include +#include +#include + +#if defined(__CL_ENABLE_EXCEPTIONS) +#include +#endif // #if defined(__CL_ENABLE_EXCEPTIONS) + +#pragma push_macro("max") +#undef max +#if defined(USE_DX_INTEROP) +#include +#include +#endif +#endif // _WIN32 + +// +#if defined(USE_CL_DEVICE_FISSION) +#include +#endif + +#if defined(__APPLE__) || defined(__MACOSX) +#include +#include +#include +#else +#include +#include +#endif // !__APPLE__ + +// To avoid accidentally taking ownership of core OpenCL types +// such as cl_kernel constructors are made explicit +// under OpenCL 1.2 +#if defined(CL_VERSION_1_2) && !defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +#define __CL_EXPLICIT_CONSTRUCTORS explicit +#else // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +#define __CL_EXPLICIT_CONSTRUCTORS +#endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + +// Define deprecated prefixes and suffixes to ensure compilation +// in case they are not pre-defined +#if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) +#define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED +#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) +#if !defined(CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED) +#define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) + +#if !defined(CL_CALLBACK) +#define CL_CALLBACK +#endif //CL_CALLBACK + +#include +#include + +#if !defined(__NO_STD_VECTOR) +#include +#endif + +#if !defined(__NO_STD_STRING) +#include +#endif + +#if defined(__linux__) || defined(__APPLE__) || defined(__MACOSX) +#include + +#include +#include +#endif // __linux__ + +#include + + +/*! \namespace cl + * + * \brief The OpenCL C++ bindings are defined within this namespace. + * + */ +namespace cl { + +class Memory; + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +#define __INIT_CL_EXT_FCN_PTR(name) \ + if(!pfn_##name) { \ + pfn_##name = (PFN_##name) \ + clGetExtensionFunctionAddress(#name); \ + if(!pfn_##name) { \ + } \ + } +#endif // #if defined(CL_VERSION_1_1) + +#if defined(CL_VERSION_1_2) +#define __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, name) \ + if(!pfn_##name) { \ + pfn_##name = (PFN_##name) \ + clGetExtensionFunctionAddressForPlatform(platform, #name); \ + if(!pfn_##name) { \ + } \ + } +#endif // #if defined(CL_VERSION_1_1) + +class Program; +class Device; +class Context; +class CommandQueue; +class Memory; +class Buffer; + +#if defined(__CL_ENABLE_EXCEPTIONS) +/*! \brief Exception class + * + * This may be thrown by API functions when __CL_ENABLE_EXCEPTIONS is defined. + */ +class Error : public std::exception +{ +private: + cl_int err_; + const char * errStr_; +public: + /*! \brief Create a new CL error exception for a given error code + * and corresponding message. + * + * \param err error code value. + * + * \param errStr a descriptive string that must remain in scope until + * handling of the exception has concluded. If set, it + * will be returned by what(). + */ + Error(cl_int err, const char * errStr = NULL) : err_(err), errStr_(errStr) + {} + + ~Error() throw() {} + + /*! \brief Get error string associated with exception + * + * \return A memory pointer to the error message string. + */ + virtual const char * what() const throw () + { + if (errStr_ == NULL) { + return "empty"; + } + else { + return errStr_; + } + } + + /*! \brief Get error code associated with exception + * + * \return The error code. + */ + cl_int err(void) const { return err_; } +}; + +#define __ERR_STR(x) #x +#else +#define __ERR_STR(x) NULL +#endif // __CL_ENABLE_EXCEPTIONS + + +namespace detail +{ +#if defined(__CL_ENABLE_EXCEPTIONS) +static inline cl_int errHandler ( + cl_int err, + const char * errStr = NULL) +{ + if (err != CL_SUCCESS) { + throw Error(err, errStr); + } + return err; +} +#else +static inline cl_int errHandler (cl_int err, const char * errStr = NULL) +{ + (void) errStr; // suppress unused variable warning + return err; +} +#endif // __CL_ENABLE_EXCEPTIONS +} + + + +//! \cond DOXYGEN_DETAIL +#if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) +#define __GET_DEVICE_INFO_ERR __ERR_STR(clGetDeviceInfo) +#define __GET_PLATFORM_INFO_ERR __ERR_STR(clGetPlatformInfo) +#define __GET_DEVICE_IDS_ERR __ERR_STR(clGetDeviceIDs) +#define __GET_PLATFORM_IDS_ERR __ERR_STR(clGetPlatformIDs) +#define __GET_CONTEXT_INFO_ERR __ERR_STR(clGetContextInfo) +#define __GET_EVENT_INFO_ERR __ERR_STR(clGetEventInfo) +#define __GET_EVENT_PROFILE_INFO_ERR __ERR_STR(clGetEventProfileInfo) +#define __GET_MEM_OBJECT_INFO_ERR __ERR_STR(clGetMemObjectInfo) +#define __GET_IMAGE_INFO_ERR __ERR_STR(clGetImageInfo) +#define __GET_SAMPLER_INFO_ERR __ERR_STR(clGetSamplerInfo) +#define __GET_KERNEL_INFO_ERR __ERR_STR(clGetKernelInfo) +#if defined(CL_VERSION_1_2) +#define __GET_KERNEL_ARG_INFO_ERR __ERR_STR(clGetKernelArgInfo) +#endif // #if defined(CL_VERSION_1_2) +#define __GET_KERNEL_WORK_GROUP_INFO_ERR __ERR_STR(clGetKernelWorkGroupInfo) +#define __GET_PROGRAM_INFO_ERR __ERR_STR(clGetProgramInfo) +#define __GET_PROGRAM_BUILD_INFO_ERR __ERR_STR(clGetProgramBuildInfo) +#define __GET_COMMAND_QUEUE_INFO_ERR __ERR_STR(clGetCommandQueueInfo) + +#define __CREATE_CONTEXT_ERR __ERR_STR(clCreateContext) +#define __CREATE_CONTEXT_FROM_TYPE_ERR __ERR_STR(clCreateContextFromType) +#define __GET_SUPPORTED_IMAGE_FORMATS_ERR __ERR_STR(clGetSupportedImageFormats) + +#define __CREATE_BUFFER_ERR __ERR_STR(clCreateBuffer) +#define __COPY_ERR __ERR_STR(cl::copy) +#define __CREATE_SUBBUFFER_ERR __ERR_STR(clCreateSubBuffer) +#define __CREATE_GL_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) +#define __CREATE_GL_RENDER_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) +#define __GET_GL_OBJECT_INFO_ERR __ERR_STR(clGetGLObjectInfo) +#if defined(CL_VERSION_1_2) +#define __CREATE_IMAGE_ERR __ERR_STR(clCreateImage) +#define __CREATE_GL_TEXTURE_ERR __ERR_STR(clCreateFromGLTexture) +#define __IMAGE_DIMENSION_ERR __ERR_STR(Incorrect image dimensions) +#endif // #if defined(CL_VERSION_1_2) +#define __CREATE_SAMPLER_ERR __ERR_STR(clCreateSampler) +#define __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR __ERR_STR(clSetMemObjectDestructorCallback) + +#define __CREATE_USER_EVENT_ERR __ERR_STR(clCreateUserEvent) +#define __SET_USER_EVENT_STATUS_ERR __ERR_STR(clSetUserEventStatus) +#define __SET_EVENT_CALLBACK_ERR __ERR_STR(clSetEventCallback) +#define __WAIT_FOR_EVENTS_ERR __ERR_STR(clWaitForEvents) + +#define __CREATE_KERNEL_ERR __ERR_STR(clCreateKernel) +#define __SET_KERNEL_ARGS_ERR __ERR_STR(clSetKernelArg) +#define __CREATE_PROGRAM_WITH_SOURCE_ERR __ERR_STR(clCreateProgramWithSource) +#define __CREATE_PROGRAM_WITH_BINARY_ERR __ERR_STR(clCreateProgramWithBinary) +#if defined(CL_VERSION_1_2) +#define __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR __ERR_STR(clCreateProgramWithBuiltInKernels) +#endif // #if defined(CL_VERSION_1_2) +#define __BUILD_PROGRAM_ERR __ERR_STR(clBuildProgram) +#if defined(CL_VERSION_1_2) +#define __COMPILE_PROGRAM_ERR __ERR_STR(clCompileProgram) + +#endif // #if defined(CL_VERSION_1_2) +#define __CREATE_KERNELS_IN_PROGRAM_ERR __ERR_STR(clCreateKernelsInProgram) + +#define __CREATE_COMMAND_QUEUE_ERR __ERR_STR(clCreateCommandQueue) +#define __SET_COMMAND_QUEUE_PROPERTY_ERR __ERR_STR(clSetCommandQueueProperty) +#define __ENQUEUE_READ_BUFFER_ERR __ERR_STR(clEnqueueReadBuffer) +#define __ENQUEUE_READ_BUFFER_RECT_ERR __ERR_STR(clEnqueueReadBufferRect) +#define __ENQUEUE_WRITE_BUFFER_ERR __ERR_STR(clEnqueueWriteBuffer) +#define __ENQUEUE_WRITE_BUFFER_RECT_ERR __ERR_STR(clEnqueueWriteBufferRect) +#define __ENQEUE_COPY_BUFFER_ERR __ERR_STR(clEnqueueCopyBuffer) +#define __ENQEUE_COPY_BUFFER_RECT_ERR __ERR_STR(clEnqueueCopyBufferRect) +#define __ENQUEUE_FILL_BUFFER_ERR __ERR_STR(clEnqueueFillBuffer) +#define __ENQUEUE_READ_IMAGE_ERR __ERR_STR(clEnqueueReadImage) +#define __ENQUEUE_WRITE_IMAGE_ERR __ERR_STR(clEnqueueWriteImage) +#define __ENQUEUE_COPY_IMAGE_ERR __ERR_STR(clEnqueueCopyImage) +#define __ENQUEUE_FILL_IMAGE_ERR __ERR_STR(clEnqueueFillImage) +#define __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR __ERR_STR(clEnqueueCopyImageToBuffer) +#define __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR __ERR_STR(clEnqueueCopyBufferToImage) +#define __ENQUEUE_MAP_BUFFER_ERR __ERR_STR(clEnqueueMapBuffer) +#define __ENQUEUE_MAP_IMAGE_ERR __ERR_STR(clEnqueueMapImage) +#define __ENQUEUE_UNMAP_MEM_OBJECT_ERR __ERR_STR(clEnqueueUnMapMemObject) +#define __ENQUEUE_NDRANGE_KERNEL_ERR __ERR_STR(clEnqueueNDRangeKernel) +#define __ENQUEUE_TASK_ERR __ERR_STR(clEnqueueTask) +#define __ENQUEUE_NATIVE_KERNEL __ERR_STR(clEnqueueNativeKernel) +#if defined(CL_VERSION_1_2) +#define __ENQUEUE_MIGRATE_MEM_OBJECTS_ERR __ERR_STR(clEnqueueMigrateMemObjects) +#endif // #if defined(CL_VERSION_1_2) + +#define __ENQUEUE_ACQUIRE_GL_ERR __ERR_STR(clEnqueueAcquireGLObjects) +#define __ENQUEUE_RELEASE_GL_ERR __ERR_STR(clEnqueueReleaseGLObjects) + + +#define __RETAIN_ERR __ERR_STR(Retain Object) +#define __RELEASE_ERR __ERR_STR(Release Object) +#define __FLUSH_ERR __ERR_STR(clFlush) +#define __FINISH_ERR __ERR_STR(clFinish) +#define __VECTOR_CAPACITY_ERR __ERR_STR(Vector capacity error) + +/** + * CL 1.2 version that uses device fission. + */ +#if defined(CL_VERSION_1_2) +#define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevices) +#else +#define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevicesEXT) +#endif // #if defined(CL_VERSION_1_2) + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +#define __ENQUEUE_MARKER_ERR __ERR_STR(clEnqueueMarker) +#define __ENQUEUE_WAIT_FOR_EVENTS_ERR __ERR_STR(clEnqueueWaitForEvents) +#define __ENQUEUE_BARRIER_ERR __ERR_STR(clEnqueueBarrier) +#define __UNLOAD_COMPILER_ERR __ERR_STR(clUnloadCompiler) +#define __CREATE_GL_TEXTURE_2D_ERR __ERR_STR(clCreateFromGLTexture2D) +#define __CREATE_GL_TEXTURE_3D_ERR __ERR_STR(clCreateFromGLTexture3D) +#define __CREATE_IMAGE2D_ERR __ERR_STR(clCreateImage2D) +#define __CREATE_IMAGE3D_ERR __ERR_STR(clCreateImage3D) +#endif // #if defined(CL_VERSION_1_1) + +#endif // __CL_USER_OVERRIDE_ERROR_STRINGS +//! \endcond + +/** + * CL 1.2 marker and barrier commands + */ +#if defined(CL_VERSION_1_2) +#define __ENQUEUE_MARKER_WAIT_LIST_ERR __ERR_STR(clEnqueueMarkerWithWaitList) +#define __ENQUEUE_BARRIER_WAIT_LIST_ERR __ERR_STR(clEnqueueBarrierWithWaitList) +#endif // #if defined(CL_VERSION_1_2) + +#if !defined(__USE_DEV_STRING) && !defined(__NO_STD_STRING) +typedef std::string STRING_CLASS; +#elif !defined(__USE_DEV_STRING) + +/*! \class string + * \brief Simple string class, that provides a limited subset of std::string + * functionality but avoids many of the issues that come with that class. + + * \note Deprecated. Please use std::string as default or + * re-define the string class to match the std::string + * interface by defining STRING_CLASS + */ +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED string CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ +private: + ::size_t size_; + char * str_; +public: + //! \brief Constructs an empty string, allocating no memory. + string(void) : size_(0), str_(NULL) + { + } + + /*! \brief Constructs a string populated from an arbitrary value of + * specified size. + * + * An extra '\0' is added, in case none was contained in str. + * + * \param str the initial value of the string instance. Note that '\0' + * characters receive no special treatment. If NULL, + * the string is left empty, with a size of 0. + * + * \param size the number of characters to copy from str. + */ + string(const char * str, ::size_t size) : + size_(size), + str_(NULL) + { + if( size > 0 ) { + str_ = new char[size_+1]; + if (str_ != NULL) { + memcpy(str_, str, size_ * sizeof(char)); + str_[size_] = '\0'; + } + else { + size_ = 0; + } + } + } + + /*! \brief Constructs a string populated from a null-terminated value. + * + * \param str the null-terminated initial value of the string instance. + * If NULL, the string is left empty, with a size of 0. + */ + string(const char * str) : + size_(0), + str_(NULL) + { + if( str ) { + size_= ::strlen(str); + } + if( size_ > 0 ) { + str_ = new char[size_ + 1]; + if (str_ != NULL) { + memcpy(str_, str, (size_ + 1) * sizeof(char)); + } + } + } + + void resize( ::size_t n ) + { + if( size_ == n ) { + return; + } + if (n == 0) { + if( str_ ) { + delete [] str_; + } + str_ = NULL; + size_ = 0; + } + else { + char *newString = new char[n + 1]; + int copySize = n; + if( size_ < n ) { + copySize = size_; + } + size_ = n; + + if(str_) { + memcpy(newString, str_, (copySize + 1) * sizeof(char)); + } + if( copySize < size_ ) { + memset(newString + copySize, 0, size_ - copySize); + } + newString[size_] = '\0'; + + delete [] str_; + str_ = newString; + } + } + + const char& operator[] ( ::size_t pos ) const + { + return str_[pos]; + } + + char& operator[] ( ::size_t pos ) + { + return str_[pos]; + } + + /*! \brief Copies the value of another string to this one. + * + * \param rhs the string to copy. + * + * \returns a reference to the modified instance. + */ + string& operator=(const string& rhs) + { + if (this == &rhs) { + return *this; + } + + if( str_ != NULL ) { + delete [] str_; + str_ = NULL; + size_ = 0; + } + + if (rhs.size_ == 0 || rhs.str_ == NULL) { + str_ = NULL; + size_ = 0; + } + else { + str_ = new char[rhs.size_ + 1]; + size_ = rhs.size_; + + if (str_ != NULL) { + memcpy(str_, rhs.str_, (size_ + 1) * sizeof(char)); + } + else { + size_ = 0; + } + } + + return *this; + } + + /*! \brief Constructs a string by copying the value of another instance. + * + * \param rhs the string to copy. + */ + string(const string& rhs) : + size_(0), + str_(NULL) + { + *this = rhs; + } + + //! \brief Destructor - frees memory used to hold the current value. + ~string() + { + delete[] str_; + str_ = NULL; + } + + //! \brief Queries the length of the string, excluding any added '\0's. + ::size_t size(void) const { return size_; } + + //! \brief Queries the length of the string, excluding any added '\0's. + ::size_t length(void) const { return size(); } + + /*! \brief Returns a pointer to the private copy held by this instance, + * or "" if empty/unset. + */ + const char * c_str(void) const { return (str_) ? str_ : "";} +}; +typedef cl::string STRING_CLASS; +#endif // #elif !defined(__USE_DEV_STRING) + +#if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) +#define VECTOR_CLASS std::vector +#elif !defined(__USE_DEV_VECTOR) +#define VECTOR_CLASS cl::vector + +#if !defined(__MAX_DEFAULT_VECTOR_SIZE) +#define __MAX_DEFAULT_VECTOR_SIZE 10 +#endif + +/*! \class vector + * \brief Fixed sized vector implementation that mirroring + * + * \note Deprecated. Please use std::vector as default or + * re-define the vector class to match the std::vector + * interface by defining VECTOR_CLASS + + * \note Not recommended for use with custom objects as + * current implementation will construct N elements + * + * std::vector functionality. + * \brief Fixed sized vector compatible with std::vector. + * + * \note + * This differs from std::vector<> not just in memory allocation, + * but also in terms of when members are constructed, destroyed, + * and assigned instead of being copy constructed. + * + * \param T type of element contained in the vector. + * + * \param N maximum size of the vector. + */ +template +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED vector CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ +private: + T data_[N]; + unsigned int size_; + +public: + //! \brief Constructs an empty vector with no memory allocated. + vector() : + size_(static_cast(0)) + {} + + //! \brief Deallocates the vector's memory and destroys all of its elements. + ~vector() + { + clear(); + } + + //! \brief Returns the number of elements currently contained. + unsigned int size(void) const + { + return size_; + } + + /*! \brief Empties the vector of all elements. + * \note + * This does not deallocate memory but will invoke destructors + * on contained elements. + */ + void clear() + { + while(!empty()) { + pop_back(); + } + } + + /*! \brief Appends an element after the last valid element. + * Calling this on a vector that has reached capacity will throw an + * exception if exceptions are enabled. + */ + void push_back (const T& x) + { + if (size() < N) { + new (&data_[size_]) T(x); + size_++; + } else { + detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); + } + } + + /*! \brief Removes the last valid element from the vector. + * Calling this on an empty vector will throw an exception + * if exceptions are enabled. + */ + void pop_back(void) + { + if (size_ != 0) { + --size_; + data_[size_].~T(); + } else { + detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); + } + } + + /*! \brief Constructs with a value copied from another. + * + * \param vec the vector to copy. + */ + vector(const vector& vec) : + size_(vec.size_) + { + if (size_ != 0) { + assign(vec.begin(), vec.end()); + } + } + + /*! \brief Constructs with a specified number of initial elements. + * + * \param size number of initial elements. + * + * \param val value of initial elements. + */ + vector(unsigned int size, const T& val = T()) : + size_(0) + { + for (unsigned int i = 0; i < size; i++) { + push_back(val); + } + } + + /*! \brief Overwrites the current content with that copied from another + * instance. + * + * \param rhs vector to copy. + * + * \returns a reference to this. + */ + vector& operator=(const vector& rhs) + { + if (this == &rhs) { + return *this; + } + + if (rhs.size_ != 0) { + assign(rhs.begin(), rhs.end()); + } else { + clear(); + } + + return *this; + } + + /*! \brief Tests equality against another instance. + * + * \param vec the vector against which to compare. + */ + bool operator==(vector &vec) + { + if (size() != vec.size()) { + return false; + } + + for( unsigned int i = 0; i < size(); ++i ) { + if( operator[](i) != vec[i] ) { + return false; + } + } + return true; + } + + //! \brief Conversion operator to T*. + operator T* () { return data_; } + + //! \brief Conversion operator to const T*. + operator const T* () const { return data_; } + + //! \brief Tests whether this instance has any elements. + bool empty (void) const + { + return size_==0; + } + + //! \brief Returns the maximum number of elements this instance can hold. + unsigned int max_size (void) const + { + return N; + } + + //! \brief Returns the maximum number of elements this instance can hold. + unsigned int capacity () const + { + return N; + } + + /*! \brief Returns a reference to a given element. + * + * \param index which element to access. * + * \note + * The caller is responsible for ensuring index is >= 0 and < size(). + */ + T& operator[](int index) + { + return data_[index]; + } + + /*! \brief Returns a const reference to a given element. + * + * \param index which element to access. + * + * \note + * The caller is responsible for ensuring index is >= 0 and < size(). + */ + const T& operator[](int index) const + { + return data_[index]; + } + + /*! \brief Assigns elements of the vector based on a source iterator range. + * + * \param start Beginning iterator of source range + * \param end Enditerator of source range + * + * \note + * Will throw an exception if exceptions are enabled and size exceeded. + */ + template + void assign(I start, I end) + { + clear(); + while(start != end) { + push_back(*start); + start++; + } + } + + /*! \class iterator + * \brief Const iterator class for vectors + */ + class iterator + { + private: + const vector *vec_; + int index_; + + /** + * Internal iterator constructor to capture reference + * to the vector it iterates over rather than taking + * the vector by copy. + */ + iterator (const vector &vec, int index) : + vec_(&vec) + { + if( !vec.empty() ) { + index_ = index; + } else { + index_ = -1; + } + } + + public: + iterator(void) : + index_(-1), + vec_(NULL) + { + } + + iterator(const iterator& rhs) : + vec_(rhs.vec_), + index_(rhs.index_) + { + } + + ~iterator(void) {} + + static iterator begin(const cl::vector &vec) + { + iterator i(vec, 0); + + return i; + } + + static iterator end(const cl::vector &vec) + { + iterator i(vec, vec.size()); + + return i; + } + + bool operator==(iterator i) + { + return ((vec_ == i.vec_) && + (index_ == i.index_)); + } + + bool operator!=(iterator i) + { + return (!(*this==i)); + } + + iterator& operator++() + { + ++index_; + return *this; + } + + iterator operator++(int) + { + iterator retVal(*this); + ++index_; + return retVal; + } + + iterator& operator--() + { + --index_; + return *this; + } + + iterator operator--(int) + { + iterator retVal(*this); + --index_; + return retVal; + } + + const T& operator *() const + { + return (*vec_)[index_]; + } + }; + + iterator begin(void) + { + return iterator::begin(*this); + } + + iterator begin(void) const + { + return iterator::begin(*this); + } + + iterator end(void) + { + return iterator::end(*this); + } + + iterator end(void) const + { + return iterator::end(*this); + } + + T& front(void) + { + return data_[0]; + } + + T& back(void) + { + return data_[size_]; + } + + const T& front(void) const + { + return data_[0]; + } + + const T& back(void) const + { + return data_[size_-1]; + } +}; +#endif // #if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) + + + + + +namespace detail { +#define __DEFAULT_NOT_INITIALIZED 1 +#define __DEFAULT_BEING_INITIALIZED 2 +#define __DEFAULT_INITIALIZED 4 + + /* + * Compare and exchange primitives are needed for handling of defaults + */ + inline int compare_exchange(volatile int * dest, int exchange, int comparand) + { +#ifdef _WIN32 + return (int)(InterlockedCompareExchange( + (volatile long*)dest, + (long)exchange, + (long)comparand)); +#elif defined(__APPLE__) || defined(__MACOSX) + return OSAtomicOr32Orig((uint32_t)exchange, (volatile uint32_t*)dest); +#else // !_WIN32 || defined(__APPLE__) || defined(__MACOSX) + return (__sync_val_compare_and_swap( + dest, + comparand, + exchange)); +#endif // !_WIN32 + } + + inline void fence() { _mm_mfence(); } +}; // namespace detail + + +/*! \brief class used to interface between C++ and + * OpenCL C calls that require arrays of size_t values, whose + * size is known statically. + */ +template +class size_t +{ +private: + ::size_t data_[N]; + +public: + //! \brief Initialize size_t to all 0s + size_t() + { + for( int i = 0; i < N; ++i ) { + data_[i] = 0; + } + } + + ::size_t& operator[](int index) + { + return data_[index]; + } + + const ::size_t& operator[](int index) const + { + return data_[index]; + } + + //! \brief Conversion operator to T*. + operator ::size_t* () { return data_; } + + //! \brief Conversion operator to const T*. + operator const ::size_t* () const { return data_; } +}; + +namespace detail { + +// Generic getInfoHelper. The final parameter is used to guide overload +// resolution: the actual parameter passed is an int, which makes this +// a worse conversion sequence than a specialization that declares the +// parameter as an int. +template +inline cl_int getInfoHelper(Functor f, cl_uint name, T* param, long) +{ + return f(name, sizeof(T), param, NULL); +} + +// Specialized getInfoHelper for VECTOR_CLASS params +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + T* value = (T*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + param->assign(&value[0], &value[required/sizeof(T)]); + return CL_SUCCESS; +} + +/* Specialization for reference-counted types. This depends on the + * existence of Wrapper::cl_type, and none of the other types having the + * cl_type member. Note that simplify specifying the parameter as Wrapper + * does not work, because when using a derived type (e.g. Context) the generic + * template will provide a better match. + */ +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int, typename T::cl_type = 0) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + typename T::cl_type * value = (typename T::cl_type *) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + ::size_t elements = required / sizeof(typename T::cl_type); + param->assign(&value[0], &value[elements]); + for (::size_t i = 0; i < elements; i++) + { + if (value[i] != NULL) + { + err = (*param)[i].retain(); + if (err != CL_SUCCESS) { + return err; + } + } + } + return CL_SUCCESS; +} + +// Specialized for getInfo +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int) +{ + cl_int err = f(name, param->size() * sizeof(char *), &(*param)[0], NULL); + + if (err != CL_SUCCESS) { + return err; + } + + return CL_SUCCESS; +} + +// Specialized GetInfoHelper for STRING_CLASS params +template +inline cl_int getInfoHelper(Func f, cl_uint name, STRING_CLASS* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + char* value = (char*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + *param = value; + return CL_SUCCESS; +} + +// Specialized GetInfoHelper for cl::size_t params +template +inline cl_int getInfoHelper(Func f, cl_uint name, size_t* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + ::size_t* value = (::size_t*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + for(int i = 0; i < N; ++i) { + (*param)[i] = value[i]; + } + + return CL_SUCCESS; +} + +template struct ReferenceHandler; + +/* Specialization for reference-counted types. This depends on the + * existence of Wrapper::cl_type, and none of the other types having the + * cl_type member. Note that simplify specifying the parameter as Wrapper + * does not work, because when using a derived type (e.g. Context) the generic + * template will provide a better match. + */ +template +inline cl_int getInfoHelper(Func f, cl_uint name, T* param, int, typename T::cl_type = 0) +{ + typename T::cl_type value; + cl_int err = f(name, sizeof(value), &value, NULL); + if (err != CL_SUCCESS) { + return err; + } + *param = value; + if (value != NULL) + { + err = param->retain(); + if (err != CL_SUCCESS) { + return err; + } + } + return CL_SUCCESS; +} + +#define __PARAM_NAME_INFO_1_0(F) \ + F(cl_platform_info, CL_PLATFORM_PROFILE, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_VERSION, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_NAME, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_VENDOR, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_EXTENSIONS, STRING_CLASS) \ + \ + F(cl_device_info, CL_DEVICE_TYPE, cl_device_type) \ + F(cl_device_info, CL_DEVICE_VENDOR_ID, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_COMPUTE_UNITS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_GROUP_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_SIZES, VECTOR_CLASS< ::size_t>) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_CLOCK_FREQUENCY, cl_uint) \ + F(cl_device_info, CL_DEVICE_ADDRESS_BITS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_READ_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WRITE_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_MEM_ALLOC_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_WIDTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_HEIGHT, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_WIDTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_HEIGHT, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_DEPTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE_SUPPORT, cl_bool) \ + F(cl_device_info, CL_DEVICE_MAX_PARAMETER_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_MAX_SAMPLERS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MEM_BASE_ADDR_ALIGN, cl_uint) \ + F(cl_device_info, CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE, cl_uint) \ + F(cl_device_info, CL_DEVICE_SINGLE_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, cl_device_mem_cache_type) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE, cl_uint)\ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_MAX_CONSTANT_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_LOCAL_MEM_TYPE, cl_device_local_mem_type) \ + F(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_ERROR_CORRECTION_SUPPORT, cl_bool) \ + F(cl_device_info, CL_DEVICE_PROFILING_TIMER_RESOLUTION, ::size_t) \ + F(cl_device_info, CL_DEVICE_ENDIAN_LITTLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_AVAILABLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_COMPILER_AVAILABLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_EXECUTION_CAPABILITIES, cl_device_exec_capabilities) \ + F(cl_device_info, CL_DEVICE_QUEUE_PROPERTIES, cl_command_queue_properties) \ + F(cl_device_info, CL_DEVICE_PLATFORM, cl_platform_id) \ + F(cl_device_info, CL_DEVICE_NAME, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_VENDOR, STRING_CLASS) \ + F(cl_device_info, CL_DRIVER_VERSION, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_PROFILE, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_VERSION, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_EXTENSIONS, STRING_CLASS) \ + \ + F(cl_context_info, CL_CONTEXT_REFERENCE_COUNT, cl_uint) \ + F(cl_context_info, CL_CONTEXT_DEVICES, VECTOR_CLASS) \ + F(cl_context_info, CL_CONTEXT_PROPERTIES, VECTOR_CLASS) \ + \ + F(cl_event_info, CL_EVENT_COMMAND_QUEUE, cl::CommandQueue) \ + F(cl_event_info, CL_EVENT_COMMAND_TYPE, cl_command_type) \ + F(cl_event_info, CL_EVENT_REFERENCE_COUNT, cl_uint) \ + F(cl_event_info, CL_EVENT_COMMAND_EXECUTION_STATUS, cl_uint) \ + \ + F(cl_profiling_info, CL_PROFILING_COMMAND_QUEUED, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_SUBMIT, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_START, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_END, cl_ulong) \ + \ + F(cl_mem_info, CL_MEM_TYPE, cl_mem_object_type) \ + F(cl_mem_info, CL_MEM_FLAGS, cl_mem_flags) \ + F(cl_mem_info, CL_MEM_SIZE, ::size_t) \ + F(cl_mem_info, CL_MEM_HOST_PTR, void*) \ + F(cl_mem_info, CL_MEM_MAP_COUNT, cl_uint) \ + F(cl_mem_info, CL_MEM_REFERENCE_COUNT, cl_uint) \ + F(cl_mem_info, CL_MEM_CONTEXT, cl::Context) \ + \ + F(cl_image_info, CL_IMAGE_FORMAT, cl_image_format) \ + F(cl_image_info, CL_IMAGE_ELEMENT_SIZE, ::size_t) \ + F(cl_image_info, CL_IMAGE_ROW_PITCH, ::size_t) \ + F(cl_image_info, CL_IMAGE_SLICE_PITCH, ::size_t) \ + F(cl_image_info, CL_IMAGE_WIDTH, ::size_t) \ + F(cl_image_info, CL_IMAGE_HEIGHT, ::size_t) \ + F(cl_image_info, CL_IMAGE_DEPTH, ::size_t) \ + \ + F(cl_sampler_info, CL_SAMPLER_REFERENCE_COUNT, cl_uint) \ + F(cl_sampler_info, CL_SAMPLER_CONTEXT, cl::Context) \ + F(cl_sampler_info, CL_SAMPLER_NORMALIZED_COORDS, cl_addressing_mode) \ + F(cl_sampler_info, CL_SAMPLER_ADDRESSING_MODE, cl_filter_mode) \ + F(cl_sampler_info, CL_SAMPLER_FILTER_MODE, cl_bool) \ + \ + F(cl_program_info, CL_PROGRAM_REFERENCE_COUNT, cl_uint) \ + F(cl_program_info, CL_PROGRAM_CONTEXT, cl::Context) \ + F(cl_program_info, CL_PROGRAM_NUM_DEVICES, cl_uint) \ + F(cl_program_info, CL_PROGRAM_DEVICES, VECTOR_CLASS) \ + F(cl_program_info, CL_PROGRAM_SOURCE, STRING_CLASS) \ + F(cl_program_info, CL_PROGRAM_BINARY_SIZES, VECTOR_CLASS< ::size_t>) \ + F(cl_program_info, CL_PROGRAM_BINARIES, VECTOR_CLASS) \ + \ + F(cl_program_build_info, CL_PROGRAM_BUILD_STATUS, cl_build_status) \ + F(cl_program_build_info, CL_PROGRAM_BUILD_OPTIONS, STRING_CLASS) \ + F(cl_program_build_info, CL_PROGRAM_BUILD_LOG, STRING_CLASS) \ + \ + F(cl_kernel_info, CL_KERNEL_FUNCTION_NAME, STRING_CLASS) \ + F(cl_kernel_info, CL_KERNEL_NUM_ARGS, cl_uint) \ + F(cl_kernel_info, CL_KERNEL_REFERENCE_COUNT, cl_uint) \ + F(cl_kernel_info, CL_KERNEL_CONTEXT, cl::Context) \ + F(cl_kernel_info, CL_KERNEL_PROGRAM, cl::Program) \ + \ + F(cl_kernel_work_group_info, CL_KERNEL_WORK_GROUP_SIZE, ::size_t) \ + F(cl_kernel_work_group_info, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, cl::size_t<3>) \ + F(cl_kernel_work_group_info, CL_KERNEL_LOCAL_MEM_SIZE, cl_ulong) \ + \ + F(cl_command_queue_info, CL_QUEUE_CONTEXT, cl::Context) \ + F(cl_command_queue_info, CL_QUEUE_DEVICE, cl::Device) \ + F(cl_command_queue_info, CL_QUEUE_REFERENCE_COUNT, cl_uint) \ + F(cl_command_queue_info, CL_QUEUE_PROPERTIES, cl_command_queue_properties) + +#if defined(CL_VERSION_1_1) +#define __PARAM_NAME_INFO_1_1(F) \ + F(cl_context_info, CL_CONTEXT_NUM_DEVICES, cl_uint)\ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_INT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF, cl_uint) \ + F(cl_device_info, CL_DEVICE_DOUBLE_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_HALF_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_HOST_UNIFIED_MEMORY, cl_bool) \ + F(cl_device_info, CL_DEVICE_OPENCL_C_VERSION, STRING_CLASS) \ + \ + F(cl_mem_info, CL_MEM_ASSOCIATED_MEMOBJECT, cl::Memory) \ + F(cl_mem_info, CL_MEM_OFFSET, ::size_t) \ + \ + F(cl_kernel_work_group_info, CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE, ::size_t) \ + F(cl_kernel_work_group_info, CL_KERNEL_PRIVATE_MEM_SIZE, cl_ulong) \ + \ + F(cl_event_info, CL_EVENT_CONTEXT, cl::Context) +#endif // CL_VERSION_1_1 + + +#if defined(CL_VERSION_1_2) +#define __PARAM_NAME_INFO_1_2(F) \ + F(cl_image_info, CL_IMAGE_BUFFER, cl::Buffer) \ + \ + F(cl_program_info, CL_PROGRAM_NUM_KERNELS, ::size_t) \ + F(cl_program_info, CL_PROGRAM_KERNEL_NAMES, STRING_CLASS) \ + \ + F(cl_program_build_info, CL_PROGRAM_BINARY_TYPE, cl_program_binary_type) \ + \ + F(cl_kernel_info, CL_KERNEL_ATTRIBUTES, STRING_CLASS) \ + \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_ADDRESS_QUALIFIER, cl_kernel_arg_address_qualifier) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_ACCESS_QUALIFIER, cl_kernel_arg_access_qualifier) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_TYPE_NAME, STRING_CLASS) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_NAME, STRING_CLASS) \ + \ + F(cl_device_info, CL_DEVICE_PARENT_DEVICE, cl_device_id) \ + F(cl_device_info, CL_DEVICE_PARTITION_PROPERTIES, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_PARTITION_TYPE, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_REFERENCE_COUNT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, ::size_t) \ + F(cl_device_info, CL_DEVICE_PARTITION_AFFINITY_DOMAIN, cl_device_affinity_domain) \ + F(cl_device_info, CL_DEVICE_BUILT_IN_KERNELS, STRING_CLASS) +#endif // #if defined(CL_VERSION_1_2) + +#if defined(USE_CL_DEVICE_FISSION) +#define __PARAM_NAME_DEVICE_FISSION(F) \ + F(cl_device_info, CL_DEVICE_PARENT_DEVICE_EXT, cl_device_id) \ + F(cl_device_info, CL_DEVICE_PARTITION_TYPES_EXT, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \ + F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, VECTOR_CLASS) +#endif // USE_CL_DEVICE_FISSION + +template +struct param_traits {}; + +#define __CL_DECLARE_PARAM_TRAITS(token, param_name, T) \ +struct token; \ +template<> \ +struct param_traits \ +{ \ + enum { value = param_name }; \ + typedef T param_type; \ +}; + +__PARAM_NAME_INFO_1_0(__CL_DECLARE_PARAM_TRAITS) +#if defined(CL_VERSION_1_1) +__PARAM_NAME_INFO_1_1(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_1_1 +#if defined(CL_VERSION_1_2) +__PARAM_NAME_INFO_1_2(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_1_1 + +#if defined(USE_CL_DEVICE_FISSION) +__PARAM_NAME_DEVICE_FISSION(__CL_DECLARE_PARAM_TRAITS); +#endif // USE_CL_DEVICE_FISSION + +#ifdef CL_PLATFORM_ICD_SUFFIX_KHR +__CL_DECLARE_PARAM_TRAITS(cl_platform_info, CL_PLATFORM_ICD_SUFFIX_KHR, STRING_CLASS) +#endif + +#ifdef CL_DEVICE_PROFILING_TIMER_OFFSET_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_PROFILING_TIMER_OFFSET_AMD, cl_ulong) +#endif + +#ifdef CL_DEVICE_GLOBAL_FREE_MEMORY_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_FREE_MEMORY_AMD, VECTOR_CLASS< ::size_t>) +#endif +#ifdef CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_SIMD_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_WAVEFRONT_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WAVEFRONT_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_LOCAL_MEM_BANKS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_BANKS_AMD, cl_uint) +#endif + +#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV, cl_uint) +#endif +#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV, cl_uint) +#endif +#ifdef CL_DEVICE_REGISTERS_PER_BLOCK_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_REGISTERS_PER_BLOCK_NV, cl_uint) +#endif +#ifdef CL_DEVICE_WARP_SIZE_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WARP_SIZE_NV, cl_uint) +#endif +#ifdef CL_DEVICE_GPU_OVERLAP_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GPU_OVERLAP_NV, cl_bool) +#endif +#ifdef CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV, cl_bool) +#endif +#ifdef CL_DEVICE_INTEGRATED_MEMORY_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_INTEGRATED_MEMORY_NV, cl_bool) +#endif + +// Convenience functions + +template +inline cl_int +getInfo(Func f, cl_uint name, T* param) +{ + return getInfoHelper(f, name, param, 0); +} + +template +struct GetInfoFunctor0 +{ + Func f_; const Arg0& arg0_; + cl_int operator ()( + cl_uint param, ::size_t size, void* value, ::size_t* size_ret) + { return f_(arg0_, param, size, value, size_ret); } +}; + +template +struct GetInfoFunctor1 +{ + Func f_; const Arg0& arg0_; const Arg1& arg1_; + cl_int operator ()( + cl_uint param, ::size_t size, void* value, ::size_t* size_ret) + { return f_(arg0_, arg1_, param, size, value, size_ret); } +}; + +template +inline cl_int +getInfo(Func f, const Arg0& arg0, cl_uint name, T* param) +{ + GetInfoFunctor0 f0 = { f, arg0 }; + return getInfoHelper(f0, name, param, 0); +} + +template +inline cl_int +getInfo(Func f, const Arg0& arg0, const Arg1& arg1, cl_uint name, T* param) +{ + GetInfoFunctor1 f0 = { f, arg0, arg1 }; + return getInfoHelper(f0, name, param, 0); +} + +template +struct ReferenceHandler +{ }; + +#if defined(CL_VERSION_1_2) +/** + * OpenCL 1.2 devices do have retain/release. + */ +template <> +struct ReferenceHandler +{ + /** + * Retain the device. + * \param device A valid device created using createSubDevices + * \return + * CL_SUCCESS if the function executed successfully. + * CL_INVALID_DEVICE if device was not a valid subdevice + * CL_OUT_OF_RESOURCES + * CL_OUT_OF_HOST_MEMORY + */ + static cl_int retain(cl_device_id device) + { return ::clRetainDevice(device); } + /** + * Retain the device. + * \param device A valid device created using createSubDevices + * \return + * CL_SUCCESS if the function executed successfully. + * CL_INVALID_DEVICE if device was not a valid subdevice + * CL_OUT_OF_RESOURCES + * CL_OUT_OF_HOST_MEMORY + */ + static cl_int release(cl_device_id device) + { return ::clReleaseDevice(device); } +}; +#else // #if defined(CL_VERSION_1_2) +/** + * OpenCL 1.1 devices do not have retain/release. + */ +template <> +struct ReferenceHandler +{ + // cl_device_id does not have retain(). + static cl_int retain(cl_device_id) + { return CL_SUCCESS; } + // cl_device_id does not have release(). + static cl_int release(cl_device_id) + { return CL_SUCCESS; } +}; +#endif // #if defined(CL_VERSION_1_2) + +template <> +struct ReferenceHandler +{ + // cl_platform_id does not have retain(). + static cl_int retain(cl_platform_id) + { return CL_SUCCESS; } + // cl_platform_id does not have release(). + static cl_int release(cl_platform_id) + { return CL_SUCCESS; } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_context context) + { return ::clRetainContext(context); } + static cl_int release(cl_context context) + { return ::clReleaseContext(context); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_command_queue queue) + { return ::clRetainCommandQueue(queue); } + static cl_int release(cl_command_queue queue) + { return ::clReleaseCommandQueue(queue); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_mem memory) + { return ::clRetainMemObject(memory); } + static cl_int release(cl_mem memory) + { return ::clReleaseMemObject(memory); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_sampler sampler) + { return ::clRetainSampler(sampler); } + static cl_int release(cl_sampler sampler) + { return ::clReleaseSampler(sampler); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_program program) + { return ::clRetainProgram(program); } + static cl_int release(cl_program program) + { return ::clReleaseProgram(program); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_kernel kernel) + { return ::clRetainKernel(kernel); } + static cl_int release(cl_kernel kernel) + { return ::clReleaseKernel(kernel); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_event event) + { return ::clRetainEvent(event); } + static cl_int release(cl_event event) + { return ::clReleaseEvent(event); } +}; + + +// Extracts version number with major in the upper 16 bits, minor in the lower 16 +static cl_uint getVersion(const char *versionInfo) +{ + int highVersion = 0; + int lowVersion = 0; + int index = 7; + while(versionInfo[index] != '.' ) { + highVersion *= 10; + highVersion += versionInfo[index]-'0'; + ++index; + } + ++index; + while(versionInfo[index] != ' ' ) { + lowVersion *= 10; + lowVersion += versionInfo[index]-'0'; + ++index; + } + return (highVersion << 16) | lowVersion; +} + +static cl_uint getPlatformVersion(cl_platform_id platform) +{ + ::size_t size = 0; + clGetPlatformInfo(platform, CL_PLATFORM_VERSION, 0, NULL, &size); + char *versionInfo = (char *) alloca(size); + clGetPlatformInfo(platform, CL_PLATFORM_VERSION, size, &versionInfo[0], &size); + return getVersion(versionInfo); +} + +static cl_uint getDevicePlatformVersion(cl_device_id device) +{ + cl_platform_id platform; + clGetDeviceInfo(device, CL_DEVICE_PLATFORM, sizeof(platform), &platform, NULL); + return getPlatformVersion(platform); +} + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +static cl_uint getContextPlatformVersion(cl_context context) +{ + // The platform cannot be queried directly, so we first have to grab a + // device and obtain its context + ::size_t size = 0; + clGetContextInfo(context, CL_CONTEXT_DEVICES, 0, NULL, &size); + if (size == 0) + return 0; + cl_device_id *devices = (cl_device_id *) alloca(size); + clGetContextInfo(context, CL_CONTEXT_DEVICES, size, devices, NULL); + return getDevicePlatformVersion(devices[0]); +} +#endif // #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + +template +class Wrapper +{ +public: + typedef T cl_type; + +protected: + cl_type object_; + +public: + Wrapper() : object_(NULL) { } + + Wrapper(const cl_type &obj) : object_(obj) { } + + ~Wrapper() + { + if (object_ != NULL) { release(); } + } + + Wrapper(const Wrapper& rhs) + { + object_ = rhs.object_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + } + + Wrapper& operator = (const Wrapper& rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs.object_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + return *this; + } + + Wrapper& operator = (const cl_type &rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs; + return *this; + } + + cl_type operator ()() const { return object_; } + + cl_type& operator ()() { return object_; } + +protected: + template + friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); + + cl_int retain() const + { + return ReferenceHandler::retain(object_); + } + + cl_int release() const + { + return ReferenceHandler::release(object_); + } +}; + +template <> +class Wrapper +{ +public: + typedef cl_device_id cl_type; + +protected: + cl_type object_; + bool referenceCountable_; + + static bool isReferenceCountable(cl_device_id device) + { + bool retVal = false; + if (device != NULL) { + int version = getDevicePlatformVersion(device); + if(version > ((1 << 16) + 1)) { + retVal = true; + } + } + return retVal; + } + +public: + Wrapper() : object_(NULL), referenceCountable_(false) + { + } + + Wrapper(const cl_type &obj) : object_(obj), referenceCountable_(false) + { + referenceCountable_ = isReferenceCountable(obj); + } + + ~Wrapper() + { + if (object_ != NULL) { release(); } + } + + Wrapper(const Wrapper& rhs) + { + object_ = rhs.object_; + referenceCountable_ = isReferenceCountable(object_); + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + } + + Wrapper& operator = (const Wrapper& rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs.object_; + referenceCountable_ = rhs.referenceCountable_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + return *this; + } + + Wrapper& operator = (const cl_type &rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs; + referenceCountable_ = isReferenceCountable(object_); + return *this; + } + + cl_type operator ()() const { return object_; } + + cl_type& operator ()() { return object_; } + +protected: + template + friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); + + template + friend inline cl_int getInfoHelper(Func, cl_uint, VECTOR_CLASS*, int, typename U::cl_type); + + cl_int retain() const + { + if( referenceCountable_ ) { + return ReferenceHandler::retain(object_); + } + else { + return CL_SUCCESS; + } + } + + cl_int release() const + { + if( referenceCountable_ ) { + return ReferenceHandler::release(object_); + } + else { + return CL_SUCCESS; + } + } +}; + +} // namespace detail +//! \endcond + +/*! \stuct ImageFormat + * \brief Adds constructors and member functions for cl_image_format. + * + * \see cl_image_format + */ +struct ImageFormat : public cl_image_format +{ + //! \brief Default constructor - performs no initialization. + ImageFormat(){} + + //! \brief Initializing constructor. + ImageFormat(cl_channel_order order, cl_channel_type type) + { + image_channel_order = order; + image_channel_data_type = type; + } + + //! \brief Assignment operator. + ImageFormat& operator = (const ImageFormat& rhs) + { + if (this != &rhs) { + this->image_channel_data_type = rhs.image_channel_data_type; + this->image_channel_order = rhs.image_channel_order; + } + return *this; + } +}; + +/*! \brief Class interface for cl_device_id. + * + * \note Copies of these objects are inexpensive, since they don't 'own' + * any underlying resources or data structures. + * + * \see cl_device_id + */ +class Device : public detail::Wrapper +{ +public: + //! \brief Default constructor - initializes to NULL. + Device() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device(const Device& device) : detail::Wrapper(device) { } + + /*! \brief Constructor from cl_device_id. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device(const cl_device_id &device) : detail::Wrapper(device) { } + + /*! \brief Returns the first device on the default context. + * + * \see Context::getDefault() + */ + static Device getDefault(cl_int * err = NULL); + + /*! \brief Assignment operator from Device. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device& operator = (const Device& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_device_id. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device& operator = (const cl_device_id& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetDeviceInfo(). + template + cl_int getInfo(cl_device_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetDeviceInfo, object_, name, param), + __GET_DEVICE_INFO_ERR); + } + + //! \brief Wrapper for clGetDeviceInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_device_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /** + * CL 1.2 version + */ +#if defined(CL_VERSION_1_2) + //! \brief Wrapper for clCreateSubDevicesEXT(). + cl_int createSubDevices( + const cl_device_partition_property * properties, + VECTOR_CLASS* devices) + { + cl_uint n = 0; + cl_int err = clCreateSubDevices(object_, properties, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = clCreateSubDevices(object_, properties, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif // #if defined(CL_VERSION_1_2) + +/** + * CL 1.1 version that uses device fission. + */ +#if defined(CL_VERSION_1_1) +#if defined(USE_CL_DEVICE_FISSION) + cl_int createSubDevices( + const cl_device_partition_property_ext * properties, + VECTOR_CLASS* devices) + { + typedef CL_API_ENTRY cl_int + ( CL_API_CALL * PFN_clCreateSubDevicesEXT)( + cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + static PFN_clCreateSubDevicesEXT pfn_clCreateSubDevicesEXT = NULL; + __INIT_CL_EXT_FCN_PTR(clCreateSubDevicesEXT); + + cl_uint n = 0; + cl_int err = pfn_clCreateSubDevicesEXT(object_, properties, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = pfn_clCreateSubDevicesEXT(object_, properties, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif // #if defined(USE_CL_DEVICE_FISSION) +#endif // #if defined(CL_VERSION_1_1) +}; + +/*! \brief Class interface for cl_platform_id. + * + * \note Copies of these objects are inexpensive, since they don't 'own' + * any underlying resources or data structures. + * + * \see cl_platform_id + */ +class Platform : public detail::Wrapper +{ +public: + //! \brief Default constructor - initializes to NULL. + Platform() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform(const Platform& platform) : detail::Wrapper(platform) { } + + /*! \brief Constructor from cl_platform_id. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform(const cl_platform_id &platform) : detail::Wrapper(platform) { } + + /*! \brief Assignment operator from Platform. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform& operator = (const Platform& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_platform_id. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform& operator = (const cl_platform_id& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetPlatformInfo(). + cl_int getInfo(cl_platform_info name, STRING_CLASS* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetPlatformInfo, object_, name, param), + __GET_PLATFORM_INFO_ERR); + } + + //! \brief Wrapper for clGetPlatformInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_platform_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Gets a list of devices for this platform. + * + * Wraps clGetDeviceIDs(). + */ + cl_int getDevices( + cl_device_type type, + VECTOR_CLASS* devices) const + { + cl_uint n = 0; + if( devices == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); + } + cl_int err = ::clGetDeviceIDs(object_, type, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = ::clGetDeviceIDs(object_, type, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } + +#if defined(USE_DX_INTEROP) + /*! \brief Get the list of available D3D10 devices. + * + * \param d3d_device_source. + * + * \param d3d_object. + * + * \param d3d_device_set. + * + * \param devices returns a vector of OpenCL D3D10 devices found. The cl::Device + * values returned in devices can be used to identify a specific OpenCL + * device. If \a devices argument is NULL, this argument is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * + * The application can query specific capabilities of the OpenCL device(s) + * returned by cl::getDevices. This can be used by the application to + * determine which device(s) to use. + * + * \note In the case that exceptions are enabled and a return value + * other than CL_SUCCESS is generated, then cl::Error exception is + * generated. + */ + cl_int getDevices( + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + VECTOR_CLASS* devices) const + { + typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clGetDeviceIDsFromD3D10KHR)( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint* num_devices); + + if( devices == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); + } + + static PFN_clGetDeviceIDsFromD3D10KHR pfn_clGetDeviceIDsFromD3D10KHR = NULL; + __INIT_CL_EXT_FCN_PTR_PLATFORM(object_, clGetDeviceIDsFromD3D10KHR); + + cl_uint n = 0; + cl_int err = pfn_clGetDeviceIDsFromD3D10KHR( + object_, + d3d_device_source, + d3d_object, + d3d_device_set, + 0, + NULL, + &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = pfn_clGetDeviceIDsFromD3D10KHR( + object_, + d3d_device_source, + d3d_object, + d3d_device_set, + n, + ids, + NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif + + /*! \brief Gets a list of available platforms. + * + * Wraps clGetPlatformIDs(). + */ + static cl_int get( + VECTOR_CLASS* platforms) + { + cl_uint n = 0; + + if( platforms == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); + } + + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + platforms->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } + + /*! \brief Gets the first available platform. + * + * Wraps clGetPlatformIDs(), returning the first result. + */ + static cl_int get( + Platform * platform) + { + cl_uint n = 0; + + if( platform == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); + } + + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + *platform = ids[0]; + return CL_SUCCESS; + } + + /*! \brief Gets the first available platform, returning it by value. + * + * Wraps clGetPlatformIDs(), returning the first result. + */ + static Platform get( + cl_int * errResult = NULL) + { + Platform platform; + cl_uint n = 0; + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + if (errResult != NULL) { + *errResult = err; + } + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + + if (err != CL_SUCCESS) { + detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + if (errResult != NULL) { + *errResult = err; + } + + return ids[0]; + } + + static Platform getDefault( + cl_int *errResult = NULL ) + { + return get(errResult); + } + + +#if defined(CL_VERSION_1_2) + //! \brief Wrapper for clUnloadCompiler(). + cl_int + unloadCompiler() + { + return ::clUnloadPlatformCompiler(object_); + } +#endif // #if defined(CL_VERSION_1_2) +}; // class Platform + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +/** + * Unload the OpenCL compiler. + * \note Deprecated for OpenCL 1.2. Use Platform::unloadCompiler instead. + */ +inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int +UnloadCompiler() CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; +inline cl_int +UnloadCompiler() +{ + return ::clUnloadCompiler(); +} +#endif // #if defined(CL_VERSION_1_1) + +/*! \brief Class interface for cl_context. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_context as the original. For details, see + * clRetainContext() and clReleaseContext(). + * + * \see cl_context + */ +class Context + : public detail::Wrapper +{ +private: + static volatile int default_initialized_; + static Context default_; + static volatile cl_int default_error_; +public: + /*! \brief Destructor. + * + * This calls clReleaseContext() on the value held by this instance. + */ + ~Context() { } + + /*! \brief Constructs a context including a list of specified devices. + * + * Wraps clCreateContext(). + */ + Context( + const VECTOR_CLASS& devices, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + object_ = ::clCreateContext( + properties, (cl_uint) numDevices, + deviceIDs, + notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_ERR); + if (err != NULL) { + *err = error; + } + } + + Context( + const Device& device, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + + cl_device_id deviceID = device(); + + object_ = ::clCreateContext( + properties, 1, + &deviceID, + notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Constructs a context including all devices of a specified type. + * + * Wraps clCreateContextFromType(). + */ + Context( + cl_device_type type, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + +#if !defined(__APPLE__) || !defined(__MACOS) + cl_context_properties prop[4] = {CL_CONTEXT_PLATFORM, 0, 0, 0 }; + if (properties == NULL) { + prop[1] = (cl_context_properties)Platform::get(&error)(); + if (error != CL_SUCCESS) { + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + return; + } + } + + properties = &prop[0]; + } +#endif + object_ = ::clCreateContextFromType( + properties, type, notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Returns a singleton context including all devices of CL_DEVICE_TYPE_DEFAULT. + * + * \note All calls to this function return the same cl_context as the first. + */ + static Context getDefault(cl_int * err = NULL) + { + int state = detail::compare_exchange( + &default_initialized_, + __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); + + if (state & __DEFAULT_INITIALIZED) { + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + if (state & __DEFAULT_BEING_INITIALIZED) { + // Assume writes will propagate eventually... + while(default_initialized_ != __DEFAULT_INITIALIZED) { + detail::fence(); + } + + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + cl_int error; + default_ = Context( + CL_DEVICE_TYPE_DEFAULT, + NULL, + NULL, + NULL, + &error); + + detail::fence(); + + default_error_ = error; + // Assume writes will propagate eventually... + default_initialized_ = __DEFAULT_INITIALIZED; + + detail::fence(); + + if (err != NULL) { + *err = default_error_; + } + return default_; + + } + + //! \brief Default constructor - initializes to NULL. + Context() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This calls clRetainContext() on the parameter's cl_context. + */ + Context(const Context& context) : detail::Wrapper(context) { } + + /*! \brief Constructor from cl_context - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_context + * into the new Context object. + */ + __CL_EXPLICIT_CONSTRUCTORS Context(const cl_context& context) : detail::Wrapper(context) { } + + /*! \brief Assignment operator from Context. + * + * This calls clRetainContext() on the parameter and clReleaseContext() on + * the previous value held by this instance. + */ + Context& operator = (const Context& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_context - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseContext() on the value previously held by this instance. + */ + Context& operator = (const cl_context& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetContextInfo(). + template + cl_int getInfo(cl_context_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetContextInfo, object_, name, param), + __GET_CONTEXT_INFO_ERR); + } + + //! \brief Wrapper for clGetContextInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_context_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Gets a list of supported image formats. + * + * Wraps clGetSupportedImageFormats(). + */ + cl_int getSupportedImageFormats( + cl_mem_flags flags, + cl_mem_object_type type, + VECTOR_CLASS* formats) const + { + cl_uint numEntries; + cl_int err = ::clGetSupportedImageFormats( + object_, + flags, + type, + 0, + NULL, + &numEntries); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); + } + + ImageFormat* value = (ImageFormat*) + alloca(numEntries * sizeof(ImageFormat)); + err = ::clGetSupportedImageFormats( + object_, + flags, + type, + numEntries, + (cl_image_format*) value, + NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); + } + + formats->assign(&value[0], &value[numEntries]); + return CL_SUCCESS; + } +}; + +inline Device Device::getDefault(cl_int * err) +{ + cl_int error; + Device device; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + device = context.getInfo()[0]; + if (err != NULL) { + *err = CL_SUCCESS; + } + } + + return device; +} + + +#ifdef _WIN32 +__declspec(selectany) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__declspec(selectany) Context Context::default_; +__declspec(selectany) volatile cl_int Context::default_error_ = CL_SUCCESS; +#else +__attribute__((weak)) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__attribute__((weak)) Context Context::default_; +__attribute__((weak)) volatile cl_int Context::default_error_ = CL_SUCCESS; +#endif + +/*! \brief Class interface for cl_event. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_event as the original. For details, see + * clRetainEvent() and clReleaseEvent(). + * + * \see cl_event + */ +class Event : public detail::Wrapper +{ +public: + /*! \brief Destructor. + * + * This calls clReleaseEvent() on the value held by this instance. + */ + ~Event() { } + + //! \brief Default constructor - initializes to NULL. + Event() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This calls clRetainEvent() on the parameter's cl_event. + */ + Event(const Event& event) : detail::Wrapper(event) { } + + /*! \brief Constructor from cl_event - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_event + * into the new Event object. + */ + Event(const cl_event& event) : detail::Wrapper(event) { } + + /*! \brief Assignment operator from cl_event - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseEvent() on the value previously held by this instance. + */ + Event& operator = (const Event& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_event. + * + * This calls clRetainEvent() on the parameter and clReleaseEvent() on + * the previous value held by this instance. + */ + Event& operator = (const cl_event& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetEventInfo(). + template + cl_int getInfo(cl_event_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetEventInfo, object_, name, param), + __GET_EVENT_INFO_ERR); + } + + //! \brief Wrapper for clGetEventInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_event_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + //! \brief Wrapper for clGetEventProfilingInfo(). + template + cl_int getProfilingInfo(cl_profiling_info name, T* param) const + { + return detail::errHandler(detail::getInfo( + &::clGetEventProfilingInfo, object_, name, param), + __GET_EVENT_PROFILE_INFO_ERR); + } + + //! \brief Wrapper for clGetEventProfilingInfo() that returns by value. + template typename + detail::param_traits::param_type + getProfilingInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_profiling_info, name>::param_type param; + cl_int result = getProfilingInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Blocks the calling thread until this event completes. + * + * Wraps clWaitForEvents(). + */ + cl_int wait() const + { + return detail::errHandler( + ::clWaitForEvents(1, &object_), + __WAIT_FOR_EVENTS_ERR); + } + +#if defined(CL_VERSION_1_1) + /*! \brief Registers a user callback function for a specific command execution status. + * + * Wraps clSetEventCallback(). + */ + cl_int setCallback( + cl_int type, + void (CL_CALLBACK * pfn_notify)(cl_event, cl_int, void *), + void * user_data = NULL) + { + return detail::errHandler( + ::clSetEventCallback( + object_, + type, + pfn_notify, + user_data), + __SET_EVENT_CALLBACK_ERR); + } +#endif + + /*! \brief Blocks the calling thread until every event specified is complete. + * + * Wraps clWaitForEvents(). + */ + static cl_int + waitForEvents(const VECTOR_CLASS& events) + { + return detail::errHandler( + ::clWaitForEvents( + (cl_uint) events.size(), (cl_event*)&events.front()), + __WAIT_FOR_EVENTS_ERR); + } +}; + +#if defined(CL_VERSION_1_1) +/*! \brief Class interface for user events (a subset of cl_event's). + * + * See Event for details about copy semantics, etc. + */ +class UserEvent : public Event +{ +public: + /*! \brief Constructs a user event on a given context. + * + * Wraps clCreateUserEvent(). + */ + UserEvent( + const Context& context, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateUserEvent( + context(), + &error); + + detail::errHandler(error, __CREATE_USER_EVENT_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + UserEvent() : Event() { } + + //! \brief Copy constructor - performs shallow copy. + UserEvent(const UserEvent& event) : Event(event) { } + + //! \brief Assignment Operator - performs shallow copy. + UserEvent& operator = (const UserEvent& rhs) + { + if (this != &rhs) { + Event::operator=(rhs); + } + return *this; + } + + /*! \brief Sets the execution status of a user event object. + * + * Wraps clSetUserEventStatus(). + */ + cl_int setStatus(cl_int status) + { + return detail::errHandler( + ::clSetUserEventStatus(object_,status), + __SET_USER_EVENT_STATUS_ERR); + } +}; +#endif + +/*! \brief Blocks the calling thread until every event specified is complete. + * + * Wraps clWaitForEvents(). + */ +inline static cl_int +WaitForEvents(const VECTOR_CLASS& events) +{ + return detail::errHandler( + ::clWaitForEvents( + (cl_uint) events.size(), (cl_event*)&events.front()), + __WAIT_FOR_EVENTS_ERR); +} + +/*! \brief Class interface for cl_mem. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_mem as the original. For details, see + * clRetainMemObject() and clReleaseMemObject(). + * + * \see cl_mem + */ +class Memory : public detail::Wrapper +{ +public: + + /*! \brief Destructor. + * + * This calls clReleaseMemObject() on the value held by this instance. + */ + ~Memory() {} + + //! \brief Default constructor - initializes to NULL. + Memory() : detail::Wrapper() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainMemObject() on the parameter's cl_mem. + */ + Memory(const Memory& memory) : detail::Wrapper(memory) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_mem + * into the new Memory object. + */ + __CL_EXPLICIT_CONSTRUCTORS Memory(const cl_mem& memory) : detail::Wrapper(memory) { } + + /*! \brief Assignment operator from Memory. + * + * This calls clRetainMemObject() on the parameter and clReleaseMemObject() + * on the previous value held by this instance. + */ + Memory& operator = (const Memory& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_mem - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseMemObject() on the value previously held by this instance. + */ + Memory& operator = (const cl_mem& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetMemObjectInfo(). + template + cl_int getInfo(cl_mem_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetMemObjectInfo, object_, name, param), + __GET_MEM_OBJECT_INFO_ERR); + } + + //! \brief Wrapper for clGetMemObjectInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_mem_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + +#if defined(CL_VERSION_1_1) + /*! \brief Registers a callback function to be called when the memory object + * is no longer needed. + * + * Wraps clSetMemObjectDestructorCallback(). + * + * Repeated calls to this function, for a given cl_mem value, will append + * to the list of functions called (in reverse order) when memory object's + * resources are freed and the memory object is deleted. + * + * \note + * The registered callbacks are associated with the underlying cl_mem + * value - not the Memory class instance. + */ + cl_int setDestructorCallback( + void (CL_CALLBACK * pfn_notify)(cl_mem, void *), + void * user_data = NULL) + { + return detail::errHandler( + ::clSetMemObjectDestructorCallback( + object_, + pfn_notify, + user_data), + __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR); + } +#endif + +}; + +// Pre-declare copy functions +class Buffer; +template< typename IteratorType > +cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ); +template< typename IteratorType > +cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ); + +/*! \brief Class interface for Buffer Memory Objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Buffer : public Memory +{ +public: + + /*! \brief Constructs a Buffer in a specified context. + * + * Wraps clCreateBuffer(). + * + * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was + * specified. Note alignment & exclusivity requirements. + */ + Buffer( + const Context& context, + cl_mem_flags flags, + ::size_t size, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Constructs a Buffer in the default context. + * + * Wraps clCreateBuffer(). + * + * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was + * specified. Note alignment & exclusivity requirements. + * + * \see Context::getDefault() + */ + Buffer( + cl_mem_flags flags, + ::size_t size, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + + Context context = Context::getDefault(err); + + object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! + * \brief Construct a Buffer from a host container via iterators. + * If useHostPtr is specified iterators must be random access. + */ + template< typename IteratorType > + Buffer( + IteratorType startIterator, + IteratorType endIterator, + bool readOnly, + bool useHostPtr = false, + cl_int* err = NULL) + { + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + cl_mem_flags flags = 0; + if( readOnly ) { + flags |= CL_MEM_READ_ONLY; + } + else { + flags |= CL_MEM_READ_WRITE; + } + if( useHostPtr ) { + flags |= CL_MEM_USE_HOST_PTR; + } + + ::size_t size = sizeof(DataType)*(endIterator - startIterator); + + Context context = Context::getDefault(err); + + if( useHostPtr ) { + object_ = ::clCreateBuffer(context(), flags, size, static_cast(&*startIterator), &error); + } else { + object_ = ::clCreateBuffer(context(), flags, size, 0, &error); + } + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + if( !useHostPtr ) { + error = cl::copy(startIterator, endIterator, *this); + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + } + + //! \brief Default constructor - initializes to NULL. + Buffer() : Memory() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Buffer(const Buffer& buffer) : Memory(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Buffer(const cl_mem& buffer) : Memory(buffer) { } + + /*! \brief Assignment from Buffer - performs shallow copy. + * + * See Memory for further details. + */ + Buffer& operator = (const Buffer& rhs) + { + if (this != &rhs) { + Memory::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Buffer& operator = (const cl_mem& rhs) + { + Memory::operator=(rhs); + return *this; + } + +#if defined(CL_VERSION_1_1) + /*! \brief Creates a new buffer object from this. + * + * Wraps clCreateSubBuffer(). + */ + Buffer createSubBuffer( + cl_mem_flags flags, + cl_buffer_create_type buffer_create_type, + const void * buffer_create_info, + cl_int * err = NULL) + { + Buffer result; + cl_int error; + result.object_ = ::clCreateSubBuffer( + object_, + flags, + buffer_create_type, + buffer_create_info, + &error); + + detail::errHandler(error, __CREATE_SUBBUFFER_ERR); + if (err != NULL) { + *err = error; + } + + return result; + } +#endif +}; + +#if defined (USE_DX_INTEROP) +/*! \brief Class interface for creating OpenCL buffers from ID3D10Buffer's. + * + * This is provided to facilitate interoperability with Direct3D. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferD3D10 : public Buffer +{ +public: + typedef CL_API_ENTRY cl_mem (CL_API_CALL *PFN_clCreateFromD3D10BufferKHR)( + cl_context context, cl_mem_flags flags, ID3D10Buffer* buffer, + cl_int* errcode_ret); + + /*! \brief Constructs a BufferD3D10, in a specified context, from a + * given ID3D10Buffer. + * + * Wraps clCreateFromD3D10BufferKHR(). + */ + BufferD3D10( + const Context& context, + cl_mem_flags flags, + ID3D10Buffer* bufobj, + cl_int * err = NULL) + { + static PFN_clCreateFromD3D10BufferKHR pfn_clCreateFromD3D10BufferKHR = NULL; + +#if defined(CL_VERSION_1_2) + vector props = context.getInfo(); + cl_platform platform = -1; + for( int i = 0; i < props.size(); ++i ) { + if( props[i] == CL_CONTEXT_PLATFORM ) { + platform = props[i+1]; + } + } + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clCreateFromD3D10BufferKHR); +#endif +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clCreateFromD3D10BufferKHR); +#endif + + cl_int error; + object_ = pfn_clCreateFromD3D10BufferKHR( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferD3D10() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10(const BufferD3D10& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferD3D10(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferD3D10 - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10& operator = (const BufferD3D10& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } +}; +#endif + +/*! \brief Class interface for GL Buffer Memory Objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferGL : public Buffer +{ +public: + /*! \brief Constructs a BufferGL in a specified context, from a given + * GL buffer. + * + * Wraps clCreateFromGLBuffer(). + */ + BufferGL( + const Context& context, + cl_mem_flags flags, + GLuint bufobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLBuffer( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferGL() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL(const BufferGL& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferGL(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferGL - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL& operator = (const BufferGL& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetGLObjectInfo(). + cl_int getObjectInfo( + cl_gl_object_type *type, + GLuint * gl_object_name) + { + return detail::errHandler( + ::clGetGLObjectInfo(object_,type,gl_object_name), + __GET_GL_OBJECT_INFO_ERR); + } +}; + +/*! \brief Class interface for GL Render Buffer Memory Objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferRenderGL : public Buffer +{ +public: + /*! \brief Constructs a BufferRenderGL in a specified context, from a given + * GL Renderbuffer. + * + * Wraps clCreateFromGLRenderbuffer(). + */ + BufferRenderGL( + const Context& context, + cl_mem_flags flags, + GLuint bufobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLRenderbuffer( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_RENDER_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferRenderGL() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL(const BufferGL& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferRenderGL(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferGL - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL& operator = (const BufferRenderGL& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetGLObjectInfo(). + cl_int getObjectInfo( + cl_gl_object_type *type, + GLuint * gl_object_name) + { + return detail::errHandler( + ::clGetGLObjectInfo(object_,type,gl_object_name), + __GET_GL_OBJECT_INFO_ERR); + } +}; + +/*! \brief C++ base class for Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image : public Memory +{ +protected: + //! \brief Default constructor - initializes to NULL. + Image() : Memory() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image(const Image& image) : Memory(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image(const cl_mem& image) : Memory(image) { } + + /*! \brief Assignment from Image - performs shallow copy. + * + * See Memory for further details. + */ + Image& operator = (const Image& rhs) + { + if (this != &rhs) { + Memory::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image& operator = (const cl_mem& rhs) + { + Memory::operator=(rhs); + return *this; + } + +public: + //! \brief Wrapper for clGetImageInfo(). + template + cl_int getImageInfo(cl_image_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetImageInfo, object_, name, param), + __GET_IMAGE_INFO_ERR); + } + + //! \brief Wrapper for clGetImageInfo() that returns by value. + template typename + detail::param_traits::param_type + getImageInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_image_info, name>::param_type param; + cl_int result = getImageInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +}; + +#if defined(CL_VERSION_1_2) +/*! \brief Class interface for 1D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image1D : public Image +{ +public: + /*! \brief Constructs a 1D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image1D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE1D; + desc.image_width = width; + desc.image_row_pitch = 0; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + Image1D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image1D(const Image1D& image1D) : Image(image1D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image1D(const cl_mem& image1D) : Image(image1D) { } + + /*! \brief Assignment from Image1D - performs shallow copy. + * + * See Memory for further details. + */ + Image1D& operator = (const Image1D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image1D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +/*! \class Image1DBuffer + * \brief Image interface for 1D buffer images. + */ +class Image1DBuffer : public Image +{ +public: + Image1DBuffer( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + const Buffer &buffer, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE1D_BUFFER; + desc.image_width = width; + desc.image_row_pitch = 0; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = buffer(); + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + NULL, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image1DBuffer() { } + + Image1DBuffer(const Image1DBuffer& image1D) : Image(image1D) { } + + __CL_EXPLICIT_CONSTRUCTORS Image1DBuffer(const cl_mem& image1D) : Image(image1D) { } + + Image1DBuffer& operator = (const Image1DBuffer& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image1DBuffer& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +/*! \class Image1DArray + * \brief Image interface for arrays of 1D images. + */ +class Image1DArray : public Image +{ +public: + Image1DArray( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t arraySize, + ::size_t width, + ::size_t rowPitch, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE1D_ARRAY; + desc.image_array_size = arraySize; + desc.image_width = width; + desc.image_row_pitch = rowPitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image1DArray() { } + + Image1DArray(const Image1DArray& imageArray) : Image(imageArray) { } + + __CL_EXPLICIT_CONSTRUCTORS Image1DArray(const cl_mem& imageArray) : Image(imageArray) { } + + Image1DArray& operator = (const Image1DArray& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image1DArray& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + + +/*! \brief Class interface for 2D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image2D : public Image +{ +public: + /*! \brief Constructs a 1D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image2D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + ::size_t height, + ::size_t row_pitch = 0, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + bool useCreateImage; + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + // Run-time decision based on the actual platform + { + cl_uint version = detail::getContextPlatformVersion(context()); + useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above + } +#elif defined(CL_VERSION_1_2) + useCreateImage = true; +#else + useCreateImage = false; +#endif + +#if defined(CL_VERSION_1_2) + if (useCreateImage) + { + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE2D; + desc.image_width = width; + desc.image_height = height; + desc.image_row_pitch = row_pitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) +#if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + if (!useCreateImage) + { + object_ = ::clCreateImage2D( + context(), flags,&format, width, height, row_pitch, host_ptr, &error); + + detail::errHandler(error, __CREATE_IMAGE2D_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + } + + //! \brief Default constructor - initializes to NULL. + Image2D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image2D(const Image2D& image2D) : Image(image2D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image2D(const cl_mem& image2D) : Image(image2D) { } + + /*! \brief Assignment from Image2D - performs shallow copy. + * + * See Memory for further details. + */ + Image2D& operator = (const Image2D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image2D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + + +#if !defined(CL_VERSION_1_2) +/*! \brief Class interface for GL 2D Image Memory objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + * \note Deprecated for OpenCL 1.2. Please use ImageGL instead. + */ +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED Image2DGL CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED : public Image2D +{ +public: + /*! \brief Constructs an Image2DGL in a specified context, from a given + * GL Texture. + * + * Wraps clCreateFromGLTexture2D(). + */ + Image2DGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture2D( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_2D_ERR); + if (err != NULL) { + *err = error; + } + + } + + //! \brief Default constructor - initializes to NULL. + Image2DGL() : Image2D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL(const Image2DGL& image) : Image2D(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image2DGL(const cl_mem& image) : Image2D(image) { } + + /*! \brief Assignment from Image2DGL - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL& operator = (const Image2DGL& rhs) + { + if (this != &rhs) { + Image2D::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL& operator = (const cl_mem& rhs) + { + Image2D::operator=(rhs); + return *this; + } +}; +#endif // #if !defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_1_2) +/*! \class Image2DArray + * \brief Image interface for arrays of 2D images. + */ +class Image2DArray : public Image +{ +public: + Image2DArray( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t arraySize, + ::size_t width, + ::size_t height, + ::size_t rowPitch, + ::size_t slicePitch, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE2D_ARRAY; + desc.image_array_size = arraySize; + desc.image_width = width; + desc.image_height = height; + desc.image_row_pitch = rowPitch; + desc.image_slice_pitch = slicePitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image2DArray() { } + + Image2DArray(const Image2DArray& imageArray) : Image(imageArray) { } + + __CL_EXPLICIT_CONSTRUCTORS Image2DArray(const cl_mem& imageArray) : Image(imageArray) { } + + Image2DArray& operator = (const Image2DArray& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image2DArray& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + +/*! \brief Class interface for 3D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image3D : public Image +{ +public: + /*! \brief Constructs a 3D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image3D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + ::size_t height, + ::size_t depth, + ::size_t row_pitch = 0, + ::size_t slice_pitch = 0, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + bool useCreateImage; + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + // Run-time decision based on the actual platform + { + cl_uint version = detail::getContextPlatformVersion(context()); + useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above + } +#elif defined(CL_VERSION_1_2) + useCreateImage = true; +#else + useCreateImage = false; +#endif + +#if defined(CL_VERSION_1_2) + if (useCreateImage) + { + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE3D; + desc.image_width = width; + desc.image_height = height; + desc.image_depth = depth; + desc.image_row_pitch = row_pitch; + desc.image_slice_pitch = slice_pitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) +#if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + if (!useCreateImage) + { + object_ = ::clCreateImage3D( + context(), flags, &format, width, height, depth, row_pitch, + slice_pitch, host_ptr, &error); + + detail::errHandler(error, __CREATE_IMAGE3D_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + } + + //! \brief Default constructor - initializes to NULL. + Image3D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image3D(const Image3D& image3D) : Image(image3D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image3D(const cl_mem& image3D) : Image(image3D) { } + + /*! \brief Assignment from Image3D - performs shallow copy. + * + * See Memory for further details. + */ + Image3D& operator = (const Image3D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image3D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +#if !defined(CL_VERSION_1_2) +/*! \brief Class interface for GL 3D Image Memory objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image3DGL : public Image3D +{ +public: + /*! \brief Constructs an Image3DGL in a specified context, from a given + * GL Texture. + * + * Wraps clCreateFromGLTexture3D(). + */ + Image3DGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture3D( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_3D_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + Image3DGL() : Image3D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL(const Image3DGL& image) : Image3D(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image3DGL(const cl_mem& image) : Image3D(image) { } + + /*! \brief Assignment from Image3DGL - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL& operator = (const Image3DGL& rhs) + { + if (this != &rhs) { + Image3D::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL& operator = (const cl_mem& rhs) + { + Image3D::operator=(rhs); + return *this; + } +}; +#endif // #if !defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_1_2) +/*! \class ImageGL + * \brief general image interface for GL interop. + * We abstract the 2D and 3D GL images into a single instance here + * that wraps all GL sourced images on the grounds that setup information + * was performed by OpenCL anyway. + */ +class ImageGL : public Image +{ +public: + ImageGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_ERR); + if (err != NULL) { + *err = error; + } + } + + ImageGL() : Image() { } + + ImageGL(const ImageGL& image) : Image(image) { } + + __CL_EXPLICIT_CONSTRUCTORS ImageGL(const cl_mem& image) : Image(image) { } + + ImageGL& operator = (const ImageGL& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + ImageGL& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + +/*! \brief Class interface for cl_sampler. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_sampler as the original. For details, see + * clRetainSampler() and clReleaseSampler(). + * + * \see cl_sampler + */ +class Sampler : public detail::Wrapper +{ +public: + /*! \brief Destructor. + * + * This calls clReleaseSampler() on the value held by this instance. + */ + ~Sampler() { } + + //! \brief Default constructor - initializes to NULL. + Sampler() { } + + /*! \brief Constructs a Sampler in a specified context. + * + * Wraps clCreateSampler(). + */ + Sampler( + const Context& context, + cl_bool normalized_coords, + cl_addressing_mode addressing_mode, + cl_filter_mode filter_mode, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateSampler( + context(), + normalized_coords, + addressing_mode, + filter_mode, + &error); + + detail::errHandler(error, __CREATE_SAMPLER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainSampler() on the parameter's cl_sampler. + */ + Sampler(const Sampler& sampler) : detail::Wrapper(sampler) { } + + /*! \brief Constructor from cl_sampler - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_sampler + * into the new Sampler object. + */ + Sampler(const cl_sampler& sampler) : detail::Wrapper(sampler) { } + + /*! \brief Assignment operator from Sampler. + * + * This calls clRetainSampler() on the parameter and clReleaseSampler() + * on the previous value held by this instance. + */ + Sampler& operator = (const Sampler& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_sampler - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseSampler() on the value previously held by this instance. + */ + Sampler& operator = (const cl_sampler& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetSamplerInfo(). + template + cl_int getInfo(cl_sampler_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetSamplerInfo, object_, name, param), + __GET_SAMPLER_INFO_ERR); + } + + //! \brief Wrapper for clGetSamplerInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_sampler_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +}; + +class Program; +class CommandQueue; +class Kernel; + +//! \brief Class interface for specifying NDRange values. +class NDRange +{ +private: + size_t<3> sizes_; + cl_uint dimensions_; + +public: + //! \brief Default constructor - resulting range has zero dimensions. + NDRange() + : dimensions_(0) + { } + + //! \brief Constructs one-dimensional range. + NDRange(::size_t size0) + : dimensions_(1) + { + sizes_[0] = size0; + } + + //! \brief Constructs two-dimensional range. + NDRange(::size_t size0, ::size_t size1) + : dimensions_(2) + { + sizes_[0] = size0; + sizes_[1] = size1; + } + + //! \brief Constructs three-dimensional range. + NDRange(::size_t size0, ::size_t size1, ::size_t size2) + : dimensions_(3) + { + sizes_[0] = size0; + sizes_[1] = size1; + sizes_[2] = size2; + } + + /*! \brief Conversion operator to const ::size_t *. + * + * \returns a pointer to the size of the first dimension. + */ + operator const ::size_t*() const { + return (const ::size_t*) sizes_; + } + + //! \brief Queries the number of dimensions in the range. + ::size_t dimensions() const { return dimensions_; } +}; + +//! \brief A zero-dimensional range. +static const NDRange NullRange; + +//! \brief Local address wrapper for use with Kernel::setArg +struct LocalSpaceArg +{ + ::size_t size_; +}; + +namespace detail { + +template +struct KernelArgumentHandler +{ + static ::size_t size(const T&) { return sizeof(T); } + static T* ptr(T& value) { return &value; } +}; + +template <> +struct KernelArgumentHandler +{ + static ::size_t size(const LocalSpaceArg& value) { return value.size_; } + static void* ptr(LocalSpaceArg&) { return NULL; } +}; + +} +//! \endcond + +/*! __local + * \brief Helper function for generating LocalSpaceArg objects. + * Deprecated. Replaced with Local. + */ +inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED LocalSpaceArg +__local(::size_t size) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; +inline LocalSpaceArg +__local(::size_t size) +{ + LocalSpaceArg ret = { size }; + return ret; +} + +/*! Local + * \brief Helper function for generating LocalSpaceArg objects. + */ +inline LocalSpaceArg +Local(::size_t size) +{ + LocalSpaceArg ret = { size }; + return ret; +} + +//class KernelFunctor; + +/*! \brief Class interface for cl_kernel. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_kernel as the original. For details, see + * clRetainKernel() and clReleaseKernel(). + * + * \see cl_kernel + */ +class Kernel : public detail::Wrapper +{ +public: + inline Kernel(const Program& program, const char* name, cl_int* err = NULL); + + /*! \brief Destructor. + * + * This calls clReleaseKernel() on the value held by this instance. + */ + ~Kernel() { } + + //! \brief Default constructor - initializes to NULL. + Kernel() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainKernel() on the parameter's cl_kernel. + */ + Kernel(const Kernel& kernel) : detail::Wrapper(kernel) { } + + /*! \brief Constructor from cl_kernel - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_kernel + * into the new Kernel object. + */ + __CL_EXPLICIT_CONSTRUCTORS Kernel(const cl_kernel& kernel) : detail::Wrapper(kernel) { } + + /*! \brief Assignment operator from Kernel. + * + * This calls clRetainKernel() on the parameter and clReleaseKernel() + * on the previous value held by this instance. + */ + Kernel& operator = (const Kernel& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_kernel - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseKernel() on the value previously held by this instance. + */ + Kernel& operator = (const cl_kernel& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + template + cl_int getInfo(cl_kernel_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetKernelInfo, object_, name, param), + __GET_KERNEL_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + +#if defined(CL_VERSION_1_2) + template + cl_int getArgInfo(cl_uint argIndex, cl_kernel_arg_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetKernelArgInfo, object_, argIndex, name, param), + __GET_KERNEL_ARG_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getArgInfo(cl_uint argIndex, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_arg_info, name>::param_type param; + cl_int result = getArgInfo(argIndex, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +#endif // #if defined(CL_VERSION_1_2) + + template + cl_int getWorkGroupInfo( + const Device& device, cl_kernel_work_group_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetKernelWorkGroupInfo, object_, device(), name, param), + __GET_KERNEL_WORK_GROUP_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getWorkGroupInfo(const Device& device, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_work_group_info, name>::param_type param; + cl_int result = getWorkGroupInfo(device, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + template + cl_int setArg(cl_uint index, T value) + { + return detail::errHandler( + ::clSetKernelArg( + object_, + index, + detail::KernelArgumentHandler::size(value), + detail::KernelArgumentHandler::ptr(value)), + __SET_KERNEL_ARGS_ERR); + } + + cl_int setArg(cl_uint index, ::size_t size, void* argPtr) + { + return detail::errHandler( + ::clSetKernelArg(object_, index, size, argPtr), + __SET_KERNEL_ARGS_ERR); + } +}; + +/*! \class Program + * \brief Program interface that implements cl_program. + */ +class Program : public detail::Wrapper +{ +public: + typedef VECTOR_CLASS > Binaries; + typedef VECTOR_CLASS > Sources; + + Program( + const STRING_CLASS& source, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + Context context = Context::getDefault(err); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const STRING_CLASS& source, + bool build, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + Context context = Context::getDefault(err); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS && build) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const Context& context, + const STRING_CLASS& source, + bool build = false, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS && build) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const Context& context, + const Sources& sources, + cl_int* err = NULL) + { + cl_int error; + + const ::size_t n = (::size_t)sources.size(); + ::size_t* lengths = (::size_t*) alloca(n * sizeof(::size_t)); + const char** strings = (const char**) alloca(n * sizeof(const char*)); + + for (::size_t i = 0; i < n; ++i) { + strings[i] = sources[(int)i].first; + lengths[i] = sources[(int)i].second; + } + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)n, strings, lengths, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + if (err != NULL) { + *err = error; + } + } + + /** + * Construct a program object from a list of devices and a per-device list of binaries. + * \param context A valid OpenCL context in which to construct the program. + * \param devices A vector of OpenCL device objects for which the program will be created. + * \param binaries A vector of pairs of a pointer to a binary object and its length. + * \param binaryStatus An optional vector that on completion will be resized to + * match the size of binaries and filled with values to specify if each binary + * was successfully loaded. + * Set to CL_SUCCESS if the binary was successfully loaded. + * Set to CL_INVALID_VALUE if the length is 0 or the binary pointer is NULL. + * Set to CL_INVALID_BINARY if the binary provided is not valid for the matching device. + * \param err if non-NULL will be set to CL_SUCCESS on successful operation or one of the following errors: + * CL_INVALID_CONTEXT if context is not a valid context. + * CL_INVALID_VALUE if the length of devices is zero; or if the length of binaries does not match the length of devices; + * or if any entry in binaries is NULL or has length 0. + * CL_INVALID_DEVICE if OpenCL devices listed in devices are not in the list of devices associated with context. + * CL_INVALID_BINARY if an invalid program binary was encountered for any device. binaryStatus will return specific status for each device. + * CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the OpenCL implementation on the host. + */ + Program( + const Context& context, + const VECTOR_CLASS& devices, + const Binaries& binaries, + VECTOR_CLASS* binaryStatus = NULL, + cl_int* err = NULL) + { + cl_int error; + + const ::size_t numDevices = devices.size(); + + // Catch size mismatch early and return + if(binaries.size() != numDevices) { + error = CL_INVALID_VALUE; + detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); + if (err != NULL) { + *err = error; + } + return; + } + + ::size_t* lengths = (::size_t*) alloca(numDevices * sizeof(::size_t)); + const unsigned char** images = (const unsigned char**) alloca(numDevices * sizeof(const unsigned char**)); + + for (::size_t i = 0; i < numDevices; ++i) { + images[i] = (const unsigned char*)binaries[i].first; + lengths[i] = binaries[(int)i].second; + } + + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + if(binaryStatus) { + binaryStatus->resize(numDevices); + } + + object_ = ::clCreateProgramWithBinary( + context(), (cl_uint) devices.size(), + deviceIDs, + lengths, images, binaryStatus != NULL + ? &binaryStatus->front() + : NULL, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); + if (err != NULL) { + *err = error; + } + } + + +#if defined(CL_VERSION_1_2) + /** + * Create program using builtin kernels. + * \param kernelNames Semi-colon separated list of builtin kernel names + */ + Program( + const Context& context, + const VECTOR_CLASS& devices, + const STRING_CLASS& kernelNames, + cl_int* err = NULL) + { + cl_int error; + + + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + object_ = ::clCreateProgramWithBuiltInKernels( + context(), + (cl_uint) devices.size(), + deviceIDs, + kernelNames.c_str(), + &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) + + Program() { } + + Program(const Program& program) : detail::Wrapper(program) { } + + __CL_EXPLICIT_CONSTRUCTORS Program(const cl_program& program) : detail::Wrapper(program) { } + + Program& operator = (const Program& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + Program& operator = (const cl_program& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + cl_int build( + const VECTOR_CLASS& devices, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + return detail::errHandler( + ::clBuildProgram( + object_, + (cl_uint) + devices.size(), + deviceIDs, + options, + notifyFptr, + data), + __BUILD_PROGRAM_ERR); + } + + cl_int build( + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + return detail::errHandler( + ::clBuildProgram( + object_, + 0, + NULL, + options, + notifyFptr, + data), + __BUILD_PROGRAM_ERR); + } + +#if defined(CL_VERSION_1_2) + cl_int compile( + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + return detail::errHandler( + ::clCompileProgram( + object_, + 0, + NULL, + options, + 0, + NULL, + NULL, + notifyFptr, + data), + __COMPILE_PROGRAM_ERR); + } +#endif + + template + cl_int getInfo(cl_program_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetProgramInfo, object_, name, param), + __GET_PROGRAM_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_program_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + template + cl_int getBuildInfo( + const Device& device, cl_program_build_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetProgramBuildInfo, object_, device(), name, param), + __GET_PROGRAM_BUILD_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getBuildInfo(const Device& device, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_program_build_info, name>::param_type param; + cl_int result = getBuildInfo(device, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + cl_int createKernels(VECTOR_CLASS* kernels) + { + cl_uint numKernels; + cl_int err = ::clCreateKernelsInProgram(object_, 0, NULL, &numKernels); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); + } + + Kernel* value = (Kernel*) alloca(numKernels * sizeof(Kernel)); + err = ::clCreateKernelsInProgram( + object_, numKernels, (cl_kernel*) value, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); + } + + kernels->assign(&value[0], &value[numKernels]); + return CL_SUCCESS; + } +}; + +#if defined(CL_VERSION_1_2) +inline Program linkProgram( + Program input1, + Program input2, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL, + cl_int* err = NULL) +{ + cl_int err_local = CL_SUCCESS; + + cl_program programs[2] = { input1(), input2() }; + + Context ctx = input1.getInfo(); + + cl_program prog = ::clLinkProgram( + ctx(), + 0, + NULL, + options, + 2, + programs, + notifyFptr, + data, + &err_local); + + detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); + if (err != NULL) { + *err = err_local; + } + + return Program(prog); +} + +inline Program linkProgram( + VECTOR_CLASS inputPrograms, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL, + cl_int* err = NULL) +{ + cl_int err_local = CL_SUCCESS; + + cl_program * programs = (cl_program*) alloca(inputPrograms.size() * sizeof(cl_program)); + + if (programs != NULL) { + for (unsigned int i = 0; i < inputPrograms.size(); i++) { + programs[i] = inputPrograms[i](); + } + } + + cl_program prog = ::clLinkProgram( + Context::getDefault()(), + 0, + NULL, + options, + (cl_uint)inputPrograms.size(), + programs, + notifyFptr, + data, + &err_local); + + detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); + if (err != NULL) { + *err = err_local; + } + + return Program(prog); +} +#endif + +template<> +inline VECTOR_CLASS cl::Program::getInfo(cl_int* err) const +{ + VECTOR_CLASS< ::size_t> sizes = getInfo(); + VECTOR_CLASS binaries; + for (VECTOR_CLASS< ::size_t>::iterator s = sizes.begin(); s != sizes.end(); ++s) + { + char *ptr = NULL; + if (*s != 0) + ptr = new char[*s]; + binaries.push_back(ptr); + } + + cl_int result = getInfo(CL_PROGRAM_BINARIES, &binaries); + if (err != NULL) { + *err = result; + } + return binaries; +} + +inline Kernel::Kernel(const Program& program, const char* name, cl_int* err) +{ + cl_int error; + + object_ = ::clCreateKernel(program(), name, &error); + detail::errHandler(error, __CREATE_KERNEL_ERR); + + if (err != NULL) { + *err = error; + } + +} + +/*! \class CommandQueue + * \brief CommandQueue interface for cl_command_queue. + */ +class CommandQueue : public detail::Wrapper +{ +private: + static volatile int default_initialized_; + static CommandQueue default_; + static volatile cl_int default_error_; +public: + CommandQueue( + cl_command_queue_properties properties, + cl_int* err = NULL) + { + cl_int error; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + Device device = context.getInfo()[0]; + + object_ = ::clCreateCommandQueue( + context(), device(), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + } + + CommandQueue( + const Context& context, + const Device& device, + cl_command_queue_properties properties = 0, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateCommandQueue( + context(), device(), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + + static CommandQueue getDefault(cl_int * err = NULL) + { + int state = detail::compare_exchange( + &default_initialized_, + __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); + + if (state & __DEFAULT_INITIALIZED) { + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + if (state & __DEFAULT_BEING_INITIALIZED) { + // Assume writes will propagate eventually... + while(default_initialized_ != __DEFAULT_INITIALIZED) { + detail::fence(); + } + + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + cl_int error; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + Device device = context.getInfo()[0]; + + default_ = CommandQueue(context, device, 0, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + + detail::fence(); + + default_error_ = error; + // Assume writes will propagate eventually... + default_initialized_ = __DEFAULT_INITIALIZED; + + detail::fence(); + + if (err != NULL) { + *err = default_error_; + } + return default_; + + } + + CommandQueue() { } + + CommandQueue(const CommandQueue& commandQueue) : detail::Wrapper(commandQueue) { } + + CommandQueue(const cl_command_queue& commandQueue) : detail::Wrapper(commandQueue) { } + + CommandQueue& operator = (const CommandQueue& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + CommandQueue& operator = (const cl_command_queue& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + template + cl_int getInfo(cl_command_queue_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetCommandQueueInfo, object_, name, param), + __GET_COMMAND_QUEUE_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_command_queue_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + cl_int enqueueReadBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadBuffer( + object_, buffer(), blocking, offset, size, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + const void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteBuffer( + object_, buffer(), blocking, offset, size, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBuffer( + const Buffer& src, + const Buffer& dst, + ::size_t src_offset, + ::size_t dst_offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBuffer( + object_, src(), dst(), src_offset, dst_offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQEUE_COPY_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReadBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadBufferRect( + object_, + buffer(), + blocking, + (const ::size_t *)buffer_offset, + (const ::size_t *)host_offset, + (const ::size_t *)region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteBufferRect( + object_, + buffer(), + blocking, + (const ::size_t *)buffer_offset, + (const ::size_t *)host_offset, + (const ::size_t *)region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBufferRect( + const Buffer& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + ::size_t src_row_pitch, + ::size_t src_slice_pitch, + ::size_t dst_row_pitch, + ::size_t dst_slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBufferRect( + object_, + src(), + dst(), + (const ::size_t *)src_origin, + (const ::size_t *)dst_origin, + (const ::size_t *)region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQEUE_COPY_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueue a command to fill a buffer object with a pattern + * of a given size. The pattern is specified a as vector. + * \tparam PatternType The datatype of the pattern field. + * The pattern type must be an accepted OpenCL data type. + */ + template + cl_int enqueueFillBuffer( + const Buffer& buffer, + PatternType pattern, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillBuffer( + object_, + buffer(), + static_cast(&pattern), + sizeof(PatternType), + offset, + size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueReadImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadImage( + object_, image(), blocking, (const ::size_t *) origin, + (const ::size_t *) region, row_pitch, slice_pitch, ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteImage( + object_, image(), blocking, (const ::size_t *) origin, + (const ::size_t *) region, row_pitch, slice_pitch, ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyImage( + const Image& src, + const Image& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyImage( + object_, src(), dst(), (const ::size_t *) src_origin, + (const ::size_t *)dst_origin, (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA floating-point color value if + * the image channel data type is not an unnormalized signed or + * unsigned data type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_float4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA signed integer color value if + * the image channel data type is an unnormalized signed integer + * type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_int4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA unsigned integer color value if + * the image channel data type is an unnormalized unsigned integer + * type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_uint4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueCopyImageToBuffer( + const Image& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& region, + ::size_t dst_offset, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyImageToBuffer( + object_, src(), dst(), (const ::size_t *) src_origin, + (const ::size_t *) region, dst_offset, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBufferToImage( + const Buffer& src, + const Image& dst, + ::size_t src_offset, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBufferToImage( + object_, src(), dst(), src_offset, + (const ::size_t *) dst_origin, (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + void* enqueueMapBuffer( + const Buffer& buffer, + cl_bool blocking, + cl_map_flags flags, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) const + { + cl_int error; + void * result = ::clEnqueueMapBuffer( + object_, buffer(), blocking, flags, offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + return result; + } + + void* enqueueMapImage( + const Image& buffer, + cl_bool blocking, + cl_map_flags flags, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t * row_pitch, + ::size_t * slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) const + { + cl_int error; + void * result = ::clEnqueueMapImage( + object_, buffer(), blocking, flags, + (const ::size_t *) origin, (const ::size_t *) region, + row_pitch, slice_pitch, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + return result; + } + + cl_int enqueueUnmapMemObject( + const Memory& memory, + void* mapped_ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueUnmapMemObject( + object_, memory(), mapped_ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueues a marker command which waits for either a list of events to complete, + * or all previously enqueued commands to complete. + * + * Enqueues a marker command which waits for either a list of events to complete, + * or if the list is empty it waits for all commands previously enqueued in command_queue + * to complete before it completes. This command returns an event which can be waited on, + * i.e. this event can be waited on to insure that all events either in the event_wait_list + * or all previously enqueued commands, queued before this command to command_queue, + * have completed. + */ + cl_int enqueueMarkerWithWaitList( + const VECTOR_CLASS *events = 0, + Event *event = 0) + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueMarkerWithWaitList( + object_, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_MARKER_WAIT_LIST_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * A synchronization point that enqueues a barrier operation. + * + * Enqueues a barrier command which waits for either a list of events to complete, + * or if the list is empty it waits for all commands previously enqueued in command_queue + * to complete before it completes. This command blocks command execution, that is, any + * following commands enqueued after it do not execute until it completes. This command + * returns an event which can be waited on, i.e. this event can be waited on to insure that + * all events either in the event_wait_list or all previously enqueued commands, queued + * before this command to command_queue, have completed. + */ + cl_int enqueueBarrierWithWaitList( + const VECTOR_CLASS *events = 0, + Event *event = 0) + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueBarrierWithWaitList( + object_, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_BARRIER_WAIT_LIST_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueues a command to indicate with which device a set of memory objects + * should be associated. + */ + cl_int enqueueMigrateMemObjects( + const VECTOR_CLASS &memObjects, + cl_mem_migration_flags flags, + const VECTOR_CLASS* events = NULL, + Event* event = NULL + ) + { + cl_event tmp; + + cl_mem* localMemObjects = static_cast(alloca(memObjects.size() * sizeof(cl_mem))); + for( int i = 0; i < (int)memObjects.size(); ++i ) { + localMemObjects[i] = memObjects[i](); + } + + + cl_int err = detail::errHandler( + ::clEnqueueMigrateMemObjects( + object_, + (cl_uint)memObjects.size(), + static_cast(localMemObjects), + flags, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueNDRangeKernel( + const Kernel& kernel, + const NDRange& offset, + const NDRange& global, + const NDRange& local = NullRange, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueNDRangeKernel( + object_, kernel(), (cl_uint) global.dimensions(), + offset.dimensions() != 0 ? (const ::size_t*) offset : NULL, + (const ::size_t*) global, + local.dimensions() != 0 ? (const ::size_t*) local : NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_NDRANGE_KERNEL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueTask( + const Kernel& kernel, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueTask( + object_, kernel(), + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_TASK_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueNativeKernel( + void (CL_CALLBACK *userFptr)(void *), + std::pair args, + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* mem_locs = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_mem * mems = (mem_objects != NULL && mem_objects->size() > 0) + ? (cl_mem*) alloca(mem_objects->size() * sizeof(cl_mem)) + : NULL; + + if (mems != NULL) { + for (unsigned int i = 0; i < mem_objects->size(); i++) { + mems[i] = ((*mem_objects)[i])(); + } + } + + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueNativeKernel( + object_, userFptr, args.first, args.second, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + mems, + (mem_locs != NULL) ? (const void **) &mem_locs->front() : NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_NATIVE_KERNEL); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueMarker(Event* event = NULL) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueMarker(object_, (cl_event*) event), + __ENQUEUE_MARKER_ERR); + } + + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueWaitForEvents(const VECTOR_CLASS& events) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueWaitForEvents( + object_, + (cl_uint) events.size(), + (const cl_event*) &events.front()), + __ENQUEUE_WAIT_FOR_EVENTS_ERR); + } +#endif // #if defined(CL_VERSION_1_1) + + cl_int enqueueAcquireGLObjects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueAcquireGLObjects( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_ACQUIRE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReleaseGLObjects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReleaseGLObjects( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_RELEASE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined (USE_DX_INTEROP) +typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueAcquireD3D10ObjectsKHR)( + cl_command_queue command_queue, cl_uint num_objects, + const cl_mem* mem_objects, cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, cl_event* event); +typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)( + cl_command_queue command_queue, cl_uint num_objects, + const cl_mem* mem_objects, cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, cl_event* event); + + cl_int enqueueAcquireD3D10Objects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL; +#if defined(CL_VERSION_1_2) + cl_context context = getInfo(); + cl::Device device(getInfo()); + cl_platform_id platform = device.getInfo(); + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueAcquireD3D10ObjectsKHR); +#endif +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clEnqueueAcquireD3D10ObjectsKHR); +#endif + + cl_event tmp; + cl_int err = detail::errHandler( + pfn_clEnqueueAcquireD3D10ObjectsKHR( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_ACQUIRE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReleaseD3D10Objects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + static PFN_clEnqueueReleaseD3D10ObjectsKHR pfn_clEnqueueReleaseD3D10ObjectsKHR = NULL; +#if defined(CL_VERSION_1_2) + cl_context context = getInfo(); + cl::Device device(getInfo()); + cl_platform_id platform = device.getInfo(); + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueReleaseD3D10ObjectsKHR); +#endif // #if defined(CL_VERSION_1_2) +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clEnqueueReleaseD3D10ObjectsKHR); +#endif // #if defined(CL_VERSION_1_1) + + cl_event tmp; + cl_int err = detail::errHandler( + pfn_clEnqueueReleaseD3D10ObjectsKHR( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_RELEASE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueBarrier() const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueBarrier(object_), + __ENQUEUE_BARRIER_ERR); + } +#endif // #if defined(CL_VERSION_1_1) + + cl_int flush() const + { + return detail::errHandler(::clFlush(object_), __FLUSH_ERR); + } + + cl_int finish() const + { + return detail::errHandler(::clFinish(object_), __FINISH_ERR); + } +}; + +#ifdef _WIN32 +__declspec(selectany) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__declspec(selectany) CommandQueue CommandQueue::default_; +__declspec(selectany) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; +#else +__attribute__((weak)) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__attribute__((weak)) CommandQueue CommandQueue::default_; +__attribute__((weak)) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; +#endif + +inline cl_int enqueueReadBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadBuffer(buffer, blocking, offset, size, ptr, events, event); +} + +inline cl_int enqueueWriteBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + const void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteBuffer(buffer, blocking, offset, size, ptr, events, event); +} + +inline void* enqueueMapBuffer( + const Buffer& buffer, + cl_bool blocking, + cl_map_flags flags, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + void * result = ::clEnqueueMapBuffer( + queue(), buffer(), blocking, flags, offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + return result; +} + +inline cl_int enqueueUnmapMemObject( + const Memory& memory, + void* mapped_ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (error != CL_SUCCESS) { + return error; + } + + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueUnmapMemObject( + queue(), memory(), mapped_ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; +} + +inline cl_int enqueueCopyBuffer( + const Buffer& src, + const Buffer& dst, + ::size_t src_offset, + ::size_t dst_offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBuffer(src, dst, src_offset, dst_offset, size, events, event); +} + +/** + * Blocking copy operation between iterators and a buffer. + */ +template< typename IteratorType > +inline cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + ::size_t length = endIterator-startIterator; + ::size_t byteLength = length*sizeof(DataType); + + DataType *pointer = + static_cast(enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_WRITE, 0, byteLength, 0, 0, &error)); + // if exceptions enabled, enqueueMapBuffer will throw + if( error != CL_SUCCESS ) { + return error; + } +#if defined(_MSC_VER) + std::copy( + startIterator, + endIterator, + stdext::checked_array_iterator( + pointer, length)); +#else + std::copy(startIterator, endIterator, pointer); +#endif + Event endEvent; + error = enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); + // if exceptions enabled, enqueueUnmapMemObject will throw + if( error != CL_SUCCESS ) { + return error; + } + endEvent.wait(); + return CL_SUCCESS; +} + +/** + * Blocking copy operation between iterators and a buffer. + */ +template< typename IteratorType > +inline cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + ::size_t length = endIterator-startIterator; + ::size_t byteLength = length*sizeof(DataType); + + DataType *pointer = + static_cast(enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_READ, 0, byteLength, 0, 0, &error)); + // if exceptions enabled, enqueueMapBuffer will throw + if( error != CL_SUCCESS ) { + return error; + } + std::copy(pointer, pointer + length, startIterator); + Event endEvent; + error = enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); + // if exceptions enabled, enqueueUnmapMemObject will throw + if( error != CL_SUCCESS ) { + return error; + } + endEvent.wait(); + return CL_SUCCESS; +} + +#if defined(CL_VERSION_1_1) +inline cl_int enqueueReadBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadBufferRect( + buffer, + blocking, + buffer_offset, + host_offset, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueWriteBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteBufferRect( + buffer, + blocking, + buffer_offset, + host_offset, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueCopyBufferRect( + const Buffer& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + ::size_t src_row_pitch, + ::size_t src_slice_pitch, + ::size_t dst_row_pitch, + ::size_t dst_slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBufferRect( + src, + dst, + src_origin, + dst_origin, + region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + events, + event); +} +#endif + +inline cl_int enqueueReadImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadImage( + image, + blocking, + origin, + region, + row_pitch, + slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueWriteImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteImage( + image, + blocking, + origin, + region, + row_pitch, + slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueCopyImage( + const Image& src, + const Image& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyImage( + src, + dst, + src_origin, + dst_origin, + region, + events, + event); +} + +inline cl_int enqueueCopyImageToBuffer( + const Image& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& region, + ::size_t dst_offset, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyImageToBuffer( + src, + dst, + src_origin, + region, + dst_offset, + events, + event); +} + +inline cl_int enqueueCopyBufferToImage( + const Buffer& src, + const Image& dst, + ::size_t src_offset, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBufferToImage( + src, + dst, + src_offset, + dst_origin, + region, + events, + event); +} + + +inline cl_int flush(void) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.flush(); +} + +inline cl_int finish(void) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + + return queue.finish(); +} + +// Kernel Functor support +// New interface as of September 2011 +// Requires the C++11 std::tr1::function (note do not support TR1) +// Visual Studio 2010 and GCC 4.2 + +struct EnqueueArgs +{ + CommandQueue queue_; + const NDRange offset_; + const NDRange global_; + const NDRange local_; + VECTOR_CLASS events_; + + EnqueueArgs(NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange) + { + + } + + EnqueueArgs(NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local) + { + + } + + EnqueueArgs(NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local) + { + + } + + EnqueueArgs(Event e, NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange) + { + events_.push_back(e); + } + + EnqueueArgs(Event e, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(Event e, NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange), + events_(events) + { + + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local) + { + + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local), + events_(events) + { + + } +}; + +namespace detail { + +class NullType {}; + +template +struct SetArg +{ + static void set (Kernel kernel, T0 arg) + { + kernel.setArg(index, arg); + } +}; + +template +struct SetArg +{ + static void set (Kernel, NullType) + { + } +}; + +template < + typename T0, typename T1, typename T2, typename T3, + typename T4, typename T5, typename T6, typename T7, + typename T8, typename T9, typename T10, typename T11, + typename T12, typename T13, typename T14, typename T15, + typename T16, typename T17, typename T18, typename T19, + typename T20, typename T21, typename T22, typename T23, + typename T24, typename T25, typename T26, typename T27, + typename T28, typename T29, typename T30, typename T31 +> +class KernelFunctorGlobal +{ +private: + Kernel kernel_; + +public: + KernelFunctorGlobal( + Kernel kernel) : + kernel_(kernel) + {} + + KernelFunctorGlobal( + const Program& program, + const STRING_CLASS name, + cl_int * err = NULL) : + kernel_(program, name.c_str(), err) + {} + + Event operator() ( + const EnqueueArgs& args, + T0 t0, + T1 t1 = NullType(), + T2 t2 = NullType(), + T3 t3 = NullType(), + T4 t4 = NullType(), + T5 t5 = NullType(), + T6 t6 = NullType(), + T7 t7 = NullType(), + T8 t8 = NullType(), + T9 t9 = NullType(), + T10 t10 = NullType(), + T11 t11 = NullType(), + T12 t12 = NullType(), + T13 t13 = NullType(), + T14 t14 = NullType(), + T15 t15 = NullType(), + T16 t16 = NullType(), + T17 t17 = NullType(), + T18 t18 = NullType(), + T19 t19 = NullType(), + T20 t20 = NullType(), + T21 t21 = NullType(), + T22 t22 = NullType(), + T23 t23 = NullType(), + T24 t24 = NullType(), + T25 t25 = NullType(), + T26 t26 = NullType(), + T27 t27 = NullType(), + T28 t28 = NullType(), + T29 t29 = NullType(), + T30 t30 = NullType(), + T31 t31 = NullType() + ) + { + Event event; + SetArg<0, T0>::set(kernel_, t0); + SetArg<1, T1>::set(kernel_, t1); + SetArg<2, T2>::set(kernel_, t2); + SetArg<3, T3>::set(kernel_, t3); + SetArg<4, T4>::set(kernel_, t4); + SetArg<5, T5>::set(kernel_, t5); + SetArg<6, T6>::set(kernel_, t6); + SetArg<7, T7>::set(kernel_, t7); + SetArg<8, T8>::set(kernel_, t8); + SetArg<9, T9>::set(kernel_, t9); + SetArg<10, T10>::set(kernel_, t10); + SetArg<11, T11>::set(kernel_, t11); + SetArg<12, T12>::set(kernel_, t12); + SetArg<13, T13>::set(kernel_, t13); + SetArg<14, T14>::set(kernel_, t14); + SetArg<15, T15>::set(kernel_, t15); + SetArg<16, T16>::set(kernel_, t16); + SetArg<17, T17>::set(kernel_, t17); + SetArg<18, T18>::set(kernel_, t18); + SetArg<19, T19>::set(kernel_, t19); + SetArg<20, T20>::set(kernel_, t20); + SetArg<21, T21>::set(kernel_, t21); + SetArg<22, T22>::set(kernel_, t22); + SetArg<23, T23>::set(kernel_, t23); + SetArg<24, T24>::set(kernel_, t24); + SetArg<25, T25>::set(kernel_, t25); + SetArg<26, T26>::set(kernel_, t26); + SetArg<27, T27>::set(kernel_, t27); + SetArg<28, T28>::set(kernel_, t28); + SetArg<29, T29>::set(kernel_, t29); + SetArg<30, T30>::set(kernel_, t30); + SetArg<31, T31>::set(kernel_, t31); + + args.queue_.enqueueNDRangeKernel( + kernel_, + args.offset_, + args.global_, + args.local_, + &args.events_, + &event); + + return event; + } + +}; + +//------------------------------------------------------------------------------------------------------ + + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29, + typename T30, + typename T31> +struct functionImplementation_ +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + T31> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 32)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + T31); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29, + T30 arg30, + T31 arg31) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29, + arg30, + arg31); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29, + typename T30> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 31)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29, + T30 arg30) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29, + arg30); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 30)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 29)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 28)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 27)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 26)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 25)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 24)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 23)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 22)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 21)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 20)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 19)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 18)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 17)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 16)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 15)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 14)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 13)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 12)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 11)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 10)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 9)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 8)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 7)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 6)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 5)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 4)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3); + } + + +}; + +template< + typename T0, + typename T1, + typename T2> +struct functionImplementation_ +< T0, + T1, + T2, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 3)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2); + } + + +}; + +template< + typename T0, + typename T1> +struct functionImplementation_ +< T0, + T1, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 2)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1) + { + return functor_( + enqueueArgs, + arg0, + arg1); + } + + +}; + +template< + typename T0> +struct functionImplementation_ +< T0, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 1)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0) + { + return functor_( + enqueueArgs, + arg0); + } + + +}; + + + + + +} // namespace detail + +//---------------------------------------------------------------------------------------------- + +template < + typename T0, typename T1 = detail::NullType, typename T2 = detail::NullType, + typename T3 = detail::NullType, typename T4 = detail::NullType, + typename T5 = detail::NullType, typename T6 = detail::NullType, + typename T7 = detail::NullType, typename T8 = detail::NullType, + typename T9 = detail::NullType, typename T10 = detail::NullType, + typename T11 = detail::NullType, typename T12 = detail::NullType, + typename T13 = detail::NullType, typename T14 = detail::NullType, + typename T15 = detail::NullType, typename T16 = detail::NullType, + typename T17 = detail::NullType, typename T18 = detail::NullType, + typename T19 = detail::NullType, typename T20 = detail::NullType, + typename T21 = detail::NullType, typename T22 = detail::NullType, + typename T23 = detail::NullType, typename T24 = detail::NullType, + typename T25 = detail::NullType, typename T26 = detail::NullType, + typename T27 = detail::NullType, typename T28 = detail::NullType, + typename T29 = detail::NullType, typename T30 = detail::NullType, + typename T31 = detail::NullType +> +struct make_kernel : + public detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + > +{ +public: + typedef detail::KernelFunctorGlobal< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + > FunctorType; + + make_kernel( + const Program& program, + const STRING_CLASS name, + cl_int * err = NULL) : + detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + >( + FunctorType(program, name, err)) + {} + + make_kernel( + const Kernel kernel) : + detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + >( + FunctorType(kernel)) + {} +}; + + +//---------------------------------------------------------------------------------------------------------------------- + +#undef __ERR_STR +#if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) +#undef __GET_DEVICE_INFO_ERR +#undef __GET_PLATFORM_INFO_ERR +#undef __GET_DEVICE_IDS_ERR +#undef __GET_CONTEXT_INFO_ERR +#undef __GET_EVENT_INFO_ERR +#undef __GET_EVENT_PROFILE_INFO_ERR +#undef __GET_MEM_OBJECT_INFO_ERR +#undef __GET_IMAGE_INFO_ERR +#undef __GET_SAMPLER_INFO_ERR +#undef __GET_KERNEL_INFO_ERR +#undef __GET_KERNEL_ARG_INFO_ERR +#undef __GET_KERNEL_WORK_GROUP_INFO_ERR +#undef __GET_PROGRAM_INFO_ERR +#undef __GET_PROGRAM_BUILD_INFO_ERR +#undef __GET_COMMAND_QUEUE_INFO_ERR + +#undef __CREATE_CONTEXT_ERR +#undef __CREATE_CONTEXT_FROM_TYPE_ERR +#undef __GET_SUPPORTED_IMAGE_FORMATS_ERR + +#undef __CREATE_BUFFER_ERR +#undef __CREATE_SUBBUFFER_ERR +#undef __CREATE_IMAGE2D_ERR +#undef __CREATE_IMAGE3D_ERR +#undef __CREATE_SAMPLER_ERR +#undef __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR + +#undef __CREATE_USER_EVENT_ERR +#undef __SET_USER_EVENT_STATUS_ERR +#undef __SET_EVENT_CALLBACK_ERR +#undef __SET_PRINTF_CALLBACK_ERR + +#undef __WAIT_FOR_EVENTS_ERR + +#undef __CREATE_KERNEL_ERR +#undef __SET_KERNEL_ARGS_ERR +#undef __CREATE_PROGRAM_WITH_SOURCE_ERR +#undef __CREATE_PROGRAM_WITH_BINARY_ERR +#undef __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR +#undef __BUILD_PROGRAM_ERR +#undef __CREATE_KERNELS_IN_PROGRAM_ERR + +#undef __CREATE_COMMAND_QUEUE_ERR +#undef __SET_COMMAND_QUEUE_PROPERTY_ERR +#undef __ENQUEUE_READ_BUFFER_ERR +#undef __ENQUEUE_WRITE_BUFFER_ERR +#undef __ENQUEUE_READ_BUFFER_RECT_ERR +#undef __ENQUEUE_WRITE_BUFFER_RECT_ERR +#undef __ENQEUE_COPY_BUFFER_ERR +#undef __ENQEUE_COPY_BUFFER_RECT_ERR +#undef __ENQUEUE_READ_IMAGE_ERR +#undef __ENQUEUE_WRITE_IMAGE_ERR +#undef __ENQUEUE_COPY_IMAGE_ERR +#undef __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR +#undef __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR +#undef __ENQUEUE_MAP_BUFFER_ERR +#undef __ENQUEUE_MAP_IMAGE_ERR +#undef __ENQUEUE_UNMAP_MEM_OBJECT_ERR +#undef __ENQUEUE_NDRANGE_KERNEL_ERR +#undef __ENQUEUE_TASK_ERR +#undef __ENQUEUE_NATIVE_KERNEL + +#undef __CL_EXPLICIT_CONSTRUCTORS + +#undef __UNLOAD_COMPILER_ERR +#endif //__CL_USER_OVERRIDE_ERROR_STRINGS + +#undef __CL_FUNCTION_TYPE + +// Extensions +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_VERSION_1_1) +#undef __INIT_CL_EXT_FCN_PTR +#endif // #if defined(CL_VERSION_1_1) +#undef __CREATE_SUB_DEVICES + +#if defined(USE_CL_DEVICE_FISSION) +#undef __PARAM_NAME_DEVICE_FISSION +#endif // USE_CL_DEVICE_FISSION + +#undef __DEFAULT_NOT_INITIALIZED +#undef __DEFAULT_BEING_INITIALIZED +#undef __DEFAULT_INITIALIZED + +} // namespace cl + +#ifdef _WIN32 +#pragma pop_macro("max") +#endif // _WIN32 + +#endif // CL_HPP_ diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d10.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d10.h new file mode 100644 index 0000000000..81b0d37214 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d10.h @@ -0,0 +1,126 @@ +/********************************************************************************** + * Copyright (c) 2008-2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_CL_D3D10_H +#define __OPENCL_CL_D3D10_H + +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/****************************************************************************** + * cl_khr_d3d10_sharing */ +#define cl_khr_d3d10_sharing 1 + +typedef cl_uint cl_d3d10_device_source_khr; +typedef cl_uint cl_d3d10_device_set_khr; + +/******************************************************************************/ + +// Error Codes +#define CL_INVALID_D3D10_DEVICE_KHR -1002 +#define CL_INVALID_D3D10_RESOURCE_KHR -1003 +#define CL_D3D10_RESOURCE_ALREADY_ACQUIRED_KHR -1004 +#define CL_D3D10_RESOURCE_NOT_ACQUIRED_KHR -1005 + +// cl_d3d10_device_source_nv +#define CL_D3D10_DEVICE_KHR 0x4010 +#define CL_D3D10_DXGI_ADAPTER_KHR 0x4011 + +// cl_d3d10_device_set_nv +#define CL_PREFERRED_DEVICES_FOR_D3D10_KHR 0x4012 +#define CL_ALL_DEVICES_FOR_D3D10_KHR 0x4013 + +// cl_context_info +#define CL_CONTEXT_D3D10_DEVICE_KHR 0x4014 +#define CL_CONTEXT_D3D10_PREFER_SHARED_RESOURCES_KHR 0x402C + +// cl_mem_info +#define CL_MEM_D3D10_RESOURCE_KHR 0x4015 + +// cl_image_info +#define CL_IMAGE_D3D10_SUBRESOURCE_KHR 0x4016 + +// cl_command_type +#define CL_COMMAND_ACQUIRE_D3D10_OBJECTS_KHR 0x4017 +#define CL_COMMAND_RELEASE_D3D10_OBJECTS_KHR 0x4018 + +/******************************************************************************/ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromD3D10KHR_fn)( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10BufferKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D10Buffer * resource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10Texture2DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D10Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10Texture3DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D10Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireD3D10ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseD3D10ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +#ifdef __cplusplus +} +#endif + +#endif // __OPENCL_CL_D3D10_H + diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d11.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d11.h new file mode 100644 index 0000000000..d3c8bdc2b1 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_d3d11.h @@ -0,0 +1,126 @@ +/********************************************************************************** + * Copyright (c) 2008-2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_CL_D3D11_H +#define __OPENCL_CL_D3D11_H + +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/****************************************************************************** + * cl_khr_d3d11_sharing */ +#define cl_khr_d3d11_sharing 1 + +typedef cl_uint cl_d3d11_device_source_khr; +typedef cl_uint cl_d3d11_device_set_khr; + +/******************************************************************************/ + +// Error Codes +#define CL_INVALID_D3D11_DEVICE_KHR -1006 +#define CL_INVALID_D3D11_RESOURCE_KHR -1007 +#define CL_D3D11_RESOURCE_ALREADY_ACQUIRED_KHR -1008 +#define CL_D3D11_RESOURCE_NOT_ACQUIRED_KHR -1009 + +// cl_d3d11_device_source +#define CL_D3D11_DEVICE_KHR 0x4019 +#define CL_D3D11_DXGI_ADAPTER_KHR 0x401A + +// cl_d3d11_device_set +#define CL_PREFERRED_DEVICES_FOR_D3D11_KHR 0x401B +#define CL_ALL_DEVICES_FOR_D3D11_KHR 0x401C + +// cl_context_info +#define CL_CONTEXT_D3D11_DEVICE_KHR 0x401D +#define CL_CONTEXT_D3D11_PREFER_SHARED_RESOURCES_KHR 0x402D + +// cl_mem_info +#define CL_MEM_D3D11_RESOURCE_KHR 0x401E + +// cl_image_info +#define CL_IMAGE_D3D11_SUBRESOURCE_KHR 0x401F + +// cl_command_type +#define CL_COMMAND_ACQUIRE_D3D11_OBJECTS_KHR 0x4020 +#define CL_COMMAND_RELEASE_D3D11_OBJECTS_KHR 0x4021 + +/******************************************************************************/ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromD3D11KHR_fn)( + cl_platform_id platform, + cl_d3d11_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d11_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11BufferKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D11Buffer * resource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11Texture2DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D11Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11Texture3DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D11Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireD3D11ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseD3D11ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +#ifdef __cplusplus +} +#endif + +#endif // __OPENCL_CL_D3D11_H + diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_dx9_media_sharing.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_dx9_media_sharing.h new file mode 100644 index 0000000000..1ef543a5af --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_dx9_media_sharing.h @@ -0,0 +1,127 @@ +/********************************************************************************** + * Copyright (c) 2008-2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_CL_DX9_MEDIA_SHARING_H +#define __OPENCL_CL_DX9_MEDIA_SHARING_H + +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/****************************************************************************** +/* cl_khr_dx9_media_sharing */ +#define cl_khr_dx9_media_sharing 1 + +typedef cl_uint cl_dx9_media_adapter_type_khr; +typedef cl_uint cl_dx9_media_adapter_set_khr; + +#if defined(_WIN32) +#include +typedef struct _cl_dx9_surface_info_khr +{ + IDirect3DSurface9 *resource; + HANDLE shared_handle; +} cl_dx9_surface_info_khr; +#endif + + +/******************************************************************************/ + +// Error Codes +#define CL_INVALID_DX9_MEDIA_ADAPTER_KHR -1010 +#define CL_INVALID_DX9_MEDIA_SURFACE_KHR -1011 +#define CL_DX9_MEDIA_SURFACE_ALREADY_ACQUIRED_KHR -1012 +#define CL_DX9_MEDIA_SURFACE_NOT_ACQUIRED_KHR -1013 + +// cl_media_adapter_type_khr +#define CL_ADAPTER_D3D9_KHR 0x2020 +#define CL_ADAPTER_D3D9EX_KHR 0x2021 +#define CL_ADAPTER_DXVA_KHR 0x2022 + +// cl_media_adapter_set_khr +#define CL_PREFERRED_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR 0x2023 +#define CL_ALL_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR 0x2024 + +// cl_context_info +#define CL_CONTEXT_ADAPTER_D3D9_KHR 0x2025 +#define CL_CONTEXT_ADAPTER_D3D9EX_KHR 0x2026 +#define CL_CONTEXT_ADAPTER_DXVA_KHR 0x2027 + +// cl_mem_info +#define CL_MEM_DX9_MEDIA_ADAPTER_TYPE_KHR 0x2028 +#define CL_MEM_DX9_MEDIA_SURFACE_INFO_KHR 0x2029 + +// cl_image_info +#define CL_IMAGE_DX9_MEDIA_PLANE_KHR 0x202A + +// cl_command_type +#define CL_COMMAND_ACQUIRE_DX9_MEDIA_SURFACES_KHR 0x202B +#define CL_COMMAND_RELEASE_DX9_MEDIA_SURFACES_KHR 0x202C + +/******************************************************************************/ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromDX9MediaAdapterKHR_fn)( + cl_platform_id platform, + cl_uint num_media_adapters, + cl_dx9_media_adapter_type_khr * media_adapter_type, + void * media_adapters, + cl_dx9_media_adapter_set_khr media_adapter_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromDX9MediaSurfaceKHR_fn)( + cl_context context, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + void * surface_info, + cl_uint plane, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireDX9MediaSurfacesKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseDX9MediaSurfacesKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +#ifdef __cplusplus +} +#endif + +#endif // __OPENCL_CL_DX9_MEDIA_SHARING_H + diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_ext.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_ext.h new file mode 100644 index 0000000000..1ff16d8906 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_ext.h @@ -0,0 +1,489 @@ +/******************************************************************************* + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/* $Revision: 14835 $ on $Date: 2011-05-26 11:32:00 -0700 (Thu, 26 May 2011) $ */ + +/* cl_ext.h contains OpenCL extensions which don't have external */ +/* (OpenGL, D3D) dependencies. */ + +#ifndef __CL_EXT_H +#define __CL_EXT_H + +#ifdef __cplusplus +extern "C" { +#endif + +#ifdef __APPLE__ + #include + #include +#else + #include +#endif + +/* cl_khr_fp16 extension - no extension #define since it has no functions */ +#define CL_DEVICE_HALF_FP_CONFIG 0x1033 + +/* Memory object destruction + * + * Apple extension for use to manage externally allocated buffers used with cl_mem objects with CL_MEM_USE_HOST_PTR + * + * Registers a user callback function that will be called when the memory object is deleted and its resources + * freed. Each call to clSetMemObjectCallbackFn registers the specified user callback function on a callback + * stack associated with memobj. The registered user callback functions are called in the reverse order in + * which they were registered. The user callback functions are called and then the memory object is deleted + * and its resources freed. This provides a mechanism for the application (and libraries) using memobj to be + * notified when the memory referenced by host_ptr, specified when the memory object is created and used as + * the storage bits for the memory object, can be reused or freed. + * + * The application may not call CL api's with the cl_mem object passed to the pfn_notify. + * + * Please check for the "cl_APPLE_SetMemObjectDestructor" extension using clGetDeviceInfo(CL_DEVICE_EXTENSIONS) + * before using. + */ +#define cl_APPLE_SetMemObjectDestructor 1 +cl_int CL_API_ENTRY clSetMemObjectDestructorAPPLE( cl_mem /* memobj */, + void (* /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), + void * /*user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + + +/* Context Logging Functions + * + * The next three convenience functions are intended to be used as the pfn_notify parameter to clCreateContext(). + * Please check for the "cl_APPLE_ContextLoggingFunctions" extension using clGetDeviceInfo(CL_DEVICE_EXTENSIONS) + * before using. + * + * clLogMessagesToSystemLog fowards on all log messages to the Apple System Logger + */ +#define cl_APPLE_ContextLoggingFunctions 1 +extern void CL_API_ENTRY clLogMessagesToSystemLogAPPLE( const char * /* errstr */, + const void * /* private_info */, + size_t /* cb */, + void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + +/* clLogMessagesToStdout sends all log messages to the file descriptor stdout */ +extern void CL_API_ENTRY clLogMessagesToStdoutAPPLE( const char * /* errstr */, + const void * /* private_info */, + size_t /* cb */, + void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + +/* clLogMessagesToStderr sends all log messages to the file descriptor stderr */ +extern void CL_API_ENTRY clLogMessagesToStderrAPPLE( const char * /* errstr */, + const void * /* private_info */, + size_t /* cb */, + void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + + +/************************ +* cl_khr_icd extension * +************************/ +#define cl_khr_icd 1 + +/* cl_platform_info */ +#define CL_PLATFORM_ICD_SUFFIX_KHR 0x0920 + +/* Additional Error Codes */ +#define CL_PLATFORM_NOT_FOUND_KHR -1001 + +extern CL_API_ENTRY cl_int CL_API_CALL +clIcdGetPlatformIDsKHR(cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clIcdGetPlatformIDsKHR_fn)( + cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */); + + +/* Extension: cl_khr_image2D_buffer + * + * This extension allows a 2D image to be created from a cl_mem buffer without a copy. + * The type associated with a 2D image created from a buffer in an OpenCL program is image2d_t. + * Both the sampler and sampler-less read_image built-in functions are supported for 2D images + * and 2D images created from a buffer. Similarly, the write_image built-ins are also supported + * for 2D images created from a buffer. + * + * When the 2D image from buffer is created, the client must specify the width, + * height, image format (i.e. channel order and channel data type) and optionally the row pitch + * + * The pitch specified must be a multiple of CL_DEVICE_IMAGE_PITCH_ALIGNMENT pixels. + * The base address of the buffer must be aligned to CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT pixels. + */ + +/************************************* + * cl_khr_initalize_memory extension * + *************************************/ + +#define CL_CONTEXT_MEMORY_INITIALIZE_KHR 0x200E + + +/************************************** + * cl_khr_terminate_context extension * + **************************************/ + +#define CL_DEVICE_TERMINATE_CAPABILITY_KHR 0x200F +#define CL_CONTEXT_TERMINATE_KHR 0x2010 + +#define cl_khr_terminate_context 1 +extern CL_API_ENTRY cl_int CL_API_CALL clTerminateContextKHR(cl_context /* context */) CL_EXT_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clTerminateContextKHR_fn)(cl_context /* context */) CL_EXT_SUFFIX__VERSION_1_2; + +// +/****************************** + * cl_khr_subgroups extension * + ******************************/ + +#define cl_khr_subgroups 1 + +typedef cl_uint cl_kernel_sub_group_info; + +#define CL_KERNEL_MAX_SUB_GROUP_SIZE_FOR_NDRANGE 0x2011 +#define CL_KERNEL_SUB_GROUP_COUNT_FOR_NDRANGE 0x2012 + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelSubGroupInfoKHR(cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_sub_group_info /* param_name */, + size_t /* input_value_size */, + const void * /* input_value */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetKernelSubGroupInfoKHR_fn)( + cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_sub_group_info /* param_name */, + size_t /* input_value_size */, + const void * /* input_value */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +// + +/* + * Extension: cl_khr_spir + * + * This extension adds support to create an OpenCL program object from a + * Standard Portable Intermediate Representation (SPIR) instance + */ + +// KHR SPIR extension (Section 9.15.2 in the extension SPEC) +// TODO: The values have been approved by khronos and waiting to be updated in official header file. +#define CL_DEVICE_SPIR_VERSIONS 0x40E0 +#define CL_PROGRAM_BINARY_TYPE_INTERMEDIATE 0x40E1 + +/****************************************** +* cl_nv_device_attribute_query extension * +******************************************/ +/* cl_nv_device_attribute_query extension - no extension #define since it has no functions */ +#define CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV 0x4000 +#define CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV 0x4001 +#define CL_DEVICE_REGISTERS_PER_BLOCK_NV 0x4002 +#define CL_DEVICE_WARP_SIZE_NV 0x4003 +#define CL_DEVICE_GPU_OVERLAP_NV 0x4004 +#define CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV 0x4005 +#define CL_DEVICE_INTEGRATED_MEMORY_NV 0x4006 + +/********************************* +* cl_amd_device_memory_flags * +*********************************/ +#define cl_amd_device_memory_flags 1 +#define CL_MEM_USE_PERSISTENT_MEM_AMD (1 << 6) // Alloc from GPU's CPU visible heap + +/* cl_device_info */ +#define CL_DEVICE_MAX_ATOMIC_COUNTERS_EXT 0x4032 + +/********************************* +* cl_amd_device_attribute_query * +*********************************/ +#define CL_DEVICE_PROFILING_TIMER_OFFSET_AMD 0x4036 +#define CL_DEVICE_TOPOLOGY_AMD 0x4037 +#define CL_DEVICE_BOARD_NAME_AMD 0x4038 +#define CL_DEVICE_GLOBAL_FREE_MEMORY_AMD 0x4039 +#define CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD 0x4040 +#define CL_DEVICE_SIMD_WIDTH_AMD 0x4041 +#define CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD 0x4042 +#define CL_DEVICE_WAVEFRONT_WIDTH_AMD 0x4043 +#define CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD 0x4044 +#define CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD 0x4045 +#define CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD 0x4046 +#define CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD 0x4047 +#define CL_DEVICE_LOCAL_MEM_BANKS_AMD 0x4048 +#define CL_DEVICE_THREAD_TRACE_SUPPORTED_AMD 0x4049 +#define CL_DEVICE_GFXIP_MAJOR_AMD 0x404A +#define CL_DEVICE_GFXIP_MINOR_AMD 0x404B + +typedef union +{ + struct { cl_uint type; cl_uint data[5]; } raw; + struct { cl_uint type; cl_char unused[17]; cl_char bus; cl_char device; cl_char function; } pcie; +} cl_device_topology_amd; + +#define CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD 1 + + +// +/************* +* cl_amd_hsa * +**************/ +#define CL_HSA_ENABLED_AMD (1ull << 62) +#define CL_HSA_DISABLED_AMD (1ull << 63) + +/*************************** +* cl_amd_command_intercept * +***************************/ +#define CL_CONTEXT_COMMAND_INTERCEPT_CALLBACK_AMD 0x403D +#define CL_QUEUE_COMMAND_INTERCEPT_ENABLE_AMD (1ull << 63) + +typedef cl_int (CL_CALLBACK * intercept_callback_fn)(cl_event, cl_int *); + +/************************** +* cl_amd_command_queue_info * +**************************/ +#define CL_QUEUE_THREAD_HANDLE_AMD 0x403E + +/*************************************** +* cl-gl depth buffer interop extension * +****************************************/ + +#define CL_DEPTH 0x10BD +#define CL_DEPTH_STENCIL 0x10BE +#define CL_UNORM_INT24 0x10DF +#define CL_GL_NUM_SAMPLES 0x2012 +// + +/************************** +* cl_amd_offline_devices * +**************************/ +#define CL_CONTEXT_OFFLINE_DEVICES_AMD 0x403F + +#ifdef CL_VERSION_1_1 + /*********************************** + * cl_ext_device_fission extension * + ***********************************/ + #define cl_ext_device_fission 1 + + extern CL_API_ENTRY cl_int CL_API_CALL + clReleaseDeviceEXT( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL *clReleaseDeviceEXT_fn)( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + extern CL_API_ENTRY cl_int CL_API_CALL + clRetainDeviceEXT( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL *clRetainDeviceEXT_fn)( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef cl_ulong cl_device_partition_property_ext; + extern CL_API_ENTRY cl_int CL_API_CALL + clCreateSubDevicesEXT( cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int + ( CL_API_CALL * clCreateSubDevicesEXT_fn)( cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + /* cl_device_partition_property_ext */ + #define CL_DEVICE_PARTITION_EQUALLY_EXT 0x4050 + #define CL_DEVICE_PARTITION_BY_COUNTS_EXT 0x4051 + #define CL_DEVICE_PARTITION_BY_NAMES_EXT 0x4052 + #define CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN_EXT 0x4053 + + /* clDeviceGetInfo selectors */ + #define CL_DEVICE_PARENT_DEVICE_EXT 0x4054 + #define CL_DEVICE_PARTITION_TYPES_EXT 0x4055 + #define CL_DEVICE_AFFINITY_DOMAINS_EXT 0x4056 + #define CL_DEVICE_REFERENCE_COUNT_EXT 0x4057 + #define CL_DEVICE_PARTITION_STYLE_EXT 0x4058 + + /* clGetImageInfo enum */ + #define CL_IMAGE_BYTE_PITCH_AMD 0x4059 + + /* error codes */ + #define CL_DEVICE_PARTITION_FAILED_EXT -1057 + #define CL_INVALID_PARTITION_COUNT_EXT -1058 + #define CL_INVALID_PARTITION_NAME_EXT -1059 + + /* CL_AFFINITY_DOMAINs */ + #define CL_AFFINITY_DOMAIN_L1_CACHE_EXT 0x1 + #define CL_AFFINITY_DOMAIN_L2_CACHE_EXT 0x2 + #define CL_AFFINITY_DOMAIN_L3_CACHE_EXT 0x3 + #define CL_AFFINITY_DOMAIN_L4_CACHE_EXT 0x4 + #define CL_AFFINITY_DOMAIN_NUMA_EXT 0x10 + #define CL_AFFINITY_DOMAIN_NEXT_FISSIONABLE_EXT 0x100 + /* cl_device_partition_property_ext list terminators */ + #define CL_PROPERTIES_LIST_END_EXT ((cl_device_partition_property_ext) 0) + #define CL_PARTITION_BY_COUNTS_LIST_END_EXT ((cl_device_partition_property_ext) 0) + #define CL_PARTITION_BY_NAMES_LIST_END_EXT ((cl_device_partition_property_ext) 0 - 1) + +/********************************* +* cl_qcom_ext_host_ptr extension +*********************************/ + +#define CL_MEM_EXT_HOST_PTR_QCOM (1 << 29) + +#define CL_DEVICE_EXT_MEM_PADDING_IN_BYTES_QCOM 0x40A0 +#define CL_DEVICE_PAGE_SIZE_QCOM 0x40A1 +#define CL_IMAGE_ROW_ALIGNMENT_QCOM 0x40A2 +#define CL_IMAGE_SLICE_ALIGNMENT_QCOM 0x40A3 +#define CL_MEM_HOST_UNCACHED_QCOM 0x40A4 +#define CL_MEM_HOST_WRITEBACK_QCOM 0x40A5 +#define CL_MEM_HOST_WRITETHROUGH_QCOM 0x40A6 +#define CL_MEM_HOST_WRITE_COMBINING_QCOM 0x40A7 + +typedef cl_uint cl_image_pitch_info_qcom; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceImageInfoQCOM(cl_device_id device, + size_t image_width, + size_t image_height, + const cl_image_format *image_format, + cl_image_pitch_info_qcom param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret); + +typedef struct _cl_mem_ext_host_ptr +{ + // Type of external memory allocation. + // Legal values will be defined in layered extensions. + cl_uint allocation_type; + + // Host cache policy for this external memory allocation. + cl_uint host_cache_policy; + +} cl_mem_ext_host_ptr; + +/********************************* +* cl_qcom_ion_host_ptr extension +*********************************/ + +#define CL_MEM_ION_HOST_PTR_QCOM 0x40A8 + +typedef struct _cl_mem_ion_host_ptr +{ + // Type of external memory allocation. + // Must be CL_MEM_ION_HOST_PTR_QCOM for ION allocations. + cl_mem_ext_host_ptr ext_host_ptr; + + // ION file descriptor + int ion_filedesc; + + // Host pointer to the ION allocated memory + void* ion_hostptr; + +} cl_mem_ion_host_ptr; + + // + /************************* + * cl_amd_object_metadata * + **************************/ + #define cl_amd_object_metadata 1 + + typedef size_t cl_key_amd; + + #define CL_INVALID_OBJECT_AMD 0x403A + #define CL_INVALID_KEY_AMD 0x403B + #define CL_PLATFORM_MAX_KEYS_AMD 0x403C + + typedef CL_API_ENTRY cl_key_amd (CL_API_CALL * clCreateKeyAMD_fn)( + cl_platform_id /* platform */, + void (CL_CALLBACK * /* destructor */)( void* /* old_value */), + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int (CL_API_CALL * clObjectGetValueForKeyAMD_fn)( + void * /* object */, + cl_key_amd /* key */, + void ** /* ret_val */) CL_API_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int (CL_API_CALL * clObjectSetValueForKeyAMD_fn)( + void * /* object */, + cl_key_amd /* key */, + void * /* value */) CL_API_SUFFIX__VERSION_1_1; +// +#endif /* CL_VERSION_1_1 */ + +#ifdef CL_VERSION_1_2 + /******************************** + * cl_amd_bus_addressable_memory * + ********************************/ + + /* cl_mem flag - bitfield */ + #define CL_MEM_BUS_ADDRESSABLE_AMD (1<<30) + #define CL_MEM_EXTERNAL_PHYSICAL_AMD (1<<31) + + #define CL_COMMAND_WAIT_SIGNAL_AMD 0x4080 + #define CL_COMMAND_WRITE_SIGNAL_AMD 0x4081 + #define CL_COMMAND_MAKE_BUFFERS_RESIDENT_AMD 0x4082 + + typedef struct _cl_bus_address_amd + { + cl_ulong surface_bus_address; + cl_ulong marker_bus_address; + } cl_bus_address_amd; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL * clEnqueueWaitSignalAMD_fn)( cl_command_queue /*command_queue*/, + cl_mem /*mem_object*/, + cl_uint /*value*/, + cl_uint /*num_events*/, + const cl_event * /*event_wait_list*/, + cl_event * /*event*/) CL_EXT_SUFFIX__VERSION_1_2; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL * clEnqueueWriteSignalAMD_fn)( cl_command_queue /*command_queue*/, + cl_mem /*mem_object*/, + cl_uint /*value*/, + cl_ulong /*offset*/, + cl_uint /*num_events*/, + const cl_event * /*event_list*/, + cl_event * /*event*/) CL_EXT_SUFFIX__VERSION_1_2; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL * clEnqueueMakeBuffersResidentAMD_fn)( cl_command_queue /*command_queue*/, + cl_uint /*num_mem_objs*/, + cl_mem * /*mem_objects*/, + cl_bool /*blocking_make_resident*/, + cl_bus_address_amd * /*bus_addresses*/, + cl_uint /*num_events*/, + const cl_event * /*event_list*/, + cl_event * /*event*/) CL_EXT_SUFFIX__VERSION_1_2; + +#endif /* CL_VERSION_1_2 */ + +#ifdef __cplusplus +} +#endif + + +#endif /* __CL_EXT_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl.h new file mode 100644 index 0000000000..af2036cc99 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl.h @@ -0,0 +1,162 @@ +/********************************************************************************** + * Copyright (c) 2008 - 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +#ifndef __OPENCL_CL_GL_H +#define __OPENCL_CL_GL_H + +#ifdef __APPLE__ +#include +#else +#include +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +typedef cl_uint cl_gl_object_type; +typedef cl_uint cl_gl_texture_info; +typedef cl_uint cl_gl_platform_info; +typedef struct __GLsync *cl_GLsync; + +/* cl_gl_object_type = 0x2000 - 0x200F enum values are currently taken */ +#define CL_GL_OBJECT_BUFFER 0x2000 +#define CL_GL_OBJECT_TEXTURE2D 0x2001 +#define CL_GL_OBJECT_TEXTURE3D 0x2002 +#define CL_GL_OBJECT_RENDERBUFFER 0x2003 +#define CL_GL_OBJECT_TEXTURE2D_ARRAY 0x200E +#define CL_GL_OBJECT_TEXTURE1D 0x200F +#define CL_GL_OBJECT_TEXTURE1D_ARRAY 0x2010 +#define CL_GL_OBJECT_TEXTURE_BUFFER 0x2011 + +/* cl_gl_texture_info */ +#define CL_GL_TEXTURE_TARGET 0x2004 +#define CL_GL_MIPMAP_LEVEL 0x2005 +#define CL_GL_NUM_SAMPLES 0x2012 + + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromGLBuffer(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLuint /* bufobj */, + int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromGLTexture(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromGLRenderbuffer(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLuint /* renderbuffer */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetGLObjectInfo(cl_mem /* memobj */, + cl_gl_object_type * /* gl_object_type */, + cl_GLuint * /* gl_object_name */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetGLTextureInfo(cl_mem /* memobj */, + cl_gl_texture_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireGLObjects(cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseGLObjects(cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + + +// Deprecated OpenCL 1.1 APIs +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateFromGLTexture2D(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateFromGLTexture3D(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +/* cl_khr_gl_sharing extension */ + +#define cl_khr_gl_sharing 1 + +typedef cl_uint cl_gl_context_info; + +/* Additional Error Codes */ +#define CL_INVALID_GL_SHAREGROUP_REFERENCE_KHR -1000 + +/* cl_gl_context_info */ +#define CL_CURRENT_DEVICE_FOR_GL_CONTEXT_KHR 0x2006 +#define CL_DEVICES_FOR_GL_CONTEXT_KHR 0x2007 + +/* Additional cl_context_properties */ +#define CL_GL_CONTEXT_KHR 0x2008 +#define CL_EGL_DISPLAY_KHR 0x2009 +#define CL_GLX_DISPLAY_KHR 0x200A +#define CL_WGL_HDC_KHR 0x200B +#define CL_CGL_SHAREGROUP_KHR 0x200C + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetGLContextInfoKHR(const cl_context_properties * /* properties */, + cl_gl_context_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetGLContextInfoKHR_fn)( + const cl_context_properties * properties, + cl_gl_context_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret); + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_GL_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl_ext.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl_ext.h new file mode 100644 index 0000000000..eafab15cf1 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_gl_ext.h @@ -0,0 +1,69 @@ +/********************************************************************************** + * Copyright (c) 2008-2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 14826 $ on $Date: 2011-05-26 07:40:43 -0700 (Thu, 26 May 2011) $ */ + +/* cl_gl_ext.h contains vendor (non-KHR) OpenCL extensions which have */ +/* OpenGL dependencies. */ + +#ifndef __OPENCL_CL_GL_EXT_H +#define __OPENCL_CL_GL_EXT_H + +#ifdef __cplusplus +extern "C" { +#endif + +#ifdef __APPLE__ + #include +#else + #include +#endif + +/* + * For each extension, follow this template + * cl_VEN_extname extension */ +/* #define cl_VEN_extname 1 + * ... define new types, if any + * ... define new tokens, if any + * ... define new APIs, if any + * + * If you need GLtypes here, mirror them with a cl_GLtype, rather than including a GL header + * This allows us to avoid having to decide whether to include GL headers or GLES here. + */ + +/* + * cl_khr_gl_event extension + * See section 9.9 in the OpenCL 1.1 spec for more information + */ +#define CL_COMMAND_GL_FENCE_SYNC_OBJECT_KHR 0x200D + +extern CL_API_ENTRY cl_event CL_API_CALL +clCreateEventFromGLsyncKHR(cl_context /* context */, + cl_GLsync /* cl_GLsync */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_GL_EXT_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_platform.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_platform.h new file mode 100644 index 0000000000..e39ad9282a --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/cl_platform.h @@ -0,0 +1,1171 @@ +/********************************************************************************** + * Copyright (c) 2008-2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 14829 $ on $Date: 2011-05-26 08:22:50 -0700 (Thu, 26 May 2011) $ */ + +#ifndef __CL_PLATFORM_H +#define __CL_PLATFORM_H + +#ifdef __APPLE__ + /* Contains #defines for AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER below */ + #include +#endif + +#if !defined(_WIN32) || !defined(_MSC_VER) +# include +#endif /* !_WIN32 */ +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +#if defined(_WIN32) || defined(__CYGWIN__) + #define CL_API_ENTRY + #define CL_API_CALL __stdcall + #define CL_CALLBACK __stdcall +#else + #define CL_API_ENTRY + #define CL_API_CALL + #define CL_CALLBACK +#endif + +#ifdef __APPLE__ + #define CL_EXTENSION_WEAK_LINK __attribute__((weak_import)) + #define CL_API_SUFFIX__VERSION_1_0 AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_0 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER + #define CL_API_SUFFIX__VERSION_1_1 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define GCL_API_SUFFIX__VERSION_1_1 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_1 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER_BUT_DEPRECATED_IN_MAC_OS_X_VERSION_10_7 + + #ifdef AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define CL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define GCL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_2 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER_BUT_DEPRECATED_IN_MAC_OS_X_VERSION_10_8 + #else + #warning This path should never happen outside of internal operating system development. AvailabilityMacros do not function correctly here! + #define CL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define GCL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_2 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #endif +#else + #define CL_EXTENSION_WEAK_LINK + #define CL_API_SUFFIX__VERSION_1_0 + #define CL_EXT_SUFFIX__VERSION_1_0 + #define CL_API_SUFFIX__VERSION_1_1 + #define CL_EXT_SUFFIX__VERSION_1_1 + #define CL_API_SUFFIX__VERSION_1_2 + #define CL_EXT_SUFFIX__VERSION_1_2 + + #ifdef __GNUC__ + #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED __attribute__((deprecated)) + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + #endif + + #ifdef CL_USE_DEPRECATED_OPENCL_1_1_APIS + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED __attribute__((deprecated)) + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #endif + #elif _WIN32 + #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED __declspec(deprecated) + #endif + + #ifdef CL_USE_DEPRECATED_OPENCL_1_1_APIS + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED __declspec(deprecated) + #endif + #else + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #endif +#endif + +#if (defined (_WIN32) && defined(_MSC_VER)) + +/* scalar types */ +typedef signed __int8 cl_char; +typedef unsigned __int8 cl_uchar; +typedef signed __int16 cl_short; +typedef unsigned __int16 cl_ushort; +typedef signed __int32 cl_int; +typedef unsigned __int32 cl_uint; +typedef signed __int64 cl_long; +typedef unsigned __int64 cl_ulong; + +typedef unsigned __int16 cl_half; +#else /* !_WIN32 */ +typedef int8_t cl_char; +typedef uint8_t cl_uchar; +typedef int16_t cl_short; +typedef uint16_t cl_ushort; +typedef int32_t cl_int; +typedef uint32_t cl_uint; +typedef int64_t cl_long; +typedef uint64_t cl_ulong; + +typedef uint16_t cl_half; +#endif /* !_WIN32 */ +typedef float cl_float; +typedef double cl_double; + +#define CL_CHAR_BIT 8 +#define CL_SCHAR_MAX 127 +#define CL_SCHAR_MIN (-127-1) +#define CL_CHAR_MAX CL_SCHAR_MAX +#define CL_CHAR_MIN CL_SCHAR_MIN +#define CL_UCHAR_MAX 255 +#define CL_SHRT_MAX 32767 +#define CL_SHRT_MIN (-32767-1) +#define CL_USHRT_MAX 65535 +#define CL_INT_MAX 2147483647 +#define CL_INT_MIN (-2147483647-1) +#define CL_UINT_MAX 0xffffffffU +#define CL_LONG_MAX ((cl_long) 0x7FFFFFFFFFFFFFFFLL) +#define CL_LONG_MIN ((cl_long) -0x7FFFFFFFFFFFFFFFLL - 1LL) +#define CL_ULONG_MAX ((cl_ulong) 0xFFFFFFFFFFFFFFFFULL) + +#define CL_FLT_DIG 6 +#define CL_FLT_MANT_DIG 24 +#define CL_FLT_MAX_10_EXP +38 +#define CL_FLT_MAX_EXP +128 +#define CL_FLT_MIN_10_EXP -37 +#define CL_FLT_MIN_EXP -125 +#define CL_FLT_RADIX 2 +#define CL_FLT_MAX FLT_MAX +#define CL_FLT_MIN FLT_MIN +#define CL_FLT_EPSILON FLT_EPSILON + +#define CL_DBL_DIG 15 +#define CL_DBL_MANT_DIG 53 +#define CL_DBL_MAX_10_EXP +308 +#define CL_DBL_MAX_EXP +1024 +#define CL_DBL_MIN_10_EXP -307 +#define CL_DBL_MIN_EXP -1021 +#define CL_DBL_RADIX 2 +#define CL_DBL_MAX DBL_MAX +#define CL_DBL_MIN DBL_MIN +#define CL_DBL_EPSILON DBL_EPSILON + +#define CL_M_E 2.718281828459045090796 +#define CL_M_LOG2E 1.442695040888963387005 +#define CL_M_LOG10E 0.434294481903251816668 +#define CL_M_LN2 0.693147180559945286227 +#define CL_M_LN10 2.302585092994045901094 +#define CL_M_PI 3.141592653589793115998 +#define CL_M_PI_2 1.570796326794896557999 +#define CL_M_PI_4 0.785398163397448278999 +#define CL_M_1_PI 0.318309886183790691216 +#define CL_M_2_PI 0.636619772367581382433 +#define CL_M_2_SQRTPI 1.128379167095512558561 +#define CL_M_SQRT2 1.414213562373095145475 +#define CL_M_SQRT1_2 0.707106781186547572737 + +#define CL_M_E_F 2.71828174591064f +#define CL_M_LOG2E_F 1.44269502162933f +#define CL_M_LOG10E_F 0.43429449200630f +#define CL_M_LN2_F 0.69314718246460f +#define CL_M_LN10_F 2.30258512496948f +#define CL_M_PI_F 3.14159274101257f +#define CL_M_PI_2_F 1.57079637050629f +#define CL_M_PI_4_F 0.78539818525314f +#define CL_M_1_PI_F 0.31830987334251f +#define CL_M_2_PI_F 0.63661974668503f +#define CL_M_2_SQRTPI_F 1.12837922573090f +#define CL_M_SQRT2_F 1.41421353816986f +#define CL_M_SQRT1_2_F 0.70710676908493f + +#define CL_NAN (CL_INFINITY - CL_INFINITY) +#define CL_HUGE_VALF ((cl_float) 1e50) +#define CL_HUGE_VAL ((cl_double) 1e500) +#define CL_MAXFLOAT CL_FLT_MAX +#define CL_INFINITY CL_HUGE_VALF + +#include + +/* Mirror types to GL types. Mirror types allow us to avoid deciding which 87s to load based on whether we are using GL or GLES here. */ +typedef unsigned int cl_GLuint; +typedef int cl_GLint; +typedef unsigned int cl_GLenum; + +/* + * Vector types + * + * Note: OpenCL requires that all types be naturally aligned. + * This means that vector types must be naturally aligned. + * For example, a vector of four floats must be aligned to + * a 16 byte boundary (calculated as 4 * the natural 4-byte + * alignment of the float). The alignment qualifiers here + * will only function properly if your compiler supports them + * and if you don't actively work to defeat them. For example, + * in order for a cl_float4 to be 16 byte aligned in a struct, + * the start of the struct must itself be 16-byte aligned. + * + * Maintaining proper alignment is the user's responsibility. + */ + +/* Define basic vector types */ +#if defined( __VEC__ ) + #include /* may be omitted depending on compiler. AltiVec spec provides no way to detect whether the header is required. */ + typedef vector unsigned char __cl_uchar16; + typedef vector signed char __cl_char16; + typedef vector unsigned short __cl_ushort8; + typedef vector signed short __cl_short8; + typedef vector unsigned int __cl_uint4; + typedef vector signed int __cl_int4; + typedef vector float __cl_float4; + #define __CL_UCHAR16__ 1 + #define __CL_CHAR16__ 1 + #define __CL_USHORT8__ 1 + #define __CL_SHORT8__ 1 + #define __CL_UINT4__ 1 + #define __CL_INT4__ 1 + #define __CL_FLOAT4__ 1 +#endif + +#if defined( __SSE__ ) + #if defined( __MINGW64__ ) + #include + #else + #include + #endif + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef float __cl_float4 __attribute__((vector_size(16))); + #else + typedef __m128 __cl_float4; + #endif + #define __CL_FLOAT4__ 1 +#endif + +#if defined( __SSE2__ ) + #if defined( __MINGW64__ ) + #include + #else + #include + #endif + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef cl_uchar __cl_uchar16 __attribute__((vector_size(16))); + typedef cl_char __cl_char16 __attribute__((vector_size(16))); + typedef cl_ushort __cl_ushort8 __attribute__((vector_size(16))); + typedef cl_short __cl_short8 __attribute__((vector_size(16))); + typedef cl_uint __cl_uint4 __attribute__((vector_size(16))); + typedef cl_int __cl_int4 __attribute__((vector_size(16))); + typedef cl_ulong __cl_ulong2 __attribute__((vector_size(16))); + typedef cl_long __cl_long2 __attribute__((vector_size(16))); + typedef cl_double __cl_double2 __attribute__((vector_size(16))); + #else + typedef __m128i __cl_uchar16; + typedef __m128i __cl_char16; + typedef __m128i __cl_ushort8; + typedef __m128i __cl_short8; + typedef __m128i __cl_uint4; + typedef __m128i __cl_int4; + typedef __m128i __cl_ulong2; + typedef __m128i __cl_long2; + typedef __m128d __cl_double2; + #endif + #define __CL_UCHAR16__ 1 + #define __CL_CHAR16__ 1 + #define __CL_USHORT8__ 1 + #define __CL_SHORT8__ 1 + #define __CL_INT4__ 1 + #define __CL_UINT4__ 1 + #define __CL_ULONG2__ 1 + #define __CL_LONG2__ 1 + #define __CL_DOUBLE2__ 1 +#endif + +#if defined( __MMX__ ) + #include + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef cl_uchar __cl_uchar8 __attribute__((vector_size(8))); + typedef cl_char __cl_char8 __attribute__((vector_size(8))); + typedef cl_ushort __cl_ushort4 __attribute__((vector_size(8))); + typedef cl_short __cl_short4 __attribute__((vector_size(8))); + typedef cl_uint __cl_uint2 __attribute__((vector_size(8))); + typedef cl_int __cl_int2 __attribute__((vector_size(8))); + typedef cl_ulong __cl_ulong1 __attribute__((vector_size(8))); + typedef cl_long __cl_long1 __attribute__((vector_size(8))); + typedef cl_float __cl_float2 __attribute__((vector_size(8))); + #else + typedef __m64 __cl_uchar8; + typedef __m64 __cl_char8; + typedef __m64 __cl_ushort4; + typedef __m64 __cl_short4; + typedef __m64 __cl_uint2; + typedef __m64 __cl_int2; + typedef __m64 __cl_ulong1; + typedef __m64 __cl_long1; + typedef __m64 __cl_float2; + #endif + #define __CL_UCHAR8__ 1 + #define __CL_CHAR8__ 1 + #define __CL_USHORT4__ 1 + #define __CL_SHORT4__ 1 + #define __CL_INT2__ 1 + #define __CL_UINT2__ 1 + #define __CL_ULONG1__ 1 + #define __CL_LONG1__ 1 + #define __CL_FLOAT2__ 1 +#endif + +#if defined( __AVX__ ) + #if defined( __MINGW64__ ) + #include + #else + #include + #endif + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef cl_float __cl_float8 __attribute__((vector_size(32))); + typedef cl_double __cl_double4 __attribute__((vector_size(32))); + #else + typedef __m256 __cl_float8; + typedef __m256d __cl_double4; + #endif + #define __CL_FLOAT8__ 1 + #define __CL_DOUBLE4__ 1 +#endif + +/* Define alignment keys */ +#if (defined( __GNUC__ ) || defined( __IBMC__ )) + #define CL_ALIGNED(_x) __attribute__ ((aligned(_x))) +#elif defined( _WIN32) && (_MSC_VER) + /* Alignment keys neutered on windows because MSVC can't swallow function arguments with alignment requirements */ + /* http://msdn.microsoft.com/en-us/library/373ak2y1%28VS.71%29.aspx */ + /* #include */ + /* #define CL_ALIGNED(_x) _CRT_ALIGN(_x) */ + #define CL_ALIGNED(_x) +#else + #warning Need to implement some method to align data here + #define CL_ALIGNED(_x) +#endif + +/* Indicate whether .xyzw, .s0123 and .hi.lo are supported */ +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + /* .xyzw and .s0123...{f|F} are supported */ + #define CL_HAS_NAMED_VECTOR_FIELDS 1 + /* .hi and .lo are supported */ + #define CL_HAS_HI_LO_VECTOR_FIELDS 1 +#endif + +/* Define cl_vector types */ + +/* ---- cl_charn ---- */ +typedef union +{ + cl_char CL_ALIGNED(2) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_char x, y; }; + __extension__ struct{ cl_char s0, s1; }; + __extension__ struct{ cl_char lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2; +#endif +}cl_char2; + +typedef union +{ + cl_char CL_ALIGNED(4) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_char x, y, z, w; }; + __extension__ struct{ cl_char s0, s1, s2, s3; }; + __extension__ struct{ cl_char2 lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2[2]; +#endif +#if defined( __CL_CHAR4__) + __cl_char4 v4; +#endif +}cl_char4; + +/* cl_char3 is identical in size, alignment and behavior to cl_char4. See section 6.1.5. */ +typedef cl_char4 cl_char3; + +typedef union +{ + cl_char CL_ALIGNED(8) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_char x, y, z, w; }; + __extension__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_char4 lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2[4]; +#endif +#if defined( __CL_CHAR4__) + __cl_char4 v4[2]; +#endif +#if defined( __CL_CHAR8__ ) + __cl_char8 v8; +#endif +}cl_char8; + +typedef union +{ + cl_char CL_ALIGNED(16) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_char x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_char8 lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2[8]; +#endif +#if defined( __CL_CHAR4__) + __cl_char4 v4[4]; +#endif +#if defined( __CL_CHAR8__ ) + __cl_char8 v8[2]; +#endif +#if defined( __CL_CHAR16__ ) + __cl_char16 v16; +#endif +}cl_char16; + + +/* ---- cl_ucharn ---- */ +typedef union +{ + cl_uchar CL_ALIGNED(2) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uchar x, y; }; + __extension__ struct{ cl_uchar s0, s1; }; + __extension__ struct{ cl_uchar lo, hi; }; +#endif +#if defined( __cl_uchar2__) + __cl_uchar2 v2; +#endif +}cl_uchar2; + +typedef union +{ + cl_uchar CL_ALIGNED(4) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uchar x, y, z, w; }; + __extension__ struct{ cl_uchar s0, s1, s2, s3; }; + __extension__ struct{ cl_uchar2 lo, hi; }; +#endif +#if defined( __CL_UCHAR2__) + __cl_uchar2 v2[2]; +#endif +#if defined( __CL_UCHAR4__) + __cl_uchar4 v4; +#endif +}cl_uchar4; + +/* cl_uchar3 is identical in size, alignment and behavior to cl_uchar4. See section 6.1.5. */ +typedef cl_uchar4 cl_uchar3; + +typedef union +{ + cl_uchar CL_ALIGNED(8) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uchar x, y, z, w; }; + __extension__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_uchar4 lo, hi; }; +#endif +#if defined( __CL_UCHAR2__) + __cl_uchar2 v2[4]; +#endif +#if defined( __CL_UCHAR4__) + __cl_uchar4 v4[2]; +#endif +#if defined( __CL_UCHAR8__ ) + __cl_uchar8 v8; +#endif +}cl_uchar8; + +typedef union +{ + cl_uchar CL_ALIGNED(16) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uchar x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_uchar8 lo, hi; }; +#endif +#if defined( __CL_UCHAR2__) + __cl_uchar2 v2[8]; +#endif +#if defined( __CL_UCHAR4__) + __cl_uchar4 v4[4]; +#endif +#if defined( __CL_UCHAR8__ ) + __cl_uchar8 v8[2]; +#endif +#if defined( __CL_UCHAR16__ ) + __cl_uchar16 v16; +#endif +}cl_uchar16; + + +/* ---- cl_shortn ---- */ +typedef union +{ + cl_short CL_ALIGNED(4) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_short x, y; }; + __extension__ struct{ cl_short s0, s1; }; + __extension__ struct{ cl_short lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2; +#endif +}cl_short2; + +typedef union +{ + cl_short CL_ALIGNED(8) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_short x, y, z, w; }; + __extension__ struct{ cl_short s0, s1, s2, s3; }; + __extension__ struct{ cl_short2 lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2[2]; +#endif +#if defined( __CL_SHORT4__) + __cl_short4 v4; +#endif +}cl_short4; + +/* cl_short3 is identical in size, alignment and behavior to cl_short4. See section 6.1.5. */ +typedef cl_short4 cl_short3; + +typedef union +{ + cl_short CL_ALIGNED(16) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_short x, y, z, w; }; + __extension__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_short4 lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2[4]; +#endif +#if defined( __CL_SHORT4__) + __cl_short4 v4[2]; +#endif +#if defined( __CL_SHORT8__ ) + __cl_short8 v8; +#endif +}cl_short8; + +typedef union +{ + cl_short CL_ALIGNED(32) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_short x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_short8 lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2[8]; +#endif +#if defined( __CL_SHORT4__) + __cl_short4 v4[4]; +#endif +#if defined( __CL_SHORT8__ ) + __cl_short8 v8[2]; +#endif +#if defined( __CL_SHORT16__ ) + __cl_short16 v16; +#endif +}cl_short16; + + +/* ---- cl_ushortn ---- */ +typedef union +{ + cl_ushort CL_ALIGNED(4) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ushort x, y; }; + __extension__ struct{ cl_ushort s0, s1; }; + __extension__ struct{ cl_ushort lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2; +#endif +}cl_ushort2; + +typedef union +{ + cl_ushort CL_ALIGNED(8) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ushort x, y, z, w; }; + __extension__ struct{ cl_ushort s0, s1, s2, s3; }; + __extension__ struct{ cl_ushort2 lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2[2]; +#endif +#if defined( __CL_USHORT4__) + __cl_ushort4 v4; +#endif +}cl_ushort4; + +/* cl_ushort3 is identical in size, alignment and behavior to cl_ushort4. See section 6.1.5. */ +typedef cl_ushort4 cl_ushort3; + +typedef union +{ + cl_ushort CL_ALIGNED(16) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ushort x, y, z, w; }; + __extension__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_ushort4 lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2[4]; +#endif +#if defined( __CL_USHORT4__) + __cl_ushort4 v4[2]; +#endif +#if defined( __CL_USHORT8__ ) + __cl_ushort8 v8; +#endif +}cl_ushort8; + +typedef union +{ + cl_ushort CL_ALIGNED(32) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ushort x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_ushort8 lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2[8]; +#endif +#if defined( __CL_USHORT4__) + __cl_ushort4 v4[4]; +#endif +#if defined( __CL_USHORT8__ ) + __cl_ushort8 v8[2]; +#endif +#if defined( __CL_USHORT16__ ) + __cl_ushort16 v16; +#endif +}cl_ushort16; + +/* ---- cl_intn ---- */ +typedef union +{ + cl_int CL_ALIGNED(8) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_int x, y; }; + __extension__ struct{ cl_int s0, s1; }; + __extension__ struct{ cl_int lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2; +#endif +}cl_int2; + +typedef union +{ + cl_int CL_ALIGNED(16) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_int x, y, z, w; }; + __extension__ struct{ cl_int s0, s1, s2, s3; }; + __extension__ struct{ cl_int2 lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2[2]; +#endif +#if defined( __CL_INT4__) + __cl_int4 v4; +#endif +}cl_int4; + +/* cl_int3 is identical in size, alignment and behavior to cl_int4. See section 6.1.5. */ +typedef cl_int4 cl_int3; + +typedef union +{ + cl_int CL_ALIGNED(32) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_int x, y, z, w; }; + __extension__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_int4 lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2[4]; +#endif +#if defined( __CL_INT4__) + __cl_int4 v4[2]; +#endif +#if defined( __CL_INT8__ ) + __cl_int8 v8; +#endif +}cl_int8; + +typedef union +{ + cl_int CL_ALIGNED(64) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_int x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_int8 lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2[8]; +#endif +#if defined( __CL_INT4__) + __cl_int4 v4[4]; +#endif +#if defined( __CL_INT8__ ) + __cl_int8 v8[2]; +#endif +#if defined( __CL_INT16__ ) + __cl_int16 v16; +#endif +}cl_int16; + + +/* ---- cl_uintn ---- */ +typedef union +{ + cl_uint CL_ALIGNED(8) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uint x, y; }; + __extension__ struct{ cl_uint s0, s1; }; + __extension__ struct{ cl_uint lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2; +#endif +}cl_uint2; + +typedef union +{ + cl_uint CL_ALIGNED(16) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uint x, y, z, w; }; + __extension__ struct{ cl_uint s0, s1, s2, s3; }; + __extension__ struct{ cl_uint2 lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2[2]; +#endif +#if defined( __CL_UINT4__) + __cl_uint4 v4; +#endif +}cl_uint4; + +/* cl_uint3 is identical in size, alignment and behavior to cl_uint4. See section 6.1.5. */ +typedef cl_uint4 cl_uint3; + +typedef union +{ + cl_uint CL_ALIGNED(32) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uint x, y, z, w; }; + __extension__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_uint4 lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2[4]; +#endif +#if defined( __CL_UINT4__) + __cl_uint4 v4[2]; +#endif +#if defined( __CL_UINT8__ ) + __cl_uint8 v8; +#endif +}cl_uint8; + +typedef union +{ + cl_uint CL_ALIGNED(64) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_uint x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_uint8 lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2[8]; +#endif +#if defined( __CL_UINT4__) + __cl_uint4 v4[4]; +#endif +#if defined( __CL_UINT8__ ) + __cl_uint8 v8[2]; +#endif +#if defined( __CL_UINT16__ ) + __cl_uint16 v16; +#endif +}cl_uint16; + +/* ---- cl_longn ---- */ +typedef union +{ + cl_long CL_ALIGNED(16) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_long x, y; }; + __extension__ struct{ cl_long s0, s1; }; + __extension__ struct{ cl_long lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2; +#endif +}cl_long2; + +typedef union +{ + cl_long CL_ALIGNED(32) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_long x, y, z, w; }; + __extension__ struct{ cl_long s0, s1, s2, s3; }; + __extension__ struct{ cl_long2 lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2[2]; +#endif +#if defined( __CL_LONG4__) + __cl_long4 v4; +#endif +}cl_long4; + +/* cl_long3 is identical in size, alignment and behavior to cl_long4. See section 6.1.5. */ +typedef cl_long4 cl_long3; + +typedef union +{ + cl_long CL_ALIGNED(64) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_long x, y, z, w; }; + __extension__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_long4 lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2[4]; +#endif +#if defined( __CL_LONG4__) + __cl_long4 v4[2]; +#endif +#if defined( __CL_LONG8__ ) + __cl_long8 v8; +#endif +}cl_long8; + +typedef union +{ + cl_long CL_ALIGNED(128) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_long x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_long8 lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2[8]; +#endif +#if defined( __CL_LONG4__) + __cl_long4 v4[4]; +#endif +#if defined( __CL_LONG8__ ) + __cl_long8 v8[2]; +#endif +#if defined( __CL_LONG16__ ) + __cl_long16 v16; +#endif +}cl_long16; + + +/* ---- cl_ulongn ---- */ +typedef union +{ + cl_ulong CL_ALIGNED(16) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ulong x, y; }; + __extension__ struct{ cl_ulong s0, s1; }; + __extension__ struct{ cl_ulong lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2; +#endif +}cl_ulong2; + +typedef union +{ + cl_ulong CL_ALIGNED(32) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ulong x, y, z, w; }; + __extension__ struct{ cl_ulong s0, s1, s2, s3; }; + __extension__ struct{ cl_ulong2 lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2[2]; +#endif +#if defined( __CL_ULONG4__) + __cl_ulong4 v4; +#endif +}cl_ulong4; + +/* cl_ulong3 is identical in size, alignment and behavior to cl_ulong4. See section 6.1.5. */ +typedef cl_ulong4 cl_ulong3; + +typedef union +{ + cl_ulong CL_ALIGNED(64) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ulong x, y, z, w; }; + __extension__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_ulong4 lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2[4]; +#endif +#if defined( __CL_ULONG4__) + __cl_ulong4 v4[2]; +#endif +#if defined( __CL_ULONG8__ ) + __cl_ulong8 v8; +#endif +}cl_ulong8; + +typedef union +{ + cl_ulong CL_ALIGNED(128) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_ulong x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_ulong8 lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2[8]; +#endif +#if defined( __CL_ULONG4__) + __cl_ulong4 v4[4]; +#endif +#if defined( __CL_ULONG8__ ) + __cl_ulong8 v8[2]; +#endif +#if defined( __CL_ULONG16__ ) + __cl_ulong16 v16; +#endif +}cl_ulong16; + + +/* --- cl_floatn ---- */ + +typedef union +{ + cl_float CL_ALIGNED(8) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_float x, y; }; + __extension__ struct{ cl_float s0, s1; }; + __extension__ struct{ cl_float lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2; +#endif +}cl_float2; + +typedef union +{ + cl_float CL_ALIGNED(16) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_float x, y, z, w; }; + __extension__ struct{ cl_float s0, s1, s2, s3; }; + __extension__ struct{ cl_float2 lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2[2]; +#endif +#if defined( __CL_FLOAT4__) + __cl_float4 v4; +#endif +}cl_float4; + +/* cl_float3 is identical in size, alignment and behavior to cl_float4. See section 6.1.5. */ +typedef cl_float4 cl_float3; + +typedef union +{ + cl_float CL_ALIGNED(32) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_float x, y, z, w; }; + __extension__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_float4 lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2[4]; +#endif +#if defined( __CL_FLOAT4__) + __cl_float4 v4[2]; +#endif +#if defined( __CL_FLOAT8__ ) + __cl_float8 v8; +#endif +}cl_float8; + +typedef union +{ + cl_float CL_ALIGNED(64) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_float x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_float8 lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2[8]; +#endif +#if defined( __CL_FLOAT4__) + __cl_float4 v4[4]; +#endif +#if defined( __CL_FLOAT8__ ) + __cl_float8 v8[2]; +#endif +#if defined( __CL_FLOAT16__ ) + __cl_float16 v16; +#endif +}cl_float16; + +/* --- cl_doublen ---- */ + +typedef union +{ + cl_double CL_ALIGNED(16) s[2]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_double x, y; }; + __extension__ struct{ cl_double s0, s1; }; + __extension__ struct{ cl_double lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2; +#endif +}cl_double2; + +typedef union +{ + cl_double CL_ALIGNED(32) s[4]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_double x, y, z, w; }; + __extension__ struct{ cl_double s0, s1, s2, s3; }; + __extension__ struct{ cl_double2 lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2[2]; +#endif +#if defined( __CL_DOUBLE4__) + __cl_double4 v4; +#endif +}cl_double4; + +/* cl_double3 is identical in size, alignment and behavior to cl_double4. See section 6.1.5. */ +typedef cl_double4 cl_double3; + +typedef union +{ + cl_double CL_ALIGNED(64) s[8]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_double x, y, z, w; }; + __extension__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7; }; + __extension__ struct{ cl_double4 lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2[4]; +#endif +#if defined( __CL_DOUBLE4__) + __cl_double4 v4[2]; +#endif +#if defined( __CL_DOUBLE8__ ) + __cl_double8 v8; +#endif +}cl_double8; + +typedef union +{ + cl_double CL_ALIGNED(128) s[16]; +#if (defined( __GNUC__) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) + __extension__ struct{ cl_double x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __extension__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __extension__ struct{ cl_double8 lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2[8]; +#endif +#if defined( __CL_DOUBLE4__) + __cl_double4 v4[4]; +#endif +#if defined( __CL_DOUBLE8__ ) + __cl_double8 v8[2]; +#endif +#if defined( __CL_DOUBLE16__ ) + __cl_double16 v16; +#endif +}cl_double16; + +/* Macro to facilitate debugging + * Usage: + * Place CL_PROGRAM_STRING_DEBUG_INFO on the line before the first line of your source. + * The first line ends with: CL_PROGRAM_STRING_DEBUG_INFO \" + * Each line thereafter of OpenCL C source must end with: \n\ + * The last line ends in "; + * + * Example: + * + * const char *my_program = CL_PROGRAM_STRING_DEBUG_INFO "\ + * kernel void foo( int a, float * b ) \n\ + * { \n\ + * // my comment \n\ + * *b[ get_global_id(0)] = a; \n\ + * } \n\ + * "; + * + * This should correctly set up the line, (column) and file information for your source + * string so you can do source level debugging. + */ +#define __CL_STRINGIFY( _x ) # _x +#define _CL_STRINGIFY( _x ) __CL_STRINGIFY( _x ) +#define CL_PROGRAM_STRING_DEBUG_INFO "#line " _CL_STRINGIFY(__LINE__) " \"" __FILE__ "\" \n\n" + +#ifdef __cplusplus +} +#endif + +#endif /* __CL_PLATFORM_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/opencl.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/opencl.h new file mode 100644 index 0000000000..3f00524719 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/opencl.h @@ -0,0 +1,54 @@ +/******************************************************************************* + * Copyright (c) 2008-2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_H +#define __OPENCL_H + +#ifdef __cplusplus +extern "C" { +#endif + +#ifdef __APPLE__ + +#include +#include +#include +#include + +#else + +#include +#include +#include +#include + +#endif + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl1.2/CL/spir.h b/opencl/api/opencl/khronos/headers/opencl1.2/CL/spir.h new file mode 100644 index 0000000000..5f0dbf6dea --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl1.2/CL/spir.h @@ -0,0 +1,118 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. All rights reserved. + * + * NOTICE TO KHRONOS MEMBER: + * + * AMD has assigned the copyright for this object code to Khronos. + * This object code is subject to Khronos ownership rights under U.S. and + * international Copyright laws. + * + * Permission is hereby granted, free of charge, to any Khronos Member + * obtaining a copy of this software and/or associated documentation files + * (the "Materials"), to use, copy, modify and merge the Materials in object + * form only and to publish, distribute and/or sell copies of the Materials + * solely in object code form as part of conformant OpenCL API implementations, + * subject to the following conditions: + * + * Khronos Members shall ensure that their respective ICD implementation, + * that is installed over another Khronos Members' ICD implementation, will + * continue to support all OpenCL devices (hardware and software) supported + * by the replaced ICD implementation. For the purposes of this notice, "ICD" + * shall mean a library that presents an implementation of the OpenCL API for + * the purpose routing API calls to different vendor implementation. + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * KHRONOS AND AMD MAKE NO REPRESENTATION ABOUT THE SUITABILITY OF THIS + * SOURCE CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" WITHOUT EXPRESS OR + * IMPLIED WARRANTY OF ANY KIND. KHRONOS AND AMD DISCLAIM ALL WARRANTIES + * WITH REGARD TO THIS SOURCE CODE, INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY, NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE. + * IN NO EVENT SHALL KHRONOS OR AMD BE LIABLE FOR ANY SPECIAL, INDIRECT, + * INCIDENTAL, OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING + * FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, + * NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH + * THE USE OR PERFORMANCE OF THIS SOURCE CODE. + * + * U.S. Government End Users. This source code is a "commercial item" as + * that term is defined at 48 C.F.R. 2.101 (OCT 1995), consisting of + * "commercial computer software" and "commercial computer software + * documentation" as such terms are used in 48 C.F.R. 12.212 (SEPT 1995) + * and is provided to the U.S. Government only as a commercial end item. + * Consistent with 48 C.F.R.12.212 and 48 C.F.R. 227.7202-1 through + * 227.7202-4 (JUNE 1995), all U.S. Government End Users acquire the + * source code with only those rights set forth herein. + */ +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ +#ifndef __SPIR_H +#define __SPIR_H + +#ifdef __cplusplus +extern "C" { +#endif +// CLS is for CL SPIR. +typedef enum { + CLS_PRIVATE_AS = 0, + CLS_GLOBAL_AS = 1, + CLS_CONSTANT_AS = 2, + CLS_LOCAL_AS = 3, + CLS_GLOBAL_HOST_AS = 4, + CLS_CONSTANT_HOST_AS = 5, + CLS_REGION_AS_AMD = 16, +} CLS_ADDRESS_SPACES; + +// Table 13 +typedef enum { + CLS_ADDRESS_MIRRORED_REPEAT = 0, + CLS_ADDRESS_REPEAT = 1, + CLS_ADDRESS_CLAMP_TO_EDGE = 2, + CLS_ADDRESS_CLAMP = 3, + CLS_ADDRESS_NONE = 4 +} CLS_ADDRESSING_MODES; + +// Table 13 +typedef enum { + CLS_FILTER_NEAREST = 0, + CLS_FILTER_LINEAR = 1 +} CLS_FILTER_MODE; + +// Table 13 +typedef enum { + CLS_NORMALIZED_COORDS_TRUE = 0, + CLS_NORMALIZED_COORDS_FALSE = 1 +} CLS_NORMALIZED_COORDS; + +// Section 2.3 +typedef enum { + CLS_READ_ONLY = 0, + CLS_WRITE_ONLY = 1, + CLS_READ_WRITE = 2, + CLS_NONE = 3 +} CLS_ACCESS_QUALIFIERS; + +// Table 14 +typedef enum { + CLS_ARG_CONST = 1, + CLS_ARG_RESTRICT = 2, + CLS_ARG_VOLATILE = 4, + CLS_ARG_NONE = 0, + CLS_ARG_MASK = 0xf, + CLS_ARG_PIPE = 8 +} CLS_ARGTYPE_QUALIFIERS; + +static const char* SPIR_TRIPLE_32BIT = "spir-unknown-unknown"; +static const char* SPIR_TRIPLE_64BIT = "spir64-unknown-unknown"; +static const char* SPIR_DATA_LAYOUT_32BIT = "p:32:32:32-i1:8:8-i8:8:8" + "-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16" + "-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128" + "-v192:256:256-v256:256:256-v512:512:512-v1024:1024:1024"; +static const char* SPIR_DATA_LAYOUT_64BIT = "p:64:64:64-i1:8:8-i8:8:8" + "-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16" + "-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128" + "-v192:256:256-v256:256:256-v512:512:512-v1024:1024:1024"; + +#ifdef __cplusplus +} +#endif +#endif // __SPIR_H diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.h new file mode 100644 index 0000000000..f09d9f4dc7 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.h @@ -0,0 +1,1380 @@ +/******************************************************************************* + * Copyright (c) 2008 - 2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +#ifndef __OPENCL_CL_H +#define __OPENCL_CL_H + +#ifdef __APPLE__ +#include +#else +#include +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +/******************************************************************************/ + +typedef struct _cl_platform_id * cl_platform_id; +typedef struct _cl_device_id * cl_device_id; +typedef struct _cl_context * cl_context; +typedef struct _cl_command_queue * cl_command_queue; +typedef struct _cl_mem * cl_mem; +typedef struct _cl_program * cl_program; +typedef struct _cl_kernel * cl_kernel; +typedef struct _cl_event * cl_event; +typedef struct _cl_sampler * cl_sampler; + +typedef cl_uint cl_bool; /* WARNING! Unlike cl_ types in cl_platform.h, cl_bool is not guaranteed to be the same size as the bool in kernels. */ +typedef cl_ulong cl_bitfield; +typedef cl_bitfield cl_device_type; +typedef cl_uint cl_platform_info; +typedef cl_uint cl_device_info; +typedef cl_bitfield cl_device_fp_config; +typedef cl_uint cl_device_mem_cache_type; +typedef cl_uint cl_device_local_mem_type; +typedef cl_bitfield cl_device_exec_capabilities; +typedef cl_bitfield cl_device_svm_capabilities; +typedef cl_bitfield cl_command_queue_properties; +typedef intptr_t cl_device_partition_property; +typedef cl_bitfield cl_device_affinity_domain; + +typedef intptr_t cl_context_properties; +typedef cl_uint cl_context_info; +typedef cl_bitfield cl_queue_properties; +typedef cl_uint cl_command_queue_info; +typedef cl_uint cl_channel_order; +typedef cl_uint cl_channel_type; +typedef cl_bitfield cl_mem_flags; +typedef cl_bitfield cl_svm_mem_flags; +typedef cl_uint cl_mem_object_type; +typedef cl_uint cl_mem_info; +typedef cl_bitfield cl_mem_migration_flags; +typedef cl_uint cl_image_info; +typedef cl_uint cl_buffer_create_type; +typedef cl_uint cl_addressing_mode; +typedef cl_uint cl_filter_mode; +typedef cl_uint cl_sampler_info; +typedef cl_bitfield cl_map_flags; +typedef intptr_t cl_pipe_properties; +typedef cl_uint cl_pipe_info; +typedef cl_uint cl_program_info; +typedef cl_uint cl_program_build_info; +typedef cl_uint cl_program_binary_type; +typedef cl_int cl_build_status; +typedef cl_uint cl_kernel_info; +typedef cl_uint cl_kernel_arg_info; +typedef cl_uint cl_kernel_arg_address_qualifier; +typedef cl_uint cl_kernel_arg_access_qualifier; +typedef cl_bitfield cl_kernel_arg_type_qualifier; +typedef cl_uint cl_kernel_work_group_info; +typedef cl_uint cl_event_info; +typedef cl_uint cl_command_type; +typedef cl_uint cl_profiling_info; +typedef cl_bitfield cl_sampler_properties; +typedef cl_uint cl_kernel_exec_info; + +typedef struct _cl_image_format { + cl_channel_order image_channel_order; + cl_channel_type image_channel_data_type; +} cl_image_format; + +typedef struct _cl_image_desc { + cl_mem_object_type image_type; + size_t image_width; + size_t image_height; + size_t image_depth; + size_t image_array_size; + size_t image_row_pitch; + size_t image_slice_pitch; + cl_uint num_mip_levels; + cl_uint num_samples; + union { + cl_mem buffer; + cl_mem mem_object; + }; +} cl_image_desc; + +typedef struct _cl_buffer_region { + size_t origin; + size_t size; +} cl_buffer_region; + + +/******************************************************************************/ + +/* Error Codes */ +#define CL_SUCCESS 0 +#define CL_DEVICE_NOT_FOUND -1 +#define CL_DEVICE_NOT_AVAILABLE -2 +#define CL_COMPILER_NOT_AVAILABLE -3 +#define CL_MEM_OBJECT_ALLOCATION_FAILURE -4 +#define CL_OUT_OF_RESOURCES -5 +#define CL_OUT_OF_HOST_MEMORY -6 +#define CL_PROFILING_INFO_NOT_AVAILABLE -7 +#define CL_MEM_COPY_OVERLAP -8 +#define CL_IMAGE_FORMAT_MISMATCH -9 +#define CL_IMAGE_FORMAT_NOT_SUPPORTED -10 +#define CL_BUILD_PROGRAM_FAILURE -11 +#define CL_MAP_FAILURE -12 +#define CL_MISALIGNED_SUB_BUFFER_OFFSET -13 +#define CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST -14 +#define CL_COMPILE_PROGRAM_FAILURE -15 +#define CL_LINKER_NOT_AVAILABLE -16 +#define CL_LINK_PROGRAM_FAILURE -17 +#define CL_DEVICE_PARTITION_FAILED -18 +#define CL_KERNEL_ARG_INFO_NOT_AVAILABLE -19 + +#define CL_INVALID_VALUE -30 +#define CL_INVALID_DEVICE_TYPE -31 +#define CL_INVALID_PLATFORM -32 +#define CL_INVALID_DEVICE -33 +#define CL_INVALID_CONTEXT -34 +#define CL_INVALID_QUEUE_PROPERTIES -35 +#define CL_INVALID_COMMAND_QUEUE -36 +#define CL_INVALID_HOST_PTR -37 +#define CL_INVALID_MEM_OBJECT -38 +#define CL_INVALID_IMAGE_FORMAT_DESCRIPTOR -39 +#define CL_INVALID_IMAGE_SIZE -40 +#define CL_INVALID_SAMPLER -41 +#define CL_INVALID_BINARY -42 +#define CL_INVALID_BUILD_OPTIONS -43 +#define CL_INVALID_PROGRAM -44 +#define CL_INVALID_PROGRAM_EXECUTABLE -45 +#define CL_INVALID_KERNEL_NAME -46 +#define CL_INVALID_KERNEL_DEFINITION -47 +#define CL_INVALID_KERNEL -48 +#define CL_INVALID_ARG_INDEX -49 +#define CL_INVALID_ARG_VALUE -50 +#define CL_INVALID_ARG_SIZE -51 +#define CL_INVALID_KERNEL_ARGS -52 +#define CL_INVALID_WORK_DIMENSION -53 +#define CL_INVALID_WORK_GROUP_SIZE -54 +#define CL_INVALID_WORK_ITEM_SIZE -55 +#define CL_INVALID_GLOBAL_OFFSET -56 +#define CL_INVALID_EVENT_WAIT_LIST -57 +#define CL_INVALID_EVENT -58 +#define CL_INVALID_OPERATION -59 +#define CL_INVALID_GL_OBJECT -60 +#define CL_INVALID_BUFFER_SIZE -61 +#define CL_INVALID_MIP_LEVEL -62 +#define CL_INVALID_GLOBAL_WORK_SIZE -63 +#define CL_INVALID_PROPERTY -64 +#define CL_INVALID_IMAGE_DESCRIPTOR -65 +#define CL_INVALID_COMPILER_OPTIONS -66 +#define CL_INVALID_LINKER_OPTIONS -67 +#define CL_INVALID_DEVICE_PARTITION_COUNT -68 +#define CL_INVALID_PIPE_SIZE -69 +#define CL_INVALID_DEVICE_QUEUE -70 + +/* OpenCL Version */ +#define CL_VERSION_1_0 1 +#define CL_VERSION_1_1 1 +#define CL_VERSION_1_2 1 +#define CL_VERSION_2_0 1 + +/* cl_bool */ +#define CL_FALSE 0 +#define CL_TRUE 1 +#define CL_BLOCKING CL_TRUE +#define CL_NON_BLOCKING CL_FALSE + +/* cl_platform_info */ +#define CL_PLATFORM_PROFILE 0x0900 +#define CL_PLATFORM_VERSION 0x0901 +#define CL_PLATFORM_NAME 0x0902 +#define CL_PLATFORM_VENDOR 0x0903 +#define CL_PLATFORM_EXTENSIONS 0x0904 + +/* cl_device_type - bitfield */ +#define CL_DEVICE_TYPE_DEFAULT (1 << 0) +#define CL_DEVICE_TYPE_CPU (1 << 1) +#define CL_DEVICE_TYPE_GPU (1 << 2) +#define CL_DEVICE_TYPE_ACCELERATOR (1 << 3) +#define CL_DEVICE_TYPE_CUSTOM (1 << 4) +#define CL_DEVICE_TYPE_ALL 0xFFFFFFFF + +/* cl_device_info */ +#define CL_DEVICE_TYPE 0x1000 +#define CL_DEVICE_VENDOR_ID 0x1001 +#define CL_DEVICE_MAX_COMPUTE_UNITS 0x1002 +#define CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 0x1003 +#define CL_DEVICE_MAX_WORK_GROUP_SIZE 0x1004 +#define CL_DEVICE_MAX_WORK_ITEM_SIZES 0x1005 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 0x1006 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT 0x1007 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT 0x1008 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG 0x1009 +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT 0x100A +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE 0x100B +#define CL_DEVICE_MAX_CLOCK_FREQUENCY 0x100C +#define CL_DEVICE_ADDRESS_BITS 0x100D +#define CL_DEVICE_MAX_READ_IMAGE_ARGS 0x100E +#define CL_DEVICE_MAX_WRITE_IMAGE_ARGS 0x100F +#define CL_DEVICE_MAX_MEM_ALLOC_SIZE 0x1010 +#define CL_DEVICE_IMAGE2D_MAX_WIDTH 0x1011 +#define CL_DEVICE_IMAGE2D_MAX_HEIGHT 0x1012 +#define CL_DEVICE_IMAGE3D_MAX_WIDTH 0x1013 +#define CL_DEVICE_IMAGE3D_MAX_HEIGHT 0x1014 +#define CL_DEVICE_IMAGE3D_MAX_DEPTH 0x1015 +#define CL_DEVICE_IMAGE_SUPPORT 0x1016 +#define CL_DEVICE_MAX_PARAMETER_SIZE 0x1017 +#define CL_DEVICE_MAX_SAMPLERS 0x1018 +#define CL_DEVICE_MEM_BASE_ADDR_ALIGN 0x1019 +#define CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE 0x101A +#define CL_DEVICE_SINGLE_FP_CONFIG 0x101B +#define CL_DEVICE_GLOBAL_MEM_CACHE_TYPE 0x101C +#define CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE 0x101D +#define CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 0x101E +#define CL_DEVICE_GLOBAL_MEM_SIZE 0x101F +#define CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 0x1020 +#define CL_DEVICE_MAX_CONSTANT_ARGS 0x1021 +#define CL_DEVICE_LOCAL_MEM_TYPE 0x1022 +#define CL_DEVICE_LOCAL_MEM_SIZE 0x1023 +#define CL_DEVICE_ERROR_CORRECTION_SUPPORT 0x1024 +#define CL_DEVICE_PROFILING_TIMER_RESOLUTION 0x1025 +#define CL_DEVICE_ENDIAN_LITTLE 0x1026 +#define CL_DEVICE_AVAILABLE 0x1027 +#define CL_DEVICE_COMPILER_AVAILABLE 0x1028 +#define CL_DEVICE_EXECUTION_CAPABILITIES 0x1029 +#define CL_DEVICE_QUEUE_PROPERTIES 0x102A /* deprecated */ +#define CL_DEVICE_QUEUE_ON_HOST_PROPERTIES 0x102A +#define CL_DEVICE_NAME 0x102B +#define CL_DEVICE_VENDOR 0x102C +#define CL_DRIVER_VERSION 0x102D +#define CL_DEVICE_PROFILE 0x102E +#define CL_DEVICE_VERSION 0x102F +#define CL_DEVICE_EXTENSIONS 0x1030 +#define CL_DEVICE_PLATFORM 0x1031 +#define CL_DEVICE_DOUBLE_FP_CONFIG 0x1032 +/* 0x1033 reserved for CL_DEVICE_HALF_FP_CONFIG */ +#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF 0x1034 +#define CL_DEVICE_HOST_UNIFIED_MEMORY 0x1035 /* deprecated */ +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR 0x1036 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT 0x1037 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 0x1038 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 0x1039 +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 0x103A +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 0x103B +#define CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 0x103C +#define CL_DEVICE_OPENCL_C_VERSION 0x103D +#define CL_DEVICE_LINKER_AVAILABLE 0x103E +#define CL_DEVICE_BUILT_IN_KERNELS 0x103F +#define CL_DEVICE_IMAGE_MAX_BUFFER_SIZE 0x1040 +#define CL_DEVICE_IMAGE_MAX_ARRAY_SIZE 0x1041 +#define CL_DEVICE_PARENT_DEVICE 0x1042 +#define CL_DEVICE_PARTITION_MAX_SUB_DEVICES 0x1043 +#define CL_DEVICE_PARTITION_PROPERTIES 0x1044 +#define CL_DEVICE_PARTITION_AFFINITY_DOMAIN 0x1045 +#define CL_DEVICE_PARTITION_TYPE 0x1046 +#define CL_DEVICE_REFERENCE_COUNT 0x1047 +#define CL_DEVICE_PREFERRED_INTEROP_USER_SYNC 0x1048 +#define CL_DEVICE_PRINTF_BUFFER_SIZE 0x1049 +#define CL_DEVICE_IMAGE_PITCH_ALIGNMENT 0x104A +#define CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT 0x104B +#define CL_DEVICE_MAX_READ_WRITE_IMAGE_ARGS 0x104C +#define CL_DEVICE_MAX_GLOBAL_VARIABLE_SIZE 0x104D +#define CL_DEVICE_QUEUE_ON_DEVICE_PROPERTIES 0x104E +#define CL_DEVICE_QUEUE_ON_DEVICE_PREFERRED_SIZE 0x104F +#define CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE 0x1050 +#define CL_DEVICE_MAX_ON_DEVICE_QUEUES 0x1051 +#define CL_DEVICE_MAX_ON_DEVICE_EVENTS 0x1052 +#define CL_DEVICE_SVM_CAPABILITIES 0x1053 +#define CL_DEVICE_GLOBAL_VARIABLE_PREFERRED_TOTAL_SIZE 0x1054 +#define CL_DEVICE_MAX_PIPE_ARGS 0x1055 +#define CL_DEVICE_PIPE_MAX_ACTIVE_RESERVATIONS 0x1056 +#define CL_DEVICE_PIPE_MAX_PACKET_SIZE 0x1057 +#define CL_DEVICE_PREFERRED_PLATFORM_ATOMIC_ALIGNMENT 0x1058 +#define CL_DEVICE_PREFERRED_GLOBAL_ATOMIC_ALIGNMENT 0x1059 +#define CL_DEVICE_PREFERRED_LOCAL_ATOMIC_ALIGNMENT 0x105A + +/* cl_device_fp_config - bitfield */ +#define CL_FP_DENORM (1 << 0) +#define CL_FP_INF_NAN (1 << 1) +#define CL_FP_ROUND_TO_NEAREST (1 << 2) +#define CL_FP_ROUND_TO_ZERO (1 << 3) +#define CL_FP_ROUND_TO_INF (1 << 4) +#define CL_FP_FMA (1 << 5) +#define CL_FP_SOFT_FLOAT (1 << 6) +#define CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT (1 << 7) + +/* cl_device_mem_cache_type */ +#define CL_NONE 0x0 +#define CL_READ_ONLY_CACHE 0x1 +#define CL_READ_WRITE_CACHE 0x2 + +/* cl_device_local_mem_type */ +#define CL_LOCAL 0x1 +#define CL_GLOBAL 0x2 + +/* cl_device_exec_capabilities - bitfield */ +#define CL_EXEC_KERNEL (1 << 0) +#define CL_EXEC_NATIVE_KERNEL (1 << 1) + +/* cl_command_queue_properties - bitfield */ +#define CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE (1 << 0) +#define CL_QUEUE_PROFILING_ENABLE (1 << 1) +#define CL_QUEUE_ON_DEVICE (1 << 2) +#define CL_QUEUE_ON_DEVICE_DEFAULT (1 << 3) + +/* cl_context_info */ +#define CL_CONTEXT_REFERENCE_COUNT 0x1080 +#define CL_CONTEXT_DEVICES 0x1081 +#define CL_CONTEXT_PROPERTIES 0x1082 +#define CL_CONTEXT_NUM_DEVICES 0x1083 + +/* cl_context_properties */ +#define CL_CONTEXT_PLATFORM 0x1084 +#define CL_CONTEXT_INTEROP_USER_SYNC 0x1085 + +/* cl_device_partition_property */ +#define CL_DEVICE_PARTITION_EQUALLY 0x1086 +#define CL_DEVICE_PARTITION_BY_COUNTS 0x1087 +#define CL_DEVICE_PARTITION_BY_COUNTS_LIST_END 0x0 +#define CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN 0x1088 + +/* cl_device_affinity_domain */ +#define CL_DEVICE_AFFINITY_DOMAIN_NUMA (1 << 0) +#define CL_DEVICE_AFFINITY_DOMAIN_L4_CACHE (1 << 1) +#define CL_DEVICE_AFFINITY_DOMAIN_L3_CACHE (1 << 2) +#define CL_DEVICE_AFFINITY_DOMAIN_L2_CACHE (1 << 3) +#define CL_DEVICE_AFFINITY_DOMAIN_L1_CACHE (1 << 4) +#define CL_DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE (1 << 5) + +/* cl_device_svm_capabilities */ +#define CL_DEVICE_SVM_COARSE_GRAIN_BUFFER (1 << 0) +#define CL_DEVICE_SVM_FINE_GRAIN_BUFFER (1 << 1) +#define CL_DEVICE_SVM_FINE_GRAIN_SYSTEM (1 << 2) +#define CL_DEVICE_SVM_ATOMICS (1 << 3) + +/* cl_command_queue_info */ +#define CL_QUEUE_CONTEXT 0x1090 +#define CL_QUEUE_DEVICE 0x1091 +#define CL_QUEUE_REFERENCE_COUNT 0x1092 +#define CL_QUEUE_PROPERTIES 0x1093 +#define CL_QUEUE_SIZE 0x1094 + +/* cl_mem_flags and cl_svm_mem_flags - bitfield */ +#define CL_MEM_READ_WRITE (1 << 0) +#define CL_MEM_WRITE_ONLY (1 << 1) +#define CL_MEM_READ_ONLY (1 << 2) +#define CL_MEM_USE_HOST_PTR (1 << 3) +#define CL_MEM_ALLOC_HOST_PTR (1 << 4) +#define CL_MEM_COPY_HOST_PTR (1 << 5) +/* reserved (1 << 6) */ +#define CL_MEM_HOST_WRITE_ONLY (1 << 7) +#define CL_MEM_HOST_READ_ONLY (1 << 8) +#define CL_MEM_HOST_NO_ACCESS (1 << 9) +#define CL_MEM_SVM_FINE_GRAIN_BUFFER (1 << 10) /* used by cl_svm_mem_flags only */ +#define CL_MEM_SVM_ATOMICS (1 << 11) /* used by cl_svm_mem_flags only */ + +/* cl_mem_migration_flags - bitfield */ +#define CL_MIGRATE_MEM_OBJECT_HOST (1 << 0) +#define CL_MIGRATE_MEM_OBJECT_CONTENT_UNDEFINED (1 << 1) + +/* cl_channel_order */ +#define CL_R 0x10B0 +#define CL_A 0x10B1 +#define CL_RG 0x10B2 +#define CL_RA 0x10B3 +#define CL_RGB 0x10B4 +#define CL_RGBA 0x10B5 +#define CL_BGRA 0x10B6 +#define CL_ARGB 0x10B7 +#define CL_INTENSITY 0x10B8 +#define CL_LUMINANCE 0x10B9 +#define CL_Rx 0x10BA +#define CL_RGx 0x10BB +#define CL_RGBx 0x10BC +#define CL_DEPTH 0x10BD +#define CL_DEPTH_STENCIL 0x10BE +#define CL_sRGB 0x10BF +#define CL_sRGBx 0x10C0 +#define CL_sRGBA 0x10C1 +#define CL_sBGRA 0x10C2 +#define CL_ABGR 0x10C3 + +/* cl_channel_type */ +#define CL_SNORM_INT8 0x10D0 +#define CL_SNORM_INT16 0x10D1 +#define CL_UNORM_INT8 0x10D2 +#define CL_UNORM_INT16 0x10D3 +#define CL_UNORM_SHORT_565 0x10D4 +#define CL_UNORM_SHORT_555 0x10D5 +#define CL_UNORM_INT_101010 0x10D6 +#define CL_SIGNED_INT8 0x10D7 +#define CL_SIGNED_INT16 0x10D8 +#define CL_SIGNED_INT32 0x10D9 +#define CL_UNSIGNED_INT8 0x10DA +#define CL_UNSIGNED_INT16 0x10DB +#define CL_UNSIGNED_INT32 0x10DC +#define CL_HALF_FLOAT 0x10DD +#define CL_FLOAT 0x10DE +#define CL_UNORM_INT24 0x10DF + +/* cl_mem_object_type */ +#define CL_MEM_OBJECT_BUFFER 0x10F0 +#define CL_MEM_OBJECT_IMAGE2D 0x10F1 +#define CL_MEM_OBJECT_IMAGE3D 0x10F2 +#define CL_MEM_OBJECT_IMAGE2D_ARRAY 0x10F3 +#define CL_MEM_OBJECT_IMAGE1D 0x10F4 +#define CL_MEM_OBJECT_IMAGE1D_ARRAY 0x10F5 +#define CL_MEM_OBJECT_IMAGE1D_BUFFER 0x10F6 +#define CL_MEM_OBJECT_PIPE 0x10F7 + +/* cl_mem_info */ +#define CL_MEM_TYPE 0x1100 +#define CL_MEM_FLAGS 0x1101 +#define CL_MEM_SIZE 0x1102 +#define CL_MEM_HOST_PTR 0x1103 +#define CL_MEM_MAP_COUNT 0x1104 +#define CL_MEM_REFERENCE_COUNT 0x1105 +#define CL_MEM_CONTEXT 0x1106 +#define CL_MEM_ASSOCIATED_MEMOBJECT 0x1107 +#define CL_MEM_OFFSET 0x1108 +#define CL_MEM_USES_SVM_POINTER 0x1109 + +/* cl_image_info */ +#define CL_IMAGE_FORMAT 0x1110 +#define CL_IMAGE_ELEMENT_SIZE 0x1111 +#define CL_IMAGE_ROW_PITCH 0x1112 +#define CL_IMAGE_SLICE_PITCH 0x1113 +#define CL_IMAGE_WIDTH 0x1114 +#define CL_IMAGE_HEIGHT 0x1115 +#define CL_IMAGE_DEPTH 0x1116 +#define CL_IMAGE_ARRAY_SIZE 0x1117 +#define CL_IMAGE_BUFFER 0x1118 +#define CL_IMAGE_NUM_MIP_LEVELS 0x1119 +#define CL_IMAGE_NUM_SAMPLES 0x111A + +/* cl_pipe_info */ +#define CL_PIPE_PACKET_SIZE 0x1120 +#define CL_PIPE_MAX_PACKETS 0x1121 + +/* cl_addressing_mode */ +#define CL_ADDRESS_NONE 0x1130 +#define CL_ADDRESS_CLAMP_TO_EDGE 0x1131 +#define CL_ADDRESS_CLAMP 0x1132 +#define CL_ADDRESS_REPEAT 0x1133 +#define CL_ADDRESS_MIRRORED_REPEAT 0x1134 + +/* cl_filter_mode */ +#define CL_FILTER_NEAREST 0x1140 +#define CL_FILTER_LINEAR 0x1141 + +/* cl_sampler_info */ +#define CL_SAMPLER_REFERENCE_COUNT 0x1150 +#define CL_SAMPLER_CONTEXT 0x1151 +#define CL_SAMPLER_NORMALIZED_COORDS 0x1152 +#define CL_SAMPLER_ADDRESSING_MODE 0x1153 +#define CL_SAMPLER_FILTER_MODE 0x1154 +#define CL_SAMPLER_MIP_FILTER_MODE 0x1155 +#define CL_SAMPLER_LOD_MIN 0x1156 +#define CL_SAMPLER_LOD_MAX 0x1157 + +/* cl_map_flags - bitfield */ +#define CL_MAP_READ (1 << 0) +#define CL_MAP_WRITE (1 << 1) +#define CL_MAP_WRITE_INVALIDATE_REGION (1 << 2) + +/* cl_program_info */ +#define CL_PROGRAM_REFERENCE_COUNT 0x1160 +#define CL_PROGRAM_CONTEXT 0x1161 +#define CL_PROGRAM_NUM_DEVICES 0x1162 +#define CL_PROGRAM_DEVICES 0x1163 +#define CL_PROGRAM_SOURCE 0x1164 +#define CL_PROGRAM_BINARY_SIZES 0x1165 +#define CL_PROGRAM_BINARIES 0x1166 +#define CL_PROGRAM_NUM_KERNELS 0x1167 +#define CL_PROGRAM_KERNEL_NAMES 0x1168 + +/* cl_program_build_info */ +#define CL_PROGRAM_BUILD_STATUS 0x1181 +#define CL_PROGRAM_BUILD_OPTIONS 0x1182 +#define CL_PROGRAM_BUILD_LOG 0x1183 +#define CL_PROGRAM_BINARY_TYPE 0x1184 +#define CL_PROGRAM_BUILD_GLOBAL_VARIABLE_TOTAL_SIZE 0x1185 + +/* cl_program_binary_type */ +#define CL_PROGRAM_BINARY_TYPE_NONE 0x0 +#define CL_PROGRAM_BINARY_TYPE_COMPILED_OBJECT 0x1 +#define CL_PROGRAM_BINARY_TYPE_LIBRARY 0x2 +#define CL_PROGRAM_BINARY_TYPE_EXECUTABLE 0x4 + +/* cl_build_status */ +#define CL_BUILD_SUCCESS 0 +#define CL_BUILD_NONE -1 +#define CL_BUILD_ERROR -2 +#define CL_BUILD_IN_PROGRESS -3 + +/* cl_kernel_info */ +#define CL_KERNEL_FUNCTION_NAME 0x1190 +#define CL_KERNEL_NUM_ARGS 0x1191 +#define CL_KERNEL_REFERENCE_COUNT 0x1192 +#define CL_KERNEL_CONTEXT 0x1193 +#define CL_KERNEL_PROGRAM 0x1194 +#define CL_KERNEL_ATTRIBUTES 0x1195 + +/* cl_kernel_arg_info */ +#define CL_KERNEL_ARG_ADDRESS_QUALIFIER 0x1196 +#define CL_KERNEL_ARG_ACCESS_QUALIFIER 0x1197 +#define CL_KERNEL_ARG_TYPE_NAME 0x1198 +#define CL_KERNEL_ARG_TYPE_QUALIFIER 0x1199 +#define CL_KERNEL_ARG_NAME 0x119A + +/* cl_kernel_arg_address_qualifier */ +#define CL_KERNEL_ARG_ADDRESS_GLOBAL 0x119B +#define CL_KERNEL_ARG_ADDRESS_LOCAL 0x119C +#define CL_KERNEL_ARG_ADDRESS_CONSTANT 0x119D +#define CL_KERNEL_ARG_ADDRESS_PRIVATE 0x119E + +/* cl_kernel_arg_access_qualifier */ +#define CL_KERNEL_ARG_ACCESS_READ_ONLY 0x11A0 +#define CL_KERNEL_ARG_ACCESS_WRITE_ONLY 0x11A1 +#define CL_KERNEL_ARG_ACCESS_READ_WRITE 0x11A2 +#define CL_KERNEL_ARG_ACCESS_NONE 0x11A3 + +/* cl_kernel_arg_type_qualifer */ +#define CL_KERNEL_ARG_TYPE_NONE 0 +#define CL_KERNEL_ARG_TYPE_CONST (1 << 0) +#define CL_KERNEL_ARG_TYPE_RESTRICT (1 << 1) +#define CL_KERNEL_ARG_TYPE_VOLATILE (1 << 2) +#define CL_KERNEL_ARG_TYPE_PIPE (1 << 3) + +/* cl_kernel_work_group_info */ +#define CL_KERNEL_WORK_GROUP_SIZE 0x11B0 +#define CL_KERNEL_COMPILE_WORK_GROUP_SIZE 0x11B1 +#define CL_KERNEL_LOCAL_MEM_SIZE 0x11B2 +#define CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 0x11B3 +#define CL_KERNEL_PRIVATE_MEM_SIZE 0x11B4 +#define CL_KERNEL_GLOBAL_WORK_SIZE 0x11B5 + +/* cl_kernel_exec_info */ +#define CL_KERNEL_EXEC_INFO_SVM_PTRS 0x11B6 +#define CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM 0x11B7 + +/* cl_event_info */ +#define CL_EVENT_COMMAND_QUEUE 0x11D0 +#define CL_EVENT_COMMAND_TYPE 0x11D1 +#define CL_EVENT_REFERENCE_COUNT 0x11D2 +#define CL_EVENT_COMMAND_EXECUTION_STATUS 0x11D3 +#define CL_EVENT_CONTEXT 0x11D4 + +/* cl_command_type */ +#define CL_COMMAND_NDRANGE_KERNEL 0x11F0 +#define CL_COMMAND_TASK 0x11F1 +#define CL_COMMAND_NATIVE_KERNEL 0x11F2 +#define CL_COMMAND_READ_BUFFER 0x11F3 +#define CL_COMMAND_WRITE_BUFFER 0x11F4 +#define CL_COMMAND_COPY_BUFFER 0x11F5 +#define CL_COMMAND_READ_IMAGE 0x11F6 +#define CL_COMMAND_WRITE_IMAGE 0x11F7 +#define CL_COMMAND_COPY_IMAGE 0x11F8 +#define CL_COMMAND_COPY_IMAGE_TO_BUFFER 0x11F9 +#define CL_COMMAND_COPY_BUFFER_TO_IMAGE 0x11FA +#define CL_COMMAND_MAP_BUFFER 0x11FB +#define CL_COMMAND_MAP_IMAGE 0x11FC +#define CL_COMMAND_UNMAP_MEM_OBJECT 0x11FD +#define CL_COMMAND_MARKER 0x11FE +#define CL_COMMAND_ACQUIRE_GL_OBJECTS 0x11FF +#define CL_COMMAND_RELEASE_GL_OBJECTS 0x1200 +#define CL_COMMAND_READ_BUFFER_RECT 0x1201 +#define CL_COMMAND_WRITE_BUFFER_RECT 0x1202 +#define CL_COMMAND_COPY_BUFFER_RECT 0x1203 +#define CL_COMMAND_USER 0x1204 +#define CL_COMMAND_BARRIER 0x1205 +#define CL_COMMAND_MIGRATE_MEM_OBJECTS 0x1206 +#define CL_COMMAND_FILL_BUFFER 0x1207 +#define CL_COMMAND_FILL_IMAGE 0x1208 +#define CL_COMMAND_SVM_FREE 0x1209 +#define CL_COMMAND_SVM_MEMCPY 0x120A +#define CL_COMMAND_SVM_MEMFILL 0x120B +#define CL_COMMAND_SVM_MAP 0x120C +#define CL_COMMAND_SVM_UNMAP 0x120D + +/* command execution status */ +#define CL_COMPLETE 0x0 +#define CL_RUNNING 0x1 +#define CL_SUBMITTED 0x2 +#define CL_QUEUED 0x3 + +/* cl_buffer_create_type */ +#define CL_BUFFER_CREATE_TYPE_REGION 0x1220 + +/* cl_profiling_info */ +#define CL_PROFILING_COMMAND_QUEUED 0x1280 +#define CL_PROFILING_COMMAND_SUBMIT 0x1281 +#define CL_PROFILING_COMMAND_START 0x1282 +#define CL_PROFILING_COMMAND_END 0x1283 +#define CL_PROFILING_COMMAND_COMPLETE 0x1284 + +/********************************************************************************************************/ + +/* Platform API */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetPlatformIDs(cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetPlatformInfo(cl_platform_id /* platform */, + cl_platform_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Device APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDs(cl_platform_id /* platform */, + cl_device_type /* device_type */, + cl_uint /* num_entries */, + cl_device_id * /* devices */, + cl_uint * /* num_devices */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceInfo(cl_device_id /* device */, + cl_device_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clCreateSubDevices(cl_device_id /* in_device */, + const cl_device_partition_property * /* properties */, + cl_uint /* num_devices */, + cl_device_id * /* out_devices */, + cl_uint * /* num_devices_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; + +/* Context APIs */ +extern CL_API_ENTRY cl_context CL_API_CALL +clCreateContext(const cl_context_properties * /* properties */, + cl_uint /* num_devices */, + const cl_device_id * /* devices */, + void (CL_CALLBACK * /* pfn_notify */)(const char *, const void *, size_t, void *), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_context CL_API_CALL +clCreateContextFromType(const cl_context_properties * /* properties */, + cl_device_type /* device_type */, + void (CL_CALLBACK * /* pfn_notify*/ )(const char *, const void *, size_t, void *), + void * /* user_data */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainContext(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseContext(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetContextInfo(cl_context /* context */, + cl_context_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Command Queue APIs */ +extern CL_API_ENTRY cl_command_queue CL_API_CALL +clCreateCommandQueueWithProperties(cl_context /* context */, + cl_device_id /* device */, + const cl_queue_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetCommandQueueInfo(cl_command_queue /* command_queue */, + cl_command_queue_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Memory Object APIs */ +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateBuffer(cl_context /* context */, + cl_mem_flags /* flags */, + size_t /* size */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateSubBuffer(cl_mem /* buffer */, + cl_mem_flags /* flags */, + cl_buffer_create_type /* buffer_create_type */, + const void * /* buffer_create_info */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateImage(cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + const cl_image_desc * /* image_desc */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreatePipe(cl_context /* context */, + cl_mem_flags /* flags */, + cl_uint /* pipe_packet_size */, + cl_uint /* pipe_max_packets */, + const cl_pipe_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetSupportedImageFormats(cl_context /* context */, + cl_mem_flags /* flags */, + cl_mem_object_type /* image_type */, + cl_uint /* num_entries */, + cl_image_format * /* image_formats */, + cl_uint * /* num_image_formats */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetMemObjectInfo(cl_mem /* memobj */, + cl_mem_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetImageInfo(cl_mem /* image */, + cl_image_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetPipeInfo(cl_mem /* pipe */, + cl_pipe_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_2_0; + + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetMemObjectDestructorCallback(cl_mem /* memobj */, + void (CL_CALLBACK * /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), + void * /*user_data */ ) CL_API_SUFFIX__VERSION_1_1; + +/* SVM Allocation APIs */ +extern CL_API_ENTRY void * CL_API_CALL +clSVMAlloc(cl_context /* context */, + cl_svm_mem_flags /* flags */, + size_t /* size */, + cl_uint /* alignment */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY void CL_API_CALL +clSVMFree(cl_context /* context */, + void * /* svm_pointer */) CL_API_SUFFIX__VERSION_2_0; + +/* Sampler APIs */ +extern CL_API_ENTRY cl_sampler CL_API_CALL +clCreateSamplerWithProperties(cl_context /* context */, + const cl_sampler_properties * /* normalized_coords */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetSamplerInfo(cl_sampler /* sampler */, + cl_sampler_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Program Object APIs */ +extern CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithSource(cl_context /* context */, + cl_uint /* count */, + const char ** /* strings */, + const size_t * /* lengths */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithBinary(cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const size_t * /* lengths */, + const unsigned char ** /* binaries */, + cl_int * /* binary_status */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithBuiltInKernels(cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* kernel_names */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clBuildProgram(cl_program /* program */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clCompileProgram(cl_program /* program */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + cl_uint /* num_input_headers */, + const cl_program * /* input_headers */, + const char ** /* header_include_names */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_program CL_API_CALL +clLinkProgram(cl_context /* context */, + cl_uint /* num_devices */, + const cl_device_id * /* device_list */, + const char * /* options */, + cl_uint /* num_input_programs */, + const cl_program * /* input_programs */, + void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), + void * /* user_data */, + cl_int * /* errcode_ret */ ) CL_API_SUFFIX__VERSION_1_2; + + +extern CL_API_ENTRY cl_int CL_API_CALL +clUnloadPlatformCompiler(cl_platform_id /* platform */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetProgramInfo(cl_program /* program */, + cl_program_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetProgramBuildInfo(cl_program /* program */, + cl_device_id /* device */, + cl_program_build_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Kernel Object APIs */ +extern CL_API_ENTRY cl_kernel CL_API_CALL +clCreateKernel(cl_program /* program */, + const char * /* kernel_name */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clCreateKernelsInProgram(cl_program /* program */, + cl_uint /* num_kernels */, + cl_kernel * /* kernels */, + cl_uint * /* num_kernels_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetKernelArg(cl_kernel /* kernel */, + cl_uint /* arg_index */, + size_t /* arg_size */, + const void * /* arg_value */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetKernelArgSVMPointer(cl_kernel /* kernel */, + cl_uint /* arg_index */, + const void * /* arg_value */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetKernelExecInfo(cl_kernel /* kernel */, + cl_kernel_exec_info /* param_name */, + size_t /* param_value_size */, + const void * /* param_value */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelInfo(cl_kernel /* kernel */, + cl_kernel_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelArgInfo(cl_kernel /* kernel */, + cl_uint /* arg_indx */, + cl_kernel_arg_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelWorkGroupInfo(cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_work_group_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Event Object APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clWaitForEvents(cl_uint /* num_events */, + const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetEventInfo(cl_event /* event */, + cl_event_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_event CL_API_CALL +clCreateUserEvent(cl_context /* context */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clRetainEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clReleaseEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetUserEventStatus(cl_event /* event */, + cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clSetEventCallback( cl_event /* event */, + cl_int /* command_exec_callback_type */, + void (CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_1; + +/* Profiling APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetEventProfilingInfo(cl_event /* event */, + cl_profiling_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +/* Flush and Finish APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clFlush(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clFinish(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; + +/* Enqueued Commands APIs */ +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + size_t /* offset */, + size_t /* size */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadBufferRect(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_read */, + const size_t * /* buffer_offset */, + const size_t * /* host_offset */, + const size_t * /* region */, + size_t /* buffer_row_pitch */, + size_t /* buffer_slice_pitch */, + size_t /* host_row_pitch */, + size_t /* host_slice_pitch */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_write */, + size_t /* offset */, + size_t /* size */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteBufferRect(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_write */, + const size_t * /* buffer_offset */, + const size_t * /* host_offset */, + const size_t * /* region */, + size_t /* buffer_row_pitch */, + size_t /* buffer_slice_pitch */, + size_t /* host_row_pitch */, + size_t /* host_slice_pitch */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueFillBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + const void * /* pattern */, + size_t /* pattern_size */, + size_t /* offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBuffer(cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_buffer */, + size_t /* src_offset */, + size_t /* dst_offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBufferRect(cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_buffer */, + const size_t * /* src_origin */, + const size_t * /* dst_origin */, + const size_t * /* region */, + size_t /* src_row_pitch */, + size_t /* src_slice_pitch */, + size_t /* dst_row_pitch */, + size_t /* dst_slice_pitch */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_read */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t /* row_pitch */, + size_t /* slice_pitch */, + void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_write */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t /* input_row_pitch */, + size_t /* input_slice_pitch */, + const void * /* ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueFillImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + const void * /* fill_color */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyImage(cl_command_queue /* command_queue */, + cl_mem /* src_image */, + cl_mem /* dst_image */, + const size_t * /* src_origin[3] */, + const size_t * /* dst_origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyImageToBuffer(cl_command_queue /* command_queue */, + cl_mem /* src_image */, + cl_mem /* dst_buffer */, + const size_t * /* src_origin[3] */, + const size_t * /* region[3] */, + size_t /* dst_offset */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBufferToImage(cl_command_queue /* command_queue */, + cl_mem /* src_buffer */, + cl_mem /* dst_image */, + size_t /* src_offset */, + const size_t * /* dst_origin[3] */, + const size_t * /* region[3] */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY void * CL_API_CALL +clEnqueueMapBuffer(cl_command_queue /* command_queue */, + cl_mem /* buffer */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + size_t /* offset */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY void * CL_API_CALL +clEnqueueMapImage(cl_command_queue /* command_queue */, + cl_mem /* image */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + const size_t * /* origin[3] */, + const size_t * /* region[3] */, + size_t * /* image_row_pitch */, + size_t * /* image_slice_pitch */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueUnmapMemObject(cl_command_queue /* command_queue */, + cl_mem /* memobj */, + void * /* mapped_ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMigrateMemObjects(cl_command_queue /* command_queue */, + cl_uint /* num_mem_objects */, + const cl_mem * /* mem_objects */, + cl_mem_migration_flags /* flags */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueNDRangeKernel(cl_command_queue /* command_queue */, + cl_kernel /* kernel */, + cl_uint /* work_dim */, + const size_t * /* global_work_offset */, + const size_t * /* global_work_size */, + const size_t * /* local_work_size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueNativeKernel(cl_command_queue /* command_queue */, + void (CL_CALLBACK * /*user_func*/)(void *), + void * /* args */, + size_t /* cb_args */, + cl_uint /* num_mem_objects */, + const cl_mem * /* mem_list */, + const void ** /* args_mem_loc */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMarkerWithWaitList(cl_command_queue /* command_queue */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueBarrierWithWaitList(cl_command_queue /* command_queue */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMFree(cl_command_queue /* command_queue */, + cl_uint /* num_svm_pointers */, + void *[] /* svm_pointers[] */, + void (CL_CALLBACK * /*pfn_free_func*/)(cl_command_queue /* queue */, + cl_uint /* num_svm_pointers */, + void *[] /* svm_pointers[] */, + void * /* user_data */), + void * /* user_data */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMMemcpy(cl_command_queue /* command_queue */, + cl_bool /* blocking_copy */, + void * /* dst_ptr */, + const void * /* src_ptr */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMMemFill(cl_command_queue /* command_queue */, + void * /* svm_ptr */, + const void * /* pattern */, + size_t /* pattern_size */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMMap(cl_command_queue /* command_queue */, + cl_bool /* blocking_map */, + cl_map_flags /* flags */, + void * /* svm_ptr */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMUnmap(cl_command_queue /* command_queue */, + void * /* svm_ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + + +/* Extension function access + * + * Returns the extension function address for the given function name, + * or NULL if a valid function can not be found. The client must + * check to make sure the address is not NULL, before using or + * calling the returned function address. + */ +extern CL_API_ENTRY void * CL_API_CALL +clGetExtensionFunctionAddressForPlatform(cl_platform_id /* platform */, + const char * /* func_name */) CL_API_SUFFIX__VERSION_1_2; + + +/* Deprecated OpenCL 1.1 APIs */ +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateImage2D(cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + size_t /* image_width */, + size_t /* image_height */, + size_t /* image_row_pitch */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateImage3D(cl_context /* context */, + cl_mem_flags /* flags */, + const cl_image_format * /* image_format */, + size_t /* image_width */, + size_t /* image_height */, + size_t /* image_depth */, + size_t /* image_row_pitch */, + size_t /* image_slice_pitch */, + void * /* host_ptr */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clEnqueueMarker(cl_command_queue /* command_queue */, + cl_event * /* event */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clEnqueueWaitForEvents(cl_command_queue /* command_queue */, + cl_uint /* num_events */, + const cl_event * /* event_list */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clEnqueueBarrier(cl_command_queue /* command_queue */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL +clUnloadCompiler(void) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED void * CL_API_CALL +clGetExtensionFunctionAddress(const char * /* func_name */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +/* Deprecated OpenCL 2.0 APIs */ +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_2_0_DEPRECATED cl_command_queue CL_API_CALL +clCreateCommandQueue(cl_context /* context */, + cl_device_id /* device */, + cl_command_queue_properties /* properties */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED; + + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_2_0_DEPRECATED cl_sampler CL_API_CALL +clCreateSampler(cl_context /* context */, + cl_bool /* normalized_coords */, + cl_addressing_mode /* addressing_mode */, + cl_filter_mode /* filter_mode */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_2_0_DEPRECATED cl_int CL_API_CALL +clEnqueueTask(cl_command_queue /* command_queue */, + cl_kernel /* kernel */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.hpp b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.hpp new file mode 100644 index 0000000000..667526e341 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl.hpp @@ -0,0 +1,12305 @@ +/******************************************************************************* + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/*! \file + * + * \brief C++ bindings for OpenCL 1.0 (rev 48), OpenCL 1.1 (rev 33) and + * OpenCL 1.2 (rev 15) + * \author Benedict R. Gaster, Laurent Morichetti and Lee Howes + * + * Additions and fixes from: + * Brian Cole, March 3rd 2010 and April 2012 + * Matt Gruenke, April 2012. + * Bruce Merry, February 2013. + * + * \version 1.2.5 + * \date June 2013 + * + * Optional extension support + * + * cl + * cl_ext_device_fission + * #define USE_CL_DEVICE_FISSION + */ + +/*! \mainpage + * \section intro Introduction + * For many large applications C++ is the language of choice and so it seems + * reasonable to define C++ bindings for OpenCL. + * + * + * The interface is contained with a single C++ header file \em cl.hpp and all + * definitions are contained within the namespace \em cl. There is no additional + * requirement to include \em cl.h and to use either the C++ or original C + * bindings it is enough to simply include \em cl.hpp. + * + * The bindings themselves are lightweight and correspond closely to the + * underlying C API. Using the C++ bindings introduces no additional execution + * overhead. + * + * For detail documentation on the bindings see: + * + * The OpenCL C++ Wrapper API 1.2 (revision 09) + * http://www.khronos.org/registry/cl/specs/opencl-cplusplus-1.2.pdf + * + * \section example Example + * + * The following example shows a general use case for the C++ + * bindings, including support for the optional exception feature and + * also the supplied vector and string classes, see following sections for + * decriptions of these features. + * + * \code + * #define __CL_ENABLE_EXCEPTIONS + * + * #if defined(__APPLE__) || defined(__MACOSX) + * #include + * #else + * #include + * #endif + * #include + * #include + * #include + * + * const char * helloStr = "__kernel void " + * "hello(void) " + * "{ " + * " " + * "} "; + * + * int + * main(void) + * { + * cl_int err = CL_SUCCESS; + * try { + * + * std::vector platforms; + * cl::Platform::get(&platforms); + * if (platforms.size() == 0) { + * std::cout << "Platform size 0\n"; + * return -1; + * } + * + * cl_context_properties properties[] = + * { CL_CONTEXT_PLATFORM, (cl_context_properties)(platforms[0])(), 0}; + * cl::Context context(CL_DEVICE_TYPE_CPU, properties); + * + * std::vector devices = context.getInfo(); + * + * cl::Program::Sources source(1, + * std::make_pair(helloStr,strlen(helloStr))); + * cl::Program program_ = cl::Program(context, source); + * program_.build(devices); + * + * cl::Kernel kernel(program_, "hello", &err); + * + * cl::Event event; + * cl::CommandQueue queue(context, devices[0], 0, &err); + * queue.enqueueNDRangeKernel( + * kernel, + * cl::NullRange, + * cl::NDRange(4,4), + * cl::NullRange, + * NULL, + * &event); + * + * event.wait(); + * } + * catch (cl::Error err) { + * std::cerr + * << "ERROR: " + * << err.what() + * << "(" + * << err.err() + * << ")" + * << std::endl; + * } + * + * return EXIT_SUCCESS; + * } + * + * \endcode + * + */ +#ifndef CL_HPP_ +#define CL_HPP_ + +#ifdef _WIN32 + +#include +#include +#include +#include + +#if defined(__CL_ENABLE_EXCEPTIONS) +#include +#endif // #if defined(__CL_ENABLE_EXCEPTIONS) + +#pragma push_macro("max") +#undef max +#if defined(USE_DX_INTEROP) +#include +#include +#endif +#endif // _WIN32 + +// +#if defined(USE_CL_DEVICE_FISSION) +#include +#endif + +#if defined(__APPLE__) || defined(__MACOSX) +#include +#include +#include +#else +#include +#include +#endif // !__APPLE__ + +// To avoid accidentally taking ownership of core OpenCL types +// such as cl_kernel constructors are made explicit +// under OpenCL 1.2 +#if defined(CL_VERSION_1_2) && !defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +#define __CL_EXPLICIT_CONSTRUCTORS explicit +#else // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +#define __CL_EXPLICIT_CONSTRUCTORS +#endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + +// Define deprecated prefixes and suffixes to ensure compilation +// in case they are not pre-defined +#if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) +#define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED +#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) +#if !defined(CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED) +#define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) + +#if !defined(CL_CALLBACK) +#define CL_CALLBACK +#endif //CL_CALLBACK + +#include +#include + +#if !defined(__NO_STD_VECTOR) +#include +#endif + +#if !defined(__NO_STD_STRING) +#include +#endif + +#if defined(__linux__) || defined(__APPLE__) || defined(__MACOSX) +#include + +#include +#include +#endif // __linux__ + +#include + + +/*! \namespace cl + * + * \brief The OpenCL C++ bindings are defined within this namespace. + * + */ +namespace cl { + +class Memory; + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +#define __INIT_CL_EXT_FCN_PTR(name) \ + if(!pfn_##name) { \ + pfn_##name = (PFN_##name) \ + clGetExtensionFunctionAddress(#name); \ + if(!pfn_##name) { \ + } \ + } +#endif // #if defined(CL_VERSION_1_1) + +#if defined(CL_VERSION_1_2) +#define __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, name) \ + if(!pfn_##name) { \ + pfn_##name = (PFN_##name) \ + clGetExtensionFunctionAddressForPlatform(platform, #name); \ + if(!pfn_##name) { \ + } \ + } +#endif // #if defined(CL_VERSION_1_1) + +class Program; +class Device; +class Context; +class CommandQueue; +class Memory; +class Buffer; + +#if defined(__CL_ENABLE_EXCEPTIONS) +/*! \brief Exception class + * + * This may be thrown by API functions when __CL_ENABLE_EXCEPTIONS is defined. + */ +class Error : public std::exception +{ +private: + cl_int err_; + const char * errStr_; +public: + /*! \brief Create a new CL error exception for a given error code + * and corresponding message. + * + * \param err error code value. + * + * \param errStr a descriptive string that must remain in scope until + * handling of the exception has concluded. If set, it + * will be returned by what(). + */ + Error(cl_int err, const char * errStr = NULL) : err_(err), errStr_(errStr) + {} + + ~Error() throw() {} + + /*! \brief Get error string associated with exception + * + * \return A memory pointer to the error message string. + */ + virtual const char * what() const throw () + { + if (errStr_ == NULL) { + return "empty"; + } + else { + return errStr_; + } + } + + /*! \brief Get error code associated with exception + * + * \return The error code. + */ + cl_int err(void) const { return err_; } +}; + +#define __ERR_STR(x) #x +#else +#define __ERR_STR(x) NULL +#endif // __CL_ENABLE_EXCEPTIONS + + +namespace detail +{ +#if defined(__CL_ENABLE_EXCEPTIONS) +static inline cl_int errHandler ( + cl_int err, + const char * errStr = NULL) +{ + if (err != CL_SUCCESS) { + throw Error(err, errStr); + } + return err; +} +#else +static inline cl_int errHandler (cl_int err, const char * errStr = NULL) +{ + (void) errStr; // suppress unused variable warning + return err; +} +#endif // __CL_ENABLE_EXCEPTIONS +} + + + +//! \cond DOXYGEN_DETAIL +#if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) +#define __GET_DEVICE_INFO_ERR __ERR_STR(clGetDeviceInfo) +#define __GET_PLATFORM_INFO_ERR __ERR_STR(clGetPlatformInfo) +#define __GET_DEVICE_IDS_ERR __ERR_STR(clGetDeviceIDs) +#define __GET_PLATFORM_IDS_ERR __ERR_STR(clGetPlatformIDs) +#define __GET_CONTEXT_INFO_ERR __ERR_STR(clGetContextInfo) +#define __GET_EVENT_INFO_ERR __ERR_STR(clGetEventInfo) +#define __GET_EVENT_PROFILE_INFO_ERR __ERR_STR(clGetEventProfileInfo) +#define __GET_MEM_OBJECT_INFO_ERR __ERR_STR(clGetMemObjectInfo) +#define __GET_IMAGE_INFO_ERR __ERR_STR(clGetImageInfo) +#define __GET_SAMPLER_INFO_ERR __ERR_STR(clGetSamplerInfo) +#define __GET_KERNEL_INFO_ERR __ERR_STR(clGetKernelInfo) +#if defined(CL_VERSION_1_2) +#define __GET_KERNEL_ARG_INFO_ERR __ERR_STR(clGetKernelArgInfo) +#endif // #if defined(CL_VERSION_1_2) +#define __GET_KERNEL_WORK_GROUP_INFO_ERR __ERR_STR(clGetKernelWorkGroupInfo) +#define __GET_PROGRAM_INFO_ERR __ERR_STR(clGetProgramInfo) +#define __GET_PROGRAM_BUILD_INFO_ERR __ERR_STR(clGetProgramBuildInfo) +#define __GET_COMMAND_QUEUE_INFO_ERR __ERR_STR(clGetCommandQueueInfo) + +#define __CREATE_CONTEXT_ERR __ERR_STR(clCreateContext) +#define __CREATE_CONTEXT_FROM_TYPE_ERR __ERR_STR(clCreateContextFromType) +#define __GET_SUPPORTED_IMAGE_FORMATS_ERR __ERR_STR(clGetSupportedImageFormats) + +#define __CREATE_BUFFER_ERR __ERR_STR(clCreateBuffer) +#define __COPY_ERR __ERR_STR(cl::copy) +#define __CREATE_SUBBUFFER_ERR __ERR_STR(clCreateSubBuffer) +#define __CREATE_GL_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) +#define __CREATE_GL_RENDER_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) +#define __GET_GL_OBJECT_INFO_ERR __ERR_STR(clGetGLObjectInfo) +#if defined(CL_VERSION_1_2) +#define __CREATE_IMAGE_ERR __ERR_STR(clCreateImage) +#define __CREATE_GL_TEXTURE_ERR __ERR_STR(clCreateFromGLTexture) +#define __IMAGE_DIMENSION_ERR __ERR_STR(Incorrect image dimensions) +#endif // #if defined(CL_VERSION_1_2) +#define __CREATE_SAMPLER_ERR __ERR_STR(clCreateSampler) +#define __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR __ERR_STR(clSetMemObjectDestructorCallback) + +#define __CREATE_USER_EVENT_ERR __ERR_STR(clCreateUserEvent) +#define __SET_USER_EVENT_STATUS_ERR __ERR_STR(clSetUserEventStatus) +#define __SET_EVENT_CALLBACK_ERR __ERR_STR(clSetEventCallback) +#define __WAIT_FOR_EVENTS_ERR __ERR_STR(clWaitForEvents) + +#define __CREATE_KERNEL_ERR __ERR_STR(clCreateKernel) +#define __SET_KERNEL_ARGS_ERR __ERR_STR(clSetKernelArg) +#define __CREATE_PROGRAM_WITH_SOURCE_ERR __ERR_STR(clCreateProgramWithSource) +#define __CREATE_PROGRAM_WITH_BINARY_ERR __ERR_STR(clCreateProgramWithBinary) +#if defined(CL_VERSION_1_2) +#define __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR __ERR_STR(clCreateProgramWithBuiltInKernels) +#endif // #if defined(CL_VERSION_1_2) +#define __BUILD_PROGRAM_ERR __ERR_STR(clBuildProgram) +#if defined(CL_VERSION_1_2) +#define __COMPILE_PROGRAM_ERR __ERR_STR(clCompileProgram) + +#endif // #if defined(CL_VERSION_1_2) +#define __CREATE_KERNELS_IN_PROGRAM_ERR __ERR_STR(clCreateKernelsInProgram) + +#define __CREATE_COMMAND_QUEUE_ERR __ERR_STR(clCreateCommandQueue) +#define __SET_COMMAND_QUEUE_PROPERTY_ERR __ERR_STR(clSetCommandQueueProperty) +#define __ENQUEUE_READ_BUFFER_ERR __ERR_STR(clEnqueueReadBuffer) +#define __ENQUEUE_READ_BUFFER_RECT_ERR __ERR_STR(clEnqueueReadBufferRect) +#define __ENQUEUE_WRITE_BUFFER_ERR __ERR_STR(clEnqueueWriteBuffer) +#define __ENQUEUE_WRITE_BUFFER_RECT_ERR __ERR_STR(clEnqueueWriteBufferRect) +#define __ENQEUE_COPY_BUFFER_ERR __ERR_STR(clEnqueueCopyBuffer) +#define __ENQEUE_COPY_BUFFER_RECT_ERR __ERR_STR(clEnqueueCopyBufferRect) +#define __ENQUEUE_FILL_BUFFER_ERR __ERR_STR(clEnqueueFillBuffer) +#define __ENQUEUE_READ_IMAGE_ERR __ERR_STR(clEnqueueReadImage) +#define __ENQUEUE_WRITE_IMAGE_ERR __ERR_STR(clEnqueueWriteImage) +#define __ENQUEUE_COPY_IMAGE_ERR __ERR_STR(clEnqueueCopyImage) +#define __ENQUEUE_FILL_IMAGE_ERR __ERR_STR(clEnqueueFillImage) +#define __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR __ERR_STR(clEnqueueCopyImageToBuffer) +#define __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR __ERR_STR(clEnqueueCopyBufferToImage) +#define __ENQUEUE_MAP_BUFFER_ERR __ERR_STR(clEnqueueMapBuffer) +#define __ENQUEUE_MAP_IMAGE_ERR __ERR_STR(clEnqueueMapImage) +#define __ENQUEUE_UNMAP_MEM_OBJECT_ERR __ERR_STR(clEnqueueUnMapMemObject) +#define __ENQUEUE_NDRANGE_KERNEL_ERR __ERR_STR(clEnqueueNDRangeKernel) +#define __ENQUEUE_TASK_ERR __ERR_STR(clEnqueueTask) +#define __ENQUEUE_NATIVE_KERNEL __ERR_STR(clEnqueueNativeKernel) +#if defined(CL_VERSION_1_2) +#define __ENQUEUE_MIGRATE_MEM_OBJECTS_ERR __ERR_STR(clEnqueueMigrateMemObjects) +#endif // #if defined(CL_VERSION_1_2) + +#define __ENQUEUE_ACQUIRE_GL_ERR __ERR_STR(clEnqueueAcquireGLObjects) +#define __ENQUEUE_RELEASE_GL_ERR __ERR_STR(clEnqueueReleaseGLObjects) + + +#define __RETAIN_ERR __ERR_STR(Retain Object) +#define __RELEASE_ERR __ERR_STR(Release Object) +#define __FLUSH_ERR __ERR_STR(clFlush) +#define __FINISH_ERR __ERR_STR(clFinish) +#define __VECTOR_CAPACITY_ERR __ERR_STR(Vector capacity error) + +/** + * CL 1.2 version that uses device fission. + */ +#if defined(CL_VERSION_1_2) +#define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevices) +#else +#define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevicesEXT) +#endif // #if defined(CL_VERSION_1_2) + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +#define __ENQUEUE_MARKER_ERR __ERR_STR(clEnqueueMarker) +#define __ENQUEUE_WAIT_FOR_EVENTS_ERR __ERR_STR(clEnqueueWaitForEvents) +#define __ENQUEUE_BARRIER_ERR __ERR_STR(clEnqueueBarrier) +#define __UNLOAD_COMPILER_ERR __ERR_STR(clUnloadCompiler) +#define __CREATE_GL_TEXTURE_2D_ERR __ERR_STR(clCreateFromGLTexture2D) +#define __CREATE_GL_TEXTURE_3D_ERR __ERR_STR(clCreateFromGLTexture3D) +#define __CREATE_IMAGE2D_ERR __ERR_STR(clCreateImage2D) +#define __CREATE_IMAGE3D_ERR __ERR_STR(clCreateImage3D) +#endif // #if defined(CL_VERSION_1_1) + +#endif // __CL_USER_OVERRIDE_ERROR_STRINGS +//! \endcond + +/** + * CL 1.2 marker and barrier commands + */ +#if defined(CL_VERSION_1_2) +#define __ENQUEUE_MARKER_WAIT_LIST_ERR __ERR_STR(clEnqueueMarkerWithWaitList) +#define __ENQUEUE_BARRIER_WAIT_LIST_ERR __ERR_STR(clEnqueueBarrierWithWaitList) +#endif // #if defined(CL_VERSION_1_2) + +#if !defined(__USE_DEV_STRING) && !defined(__NO_STD_STRING) +typedef std::string STRING_CLASS; +#elif !defined(__USE_DEV_STRING) + +/*! \class string + * \brief Simple string class, that provides a limited subset of std::string + * functionality but avoids many of the issues that come with that class. + + * \note Deprecated. Please use std::string as default or + * re-define the string class to match the std::string + * interface by defining STRING_CLASS + */ +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED string CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ +private: + ::size_t size_; + char * str_; +public: + //! \brief Constructs an empty string, allocating no memory. + string(void) : size_(0), str_(NULL) + { + } + + /*! \brief Constructs a string populated from an arbitrary value of + * specified size. + * + * An extra '\0' is added, in case none was contained in str. + * + * \param str the initial value of the string instance. Note that '\0' + * characters receive no special treatment. If NULL, + * the string is left empty, with a size of 0. + * + * \param size the number of characters to copy from str. + */ + string(const char * str, ::size_t size) : + size_(size), + str_(NULL) + { + if( size > 0 ) { + str_ = new char[size_+1]; + if (str_ != NULL) { + memcpy(str_, str, size_ * sizeof(char)); + str_[size_] = '\0'; + } + else { + size_ = 0; + } + } + } + + /*! \brief Constructs a string populated from a null-terminated value. + * + * \param str the null-terminated initial value of the string instance. + * If NULL, the string is left empty, with a size of 0. + */ + string(const char * str) : + size_(0), + str_(NULL) + { + if( str ) { + size_= ::strlen(str); + } + if( size_ > 0 ) { + str_ = new char[size_ + 1]; + if (str_ != NULL) { + memcpy(str_, str, (size_ + 1) * sizeof(char)); + } + } + } + + void resize( ::size_t n ) + { + if( size_ == n ) { + return; + } + if (n == 0) { + if( str_ ) { + delete [] str_; + } + str_ = NULL; + size_ = 0; + } + else { + char *newString = new char[n + 1]; + int copySize = n; + if( size_ < n ) { + copySize = size_; + } + size_ = n; + + if(str_) { + memcpy(newString, str_, (copySize + 1) * sizeof(char)); + } + if( copySize < size_ ) { + memset(newString + copySize, 0, size_ - copySize); + } + newString[size_] = '\0'; + + delete [] str_; + str_ = newString; + } + } + + const char& operator[] ( ::size_t pos ) const + { + return str_[pos]; + } + + char& operator[] ( ::size_t pos ) + { + return str_[pos]; + } + + /*! \brief Copies the value of another string to this one. + * + * \param rhs the string to copy. + * + * \returns a reference to the modified instance. + */ + string& operator=(const string& rhs) + { + if (this == &rhs) { + return *this; + } + + if( str_ != NULL ) { + delete [] str_; + str_ = NULL; + size_ = 0; + } + + if (rhs.size_ == 0 || rhs.str_ == NULL) { + str_ = NULL; + size_ = 0; + } + else { + str_ = new char[rhs.size_ + 1]; + size_ = rhs.size_; + + if (str_ != NULL) { + memcpy(str_, rhs.str_, (size_ + 1) * sizeof(char)); + } + else { + size_ = 0; + } + } + + return *this; + } + + /*! \brief Constructs a string by copying the value of another instance. + * + * \param rhs the string to copy. + */ + string(const string& rhs) : + size_(0), + str_(NULL) + { + *this = rhs; + } + + //! \brief Destructor - frees memory used to hold the current value. + ~string() + { + delete[] str_; + str_ = NULL; + } + + //! \brief Queries the length of the string, excluding any added '\0's. + ::size_t size(void) const { return size_; } + + //! \brief Queries the length of the string, excluding any added '\0's. + ::size_t length(void) const { return size(); } + + /*! \brief Returns a pointer to the private copy held by this instance, + * or "" if empty/unset. + */ + const char * c_str(void) const { return (str_) ? str_ : "";} +}; +typedef cl::string STRING_CLASS; +#endif // #elif !defined(__USE_DEV_STRING) + +#if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) +#define VECTOR_CLASS std::vector +#elif !defined(__USE_DEV_VECTOR) +#define VECTOR_CLASS cl::vector + +#if !defined(__MAX_DEFAULT_VECTOR_SIZE) +#define __MAX_DEFAULT_VECTOR_SIZE 10 +#endif + +/*! \class vector + * \brief Fixed sized vector implementation that mirroring + * + * \note Deprecated. Please use std::vector as default or + * re-define the vector class to match the std::vector + * interface by defining VECTOR_CLASS + + * \note Not recommended for use with custom objects as + * current implementation will construct N elements + * + * std::vector functionality. + * \brief Fixed sized vector compatible with std::vector. + * + * \note + * This differs from std::vector<> not just in memory allocation, + * but also in terms of when members are constructed, destroyed, + * and assigned instead of being copy constructed. + * + * \param T type of element contained in the vector. + * + * \param N maximum size of the vector. + */ +template +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED vector CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ +private: + T data_[N]; + unsigned int size_; + +public: + //! \brief Constructs an empty vector with no memory allocated. + vector() : + size_(static_cast(0)) + {} + + //! \brief Deallocates the vector's memory and destroys all of its elements. + ~vector() + { + clear(); + } + + //! \brief Returns the number of elements currently contained. + unsigned int size(void) const + { + return size_; + } + + /*! \brief Empties the vector of all elements. + * \note + * This does not deallocate memory but will invoke destructors + * on contained elements. + */ + void clear() + { + while(!empty()) { + pop_back(); + } + } + + /*! \brief Appends an element after the last valid element. + * Calling this on a vector that has reached capacity will throw an + * exception if exceptions are enabled. + */ + void push_back (const T& x) + { + if (size() < N) { + new (&data_[size_]) T(x); + size_++; + } else { + detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); + } + } + + /*! \brief Removes the last valid element from the vector. + * Calling this on an empty vector will throw an exception + * if exceptions are enabled. + */ + void pop_back(void) + { + if (size_ != 0) { + --size_; + data_[size_].~T(); + } else { + detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); + } + } + + /*! \brief Constructs with a value copied from another. + * + * \param vec the vector to copy. + */ + vector(const vector& vec) : + size_(vec.size_) + { + if (size_ != 0) { + assign(vec.begin(), vec.end()); + } + } + + /*! \brief Constructs with a specified number of initial elements. + * + * \param size number of initial elements. + * + * \param val value of initial elements. + */ + vector(unsigned int size, const T& val = T()) : + size_(0) + { + for (unsigned int i = 0; i < size; i++) { + push_back(val); + } + } + + /*! \brief Overwrites the current content with that copied from another + * instance. + * + * \param rhs vector to copy. + * + * \returns a reference to this. + */ + vector& operator=(const vector& rhs) + { + if (this == &rhs) { + return *this; + } + + if (rhs.size_ != 0) { + assign(rhs.begin(), rhs.end()); + } else { + clear(); + } + + return *this; + } + + /*! \brief Tests equality against another instance. + * + * \param vec the vector against which to compare. + */ + bool operator==(vector &vec) + { + if (size() != vec.size()) { + return false; + } + + for( unsigned int i = 0; i < size(); ++i ) { + if( operator[](i) != vec[i] ) { + return false; + } + } + return true; + } + + //! \brief Conversion operator to T*. + operator T* () { return data_; } + + //! \brief Conversion operator to const T*. + operator const T* () const { return data_; } + + //! \brief Tests whether this instance has any elements. + bool empty (void) const + { + return size_==0; + } + + //! \brief Returns the maximum number of elements this instance can hold. + unsigned int max_size (void) const + { + return N; + } + + //! \brief Returns the maximum number of elements this instance can hold. + unsigned int capacity () const + { + return N; + } + + /*! \brief Returns a reference to a given element. + * + * \param index which element to access. * + * \note + * The caller is responsible for ensuring index is >= 0 and < size(). + */ + T& operator[](int index) + { + return data_[index]; + } + + /*! \brief Returns a const reference to a given element. + * + * \param index which element to access. + * + * \note + * The caller is responsible for ensuring index is >= 0 and < size(). + */ + const T& operator[](int index) const + { + return data_[index]; + } + + /*! \brief Assigns elements of the vector based on a source iterator range. + * + * \param start Beginning iterator of source range + * \param end Enditerator of source range + * + * \note + * Will throw an exception if exceptions are enabled and size exceeded. + */ + template + void assign(I start, I end) + { + clear(); + while(start != end) { + push_back(*start); + start++; + } + } + + /*! \class iterator + * \brief Const iterator class for vectors + */ + class iterator + { + private: + const vector *vec_; + int index_; + + /** + * Internal iterator constructor to capture reference + * to the vector it iterates over rather than taking + * the vector by copy. + */ + iterator (const vector &vec, int index) : + vec_(&vec) + { + if( !vec.empty() ) { + index_ = index; + } else { + index_ = -1; + } + } + + public: + iterator(void) : + index_(-1), + vec_(NULL) + { + } + + iterator(const iterator& rhs) : + vec_(rhs.vec_), + index_(rhs.index_) + { + } + + ~iterator(void) {} + + static iterator begin(const cl::vector &vec) + { + iterator i(vec, 0); + + return i; + } + + static iterator end(const cl::vector &vec) + { + iterator i(vec, vec.size()); + + return i; + } + + bool operator==(iterator i) + { + return ((vec_ == i.vec_) && + (index_ == i.index_)); + } + + bool operator!=(iterator i) + { + return (!(*this==i)); + } + + iterator& operator++() + { + ++index_; + return *this; + } + + iterator operator++(int) + { + iterator retVal(*this); + ++index_; + return retVal; + } + + iterator& operator--() + { + --index_; + return *this; + } + + iterator operator--(int) + { + iterator retVal(*this); + --index_; + return retVal; + } + + const T& operator *() const + { + return (*vec_)[index_]; + } + }; + + iterator begin(void) + { + return iterator::begin(*this); + } + + iterator begin(void) const + { + return iterator::begin(*this); + } + + iterator end(void) + { + return iterator::end(*this); + } + + iterator end(void) const + { + return iterator::end(*this); + } + + T& front(void) + { + return data_[0]; + } + + T& back(void) + { + return data_[size_]; + } + + const T& front(void) const + { + return data_[0]; + } + + const T& back(void) const + { + return data_[size_-1]; + } +}; +#endif // #if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) + + + + + +namespace detail { +#define __DEFAULT_NOT_INITIALIZED 1 +#define __DEFAULT_BEING_INITIALIZED 2 +#define __DEFAULT_INITIALIZED 4 + + /* + * Compare and exchange primitives are needed for handling of defaults + */ + inline int compare_exchange(volatile int * dest, int exchange, int comparand) + { +#ifdef _WIN32 + return (int)(InterlockedCompareExchange( + (volatile long*)dest, + (long)exchange, + (long)comparand)); +#elif defined(__APPLE__) || defined(__MACOSX) + return OSAtomicOr32Orig((uint32_t)exchange, (volatile uint32_t*)dest); +#else // !_WIN32 || defined(__APPLE__) || defined(__MACOSX) + return (__sync_val_compare_and_swap( + dest, + comparand, + exchange)); +#endif // !_WIN32 + } + + inline void fence() { _mm_mfence(); } +}; // namespace detail + + +/*! \brief class used to interface between C++ and + * OpenCL C calls that require arrays of size_t values, whose + * size is known statically. + */ +template +class size_t +{ +private: + ::size_t data_[N]; + +public: + //! \brief Initialize size_t to all 0s + size_t() + { + for( int i = 0; i < N; ++i ) { + data_[i] = 0; + } + } + + ::size_t& operator[](int index) + { + return data_[index]; + } + + const ::size_t& operator[](int index) const + { + return data_[index]; + } + + //! \brief Conversion operator to T*. + operator ::size_t* () { return data_; } + + //! \brief Conversion operator to const T*. + operator const ::size_t* () const { return data_; } +}; + +namespace detail { + +// Generic getInfoHelper. The final parameter is used to guide overload +// resolution: the actual parameter passed is an int, which makes this +// a worse conversion sequence than a specialization that declares the +// parameter as an int. +template +inline cl_int getInfoHelper(Functor f, cl_uint name, T* param, long) +{ + return f(name, sizeof(T), param, NULL); +} + +// Specialized getInfoHelper for VECTOR_CLASS params +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + T* value = (T*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + param->assign(&value[0], &value[required/sizeof(T)]); + return CL_SUCCESS; +} + +/* Specialization for reference-counted types. This depends on the + * existence of Wrapper::cl_type, and none of the other types having the + * cl_type member. Note that simplify specifying the parameter as Wrapper + * does not work, because when using a derived type (e.g. Context) the generic + * template will provide a better match. + */ +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int, typename T::cl_type = 0) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + typename T::cl_type * value = (typename T::cl_type *) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + ::size_t elements = required / sizeof(typename T::cl_type); + param->assign(&value[0], &value[elements]); + for (::size_t i = 0; i < elements; i++) + { + if (value[i] != NULL) + { + err = (*param)[i].retain(); + if (err != CL_SUCCESS) { + return err; + } + } + } + return CL_SUCCESS; +} + +// Specialized for getInfo +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int) +{ + cl_int err = f(name, param->size() * sizeof(char *), &(*param)[0], NULL); + + if (err != CL_SUCCESS) { + return err; + } + + return CL_SUCCESS; +} + +// Specialized GetInfoHelper for STRING_CLASS params +template +inline cl_int getInfoHelper(Func f, cl_uint name, STRING_CLASS* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + char* value = (char*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + *param = value; + return CL_SUCCESS; +} + +// Specialized GetInfoHelper for cl::size_t params +template +inline cl_int getInfoHelper(Func f, cl_uint name, size_t* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + ::size_t* value = (::size_t*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + for(int i = 0; i < N; ++i) { + (*param)[i] = value[i]; + } + + return CL_SUCCESS; +} + +template struct ReferenceHandler; + +/* Specialization for reference-counted types. This depends on the + * existence of Wrapper::cl_type, and none of the other types having the + * cl_type member. Note that simplify specifying the parameter as Wrapper + * does not work, because when using a derived type (e.g. Context) the generic + * template will provide a better match. + */ +template +inline cl_int getInfoHelper(Func f, cl_uint name, T* param, int, typename T::cl_type = 0) +{ + typename T::cl_type value; + cl_int err = f(name, sizeof(value), &value, NULL); + if (err != CL_SUCCESS) { + return err; + } + *param = value; + if (value != NULL) + { + err = param->retain(); + if (err != CL_SUCCESS) { + return err; + } + } + return CL_SUCCESS; +} + +#define __PARAM_NAME_INFO_1_0(F) \ + F(cl_platform_info, CL_PLATFORM_PROFILE, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_VERSION, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_NAME, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_VENDOR, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_EXTENSIONS, STRING_CLASS) \ + \ + F(cl_device_info, CL_DEVICE_TYPE, cl_device_type) \ + F(cl_device_info, CL_DEVICE_VENDOR_ID, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_COMPUTE_UNITS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_GROUP_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_SIZES, VECTOR_CLASS< ::size_t>) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_CLOCK_FREQUENCY, cl_uint) \ + F(cl_device_info, CL_DEVICE_ADDRESS_BITS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_READ_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WRITE_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_MEM_ALLOC_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_WIDTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_HEIGHT, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_WIDTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_HEIGHT, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_DEPTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE_SUPPORT, cl_bool) \ + F(cl_device_info, CL_DEVICE_MAX_PARAMETER_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_MAX_SAMPLERS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MEM_BASE_ADDR_ALIGN, cl_uint) \ + F(cl_device_info, CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE, cl_uint) \ + F(cl_device_info, CL_DEVICE_SINGLE_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, cl_device_mem_cache_type) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE, cl_uint)\ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_MAX_CONSTANT_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_LOCAL_MEM_TYPE, cl_device_local_mem_type) \ + F(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_ERROR_CORRECTION_SUPPORT, cl_bool) \ + F(cl_device_info, CL_DEVICE_PROFILING_TIMER_RESOLUTION, ::size_t) \ + F(cl_device_info, CL_DEVICE_ENDIAN_LITTLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_AVAILABLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_COMPILER_AVAILABLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_EXECUTION_CAPABILITIES, cl_device_exec_capabilities) \ + F(cl_device_info, CL_DEVICE_QUEUE_PROPERTIES, cl_command_queue_properties) \ + F(cl_device_info, CL_DEVICE_PLATFORM, cl_platform_id) \ + F(cl_device_info, CL_DEVICE_NAME, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_VENDOR, STRING_CLASS) \ + F(cl_device_info, CL_DRIVER_VERSION, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_PROFILE, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_VERSION, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_EXTENSIONS, STRING_CLASS) \ + \ + F(cl_context_info, CL_CONTEXT_REFERENCE_COUNT, cl_uint) \ + F(cl_context_info, CL_CONTEXT_DEVICES, VECTOR_CLASS) \ + F(cl_context_info, CL_CONTEXT_PROPERTIES, VECTOR_CLASS) \ + \ + F(cl_event_info, CL_EVENT_COMMAND_QUEUE, cl::CommandQueue) \ + F(cl_event_info, CL_EVENT_COMMAND_TYPE, cl_command_type) \ + F(cl_event_info, CL_EVENT_REFERENCE_COUNT, cl_uint) \ + F(cl_event_info, CL_EVENT_COMMAND_EXECUTION_STATUS, cl_uint) \ + \ + F(cl_profiling_info, CL_PROFILING_COMMAND_QUEUED, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_SUBMIT, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_START, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_END, cl_ulong) \ + \ + F(cl_mem_info, CL_MEM_TYPE, cl_mem_object_type) \ + F(cl_mem_info, CL_MEM_FLAGS, cl_mem_flags) \ + F(cl_mem_info, CL_MEM_SIZE, ::size_t) \ + F(cl_mem_info, CL_MEM_HOST_PTR, void*) \ + F(cl_mem_info, CL_MEM_MAP_COUNT, cl_uint) \ + F(cl_mem_info, CL_MEM_REFERENCE_COUNT, cl_uint) \ + F(cl_mem_info, CL_MEM_CONTEXT, cl::Context) \ + \ + F(cl_image_info, CL_IMAGE_FORMAT, cl_image_format) \ + F(cl_image_info, CL_IMAGE_ELEMENT_SIZE, ::size_t) \ + F(cl_image_info, CL_IMAGE_ROW_PITCH, ::size_t) \ + F(cl_image_info, CL_IMAGE_SLICE_PITCH, ::size_t) \ + F(cl_image_info, CL_IMAGE_WIDTH, ::size_t) \ + F(cl_image_info, CL_IMAGE_HEIGHT, ::size_t) \ + F(cl_image_info, CL_IMAGE_DEPTH, ::size_t) \ + \ + F(cl_sampler_info, CL_SAMPLER_REFERENCE_COUNT, cl_uint) \ + F(cl_sampler_info, CL_SAMPLER_CONTEXT, cl::Context) \ + F(cl_sampler_info, CL_SAMPLER_NORMALIZED_COORDS, cl_addressing_mode) \ + F(cl_sampler_info, CL_SAMPLER_ADDRESSING_MODE, cl_filter_mode) \ + F(cl_sampler_info, CL_SAMPLER_FILTER_MODE, cl_bool) \ + \ + F(cl_program_info, CL_PROGRAM_REFERENCE_COUNT, cl_uint) \ + F(cl_program_info, CL_PROGRAM_CONTEXT, cl::Context) \ + F(cl_program_info, CL_PROGRAM_NUM_DEVICES, cl_uint) \ + F(cl_program_info, CL_PROGRAM_DEVICES, VECTOR_CLASS) \ + F(cl_program_info, CL_PROGRAM_SOURCE, STRING_CLASS) \ + F(cl_program_info, CL_PROGRAM_BINARY_SIZES, VECTOR_CLASS< ::size_t>) \ + F(cl_program_info, CL_PROGRAM_BINARIES, VECTOR_CLASS) \ + \ + F(cl_program_build_info, CL_PROGRAM_BUILD_STATUS, cl_build_status) \ + F(cl_program_build_info, CL_PROGRAM_BUILD_OPTIONS, STRING_CLASS) \ + F(cl_program_build_info, CL_PROGRAM_BUILD_LOG, STRING_CLASS) \ + \ + F(cl_kernel_info, CL_KERNEL_FUNCTION_NAME, STRING_CLASS) \ + F(cl_kernel_info, CL_KERNEL_NUM_ARGS, cl_uint) \ + F(cl_kernel_info, CL_KERNEL_REFERENCE_COUNT, cl_uint) \ + F(cl_kernel_info, CL_KERNEL_CONTEXT, cl::Context) \ + F(cl_kernel_info, CL_KERNEL_PROGRAM, cl::Program) \ + \ + F(cl_kernel_work_group_info, CL_KERNEL_WORK_GROUP_SIZE, ::size_t) \ + F(cl_kernel_work_group_info, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, cl::size_t<3>) \ + F(cl_kernel_work_group_info, CL_KERNEL_LOCAL_MEM_SIZE, cl_ulong) \ + \ + F(cl_command_queue_info, CL_QUEUE_CONTEXT, cl::Context) \ + F(cl_command_queue_info, CL_QUEUE_DEVICE, cl::Device) \ + F(cl_command_queue_info, CL_QUEUE_REFERENCE_COUNT, cl_uint) \ + F(cl_command_queue_info, CL_QUEUE_PROPERTIES, cl_command_queue_properties) + +#if defined(CL_VERSION_1_1) +#define __PARAM_NAME_INFO_1_1(F) \ + F(cl_context_info, CL_CONTEXT_NUM_DEVICES, cl_uint)\ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_INT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF, cl_uint) \ + F(cl_device_info, CL_DEVICE_DOUBLE_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_HALF_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_HOST_UNIFIED_MEMORY, cl_bool) \ + F(cl_device_info, CL_DEVICE_OPENCL_C_VERSION, STRING_CLASS) \ + \ + F(cl_mem_info, CL_MEM_ASSOCIATED_MEMOBJECT, cl::Memory) \ + F(cl_mem_info, CL_MEM_OFFSET, ::size_t) \ + \ + F(cl_kernel_work_group_info, CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE, ::size_t) \ + F(cl_kernel_work_group_info, CL_KERNEL_PRIVATE_MEM_SIZE, cl_ulong) \ + \ + F(cl_event_info, CL_EVENT_CONTEXT, cl::Context) +#endif // CL_VERSION_1_1 + + +#if defined(CL_VERSION_1_2) +#define __PARAM_NAME_INFO_1_2(F) \ + F(cl_image_info, CL_IMAGE_BUFFER, cl::Buffer) \ + \ + F(cl_program_info, CL_PROGRAM_NUM_KERNELS, ::size_t) \ + F(cl_program_info, CL_PROGRAM_KERNEL_NAMES, STRING_CLASS) \ + \ + F(cl_program_build_info, CL_PROGRAM_BINARY_TYPE, cl_program_binary_type) \ + \ + F(cl_kernel_info, CL_KERNEL_ATTRIBUTES, STRING_CLASS) \ + \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_ADDRESS_QUALIFIER, cl_kernel_arg_address_qualifier) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_ACCESS_QUALIFIER, cl_kernel_arg_access_qualifier) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_TYPE_NAME, STRING_CLASS) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_NAME, STRING_CLASS) \ + \ + F(cl_device_info, CL_DEVICE_PARENT_DEVICE, cl_device_id) \ + F(cl_device_info, CL_DEVICE_PARTITION_PROPERTIES, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_PARTITION_TYPE, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_REFERENCE_COUNT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, ::size_t) \ + F(cl_device_info, CL_DEVICE_PARTITION_AFFINITY_DOMAIN, cl_device_affinity_domain) \ + F(cl_device_info, CL_DEVICE_BUILT_IN_KERNELS, STRING_CLASS) +#endif // #if defined(CL_VERSION_1_2) + +#if defined(USE_CL_DEVICE_FISSION) +#define __PARAM_NAME_DEVICE_FISSION(F) \ + F(cl_device_info, CL_DEVICE_PARENT_DEVICE_EXT, cl_device_id) \ + F(cl_device_info, CL_DEVICE_PARTITION_TYPES_EXT, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \ + F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, VECTOR_CLASS) +#endif // USE_CL_DEVICE_FISSION + +template +struct param_traits {}; + +#define __CL_DECLARE_PARAM_TRAITS(token, param_name, T) \ +struct token; \ +template<> \ +struct param_traits \ +{ \ + enum { value = param_name }; \ + typedef T param_type; \ +}; + +__PARAM_NAME_INFO_1_0(__CL_DECLARE_PARAM_TRAITS) +#if defined(CL_VERSION_1_1) +__PARAM_NAME_INFO_1_1(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_1_1 +#if defined(CL_VERSION_1_2) +__PARAM_NAME_INFO_1_2(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_1_1 + +#if defined(USE_CL_DEVICE_FISSION) +__PARAM_NAME_DEVICE_FISSION(__CL_DECLARE_PARAM_TRAITS); +#endif // USE_CL_DEVICE_FISSION + +#ifdef CL_PLATFORM_ICD_SUFFIX_KHR +__CL_DECLARE_PARAM_TRAITS(cl_platform_info, CL_PLATFORM_ICD_SUFFIX_KHR, STRING_CLASS) +#endif + +#ifdef CL_DEVICE_PROFILING_TIMER_OFFSET_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_PROFILING_TIMER_OFFSET_AMD, cl_ulong) +#endif + +#ifdef CL_DEVICE_GLOBAL_FREE_MEMORY_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_FREE_MEMORY_AMD, VECTOR_CLASS< ::size_t>) +#endif +#ifdef CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_SIMD_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_WAVEFRONT_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WAVEFRONT_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_LOCAL_MEM_BANKS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_BANKS_AMD, cl_uint) +#endif + +#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV, cl_uint) +#endif +#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV, cl_uint) +#endif +#ifdef CL_DEVICE_REGISTERS_PER_BLOCK_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_REGISTERS_PER_BLOCK_NV, cl_uint) +#endif +#ifdef CL_DEVICE_WARP_SIZE_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WARP_SIZE_NV, cl_uint) +#endif +#ifdef CL_DEVICE_GPU_OVERLAP_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GPU_OVERLAP_NV, cl_bool) +#endif +#ifdef CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV, cl_bool) +#endif +#ifdef CL_DEVICE_INTEGRATED_MEMORY_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_INTEGRATED_MEMORY_NV, cl_bool) +#endif + +// Convenience functions + +template +inline cl_int +getInfo(Func f, cl_uint name, T* param) +{ + return getInfoHelper(f, name, param, 0); +} + +template +struct GetInfoFunctor0 +{ + Func f_; const Arg0& arg0_; + cl_int operator ()( + cl_uint param, ::size_t size, void* value, ::size_t* size_ret) + { return f_(arg0_, param, size, value, size_ret); } +}; + +template +struct GetInfoFunctor1 +{ + Func f_; const Arg0& arg0_; const Arg1& arg1_; + cl_int operator ()( + cl_uint param, ::size_t size, void* value, ::size_t* size_ret) + { return f_(arg0_, arg1_, param, size, value, size_ret); } +}; + +template +inline cl_int +getInfo(Func f, const Arg0& arg0, cl_uint name, T* param) +{ + GetInfoFunctor0 f0 = { f, arg0 }; + return getInfoHelper(f0, name, param, 0); +} + +template +inline cl_int +getInfo(Func f, const Arg0& arg0, const Arg1& arg1, cl_uint name, T* param) +{ + GetInfoFunctor1 f0 = { f, arg0, arg1 }; + return getInfoHelper(f0, name, param, 0); +} + +template +struct ReferenceHandler +{ }; + +#if defined(CL_VERSION_1_2) +/** + * OpenCL 1.2 devices do have retain/release. + */ +template <> +struct ReferenceHandler +{ + /** + * Retain the device. + * \param device A valid device created using createSubDevices + * \return + * CL_SUCCESS if the function executed successfully. + * CL_INVALID_DEVICE if device was not a valid subdevice + * CL_OUT_OF_RESOURCES + * CL_OUT_OF_HOST_MEMORY + */ + static cl_int retain(cl_device_id device) + { return ::clRetainDevice(device); } + /** + * Retain the device. + * \param device A valid device created using createSubDevices + * \return + * CL_SUCCESS if the function executed successfully. + * CL_INVALID_DEVICE if device was not a valid subdevice + * CL_OUT_OF_RESOURCES + * CL_OUT_OF_HOST_MEMORY + */ + static cl_int release(cl_device_id device) + { return ::clReleaseDevice(device); } +}; +#else // #if defined(CL_VERSION_1_2) +/** + * OpenCL 1.1 devices do not have retain/release. + */ +template <> +struct ReferenceHandler +{ + // cl_device_id does not have retain(). + static cl_int retain(cl_device_id) + { return CL_SUCCESS; } + // cl_device_id does not have release(). + static cl_int release(cl_device_id) + { return CL_SUCCESS; } +}; +#endif // #if defined(CL_VERSION_1_2) + +template <> +struct ReferenceHandler +{ + // cl_platform_id does not have retain(). + static cl_int retain(cl_platform_id) + { return CL_SUCCESS; } + // cl_platform_id does not have release(). + static cl_int release(cl_platform_id) + { return CL_SUCCESS; } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_context context) + { return ::clRetainContext(context); } + static cl_int release(cl_context context) + { return ::clReleaseContext(context); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_command_queue queue) + { return ::clRetainCommandQueue(queue); } + static cl_int release(cl_command_queue queue) + { return ::clReleaseCommandQueue(queue); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_mem memory) + { return ::clRetainMemObject(memory); } + static cl_int release(cl_mem memory) + { return ::clReleaseMemObject(memory); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_sampler sampler) + { return ::clRetainSampler(sampler); } + static cl_int release(cl_sampler sampler) + { return ::clReleaseSampler(sampler); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_program program) + { return ::clRetainProgram(program); } + static cl_int release(cl_program program) + { return ::clReleaseProgram(program); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_kernel kernel) + { return ::clRetainKernel(kernel); } + static cl_int release(cl_kernel kernel) + { return ::clReleaseKernel(kernel); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_event event) + { return ::clRetainEvent(event); } + static cl_int release(cl_event event) + { return ::clReleaseEvent(event); } +}; + + +// Extracts version number with major in the upper 16 bits, minor in the lower 16 +static cl_uint getVersion(const char *versionInfo) +{ + int highVersion = 0; + int lowVersion = 0; + int index = 7; + while(versionInfo[index] != '.' ) { + highVersion *= 10; + highVersion += versionInfo[index]-'0'; + ++index; + } + ++index; + while(versionInfo[index] != ' ' ) { + lowVersion *= 10; + lowVersion += versionInfo[index]-'0'; + ++index; + } + return (highVersion << 16) | lowVersion; +} + +static cl_uint getPlatformVersion(cl_platform_id platform) +{ + ::size_t size = 0; + clGetPlatformInfo(platform, CL_PLATFORM_VERSION, 0, NULL, &size); + char *versionInfo = (char *) alloca(size); + clGetPlatformInfo(platform, CL_PLATFORM_VERSION, size, &versionInfo[0], &size); + return getVersion(versionInfo); +} + +static cl_uint getDevicePlatformVersion(cl_device_id device) +{ + cl_platform_id platform; + clGetDeviceInfo(device, CL_DEVICE_PLATFORM, sizeof(platform), &platform, NULL); + return getPlatformVersion(platform); +} + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +static cl_uint getContextPlatformVersion(cl_context context) +{ + // The platform cannot be queried directly, so we first have to grab a + // device and obtain its context + ::size_t size = 0; + clGetContextInfo(context, CL_CONTEXT_DEVICES, 0, NULL, &size); + if (size == 0) + return 0; + cl_device_id *devices = (cl_device_id *) alloca(size); + clGetContextInfo(context, CL_CONTEXT_DEVICES, size, devices, NULL); + return getDevicePlatformVersion(devices[0]); +} +#endif // #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + +template +class Wrapper +{ +public: + typedef T cl_type; + +protected: + cl_type object_; + +public: + Wrapper() : object_(NULL) { } + + Wrapper(const cl_type &obj) : object_(obj) { } + + ~Wrapper() + { + if (object_ != NULL) { release(); } + } + + Wrapper(const Wrapper& rhs) + { + object_ = rhs.object_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + } + + Wrapper& operator = (const Wrapper& rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs.object_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + return *this; + } + + Wrapper& operator = (const cl_type &rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs; + return *this; + } + + cl_type operator ()() const { return object_; } + + cl_type& operator ()() { return object_; } + +protected: + template + friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); + + cl_int retain() const + { + return ReferenceHandler::retain(object_); + } + + cl_int release() const + { + return ReferenceHandler::release(object_); + } +}; + +template <> +class Wrapper +{ +public: + typedef cl_device_id cl_type; + +protected: + cl_type object_; + bool referenceCountable_; + + static bool isReferenceCountable(cl_device_id device) + { + bool retVal = false; + if (device != NULL) { + int version = getDevicePlatformVersion(device); + if(version > ((1 << 16) + 1)) { + retVal = true; + } + } + return retVal; + } + +public: + Wrapper() : object_(NULL), referenceCountable_(false) + { + } + + Wrapper(const cl_type &obj) : object_(obj), referenceCountable_(false) + { + referenceCountable_ = isReferenceCountable(obj); + } + + ~Wrapper() + { + if (object_ != NULL) { release(); } + } + + Wrapper(const Wrapper& rhs) + { + object_ = rhs.object_; + referenceCountable_ = isReferenceCountable(object_); + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + } + + Wrapper& operator = (const Wrapper& rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs.object_; + referenceCountable_ = rhs.referenceCountable_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + return *this; + } + + Wrapper& operator = (const cl_type &rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs; + referenceCountable_ = isReferenceCountable(object_); + return *this; + } + + cl_type operator ()() const { return object_; } + + cl_type& operator ()() { return object_; } + +protected: + template + friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); + + template + friend inline cl_int getInfoHelper(Func, cl_uint, VECTOR_CLASS*, int, typename U::cl_type); + + cl_int retain() const + { + if( referenceCountable_ ) { + return ReferenceHandler::retain(object_); + } + else { + return CL_SUCCESS; + } + } + + cl_int release() const + { + if( referenceCountable_ ) { + return ReferenceHandler::release(object_); + } + else { + return CL_SUCCESS; + } + } +}; + +} // namespace detail +//! \endcond + +/*! \stuct ImageFormat + * \brief Adds constructors and member functions for cl_image_format. + * + * \see cl_image_format + */ +struct ImageFormat : public cl_image_format +{ + //! \brief Default constructor - performs no initialization. + ImageFormat(){} + + //! \brief Initializing constructor. + ImageFormat(cl_channel_order order, cl_channel_type type) + { + image_channel_order = order; + image_channel_data_type = type; + } + + //! \brief Assignment operator. + ImageFormat& operator = (const ImageFormat& rhs) + { + if (this != &rhs) { + this->image_channel_data_type = rhs.image_channel_data_type; + this->image_channel_order = rhs.image_channel_order; + } + return *this; + } +}; + +/*! \brief Class interface for cl_device_id. + * + * \note Copies of these objects are inexpensive, since they don't 'own' + * any underlying resources or data structures. + * + * \see cl_device_id + */ +class Device : public detail::Wrapper +{ +public: + //! \brief Default constructor - initializes to NULL. + Device() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device(const Device& device) : detail::Wrapper(device) { } + + /*! \brief Constructor from cl_device_id. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device(const cl_device_id &device) : detail::Wrapper(device) { } + + /*! \brief Returns the first device on the default context. + * + * \see Context::getDefault() + */ + static Device getDefault(cl_int * err = NULL); + + /*! \brief Assignment operator from Device. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device& operator = (const Device& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_device_id. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device& operator = (const cl_device_id& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetDeviceInfo(). + template + cl_int getInfo(cl_device_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetDeviceInfo, object_, name, param), + __GET_DEVICE_INFO_ERR); + } + + //! \brief Wrapper for clGetDeviceInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_device_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /** + * CL 1.2 version + */ +#if defined(CL_VERSION_1_2) + //! \brief Wrapper for clCreateSubDevicesEXT(). + cl_int createSubDevices( + const cl_device_partition_property * properties, + VECTOR_CLASS* devices) + { + cl_uint n = 0; + cl_int err = clCreateSubDevices(object_, properties, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = clCreateSubDevices(object_, properties, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif // #if defined(CL_VERSION_1_2) + +/** + * CL 1.1 version that uses device fission. + */ +#if defined(CL_VERSION_1_1) +#if defined(USE_CL_DEVICE_FISSION) + cl_int createSubDevices( + const cl_device_partition_property_ext * properties, + VECTOR_CLASS* devices) + { + typedef CL_API_ENTRY cl_int + ( CL_API_CALL * PFN_clCreateSubDevicesEXT)( + cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + static PFN_clCreateSubDevicesEXT pfn_clCreateSubDevicesEXT = NULL; + __INIT_CL_EXT_FCN_PTR(clCreateSubDevicesEXT); + + cl_uint n = 0; + cl_int err = pfn_clCreateSubDevicesEXT(object_, properties, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = pfn_clCreateSubDevicesEXT(object_, properties, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif // #if defined(USE_CL_DEVICE_FISSION) +#endif // #if defined(CL_VERSION_1_1) +}; + +/*! \brief Class interface for cl_platform_id. + * + * \note Copies of these objects are inexpensive, since they don't 'own' + * any underlying resources or data structures. + * + * \see cl_platform_id + */ +class Platform : public detail::Wrapper +{ +public: + //! \brief Default constructor - initializes to NULL. + Platform() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform(const Platform& platform) : detail::Wrapper(platform) { } + + /*! \brief Constructor from cl_platform_id. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform(const cl_platform_id &platform) : detail::Wrapper(platform) { } + + /*! \brief Assignment operator from Platform. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform& operator = (const Platform& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_platform_id. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform& operator = (const cl_platform_id& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetPlatformInfo(). + cl_int getInfo(cl_platform_info name, STRING_CLASS* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetPlatformInfo, object_, name, param), + __GET_PLATFORM_INFO_ERR); + } + + //! \brief Wrapper for clGetPlatformInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_platform_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Gets a list of devices for this platform. + * + * Wraps clGetDeviceIDs(). + */ + cl_int getDevices( + cl_device_type type, + VECTOR_CLASS* devices) const + { + cl_uint n = 0; + if( devices == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); + } + cl_int err = ::clGetDeviceIDs(object_, type, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = ::clGetDeviceIDs(object_, type, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } + +#if defined(USE_DX_INTEROP) + /*! \brief Get the list of available D3D10 devices. + * + * \param d3d_device_source. + * + * \param d3d_object. + * + * \param d3d_device_set. + * + * \param devices returns a vector of OpenCL D3D10 devices found. The cl::Device + * values returned in devices can be used to identify a specific OpenCL + * device. If \a devices argument is NULL, this argument is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * + * The application can query specific capabilities of the OpenCL device(s) + * returned by cl::getDevices. This can be used by the application to + * determine which device(s) to use. + * + * \note In the case that exceptions are enabled and a return value + * other than CL_SUCCESS is generated, then cl::Error exception is + * generated. + */ + cl_int getDevices( + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + VECTOR_CLASS* devices) const + { + typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clGetDeviceIDsFromD3D10KHR)( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint* num_devices); + + if( devices == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); + } + + static PFN_clGetDeviceIDsFromD3D10KHR pfn_clGetDeviceIDsFromD3D10KHR = NULL; + __INIT_CL_EXT_FCN_PTR_PLATFORM(object_, clGetDeviceIDsFromD3D10KHR); + + cl_uint n = 0; + cl_int err = pfn_clGetDeviceIDsFromD3D10KHR( + object_, + d3d_device_source, + d3d_object, + d3d_device_set, + 0, + NULL, + &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = pfn_clGetDeviceIDsFromD3D10KHR( + object_, + d3d_device_source, + d3d_object, + d3d_device_set, + n, + ids, + NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif + + /*! \brief Gets a list of available platforms. + * + * Wraps clGetPlatformIDs(). + */ + static cl_int get( + VECTOR_CLASS* platforms) + { + cl_uint n = 0; + + if( platforms == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); + } + + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + platforms->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } + + /*! \brief Gets the first available platform. + * + * Wraps clGetPlatformIDs(), returning the first result. + */ + static cl_int get( + Platform * platform) + { + cl_uint n = 0; + + if( platform == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); + } + + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + *platform = ids[0]; + return CL_SUCCESS; + } + + /*! \brief Gets the first available platform, returning it by value. + * + * Wraps clGetPlatformIDs(), returning the first result. + */ + static Platform get( + cl_int * errResult = NULL) + { + Platform platform; + cl_uint n = 0; + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + if (errResult != NULL) { + *errResult = err; + } + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + + if (err != CL_SUCCESS) { + detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + if (errResult != NULL) { + *errResult = err; + } + + return ids[0]; + } + + static Platform getDefault( + cl_int *errResult = NULL ) + { + return get(errResult); + } + + +#if defined(CL_VERSION_1_2) + //! \brief Wrapper for clUnloadCompiler(). + cl_int + unloadCompiler() + { + return ::clUnloadPlatformCompiler(object_); + } +#endif // #if defined(CL_VERSION_1_2) +}; // class Platform + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +/** + * Unload the OpenCL compiler. + * \note Deprecated for OpenCL 1.2. Use Platform::unloadCompiler instead. + */ +inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int +UnloadCompiler() CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; +inline cl_int +UnloadCompiler() +{ + return ::clUnloadCompiler(); +} +#endif // #if defined(CL_VERSION_1_1) + +/*! \brief Class interface for cl_context. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_context as the original. For details, see + * clRetainContext() and clReleaseContext(). + * + * \see cl_context + */ +class Context + : public detail::Wrapper +{ +private: + static volatile int default_initialized_; + static Context default_; + static volatile cl_int default_error_; +public: + /*! \brief Destructor. + * + * This calls clReleaseContext() on the value held by this instance. + */ + ~Context() { } + + /*! \brief Constructs a context including a list of specified devices. + * + * Wraps clCreateContext(). + */ + Context( + const VECTOR_CLASS& devices, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + object_ = ::clCreateContext( + properties, (cl_uint) numDevices, + deviceIDs, + notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_ERR); + if (err != NULL) { + *err = error; + } + } + + Context( + const Device& device, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + + cl_device_id deviceID = device(); + + object_ = ::clCreateContext( + properties, 1, + &deviceID, + notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Constructs a context including all devices of a specified type. + * + * Wraps clCreateContextFromType(). + */ + Context( + cl_device_type type, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + +#if !defined(__APPLE__) || !defined(__MACOS) + cl_context_properties prop[4] = {CL_CONTEXT_PLATFORM, 0, 0, 0 }; + if (properties == NULL) { + prop[1] = (cl_context_properties)Platform::get(&error)(); + if (error != CL_SUCCESS) { + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + return; + } + } + + properties = &prop[0]; + } +#endif + object_ = ::clCreateContextFromType( + properties, type, notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Returns a singleton context including all devices of CL_DEVICE_TYPE_DEFAULT. + * + * \note All calls to this function return the same cl_context as the first. + */ + static Context getDefault(cl_int * err = NULL) + { + int state = detail::compare_exchange( + &default_initialized_, + __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); + + if (state & __DEFAULT_INITIALIZED) { + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + if (state & __DEFAULT_BEING_INITIALIZED) { + // Assume writes will propagate eventually... + while(default_initialized_ != __DEFAULT_INITIALIZED) { + detail::fence(); + } + + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + cl_int error; + default_ = Context( + CL_DEVICE_TYPE_DEFAULT, + NULL, + NULL, + NULL, + &error); + + detail::fence(); + + default_error_ = error; + // Assume writes will propagate eventually... + default_initialized_ = __DEFAULT_INITIALIZED; + + detail::fence(); + + if (err != NULL) { + *err = default_error_; + } + return default_; + + } + + //! \brief Default constructor - initializes to NULL. + Context() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This calls clRetainContext() on the parameter's cl_context. + */ + Context(const Context& context) : detail::Wrapper(context) { } + + /*! \brief Constructor from cl_context - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_context + * into the new Context object. + */ + __CL_EXPLICIT_CONSTRUCTORS Context(const cl_context& context) : detail::Wrapper(context) { } + + /*! \brief Assignment operator from Context. + * + * This calls clRetainContext() on the parameter and clReleaseContext() on + * the previous value held by this instance. + */ + Context& operator = (const Context& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_context - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseContext() on the value previously held by this instance. + */ + Context& operator = (const cl_context& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetContextInfo(). + template + cl_int getInfo(cl_context_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetContextInfo, object_, name, param), + __GET_CONTEXT_INFO_ERR); + } + + //! \brief Wrapper for clGetContextInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_context_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Gets a list of supported image formats. + * + * Wraps clGetSupportedImageFormats(). + */ + cl_int getSupportedImageFormats( + cl_mem_flags flags, + cl_mem_object_type type, + VECTOR_CLASS* formats) const + { + cl_uint numEntries; + cl_int err = ::clGetSupportedImageFormats( + object_, + flags, + type, + 0, + NULL, + &numEntries); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); + } + + ImageFormat* value = (ImageFormat*) + alloca(numEntries * sizeof(ImageFormat)); + err = ::clGetSupportedImageFormats( + object_, + flags, + type, + numEntries, + (cl_image_format*) value, + NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); + } + + formats->assign(&value[0], &value[numEntries]); + return CL_SUCCESS; + } +}; + +inline Device Device::getDefault(cl_int * err) +{ + cl_int error; + Device device; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + device = context.getInfo()[0]; + if (err != NULL) { + *err = CL_SUCCESS; + } + } + + return device; +} + + +#ifdef _WIN32 +__declspec(selectany) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__declspec(selectany) Context Context::default_; +__declspec(selectany) volatile cl_int Context::default_error_ = CL_SUCCESS; +#else +__attribute__((weak)) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__attribute__((weak)) Context Context::default_; +__attribute__((weak)) volatile cl_int Context::default_error_ = CL_SUCCESS; +#endif + +/*! \brief Class interface for cl_event. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_event as the original. For details, see + * clRetainEvent() and clReleaseEvent(). + * + * \see cl_event + */ +class Event : public detail::Wrapper +{ +public: + /*! \brief Destructor. + * + * This calls clReleaseEvent() on the value held by this instance. + */ + ~Event() { } + + //! \brief Default constructor - initializes to NULL. + Event() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This calls clRetainEvent() on the parameter's cl_event. + */ + Event(const Event& event) : detail::Wrapper(event) { } + + /*! \brief Constructor from cl_event - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_event + * into the new Event object. + */ + Event(const cl_event& event) : detail::Wrapper(event) { } + + /*! \brief Assignment operator from cl_event - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseEvent() on the value previously held by this instance. + */ + Event& operator = (const Event& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_event. + * + * This calls clRetainEvent() on the parameter and clReleaseEvent() on + * the previous value held by this instance. + */ + Event& operator = (const cl_event& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetEventInfo(). + template + cl_int getInfo(cl_event_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetEventInfo, object_, name, param), + __GET_EVENT_INFO_ERR); + } + + //! \brief Wrapper for clGetEventInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_event_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + //! \brief Wrapper for clGetEventProfilingInfo(). + template + cl_int getProfilingInfo(cl_profiling_info name, T* param) const + { + return detail::errHandler(detail::getInfo( + &::clGetEventProfilingInfo, object_, name, param), + __GET_EVENT_PROFILE_INFO_ERR); + } + + //! \brief Wrapper for clGetEventProfilingInfo() that returns by value. + template typename + detail::param_traits::param_type + getProfilingInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_profiling_info, name>::param_type param; + cl_int result = getProfilingInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Blocks the calling thread until this event completes. + * + * Wraps clWaitForEvents(). + */ + cl_int wait() const + { + return detail::errHandler( + ::clWaitForEvents(1, &object_), + __WAIT_FOR_EVENTS_ERR); + } + +#if defined(CL_VERSION_1_1) + /*! \brief Registers a user callback function for a specific command execution status. + * + * Wraps clSetEventCallback(). + */ + cl_int setCallback( + cl_int type, + void (CL_CALLBACK * pfn_notify)(cl_event, cl_int, void *), + void * user_data = NULL) + { + return detail::errHandler( + ::clSetEventCallback( + object_, + type, + pfn_notify, + user_data), + __SET_EVENT_CALLBACK_ERR); + } +#endif + + /*! \brief Blocks the calling thread until every event specified is complete. + * + * Wraps clWaitForEvents(). + */ + static cl_int + waitForEvents(const VECTOR_CLASS& events) + { + return detail::errHandler( + ::clWaitForEvents( + (cl_uint) events.size(), (cl_event*)&events.front()), + __WAIT_FOR_EVENTS_ERR); + } +}; + +#if defined(CL_VERSION_1_1) +/*! \brief Class interface for user events (a subset of cl_event's). + * + * See Event for details about copy semantics, etc. + */ +class UserEvent : public Event +{ +public: + /*! \brief Constructs a user event on a given context. + * + * Wraps clCreateUserEvent(). + */ + UserEvent( + const Context& context, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateUserEvent( + context(), + &error); + + detail::errHandler(error, __CREATE_USER_EVENT_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + UserEvent() : Event() { } + + //! \brief Copy constructor - performs shallow copy. + UserEvent(const UserEvent& event) : Event(event) { } + + //! \brief Assignment Operator - performs shallow copy. + UserEvent& operator = (const UserEvent& rhs) + { + if (this != &rhs) { + Event::operator=(rhs); + } + return *this; + } + + /*! \brief Sets the execution status of a user event object. + * + * Wraps clSetUserEventStatus(). + */ + cl_int setStatus(cl_int status) + { + return detail::errHandler( + ::clSetUserEventStatus(object_,status), + __SET_USER_EVENT_STATUS_ERR); + } +}; +#endif + +/*! \brief Blocks the calling thread until every event specified is complete. + * + * Wraps clWaitForEvents(). + */ +inline static cl_int +WaitForEvents(const VECTOR_CLASS& events) +{ + return detail::errHandler( + ::clWaitForEvents( + (cl_uint) events.size(), (cl_event*)&events.front()), + __WAIT_FOR_EVENTS_ERR); +} + +/*! \brief Class interface for cl_mem. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_mem as the original. For details, see + * clRetainMemObject() and clReleaseMemObject(). + * + * \see cl_mem + */ +class Memory : public detail::Wrapper +{ +public: + + /*! \brief Destructor. + * + * This calls clReleaseMemObject() on the value held by this instance. + */ + ~Memory() {} + + //! \brief Default constructor - initializes to NULL. + Memory() : detail::Wrapper() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainMemObject() on the parameter's cl_mem. + */ + Memory(const Memory& memory) : detail::Wrapper(memory) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_mem + * into the new Memory object. + */ + __CL_EXPLICIT_CONSTRUCTORS Memory(const cl_mem& memory) : detail::Wrapper(memory) { } + + /*! \brief Assignment operator from Memory. + * + * This calls clRetainMemObject() on the parameter and clReleaseMemObject() + * on the previous value held by this instance. + */ + Memory& operator = (const Memory& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_mem - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseMemObject() on the value previously held by this instance. + */ + Memory& operator = (const cl_mem& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetMemObjectInfo(). + template + cl_int getInfo(cl_mem_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetMemObjectInfo, object_, name, param), + __GET_MEM_OBJECT_INFO_ERR); + } + + //! \brief Wrapper for clGetMemObjectInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_mem_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + +#if defined(CL_VERSION_1_1) + /*! \brief Registers a callback function to be called when the memory object + * is no longer needed. + * + * Wraps clSetMemObjectDestructorCallback(). + * + * Repeated calls to this function, for a given cl_mem value, will append + * to the list of functions called (in reverse order) when memory object's + * resources are freed and the memory object is deleted. + * + * \note + * The registered callbacks are associated with the underlying cl_mem + * value - not the Memory class instance. + */ + cl_int setDestructorCallback( + void (CL_CALLBACK * pfn_notify)(cl_mem, void *), + void * user_data = NULL) + { + return detail::errHandler( + ::clSetMemObjectDestructorCallback( + object_, + pfn_notify, + user_data), + __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR); + } +#endif + +}; + +// Pre-declare copy functions +class Buffer; +template< typename IteratorType > +cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ); +template< typename IteratorType > +cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ); + +/*! \brief Class interface for Buffer Memory Objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Buffer : public Memory +{ +public: + + /*! \brief Constructs a Buffer in a specified context. + * + * Wraps clCreateBuffer(). + * + * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was + * specified. Note alignment & exclusivity requirements. + */ + Buffer( + const Context& context, + cl_mem_flags flags, + ::size_t size, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Constructs a Buffer in the default context. + * + * Wraps clCreateBuffer(). + * + * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was + * specified. Note alignment & exclusivity requirements. + * + * \see Context::getDefault() + */ + Buffer( + cl_mem_flags flags, + ::size_t size, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + + Context context = Context::getDefault(err); + + object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! + * \brief Construct a Buffer from a host container via iterators. + * If useHostPtr is specified iterators must be random access. + */ + template< typename IteratorType > + Buffer( + IteratorType startIterator, + IteratorType endIterator, + bool readOnly, + bool useHostPtr = false, + cl_int* err = NULL) + { + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + cl_mem_flags flags = 0; + if( readOnly ) { + flags |= CL_MEM_READ_ONLY; + } + else { + flags |= CL_MEM_READ_WRITE; + } + if( useHostPtr ) { + flags |= CL_MEM_USE_HOST_PTR; + } + + ::size_t size = sizeof(DataType)*(endIterator - startIterator); + + Context context = Context::getDefault(err); + + if( useHostPtr ) { + object_ = ::clCreateBuffer(context(), flags, size, static_cast(&*startIterator), &error); + } else { + object_ = ::clCreateBuffer(context(), flags, size, 0, &error); + } + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + if( !useHostPtr ) { + error = cl::copy(startIterator, endIterator, *this); + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + } + + //! \brief Default constructor - initializes to NULL. + Buffer() : Memory() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Buffer(const Buffer& buffer) : Memory(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Buffer(const cl_mem& buffer) : Memory(buffer) { } + + /*! \brief Assignment from Buffer - performs shallow copy. + * + * See Memory for further details. + */ + Buffer& operator = (const Buffer& rhs) + { + if (this != &rhs) { + Memory::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Buffer& operator = (const cl_mem& rhs) + { + Memory::operator=(rhs); + return *this; + } + +#if defined(CL_VERSION_1_1) + /*! \brief Creates a new buffer object from this. + * + * Wraps clCreateSubBuffer(). + */ + Buffer createSubBuffer( + cl_mem_flags flags, + cl_buffer_create_type buffer_create_type, + const void * buffer_create_info, + cl_int * err = NULL) + { + Buffer result; + cl_int error; + result.object_ = ::clCreateSubBuffer( + object_, + flags, + buffer_create_type, + buffer_create_info, + &error); + + detail::errHandler(error, __CREATE_SUBBUFFER_ERR); + if (err != NULL) { + *err = error; + } + + return result; + } +#endif +}; + +#if defined (USE_DX_INTEROP) +/*! \brief Class interface for creating OpenCL buffers from ID3D10Buffer's. + * + * This is provided to facilitate interoperability with Direct3D. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferD3D10 : public Buffer +{ +public: + typedef CL_API_ENTRY cl_mem (CL_API_CALL *PFN_clCreateFromD3D10BufferKHR)( + cl_context context, cl_mem_flags flags, ID3D10Buffer* buffer, + cl_int* errcode_ret); + + /*! \brief Constructs a BufferD3D10, in a specified context, from a + * given ID3D10Buffer. + * + * Wraps clCreateFromD3D10BufferKHR(). + */ + BufferD3D10( + const Context& context, + cl_mem_flags flags, + ID3D10Buffer* bufobj, + cl_int * err = NULL) + { + static PFN_clCreateFromD3D10BufferKHR pfn_clCreateFromD3D10BufferKHR = NULL; + +#if defined(CL_VERSION_1_2) + vector props = context.getInfo(); + cl_platform platform = -1; + for( int i = 0; i < props.size(); ++i ) { + if( props[i] == CL_CONTEXT_PLATFORM ) { + platform = props[i+1]; + } + } + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clCreateFromD3D10BufferKHR); +#endif +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clCreateFromD3D10BufferKHR); +#endif + + cl_int error; + object_ = pfn_clCreateFromD3D10BufferKHR( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferD3D10() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10(const BufferD3D10& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferD3D10(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferD3D10 - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10& operator = (const BufferD3D10& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } +}; +#endif + +/*! \brief Class interface for GL Buffer Memory Objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferGL : public Buffer +{ +public: + /*! \brief Constructs a BufferGL in a specified context, from a given + * GL buffer. + * + * Wraps clCreateFromGLBuffer(). + */ + BufferGL( + const Context& context, + cl_mem_flags flags, + GLuint bufobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLBuffer( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferGL() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL(const BufferGL& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferGL(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferGL - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL& operator = (const BufferGL& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetGLObjectInfo(). + cl_int getObjectInfo( + cl_gl_object_type *type, + GLuint * gl_object_name) + { + return detail::errHandler( + ::clGetGLObjectInfo(object_,type,gl_object_name), + __GET_GL_OBJECT_INFO_ERR); + } +}; + +/*! \brief Class interface for GL Render Buffer Memory Objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferRenderGL : public Buffer +{ +public: + /*! \brief Constructs a BufferRenderGL in a specified context, from a given + * GL Renderbuffer. + * + * Wraps clCreateFromGLRenderbuffer(). + */ + BufferRenderGL( + const Context& context, + cl_mem_flags flags, + GLuint bufobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLRenderbuffer( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_RENDER_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferRenderGL() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL(const BufferGL& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferRenderGL(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferGL - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL& operator = (const BufferRenderGL& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetGLObjectInfo(). + cl_int getObjectInfo( + cl_gl_object_type *type, + GLuint * gl_object_name) + { + return detail::errHandler( + ::clGetGLObjectInfo(object_,type,gl_object_name), + __GET_GL_OBJECT_INFO_ERR); + } +}; + +/*! \brief C++ base class for Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image : public Memory +{ +protected: + //! \brief Default constructor - initializes to NULL. + Image() : Memory() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image(const Image& image) : Memory(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image(const cl_mem& image) : Memory(image) { } + + /*! \brief Assignment from Image - performs shallow copy. + * + * See Memory for further details. + */ + Image& operator = (const Image& rhs) + { + if (this != &rhs) { + Memory::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image& operator = (const cl_mem& rhs) + { + Memory::operator=(rhs); + return *this; + } + +public: + //! \brief Wrapper for clGetImageInfo(). + template + cl_int getImageInfo(cl_image_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetImageInfo, object_, name, param), + __GET_IMAGE_INFO_ERR); + } + + //! \brief Wrapper for clGetImageInfo() that returns by value. + template typename + detail::param_traits::param_type + getImageInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_image_info, name>::param_type param; + cl_int result = getImageInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +}; + +#if defined(CL_VERSION_1_2) +/*! \brief Class interface for 1D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image1D : public Image +{ +public: + /*! \brief Constructs a 1D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image1D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE1D; + desc.image_width = width; + desc.image_row_pitch = 0; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + Image1D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image1D(const Image1D& image1D) : Image(image1D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image1D(const cl_mem& image1D) : Image(image1D) { } + + /*! \brief Assignment from Image1D - performs shallow copy. + * + * See Memory for further details. + */ + Image1D& operator = (const Image1D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image1D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +/*! \class Image1DBuffer + * \brief Image interface for 1D buffer images. + */ +class Image1DBuffer : public Image +{ +public: + Image1DBuffer( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + const Buffer &buffer, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE1D_BUFFER; + desc.image_width = width; + desc.image_row_pitch = 0; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = buffer(); + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + NULL, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image1DBuffer() { } + + Image1DBuffer(const Image1DBuffer& image1D) : Image(image1D) { } + + __CL_EXPLICIT_CONSTRUCTORS Image1DBuffer(const cl_mem& image1D) : Image(image1D) { } + + Image1DBuffer& operator = (const Image1DBuffer& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image1DBuffer& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +/*! \class Image1DArray + * \brief Image interface for arrays of 1D images. + */ +class Image1DArray : public Image +{ +public: + Image1DArray( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t arraySize, + ::size_t width, + ::size_t rowPitch, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE1D_ARRAY; + desc.image_array_size = arraySize; + desc.image_width = width; + desc.image_row_pitch = rowPitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image1DArray() { } + + Image1DArray(const Image1DArray& imageArray) : Image(imageArray) { } + + __CL_EXPLICIT_CONSTRUCTORS Image1DArray(const cl_mem& imageArray) : Image(imageArray) { } + + Image1DArray& operator = (const Image1DArray& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image1DArray& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + + +/*! \brief Class interface for 2D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image2D : public Image +{ +public: + /*! \brief Constructs a 1D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image2D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + ::size_t height, + ::size_t row_pitch = 0, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + bool useCreateImage; + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + // Run-time decision based on the actual platform + { + cl_uint version = detail::getContextPlatformVersion(context()); + useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above + } +#elif defined(CL_VERSION_1_2) + useCreateImage = true; +#else + useCreateImage = false; +#endif + +#if defined(CL_VERSION_1_2) + if (useCreateImage) + { + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE2D; + desc.image_width = width; + desc.image_height = height; + desc.image_row_pitch = row_pitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) +#if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + if (!useCreateImage) + { + object_ = ::clCreateImage2D( + context(), flags,&format, width, height, row_pitch, host_ptr, &error); + + detail::errHandler(error, __CREATE_IMAGE2D_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + } + + //! \brief Default constructor - initializes to NULL. + Image2D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image2D(const Image2D& image2D) : Image(image2D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image2D(const cl_mem& image2D) : Image(image2D) { } + + /*! \brief Assignment from Image2D - performs shallow copy. + * + * See Memory for further details. + */ + Image2D& operator = (const Image2D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image2D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + + +#if !defined(CL_VERSION_1_2) +/*! \brief Class interface for GL 2D Image Memory objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + * \note Deprecated for OpenCL 1.2. Please use ImageGL instead. + */ +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED Image2DGL CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED : public Image2D +{ +public: + /*! \brief Constructs an Image2DGL in a specified context, from a given + * GL Texture. + * + * Wraps clCreateFromGLTexture2D(). + */ + Image2DGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture2D( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_2D_ERR); + if (err != NULL) { + *err = error; + } + + } + + //! \brief Default constructor - initializes to NULL. + Image2DGL() : Image2D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL(const Image2DGL& image) : Image2D(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image2DGL(const cl_mem& image) : Image2D(image) { } + + /*! \brief Assignment from Image2DGL - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL& operator = (const Image2DGL& rhs) + { + if (this != &rhs) { + Image2D::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL& operator = (const cl_mem& rhs) + { + Image2D::operator=(rhs); + return *this; + } +}; +#endif // #if !defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_1_2) +/*! \class Image2DArray + * \brief Image interface for arrays of 2D images. + */ +class Image2DArray : public Image +{ +public: + Image2DArray( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t arraySize, + ::size_t width, + ::size_t height, + ::size_t rowPitch, + ::size_t slicePitch, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE2D_ARRAY; + desc.image_array_size = arraySize; + desc.image_width = width; + desc.image_height = height; + desc.image_row_pitch = rowPitch; + desc.image_slice_pitch = slicePitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image2DArray() { } + + Image2DArray(const Image2DArray& imageArray) : Image(imageArray) { } + + __CL_EXPLICIT_CONSTRUCTORS Image2DArray(const cl_mem& imageArray) : Image(imageArray) { } + + Image2DArray& operator = (const Image2DArray& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image2DArray& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + +/*! \brief Class interface for 3D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image3D : public Image +{ +public: + /*! \brief Constructs a 3D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image3D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + ::size_t height, + ::size_t depth, + ::size_t row_pitch = 0, + ::size_t slice_pitch = 0, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + bool useCreateImage; + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + // Run-time decision based on the actual platform + { + cl_uint version = detail::getContextPlatformVersion(context()); + useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above + } +#elif defined(CL_VERSION_1_2) + useCreateImage = true; +#else + useCreateImage = false; +#endif + +#if defined(CL_VERSION_1_2) + if (useCreateImage) + { + cl_image_desc desc; + desc.image_type = CL_MEM_OBJECT_IMAGE3D; + desc.image_width = width; + desc.image_height = height; + desc.image_depth = depth; + desc.image_row_pitch = row_pitch; + desc.image_slice_pitch = slice_pitch; + desc.num_mip_levels = 0; + desc.num_samples = 0; + desc.buffer = 0; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) +#if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + if (!useCreateImage) + { + object_ = ::clCreateImage3D( + context(), flags, &format, width, height, depth, row_pitch, + slice_pitch, host_ptr, &error); + + detail::errHandler(error, __CREATE_IMAGE3D_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + } + + //! \brief Default constructor - initializes to NULL. + Image3D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image3D(const Image3D& image3D) : Image(image3D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image3D(const cl_mem& image3D) : Image(image3D) { } + + /*! \brief Assignment from Image3D - performs shallow copy. + * + * See Memory for further details. + */ + Image3D& operator = (const Image3D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image3D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +#if !defined(CL_VERSION_1_2) +/*! \brief Class interface for GL 3D Image Memory objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image3DGL : public Image3D +{ +public: + /*! \brief Constructs an Image3DGL in a specified context, from a given + * GL Texture. + * + * Wraps clCreateFromGLTexture3D(). + */ + Image3DGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture3D( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_3D_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + Image3DGL() : Image3D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL(const Image3DGL& image) : Image3D(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image3DGL(const cl_mem& image) : Image3D(image) { } + + /*! \brief Assignment from Image3DGL - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL& operator = (const Image3DGL& rhs) + { + if (this != &rhs) { + Image3D::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL& operator = (const cl_mem& rhs) + { + Image3D::operator=(rhs); + return *this; + } +}; +#endif // #if !defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_1_2) +/*! \class ImageGL + * \brief general image interface for GL interop. + * We abstract the 2D and 3D GL images into a single instance here + * that wraps all GL sourced images on the grounds that setup information + * was performed by OpenCL anyway. + */ +class ImageGL : public Image +{ +public: + ImageGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_ERR); + if (err != NULL) { + *err = error; + } + } + + ImageGL() : Image() { } + + ImageGL(const ImageGL& image) : Image(image) { } + + __CL_EXPLICIT_CONSTRUCTORS ImageGL(const cl_mem& image) : Image(image) { } + + ImageGL& operator = (const ImageGL& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + ImageGL& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + +/*! \brief Class interface for cl_sampler. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_sampler as the original. For details, see + * clRetainSampler() and clReleaseSampler(). + * + * \see cl_sampler + */ +class Sampler : public detail::Wrapper +{ +public: + /*! \brief Destructor. + * + * This calls clReleaseSampler() on the value held by this instance. + */ + ~Sampler() { } + + //! \brief Default constructor - initializes to NULL. + Sampler() { } + + /*! \brief Constructs a Sampler in a specified context. + * + * Wraps clCreateSampler(). + */ + Sampler( + const Context& context, + cl_bool normalized_coords, + cl_addressing_mode addressing_mode, + cl_filter_mode filter_mode, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateSampler( + context(), + normalized_coords, + addressing_mode, + filter_mode, + &error); + + detail::errHandler(error, __CREATE_SAMPLER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainSampler() on the parameter's cl_sampler. + */ + Sampler(const Sampler& sampler) : detail::Wrapper(sampler) { } + + /*! \brief Constructor from cl_sampler - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_sampler + * into the new Sampler object. + */ + Sampler(const cl_sampler& sampler) : detail::Wrapper(sampler) { } + + /*! \brief Assignment operator from Sampler. + * + * This calls clRetainSampler() on the parameter and clReleaseSampler() + * on the previous value held by this instance. + */ + Sampler& operator = (const Sampler& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_sampler - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseSampler() on the value previously held by this instance. + */ + Sampler& operator = (const cl_sampler& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetSamplerInfo(). + template + cl_int getInfo(cl_sampler_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetSamplerInfo, object_, name, param), + __GET_SAMPLER_INFO_ERR); + } + + //! \brief Wrapper for clGetSamplerInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_sampler_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +}; + +class Program; +class CommandQueue; +class Kernel; + +//! \brief Class interface for specifying NDRange values. +class NDRange +{ +private: + size_t<3> sizes_; + cl_uint dimensions_; + +public: + //! \brief Default constructor - resulting range has zero dimensions. + NDRange() + : dimensions_(0) + { } + + //! \brief Constructs one-dimensional range. + NDRange(::size_t size0) + : dimensions_(1) + { + sizes_[0] = size0; + } + + //! \brief Constructs two-dimensional range. + NDRange(::size_t size0, ::size_t size1) + : dimensions_(2) + { + sizes_[0] = size0; + sizes_[1] = size1; + } + + //! \brief Constructs three-dimensional range. + NDRange(::size_t size0, ::size_t size1, ::size_t size2) + : dimensions_(3) + { + sizes_[0] = size0; + sizes_[1] = size1; + sizes_[2] = size2; + } + + /*! \brief Conversion operator to const ::size_t *. + * + * \returns a pointer to the size of the first dimension. + */ + operator const ::size_t*() const { + return (const ::size_t*) sizes_; + } + + //! \brief Queries the number of dimensions in the range. + ::size_t dimensions() const { return dimensions_; } +}; + +//! \brief A zero-dimensional range. +static const NDRange NullRange; + +//! \brief Local address wrapper for use with Kernel::setArg +struct LocalSpaceArg +{ + ::size_t size_; +}; + +namespace detail { + +template +struct KernelArgumentHandler +{ + static ::size_t size(const T&) { return sizeof(T); } + static T* ptr(T& value) { return &value; } +}; + +template <> +struct KernelArgumentHandler +{ + static ::size_t size(const LocalSpaceArg& value) { return value.size_; } + static void* ptr(LocalSpaceArg&) { return NULL; } +}; + +} +//! \endcond + +/*! __local + * \brief Helper function for generating LocalSpaceArg objects. + * Deprecated. Replaced with Local. + */ +inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED LocalSpaceArg +__local(::size_t size) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; +inline LocalSpaceArg +__local(::size_t size) +{ + LocalSpaceArg ret = { size }; + return ret; +} + +/*! Local + * \brief Helper function for generating LocalSpaceArg objects. + */ +inline LocalSpaceArg +Local(::size_t size) +{ + LocalSpaceArg ret = { size }; + return ret; +} + +//class KernelFunctor; + +/*! \brief Class interface for cl_kernel. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_kernel as the original. For details, see + * clRetainKernel() and clReleaseKernel(). + * + * \see cl_kernel + */ +class Kernel : public detail::Wrapper +{ +public: + inline Kernel(const Program& program, const char* name, cl_int* err = NULL); + + /*! \brief Destructor. + * + * This calls clReleaseKernel() on the value held by this instance. + */ + ~Kernel() { } + + //! \brief Default constructor - initializes to NULL. + Kernel() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainKernel() on the parameter's cl_kernel. + */ + Kernel(const Kernel& kernel) : detail::Wrapper(kernel) { } + + /*! \brief Constructor from cl_kernel - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_kernel + * into the new Kernel object. + */ + __CL_EXPLICIT_CONSTRUCTORS Kernel(const cl_kernel& kernel) : detail::Wrapper(kernel) { } + + /*! \brief Assignment operator from Kernel. + * + * This calls clRetainKernel() on the parameter and clReleaseKernel() + * on the previous value held by this instance. + */ + Kernel& operator = (const Kernel& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_kernel - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseKernel() on the value previously held by this instance. + */ + Kernel& operator = (const cl_kernel& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + template + cl_int getInfo(cl_kernel_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetKernelInfo, object_, name, param), + __GET_KERNEL_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + +#if defined(CL_VERSION_1_2) + template + cl_int getArgInfo(cl_uint argIndex, cl_kernel_arg_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetKernelArgInfo, object_, argIndex, name, param), + __GET_KERNEL_ARG_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getArgInfo(cl_uint argIndex, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_arg_info, name>::param_type param; + cl_int result = getArgInfo(argIndex, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +#endif // #if defined(CL_VERSION_1_2) + + template + cl_int getWorkGroupInfo( + const Device& device, cl_kernel_work_group_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetKernelWorkGroupInfo, object_, device(), name, param), + __GET_KERNEL_WORK_GROUP_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getWorkGroupInfo(const Device& device, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_work_group_info, name>::param_type param; + cl_int result = getWorkGroupInfo(device, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + template + cl_int setArg(cl_uint index, T value) + { + return detail::errHandler( + ::clSetKernelArg( + object_, + index, + detail::KernelArgumentHandler::size(value), + detail::KernelArgumentHandler::ptr(value)), + __SET_KERNEL_ARGS_ERR); + } + + cl_int setArg(cl_uint index, ::size_t size, void* argPtr) + { + return detail::errHandler( + ::clSetKernelArg(object_, index, size, argPtr), + __SET_KERNEL_ARGS_ERR); + } +}; + +/*! \class Program + * \brief Program interface that implements cl_program. + */ +class Program : public detail::Wrapper +{ +public: + typedef VECTOR_CLASS > Binaries; + typedef VECTOR_CLASS > Sources; + + Program( + const STRING_CLASS& source, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + Context context = Context::getDefault(err); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const STRING_CLASS& source, + bool build, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + Context context = Context::getDefault(err); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS && build) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const Context& context, + const STRING_CLASS& source, + bool build = false, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS && build) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const Context& context, + const Sources& sources, + cl_int* err = NULL) + { + cl_int error; + + const ::size_t n = (::size_t)sources.size(); + ::size_t* lengths = (::size_t*) alloca(n * sizeof(::size_t)); + const char** strings = (const char**) alloca(n * sizeof(const char*)); + + for (::size_t i = 0; i < n; ++i) { + strings[i] = sources[(int)i].first; + lengths[i] = sources[(int)i].second; + } + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)n, strings, lengths, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + if (err != NULL) { + *err = error; + } + } + + /** + * Construct a program object from a list of devices and a per-device list of binaries. + * \param context A valid OpenCL context in which to construct the program. + * \param devices A vector of OpenCL device objects for which the program will be created. + * \param binaries A vector of pairs of a pointer to a binary object and its length. + * \param binaryStatus An optional vector that on completion will be resized to + * match the size of binaries and filled with values to specify if each binary + * was successfully loaded. + * Set to CL_SUCCESS if the binary was successfully loaded. + * Set to CL_INVALID_VALUE if the length is 0 or the binary pointer is NULL. + * Set to CL_INVALID_BINARY if the binary provided is not valid for the matching device. + * \param err if non-NULL will be set to CL_SUCCESS on successful operation or one of the following errors: + * CL_INVALID_CONTEXT if context is not a valid context. + * CL_INVALID_VALUE if the length of devices is zero; or if the length of binaries does not match the length of devices; + * or if any entry in binaries is NULL or has length 0. + * CL_INVALID_DEVICE if OpenCL devices listed in devices are not in the list of devices associated with context. + * CL_INVALID_BINARY if an invalid program binary was encountered for any device. binaryStatus will return specific status for each device. + * CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the OpenCL implementation on the host. + */ + Program( + const Context& context, + const VECTOR_CLASS& devices, + const Binaries& binaries, + VECTOR_CLASS* binaryStatus = NULL, + cl_int* err = NULL) + { + cl_int error; + + const ::size_t numDevices = devices.size(); + + // Catch size mismatch early and return + if(binaries.size() != numDevices) { + error = CL_INVALID_VALUE; + detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); + if (err != NULL) { + *err = error; + } + return; + } + + ::size_t* lengths = (::size_t*) alloca(numDevices * sizeof(::size_t)); + const unsigned char** images = (const unsigned char**) alloca(numDevices * sizeof(const unsigned char**)); + + for (::size_t i = 0; i < numDevices; ++i) { + images[i] = (const unsigned char*)binaries[i].first; + lengths[i] = binaries[(int)i].second; + } + + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + if(binaryStatus) { + binaryStatus->resize(numDevices); + } + + object_ = ::clCreateProgramWithBinary( + context(), (cl_uint) devices.size(), + deviceIDs, + lengths, images, binaryStatus != NULL + ? &binaryStatus->front() + : NULL, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); + if (err != NULL) { + *err = error; + } + } + + +#if defined(CL_VERSION_1_2) + /** + * Create program using builtin kernels. + * \param kernelNames Semi-colon separated list of builtin kernel names + */ + Program( + const Context& context, + const VECTOR_CLASS& devices, + const STRING_CLASS& kernelNames, + cl_int* err = NULL) + { + cl_int error; + + + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + object_ = ::clCreateProgramWithBuiltInKernels( + context(), + (cl_uint) devices.size(), + deviceIDs, + kernelNames.c_str(), + &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) + + Program() { } + + Program(const Program& program) : detail::Wrapper(program) { } + + __CL_EXPLICIT_CONSTRUCTORS Program(const cl_program& program) : detail::Wrapper(program) { } + + Program& operator = (const Program& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + Program& operator = (const cl_program& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + cl_int build( + const VECTOR_CLASS& devices, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + return detail::errHandler( + ::clBuildProgram( + object_, + (cl_uint) + devices.size(), + deviceIDs, + options, + notifyFptr, + data), + __BUILD_PROGRAM_ERR); + } + + cl_int build( + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + return detail::errHandler( + ::clBuildProgram( + object_, + 0, + NULL, + options, + notifyFptr, + data), + __BUILD_PROGRAM_ERR); + } + +#if defined(CL_VERSION_1_2) + cl_int compile( + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + return detail::errHandler( + ::clCompileProgram( + object_, + 0, + NULL, + options, + 0, + NULL, + NULL, + notifyFptr, + data), + __COMPILE_PROGRAM_ERR); + } +#endif + + template + cl_int getInfo(cl_program_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetProgramInfo, object_, name, param), + __GET_PROGRAM_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_program_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + template + cl_int getBuildInfo( + const Device& device, cl_program_build_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetProgramBuildInfo, object_, device(), name, param), + __GET_PROGRAM_BUILD_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getBuildInfo(const Device& device, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_program_build_info, name>::param_type param; + cl_int result = getBuildInfo(device, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + cl_int createKernels(VECTOR_CLASS* kernels) + { + cl_uint numKernels; + cl_int err = ::clCreateKernelsInProgram(object_, 0, NULL, &numKernels); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); + } + + Kernel* value = (Kernel*) alloca(numKernels * sizeof(Kernel)); + err = ::clCreateKernelsInProgram( + object_, numKernels, (cl_kernel*) value, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); + } + + kernels->assign(&value[0], &value[numKernels]); + return CL_SUCCESS; + } +}; + +#if defined(CL_VERSION_1_2) +inline Program linkProgram( + Program input1, + Program input2, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL, + cl_int* err = NULL) +{ + cl_int err_local = CL_SUCCESS; + + cl_program programs[2] = { input1(), input2() }; + + Context ctx = input1.getInfo(); + + cl_program prog = ::clLinkProgram( + ctx(), + 0, + NULL, + options, + 2, + programs, + notifyFptr, + data, + &err_local); + + detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); + if (err != NULL) { + *err = err_local; + } + + return Program(prog); +} + +inline Program linkProgram( + VECTOR_CLASS inputPrograms, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL, + cl_int* err = NULL) +{ + cl_int err_local = CL_SUCCESS; + + cl_program * programs = (cl_program*) alloca(inputPrograms.size() * sizeof(cl_program)); + + if (programs != NULL) { + for (unsigned int i = 0; i < inputPrograms.size(); i++) { + programs[i] = inputPrograms[i](); + } + } + + cl_program prog = ::clLinkProgram( + Context::getDefault()(), + 0, + NULL, + options, + (cl_uint)inputPrograms.size(), + programs, + notifyFptr, + data, + &err_local); + + detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); + if (err != NULL) { + *err = err_local; + } + + return Program(prog); +} +#endif + +template<> +inline VECTOR_CLASS cl::Program::getInfo(cl_int* err) const +{ + VECTOR_CLASS< ::size_t> sizes = getInfo(); + VECTOR_CLASS binaries; + for (VECTOR_CLASS< ::size_t>::iterator s = sizes.begin(); s != sizes.end(); ++s) + { + char *ptr = NULL; + if (*s != 0) + ptr = new char[*s]; + binaries.push_back(ptr); + } + + cl_int result = getInfo(CL_PROGRAM_BINARIES, &binaries); + if (err != NULL) { + *err = result; + } + return binaries; +} + +inline Kernel::Kernel(const Program& program, const char* name, cl_int* err) +{ + cl_int error; + + object_ = ::clCreateKernel(program(), name, &error); + detail::errHandler(error, __CREATE_KERNEL_ERR); + + if (err != NULL) { + *err = error; + } + +} + +/*! \class CommandQueue + * \brief CommandQueue interface for cl_command_queue. + */ +class CommandQueue : public detail::Wrapper +{ +private: + static volatile int default_initialized_; + static CommandQueue default_; + static volatile cl_int default_error_; +public: + CommandQueue( + cl_command_queue_properties properties, + cl_int* err = NULL) + { + cl_int error; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + Device device = context.getInfo()[0]; + + object_ = ::clCreateCommandQueue( + context(), device(), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + } + + CommandQueue( + const Context& context, + const Device& device, + cl_command_queue_properties properties = 0, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateCommandQueue( + context(), device(), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + + static CommandQueue getDefault(cl_int * err = NULL) + { + int state = detail::compare_exchange( + &default_initialized_, + __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); + + if (state & __DEFAULT_INITIALIZED) { + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + if (state & __DEFAULT_BEING_INITIALIZED) { + // Assume writes will propagate eventually... + while(default_initialized_ != __DEFAULT_INITIALIZED) { + detail::fence(); + } + + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + cl_int error; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + Device device = context.getInfo()[0]; + + default_ = CommandQueue(context, device, 0, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + + detail::fence(); + + default_error_ = error; + // Assume writes will propagate eventually... + default_initialized_ = __DEFAULT_INITIALIZED; + + detail::fence(); + + if (err != NULL) { + *err = default_error_; + } + return default_; + + } + + CommandQueue() { } + + CommandQueue(const CommandQueue& commandQueue) : detail::Wrapper(commandQueue) { } + + CommandQueue(const cl_command_queue& commandQueue) : detail::Wrapper(commandQueue) { } + + CommandQueue& operator = (const CommandQueue& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + CommandQueue& operator = (const cl_command_queue& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + template + cl_int getInfo(cl_command_queue_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetCommandQueueInfo, object_, name, param), + __GET_COMMAND_QUEUE_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_command_queue_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + cl_int enqueueReadBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadBuffer( + object_, buffer(), blocking, offset, size, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + const void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteBuffer( + object_, buffer(), blocking, offset, size, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBuffer( + const Buffer& src, + const Buffer& dst, + ::size_t src_offset, + ::size_t dst_offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBuffer( + object_, src(), dst(), src_offset, dst_offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQEUE_COPY_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReadBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadBufferRect( + object_, + buffer(), + blocking, + (const ::size_t *)buffer_offset, + (const ::size_t *)host_offset, + (const ::size_t *)region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteBufferRect( + object_, + buffer(), + blocking, + (const ::size_t *)buffer_offset, + (const ::size_t *)host_offset, + (const ::size_t *)region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBufferRect( + const Buffer& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + ::size_t src_row_pitch, + ::size_t src_slice_pitch, + ::size_t dst_row_pitch, + ::size_t dst_slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBufferRect( + object_, + src(), + dst(), + (const ::size_t *)src_origin, + (const ::size_t *)dst_origin, + (const ::size_t *)region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQEUE_COPY_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueue a command to fill a buffer object with a pattern + * of a given size. The pattern is specified a as vector. + * \tparam PatternType The datatype of the pattern field. + * The pattern type must be an accepted OpenCL data type. + */ + template + cl_int enqueueFillBuffer( + const Buffer& buffer, + PatternType pattern, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillBuffer( + object_, + buffer(), + static_cast(&pattern), + sizeof(PatternType), + offset, + size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueReadImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadImage( + object_, image(), blocking, (const ::size_t *) origin, + (const ::size_t *) region, row_pitch, slice_pitch, ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteImage( + object_, image(), blocking, (const ::size_t *) origin, + (const ::size_t *) region, row_pitch, slice_pitch, ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyImage( + const Image& src, + const Image& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyImage( + object_, src(), dst(), (const ::size_t *) src_origin, + (const ::size_t *)dst_origin, (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA floating-point color value if + * the image channel data type is not an unnormalized signed or + * unsigned data type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_float4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA signed integer color value if + * the image channel data type is an unnormalized signed integer + * type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_int4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA unsigned integer color value if + * the image channel data type is an unnormalized unsigned integer + * type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_uint4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueCopyImageToBuffer( + const Image& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& region, + ::size_t dst_offset, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyImageToBuffer( + object_, src(), dst(), (const ::size_t *) src_origin, + (const ::size_t *) region, dst_offset, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBufferToImage( + const Buffer& src, + const Image& dst, + ::size_t src_offset, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBufferToImage( + object_, src(), dst(), src_offset, + (const ::size_t *) dst_origin, (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + void* enqueueMapBuffer( + const Buffer& buffer, + cl_bool blocking, + cl_map_flags flags, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) const + { + cl_int error; + void * result = ::clEnqueueMapBuffer( + object_, buffer(), blocking, flags, offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + return result; + } + + void* enqueueMapImage( + const Image& buffer, + cl_bool blocking, + cl_map_flags flags, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t * row_pitch, + ::size_t * slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) const + { + cl_int error; + void * result = ::clEnqueueMapImage( + object_, buffer(), blocking, flags, + (const ::size_t *) origin, (const ::size_t *) region, + row_pitch, slice_pitch, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + return result; + } + + cl_int enqueueUnmapMemObject( + const Memory& memory, + void* mapped_ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueUnmapMemObject( + object_, memory(), mapped_ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueues a marker command which waits for either a list of events to complete, + * or all previously enqueued commands to complete. + * + * Enqueues a marker command which waits for either a list of events to complete, + * or if the list is empty it waits for all commands previously enqueued in command_queue + * to complete before it completes. This command returns an event which can be waited on, + * i.e. this event can be waited on to insure that all events either in the event_wait_list + * or all previously enqueued commands, queued before this command to command_queue, + * have completed. + */ + cl_int enqueueMarkerWithWaitList( + const VECTOR_CLASS *events = 0, + Event *event = 0) + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueMarkerWithWaitList( + object_, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_MARKER_WAIT_LIST_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * A synchronization point that enqueues a barrier operation. + * + * Enqueues a barrier command which waits for either a list of events to complete, + * or if the list is empty it waits for all commands previously enqueued in command_queue + * to complete before it completes. This command blocks command execution, that is, any + * following commands enqueued after it do not execute until it completes. This command + * returns an event which can be waited on, i.e. this event can be waited on to insure that + * all events either in the event_wait_list or all previously enqueued commands, queued + * before this command to command_queue, have completed. + */ + cl_int enqueueBarrierWithWaitList( + const VECTOR_CLASS *events = 0, + Event *event = 0) + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueBarrierWithWaitList( + object_, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_BARRIER_WAIT_LIST_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueues a command to indicate with which device a set of memory objects + * should be associated. + */ + cl_int enqueueMigrateMemObjects( + const VECTOR_CLASS &memObjects, + cl_mem_migration_flags flags, + const VECTOR_CLASS* events = NULL, + Event* event = NULL + ) + { + cl_event tmp; + + cl_mem* localMemObjects = static_cast(alloca(memObjects.size() * sizeof(cl_mem))); + for( int i = 0; i < (int)memObjects.size(); ++i ) { + localMemObjects[i] = memObjects[i](); + } + + + cl_int err = detail::errHandler( + ::clEnqueueMigrateMemObjects( + object_, + (cl_uint)memObjects.size(), + static_cast(localMemObjects), + flags, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueNDRangeKernel( + const Kernel& kernel, + const NDRange& offset, + const NDRange& global, + const NDRange& local = NullRange, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueNDRangeKernel( + object_, kernel(), (cl_uint) global.dimensions(), + offset.dimensions() != 0 ? (const ::size_t*) offset : NULL, + (const ::size_t*) global, + local.dimensions() != 0 ? (const ::size_t*) local : NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_NDRANGE_KERNEL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueTask( + const Kernel& kernel, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueTask( + object_, kernel(), + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_TASK_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueNativeKernel( + void (CL_CALLBACK *userFptr)(void *), + std::pair args, + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* mem_locs = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_mem * mems = (mem_objects != NULL && mem_objects->size() > 0) + ? (cl_mem*) alloca(mem_objects->size() * sizeof(cl_mem)) + : NULL; + + if (mems != NULL) { + for (unsigned int i = 0; i < mem_objects->size(); i++) { + mems[i] = ((*mem_objects)[i])(); + } + } + + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueNativeKernel( + object_, userFptr, args.first, args.second, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + mems, + (mem_locs != NULL) ? (const void **) &mem_locs->front() : NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_NATIVE_KERNEL); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueMarker(Event* event = NULL) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueMarker(object_, (cl_event*) event), + __ENQUEUE_MARKER_ERR); + } + + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueWaitForEvents(const VECTOR_CLASS& events) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueWaitForEvents( + object_, + (cl_uint) events.size(), + (const cl_event*) &events.front()), + __ENQUEUE_WAIT_FOR_EVENTS_ERR); + } +#endif // #if defined(CL_VERSION_1_1) + + cl_int enqueueAcquireGLObjects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueAcquireGLObjects( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_ACQUIRE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReleaseGLObjects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReleaseGLObjects( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_RELEASE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined (USE_DX_INTEROP) +typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueAcquireD3D10ObjectsKHR)( + cl_command_queue command_queue, cl_uint num_objects, + const cl_mem* mem_objects, cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, cl_event* event); +typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)( + cl_command_queue command_queue, cl_uint num_objects, + const cl_mem* mem_objects, cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, cl_event* event); + + cl_int enqueueAcquireD3D10Objects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL; +#if defined(CL_VERSION_1_2) + cl_context context = getInfo(); + cl::Device device(getInfo()); + cl_platform_id platform = device.getInfo(); + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueAcquireD3D10ObjectsKHR); +#endif +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clEnqueueAcquireD3D10ObjectsKHR); +#endif + + cl_event tmp; + cl_int err = detail::errHandler( + pfn_clEnqueueAcquireD3D10ObjectsKHR( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_ACQUIRE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReleaseD3D10Objects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + static PFN_clEnqueueReleaseD3D10ObjectsKHR pfn_clEnqueueReleaseD3D10ObjectsKHR = NULL; +#if defined(CL_VERSION_1_2) + cl_context context = getInfo(); + cl::Device device(getInfo()); + cl_platform_id platform = device.getInfo(); + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueReleaseD3D10ObjectsKHR); +#endif // #if defined(CL_VERSION_1_2) +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clEnqueueReleaseD3D10ObjectsKHR); +#endif // #if defined(CL_VERSION_1_1) + + cl_event tmp; + cl_int err = detail::errHandler( + pfn_clEnqueueReleaseD3D10ObjectsKHR( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_RELEASE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueBarrier() const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueBarrier(object_), + __ENQUEUE_BARRIER_ERR); + } +#endif // #if defined(CL_VERSION_1_1) + + cl_int flush() const + { + return detail::errHandler(::clFlush(object_), __FLUSH_ERR); + } + + cl_int finish() const + { + return detail::errHandler(::clFinish(object_), __FINISH_ERR); + } +}; + +#ifdef _WIN32 +__declspec(selectany) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__declspec(selectany) CommandQueue CommandQueue::default_; +__declspec(selectany) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; +#else +__attribute__((weak)) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__attribute__((weak)) CommandQueue CommandQueue::default_; +__attribute__((weak)) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; +#endif + +inline cl_int enqueueReadBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadBuffer(buffer, blocking, offset, size, ptr, events, event); +} + +inline cl_int enqueueWriteBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + const void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteBuffer(buffer, blocking, offset, size, ptr, events, event); +} + +inline void* enqueueMapBuffer( + const Buffer& buffer, + cl_bool blocking, + cl_map_flags flags, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + void * result = ::clEnqueueMapBuffer( + queue(), buffer(), blocking, flags, offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + return result; +} + +inline cl_int enqueueUnmapMemObject( + const Memory& memory, + void* mapped_ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (error != CL_SUCCESS) { + return error; + } + + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueUnmapMemObject( + queue(), memory(), mapped_ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; +} + +inline cl_int enqueueCopyBuffer( + const Buffer& src, + const Buffer& dst, + ::size_t src_offset, + ::size_t dst_offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBuffer(src, dst, src_offset, dst_offset, size, events, event); +} + +/** + * Blocking copy operation between iterators and a buffer. + */ +template< typename IteratorType > +inline cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + ::size_t length = endIterator-startIterator; + ::size_t byteLength = length*sizeof(DataType); + + DataType *pointer = + static_cast(enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_WRITE, 0, byteLength, 0, 0, &error)); + // if exceptions enabled, enqueueMapBuffer will throw + if( error != CL_SUCCESS ) { + return error; + } +#if defined(_MSC_VER) + std::copy( + startIterator, + endIterator, + stdext::checked_array_iterator( + pointer, length)); +#else + std::copy(startIterator, endIterator, pointer); +#endif + Event endEvent; + error = enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); + // if exceptions enabled, enqueueUnmapMemObject will throw + if( error != CL_SUCCESS ) { + return error; + } + endEvent.wait(); + return CL_SUCCESS; +} + +/** + * Blocking copy operation between iterators and a buffer. + */ +template< typename IteratorType > +inline cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + ::size_t length = endIterator-startIterator; + ::size_t byteLength = length*sizeof(DataType); + + DataType *pointer = + static_cast(enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_READ, 0, byteLength, 0, 0, &error)); + // if exceptions enabled, enqueueMapBuffer will throw + if( error != CL_SUCCESS ) { + return error; + } + std::copy(pointer, pointer + length, startIterator); + Event endEvent; + error = enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); + // if exceptions enabled, enqueueUnmapMemObject will throw + if( error != CL_SUCCESS ) { + return error; + } + endEvent.wait(); + return CL_SUCCESS; +} + +#if defined(CL_VERSION_1_1) +inline cl_int enqueueReadBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadBufferRect( + buffer, + blocking, + buffer_offset, + host_offset, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueWriteBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteBufferRect( + buffer, + blocking, + buffer_offset, + host_offset, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueCopyBufferRect( + const Buffer& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + ::size_t src_row_pitch, + ::size_t src_slice_pitch, + ::size_t dst_row_pitch, + ::size_t dst_slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBufferRect( + src, + dst, + src_origin, + dst_origin, + region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + events, + event); +} +#endif + +inline cl_int enqueueReadImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadImage( + image, + blocking, + origin, + region, + row_pitch, + slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueWriteImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteImage( + image, + blocking, + origin, + region, + row_pitch, + slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueCopyImage( + const Image& src, + const Image& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyImage( + src, + dst, + src_origin, + dst_origin, + region, + events, + event); +} + +inline cl_int enqueueCopyImageToBuffer( + const Image& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& region, + ::size_t dst_offset, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyImageToBuffer( + src, + dst, + src_origin, + region, + dst_offset, + events, + event); +} + +inline cl_int enqueueCopyBufferToImage( + const Buffer& src, + const Image& dst, + ::size_t src_offset, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBufferToImage( + src, + dst, + src_offset, + dst_origin, + region, + events, + event); +} + + +inline cl_int flush(void) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.flush(); +} + +inline cl_int finish(void) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + + return queue.finish(); +} + +// Kernel Functor support +// New interface as of September 2011 +// Requires the C++11 std::tr1::function (note do not support TR1) +// Visual Studio 2010 and GCC 4.2 + +struct EnqueueArgs +{ + CommandQueue queue_; + const NDRange offset_; + const NDRange global_; + const NDRange local_; + VECTOR_CLASS events_; + + EnqueueArgs(NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange) + { + + } + + EnqueueArgs(NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local) + { + + } + + EnqueueArgs(NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local) + { + + } + + EnqueueArgs(Event e, NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange) + { + events_.push_back(e); + } + + EnqueueArgs(Event e, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(Event e, NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange), + events_(events) + { + + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local) + { + + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local), + events_(events) + { + + } +}; + +namespace detail { + +class NullType {}; + +template +struct SetArg +{ + static void set (Kernel kernel, T0 arg) + { + kernel.setArg(index, arg); + } +}; + +template +struct SetArg +{ + static void set (Kernel, NullType) + { + } +}; + +template < + typename T0, typename T1, typename T2, typename T3, + typename T4, typename T5, typename T6, typename T7, + typename T8, typename T9, typename T10, typename T11, + typename T12, typename T13, typename T14, typename T15, + typename T16, typename T17, typename T18, typename T19, + typename T20, typename T21, typename T22, typename T23, + typename T24, typename T25, typename T26, typename T27, + typename T28, typename T29, typename T30, typename T31 +> +class KernelFunctorGlobal +{ +private: + Kernel kernel_; + +public: + KernelFunctorGlobal( + Kernel kernel) : + kernel_(kernel) + {} + + KernelFunctorGlobal( + const Program& program, + const STRING_CLASS name, + cl_int * err = NULL) : + kernel_(program, name.c_str(), err) + {} + + Event operator() ( + const EnqueueArgs& args, + T0 t0, + T1 t1 = NullType(), + T2 t2 = NullType(), + T3 t3 = NullType(), + T4 t4 = NullType(), + T5 t5 = NullType(), + T6 t6 = NullType(), + T7 t7 = NullType(), + T8 t8 = NullType(), + T9 t9 = NullType(), + T10 t10 = NullType(), + T11 t11 = NullType(), + T12 t12 = NullType(), + T13 t13 = NullType(), + T14 t14 = NullType(), + T15 t15 = NullType(), + T16 t16 = NullType(), + T17 t17 = NullType(), + T18 t18 = NullType(), + T19 t19 = NullType(), + T20 t20 = NullType(), + T21 t21 = NullType(), + T22 t22 = NullType(), + T23 t23 = NullType(), + T24 t24 = NullType(), + T25 t25 = NullType(), + T26 t26 = NullType(), + T27 t27 = NullType(), + T28 t28 = NullType(), + T29 t29 = NullType(), + T30 t30 = NullType(), + T31 t31 = NullType() + ) + { + Event event; + SetArg<0, T0>::set(kernel_, t0); + SetArg<1, T1>::set(kernel_, t1); + SetArg<2, T2>::set(kernel_, t2); + SetArg<3, T3>::set(kernel_, t3); + SetArg<4, T4>::set(kernel_, t4); + SetArg<5, T5>::set(kernel_, t5); + SetArg<6, T6>::set(kernel_, t6); + SetArg<7, T7>::set(kernel_, t7); + SetArg<8, T8>::set(kernel_, t8); + SetArg<9, T9>::set(kernel_, t9); + SetArg<10, T10>::set(kernel_, t10); + SetArg<11, T11>::set(kernel_, t11); + SetArg<12, T12>::set(kernel_, t12); + SetArg<13, T13>::set(kernel_, t13); + SetArg<14, T14>::set(kernel_, t14); + SetArg<15, T15>::set(kernel_, t15); + SetArg<16, T16>::set(kernel_, t16); + SetArg<17, T17>::set(kernel_, t17); + SetArg<18, T18>::set(kernel_, t18); + SetArg<19, T19>::set(kernel_, t19); + SetArg<20, T20>::set(kernel_, t20); + SetArg<21, T21>::set(kernel_, t21); + SetArg<22, T22>::set(kernel_, t22); + SetArg<23, T23>::set(kernel_, t23); + SetArg<24, T24>::set(kernel_, t24); + SetArg<25, T25>::set(kernel_, t25); + SetArg<26, T26>::set(kernel_, t26); + SetArg<27, T27>::set(kernel_, t27); + SetArg<28, T28>::set(kernel_, t28); + SetArg<29, T29>::set(kernel_, t29); + SetArg<30, T30>::set(kernel_, t30); + SetArg<31, T31>::set(kernel_, t31); + + args.queue_.enqueueNDRangeKernel( + kernel_, + args.offset_, + args.global_, + args.local_, + &args.events_, + &event); + + return event; + } + +}; + +//------------------------------------------------------------------------------------------------------ + + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29, + typename T30, + typename T31> +struct functionImplementation_ +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + T31> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 32)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + T31); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29, + T30 arg30, + T31 arg31) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29, + arg30, + arg31); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29, + typename T30> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 31)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29, + T30 arg30) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29, + arg30); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 30)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 29)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 28)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 27)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 26)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 25)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 24)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 23)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 22)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 21)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 20)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 19)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 18)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 17)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 16)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 15)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 14)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 13)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 12)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 11)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 10)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 9)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 8)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 7)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 6)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 5)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 4)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3); + } + + +}; + +template< + typename T0, + typename T1, + typename T2> +struct functionImplementation_ +< T0, + T1, + T2, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 3)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2); + } + + +}; + +template< + typename T0, + typename T1> +struct functionImplementation_ +< T0, + T1, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 2)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1) + { + return functor_( + enqueueArgs, + arg0, + arg1); + } + + +}; + +template< + typename T0> +struct functionImplementation_ +< T0, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 1)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0) + { + return functor_( + enqueueArgs, + arg0); + } + + +}; + + + + + +} // namespace detail + +//---------------------------------------------------------------------------------------------- + +template < + typename T0, typename T1 = detail::NullType, typename T2 = detail::NullType, + typename T3 = detail::NullType, typename T4 = detail::NullType, + typename T5 = detail::NullType, typename T6 = detail::NullType, + typename T7 = detail::NullType, typename T8 = detail::NullType, + typename T9 = detail::NullType, typename T10 = detail::NullType, + typename T11 = detail::NullType, typename T12 = detail::NullType, + typename T13 = detail::NullType, typename T14 = detail::NullType, + typename T15 = detail::NullType, typename T16 = detail::NullType, + typename T17 = detail::NullType, typename T18 = detail::NullType, + typename T19 = detail::NullType, typename T20 = detail::NullType, + typename T21 = detail::NullType, typename T22 = detail::NullType, + typename T23 = detail::NullType, typename T24 = detail::NullType, + typename T25 = detail::NullType, typename T26 = detail::NullType, + typename T27 = detail::NullType, typename T28 = detail::NullType, + typename T29 = detail::NullType, typename T30 = detail::NullType, + typename T31 = detail::NullType +> +struct make_kernel : + public detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + > +{ +public: + typedef detail::KernelFunctorGlobal< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + > FunctorType; + + make_kernel( + const Program& program, + const STRING_CLASS name, + cl_int * err = NULL) : + detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + >( + FunctorType(program, name, err)) + {} + + make_kernel( + const Kernel kernel) : + detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + >( + FunctorType(kernel)) + {} +}; + + +//---------------------------------------------------------------------------------------------------------------------- + +#undef __ERR_STR +#if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) +#undef __GET_DEVICE_INFO_ERR +#undef __GET_PLATFORM_INFO_ERR +#undef __GET_DEVICE_IDS_ERR +#undef __GET_CONTEXT_INFO_ERR +#undef __GET_EVENT_INFO_ERR +#undef __GET_EVENT_PROFILE_INFO_ERR +#undef __GET_MEM_OBJECT_INFO_ERR +#undef __GET_IMAGE_INFO_ERR +#undef __GET_SAMPLER_INFO_ERR +#undef __GET_KERNEL_INFO_ERR +#undef __GET_KERNEL_ARG_INFO_ERR +#undef __GET_KERNEL_WORK_GROUP_INFO_ERR +#undef __GET_PROGRAM_INFO_ERR +#undef __GET_PROGRAM_BUILD_INFO_ERR +#undef __GET_COMMAND_QUEUE_INFO_ERR + +#undef __CREATE_CONTEXT_ERR +#undef __CREATE_CONTEXT_FROM_TYPE_ERR +#undef __GET_SUPPORTED_IMAGE_FORMATS_ERR + +#undef __CREATE_BUFFER_ERR +#undef __CREATE_SUBBUFFER_ERR +#undef __CREATE_IMAGE2D_ERR +#undef __CREATE_IMAGE3D_ERR +#undef __CREATE_SAMPLER_ERR +#undef __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR + +#undef __CREATE_USER_EVENT_ERR +#undef __SET_USER_EVENT_STATUS_ERR +#undef __SET_EVENT_CALLBACK_ERR +#undef __SET_PRINTF_CALLBACK_ERR + +#undef __WAIT_FOR_EVENTS_ERR + +#undef __CREATE_KERNEL_ERR +#undef __SET_KERNEL_ARGS_ERR +#undef __CREATE_PROGRAM_WITH_SOURCE_ERR +#undef __CREATE_PROGRAM_WITH_BINARY_ERR +#undef __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR +#undef __BUILD_PROGRAM_ERR +#undef __CREATE_KERNELS_IN_PROGRAM_ERR + +#undef __CREATE_COMMAND_QUEUE_ERR +#undef __SET_COMMAND_QUEUE_PROPERTY_ERR +#undef __ENQUEUE_READ_BUFFER_ERR +#undef __ENQUEUE_WRITE_BUFFER_ERR +#undef __ENQUEUE_READ_BUFFER_RECT_ERR +#undef __ENQUEUE_WRITE_BUFFER_RECT_ERR +#undef __ENQEUE_COPY_BUFFER_ERR +#undef __ENQEUE_COPY_BUFFER_RECT_ERR +#undef __ENQUEUE_READ_IMAGE_ERR +#undef __ENQUEUE_WRITE_IMAGE_ERR +#undef __ENQUEUE_COPY_IMAGE_ERR +#undef __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR +#undef __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR +#undef __ENQUEUE_MAP_BUFFER_ERR +#undef __ENQUEUE_MAP_IMAGE_ERR +#undef __ENQUEUE_UNMAP_MEM_OBJECT_ERR +#undef __ENQUEUE_NDRANGE_KERNEL_ERR +#undef __ENQUEUE_TASK_ERR +#undef __ENQUEUE_NATIVE_KERNEL + +#undef __CL_EXPLICIT_CONSTRUCTORS + +#undef __UNLOAD_COMPILER_ERR +#endif //__CL_USER_OVERRIDE_ERROR_STRINGS + +#undef __CL_FUNCTION_TYPE + +// Extensions +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_VERSION_1_1) +#undef __INIT_CL_EXT_FCN_PTR +#endif // #if defined(CL_VERSION_1_1) +#undef __CREATE_SUB_DEVICES + +#if defined(USE_CL_DEVICE_FISSION) +#undef __PARAM_NAME_DEVICE_FISSION +#endif // USE_CL_DEVICE_FISSION + +#undef __DEFAULT_NOT_INITIALIZED +#undef __DEFAULT_BEING_INITIALIZED +#undef __DEFAULT_INITIALIZED + +} // namespace cl + +#ifdef _WIN32 +#pragma pop_macro("max") +#endif // _WIN32 + +#endif // CL_HPP_ diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d10.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d10.h new file mode 100644 index 0000000000..91e4a68796 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d10.h @@ -0,0 +1,126 @@ +/********************************************************************************** + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_CL_D3D10_H +#define __OPENCL_CL_D3D10_H + +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/****************************************************************************** + * cl_khr_d3d10_sharing */ +#define cl_khr_d3d10_sharing 1 + +typedef cl_uint cl_d3d10_device_source_khr; +typedef cl_uint cl_d3d10_device_set_khr; + +/******************************************************************************/ + +/* Error Codes */ +#define CL_INVALID_D3D10_DEVICE_KHR -1002 +#define CL_INVALID_D3D10_RESOURCE_KHR -1003 +#define CL_D3D10_RESOURCE_ALREADY_ACQUIRED_KHR -1004 +#define CL_D3D10_RESOURCE_NOT_ACQUIRED_KHR -1005 + +/* cl_d3d10_device_source_nv */ +#define CL_D3D10_DEVICE_KHR 0x4010 +#define CL_D3D10_DXGI_ADAPTER_KHR 0x4011 + +/* cl_d3d10_device_set_nv */ +#define CL_PREFERRED_DEVICES_FOR_D3D10_KHR 0x4012 +#define CL_ALL_DEVICES_FOR_D3D10_KHR 0x4013 + +/* cl_context_info */ +#define CL_CONTEXT_D3D10_DEVICE_KHR 0x4014 +#define CL_CONTEXT_D3D10_PREFER_SHARED_RESOURCES_KHR 0x402C + +/* cl_mem_info */ +#define CL_MEM_D3D10_RESOURCE_KHR 0x4015 + +/* cl_image_info */ +#define CL_IMAGE_D3D10_SUBRESOURCE_KHR 0x4016 + +/* cl_command_type */ +#define CL_COMMAND_ACQUIRE_D3D10_OBJECTS_KHR 0x4017 +#define CL_COMMAND_RELEASE_D3D10_OBJECTS_KHR 0x4018 + +/******************************************************************************/ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromD3D10KHR_fn)( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10BufferKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D10Buffer * resource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10Texture2DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D10Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10Texture3DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D10Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireD3D10ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseD3D10ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_D3D10_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d11.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d11.h new file mode 100644 index 0000000000..496273b855 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_d3d11.h @@ -0,0 +1,126 @@ +/********************************************************************************** + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_CL_D3D11_H +#define __OPENCL_CL_D3D11_H + +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/****************************************************************************** + * cl_khr_d3d11_sharing */ +#define cl_khr_d3d11_sharing 1 + +typedef cl_uint cl_d3d11_device_source_khr; +typedef cl_uint cl_d3d11_device_set_khr; + +/******************************************************************************/ + +/* Error Codes */ +#define CL_INVALID_D3D11_DEVICE_KHR -1006 +#define CL_INVALID_D3D11_RESOURCE_KHR -1007 +#define CL_D3D11_RESOURCE_ALREADY_ACQUIRED_KHR -1008 +#define CL_D3D11_RESOURCE_NOT_ACQUIRED_KHR -1009 + +/* cl_d3d11_device_source */ +#define CL_D3D11_DEVICE_KHR 0x4019 +#define CL_D3D11_DXGI_ADAPTER_KHR 0x401A + +/* cl_d3d11_device_set */ +#define CL_PREFERRED_DEVICES_FOR_D3D11_KHR 0x401B +#define CL_ALL_DEVICES_FOR_D3D11_KHR 0x401C + +/* cl_context_info */ +#define CL_CONTEXT_D3D11_DEVICE_KHR 0x401D +#define CL_CONTEXT_D3D11_PREFER_SHARED_RESOURCES_KHR 0x402D + +/* cl_mem_info */ +#define CL_MEM_D3D11_RESOURCE_KHR 0x401E + +/* cl_image_info */ +#define CL_IMAGE_D3D11_SUBRESOURCE_KHR 0x401F + +/* cl_command_type */ +#define CL_COMMAND_ACQUIRE_D3D11_OBJECTS_KHR 0x4020 +#define CL_COMMAND_RELEASE_D3D11_OBJECTS_KHR 0x4021 + +/******************************************************************************/ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromD3D11KHR_fn)( + cl_platform_id platform, + cl_d3d11_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d11_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11BufferKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D11Buffer * resource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11Texture2DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D11Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11Texture3DKHR_fn)( + cl_context context, + cl_mem_flags flags, + ID3D11Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireD3D11ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseD3D11ObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_D3D11_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_dx9_media_sharing.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_dx9_media_sharing.h new file mode 100644 index 0000000000..c702dc4838 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_dx9_media_sharing.h @@ -0,0 +1,127 @@ +/********************************************************************************** + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_CL_DX9_MEDIA_SHARING_H +#define __OPENCL_CL_DX9_MEDIA_SHARING_H + +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/****************************************************************************** +/* cl_khr_dx9_media_sharing */ +#define cl_khr_dx9_media_sharing 1 + +typedef cl_uint cl_dx9_media_adapter_type_khr; +typedef cl_uint cl_dx9_media_adapter_set_khr; + +#if defined(_WIN32) +#include +typedef struct _cl_dx9_surface_info_khr +{ + IDirect3DSurface9 *resource; + HANDLE shared_handle; +} cl_dx9_surface_info_khr; +#endif + + +/******************************************************************************/ + +/* Error Codes */ +#define CL_INVALID_DX9_MEDIA_ADAPTER_KHR -1010 +#define CL_INVALID_DX9_MEDIA_SURFACE_KHR -1011 +#define CL_DX9_MEDIA_SURFACE_ALREADY_ACQUIRED_KHR -1012 +#define CL_DX9_MEDIA_SURFACE_NOT_ACQUIRED_KHR -1013 + +/* cl_media_adapter_type_khr */ +#define CL_ADAPTER_D3D9_KHR 0x2020 +#define CL_ADAPTER_D3D9EX_KHR 0x2021 +#define CL_ADAPTER_DXVA_KHR 0x2022 + +/* cl_media_adapter_set_khr */ +#define CL_PREFERRED_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR 0x2023 +#define CL_ALL_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR 0x2024 + +/* cl_context_info */ +#define CL_CONTEXT_ADAPTER_D3D9_KHR 0x2025 +#define CL_CONTEXT_ADAPTER_D3D9EX_KHR 0x2026 +#define CL_CONTEXT_ADAPTER_DXVA_KHR 0x2027 + +/* cl_mem_info */ +#define CL_MEM_DX9_MEDIA_ADAPTER_TYPE_KHR 0x2028 +#define CL_MEM_DX9_MEDIA_SURFACE_INFO_KHR 0x2029 + +/* cl_image_info */ +#define CL_IMAGE_DX9_MEDIA_PLANE_KHR 0x202A + +/* cl_command_type */ +#define CL_COMMAND_ACQUIRE_DX9_MEDIA_SURFACES_KHR 0x202B +#define CL_COMMAND_RELEASE_DX9_MEDIA_SURFACES_KHR 0x202C + +/******************************************************************************/ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromDX9MediaAdapterKHR_fn)( + cl_platform_id platform, + cl_uint num_media_adapters, + cl_dx9_media_adapter_type_khr * media_adapter_type, + void * media_adapters, + cl_dx9_media_adapter_set_khr media_adapter_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromDX9MediaSurfaceKHR_fn)( + cl_context context, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + void * surface_info, + cl_uint plane, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireDX9MediaSurfacesKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseDX9MediaSurfacesKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_DX9_MEDIA_SHARING_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_egl.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_egl.h new file mode 100644 index 0000000000..93e6c9cfba --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_egl.h @@ -0,0 +1,133 @@ +/******************************************************************************* + * Copyright (c) 2008-2010 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +#ifndef __OPENCL_CL_EGL_H +#define __OPENCL_CL_EGL_H + +#ifdef __APPLE__ + +#else +#include +#include +#include +#endif + +#ifdef __cplusplus +extern "C" { +#endif + + +/* Command type for events created with clEnqueueAcquireEGLObjectsKHR */ +#define CL_COMMAND_EGL_FENCE_SYNC_OBJECT_KHR 0x202F +#define CL_COMMAND_ACQUIRE_EGL_OBJECTS_KHR 0x202D +#define CL_COMMAND_RELEASE_EGL_OBJECTS_KHR 0x202E + +/* Error type for clCreateFromEGLImageKHR */ +#define CL_INVALID_EGL_OBJECT_KHR -1093 +#define CL_EGL_RESOURCE_NOT_ACQUIRED_KHR -1092 + +/* CLeglImageKHR is an opaque handle to an EGLImage */ +typedef void* CLeglImageKHR; + +/* CLeglDisplayKHR is an opaque handle to an EGLDisplay */ +typedef void* CLeglDisplayKHR; + +/* CLeglSyncKHR is an opaque handle to an EGLSync object */ +typedef void* CLeglSyncKHR; + +/* properties passed to clCreateFromEGLImageKHR */ +typedef intptr_t cl_egl_image_properties_khr; + + +#define cl_khr_egl_image 1 + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromEGLImageKHR(cl_context /* context */, + CLeglDisplayKHR /* egldisplay */, + CLeglImageKHR /* eglimage */, + cl_mem_flags /* flags */, + const cl_egl_image_properties_khr * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromEGLImageKHR_fn)( + cl_context context, + CLeglDisplayKHR egldisplay, + CLeglImageKHR eglimage, + cl_mem_flags flags, + const cl_egl_image_properties_khr * properties, + cl_int * errcode_ret); + + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireEGLObjectsKHR(cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireEGLObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event); + + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseEGLObjectsKHR(cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseEGLObjectsKHR_fn)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event); + + +#define cl_khr_egl_event 1 + +extern CL_API_ENTRY cl_event CL_API_CALL +clCreateEventFromEGLSyncKHR(cl_context /* context */, + CLeglSyncKHR /* sync */, + CLeglDisplayKHR /* display */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_event (CL_API_CALL *clCreateEventFromEGLSyncKHR_fn)( + cl_context context, + CLeglSyncKHR sync, + CLeglDisplayKHR display, + cl_int * errcode_ret); + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_EGL_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_ext.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_ext.h new file mode 100644 index 0000000000..640e0d5657 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_ext.h @@ -0,0 +1,492 @@ +/******************************************************************************* + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/* $Revision: 14835 $ on $Date: 2011-05-26 11:32:00 -0700 (Thu, 26 May 2011) $ */ + +/* cl_ext.h contains OpenCL extensions which don't have external */ +/* (OpenGL, D3D) dependencies. */ + +#ifndef __CL_EXT_H +#define __CL_EXT_H + +#ifdef __cplusplus +extern "C" { +#endif + +#ifdef __APPLE__ + #include + #include +#else + #include +#endif + +/* cl_khr_fp16 extension - no extension #define since it has no functions */ +#define CL_DEVICE_HALF_FP_CONFIG 0x1033 + +/* Memory object destruction + * + * Apple extension for use to manage externally allocated buffers used with cl_mem objects with CL_MEM_USE_HOST_PTR + * + * Registers a user callback function that will be called when the memory object is deleted and its resources + * freed. Each call to clSetMemObjectCallbackFn registers the specified user callback function on a callback + * stack associated with memobj. The registered user callback functions are called in the reverse order in + * which they were registered. The user callback functions are called and then the memory object is deleted + * and its resources freed. This provides a mechanism for the application (and libraries) using memobj to be + * notified when the memory referenced by host_ptr, specified when the memory object is created and used as + * the storage bits for the memory object, can be reused or freed. + * + * The application may not call CL api's with the cl_mem object passed to the pfn_notify. + * + * Please check for the "cl_APPLE_SetMemObjectDestructor" extension using clGetDeviceInfo(CL_DEVICE_EXTENSIONS) + * before using. + */ +#define cl_APPLE_SetMemObjectDestructor 1 +cl_int CL_API_ENTRY clSetMemObjectDestructorAPPLE( cl_mem /* memobj */, + void (* /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), + void * /*user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + + +/* Context Logging Functions + * + * The next three convenience functions are intended to be used as the pfn_notify parameter to clCreateContext(). + * Please check for the "cl_APPLE_ContextLoggingFunctions" extension using clGetDeviceInfo(CL_DEVICE_EXTENSIONS) + * before using. + * + * clLogMessagesToSystemLog fowards on all log messages to the Apple System Logger + */ +#define cl_APPLE_ContextLoggingFunctions 1 +extern void CL_API_ENTRY clLogMessagesToSystemLogAPPLE( const char * /* errstr */, + const void * /* private_info */, + size_t /* cb */, + void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + +/* clLogMessagesToStdout sends all log messages to the file descriptor stdout */ +extern void CL_API_ENTRY clLogMessagesToStdoutAPPLE( const char * /* errstr */, + const void * /* private_info */, + size_t /* cb */, + void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + +/* clLogMessagesToStderr sends all log messages to the file descriptor stderr */ +extern void CL_API_ENTRY clLogMessagesToStderrAPPLE( const char * /* errstr */, + const void * /* private_info */, + size_t /* cb */, + void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; + + +/************************ +* cl_khr_icd extension * +************************/ +#define cl_khr_icd 1 + +/* cl_platform_info */ +#define CL_PLATFORM_ICD_SUFFIX_KHR 0x0920 + +/* Additional Error Codes */ +#define CL_PLATFORM_NOT_FOUND_KHR -1001 + +extern CL_API_ENTRY cl_int CL_API_CALL +clIcdGetPlatformIDsKHR(cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */); + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clIcdGetPlatformIDsKHR_fn)( + cl_uint /* num_entries */, + cl_platform_id * /* platforms */, + cl_uint * /* num_platforms */); + + +/* Extension: cl_khr_image2D_buffer + * + * This extension allows a 2D image to be created from a cl_mem buffer without a copy. + * The type associated with a 2D image created from a buffer in an OpenCL program is image2d_t. + * Both the sampler and sampler-less read_image built-in functions are supported for 2D images + * and 2D images created from a buffer. Similarly, the write_image built-ins are also supported + * for 2D images created from a buffer. + * + * When the 2D image from buffer is created, the client must specify the width, + * height, image format (i.e. channel order and channel data type) and optionally the row pitch + * + * The pitch specified must be a multiple of CL_DEVICE_IMAGE_PITCH_ALIGNMENT pixels. + * The base address of the buffer must be aligned to CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT pixels. + */ + +/************************************* + * cl_khr_initalize_memory extension * + *************************************/ + +#define CL_CONTEXT_MEMORY_INITIALIZE_KHR 0x200E + + +/************************************** + * cl_khr_terminate_context extension * + **************************************/ + +#define CL_DEVICE_TERMINATE_CAPABILITY_KHR 0x200F +#define CL_CONTEXT_TERMINATE_KHR 0x2010 + +#define cl_khr_terminate_context 1 +extern CL_API_ENTRY cl_int CL_API_CALL clTerminateContextKHR(cl_context /* context */) CL_EXT_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clTerminateContextKHR_fn)(cl_context /* context */) CL_EXT_SUFFIX__VERSION_1_2; + +// +/****************************** + * cl_khr_subgroups extension * + ******************************/ + +#define cl_khr_subgroups 1 + +typedef cl_uint cl_kernel_sub_group_info; + +#define CL_KERNEL_MAX_SUB_GROUP_SIZE_FOR_NDRANGE 0x2011 +#define CL_KERNEL_SUB_GROUP_COUNT_FOR_NDRANGE 0x2012 + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetKernelSubGroupInfoKHR(cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_sub_group_info /* param_name */, + size_t /* input_value_size */, + const void * /* input_value */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetKernelSubGroupInfoKHR_fn)( + cl_kernel /* kernel */, + cl_device_id /* device */, + cl_kernel_sub_group_info /* param_name */, + size_t /* input_value_size */, + const void * /* input_value */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; + +// + +/* + * Extension: cl_khr_spir + * + * This extension adds support to create an OpenCL program object from a + * Standard Portable Intermediate Representation (SPIR) instance + */ + +#define CL_DEVICE_SPIR_VERSIONS 0x40E0 +#define CL_PROGRAM_BINARY_TYPE_INTERMEDIATE 0x40E1 + + +/****************************************** +* cl_nv_device_attribute_query extension * +******************************************/ +/* cl_nv_device_attribute_query extension - no extension #define since it has no functions */ +#define CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV 0x4000 +#define CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV 0x4001 +#define CL_DEVICE_REGISTERS_PER_BLOCK_NV 0x4002 +#define CL_DEVICE_WARP_SIZE_NV 0x4003 +#define CL_DEVICE_GPU_OVERLAP_NV 0x4004 +#define CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV 0x4005 +#define CL_DEVICE_INTEGRATED_MEMORY_NV 0x4006 + +/********************************* +* cl_amd_device_memory_flags * +*********************************/ +#define cl_amd_device_memory_flags 1 +#define CL_MEM_USE_PERSISTENT_MEM_AMD (1 << 6) // Alloc from GPU's CPU visible heap + +/* cl_device_info */ +#define CL_DEVICE_MAX_ATOMIC_COUNTERS_EXT 0x4032 + +/********************************* +* cl_amd_device_attribute_query * +*********************************/ +#define CL_DEVICE_PROFILING_TIMER_OFFSET_AMD 0x4036 +#define CL_DEVICE_TOPOLOGY_AMD 0x4037 +#define CL_DEVICE_BOARD_NAME_AMD 0x4038 +#define CL_DEVICE_GLOBAL_FREE_MEMORY_AMD 0x4039 +#define CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD 0x4040 +#define CL_DEVICE_SIMD_WIDTH_AMD 0x4041 +#define CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD 0x4042 +#define CL_DEVICE_WAVEFRONT_WIDTH_AMD 0x4043 +#define CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD 0x4044 +#define CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD 0x4045 +#define CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD 0x4046 +#define CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD 0x4047 +#define CL_DEVICE_LOCAL_MEM_BANKS_AMD 0x4048 +#define CL_DEVICE_THREAD_TRACE_SUPPORTED_AMD 0x4049 +#define CL_DEVICE_GFXIP_MAJOR_AMD 0x404A +#define CL_DEVICE_GFXIP_MINOR_AMD 0x404B + +typedef union +{ + struct { cl_uint type; cl_uint data[5]; } raw; + struct { cl_uint type; cl_char unused[17]; cl_char bus; cl_char device; cl_char function; } pcie; +} cl_device_topology_amd; + +#define CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD 1 + +/************* +* cl_amd_hsa * +**************/ +#define CL_HSA_ENABLED_AMD (1ull << 62) +#define CL_HSA_DISABLED_AMD (1ull << 63) + +// +/*************************** +* cl_amd_command_intercept * +***************************/ +#define CL_CONTEXT_COMMAND_INTERCEPT_CALLBACK_AMD 0x403D +#define CL_QUEUE_COMMAND_INTERCEPT_ENABLE_AMD (1ull << 63) + +typedef cl_int (CL_CALLBACK * intercept_callback_fn)(cl_event, cl_int *); + +/************************** +* cl_amd_command_queue_info * +**************************/ +#define CL_QUEUE_THREAD_HANDLE_AMD 0x403E + +/*************************************** +* cl-gl depth buffer interop extension * +****************************************/ + +#define CL_DEPTH 0x10BD +#define CL_DEPTH_STENCIL 0x10BE +#define CL_UNORM_INT24 0x10DF +#define CL_GL_NUM_SAMPLES 0x2012 +// + +/************************** +* cl_amd_offline_devices * +**************************/ +#define CL_CONTEXT_OFFLINE_DEVICES_AMD 0x403F + +/********************************* +* cl_arm_printf extension +*********************************/ +#define CL_PRINTF_CALLBACK_ARM 0x40B0 +#define CL_PRINTF_BUFFERSIZE_ARM 0x40B1 + +#ifdef CL_VERSION_1_1 + /*********************************** + * cl_ext_device_fission extension * + ***********************************/ + #define cl_ext_device_fission 1 + + extern CL_API_ENTRY cl_int CL_API_CALL + clReleaseDeviceEXT( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL *clReleaseDeviceEXT_fn)( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + extern CL_API_ENTRY cl_int CL_API_CALL + clRetainDeviceEXT( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL *clRetainDeviceEXT_fn)( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef cl_ulong cl_device_partition_property_ext; + extern CL_API_ENTRY cl_int CL_API_CALL + clCreateSubDevicesEXT( cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int + ( CL_API_CALL * clCreateSubDevicesEXT_fn)( cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + /* cl_device_partition_property_ext */ + #define CL_DEVICE_PARTITION_EQUALLY_EXT 0x4050 + #define CL_DEVICE_PARTITION_BY_COUNTS_EXT 0x4051 + #define CL_DEVICE_PARTITION_BY_NAMES_EXT 0x4052 + #define CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN_EXT 0x4053 + + /* clDeviceGetInfo selectors */ + #define CL_DEVICE_PARENT_DEVICE_EXT 0x4054 + #define CL_DEVICE_PARTITION_TYPES_EXT 0x4055 + #define CL_DEVICE_AFFINITY_DOMAINS_EXT 0x4056 + #define CL_DEVICE_REFERENCE_COUNT_EXT 0x4057 + #define CL_DEVICE_PARTITION_STYLE_EXT 0x4058 + + /* clGetImageInfo enum */ + #define CL_IMAGE_BYTE_PITCH_AMD 0x4059 + + /* error codes */ + #define CL_DEVICE_PARTITION_FAILED_EXT -1057 + #define CL_INVALID_PARTITION_COUNT_EXT -1058 + #define CL_INVALID_PARTITION_NAME_EXT -1059 + + /* CL_AFFINITY_DOMAINs */ + #define CL_AFFINITY_DOMAIN_L1_CACHE_EXT 0x1 + #define CL_AFFINITY_DOMAIN_L2_CACHE_EXT 0x2 + #define CL_AFFINITY_DOMAIN_L3_CACHE_EXT 0x3 + #define CL_AFFINITY_DOMAIN_L4_CACHE_EXT 0x4 + #define CL_AFFINITY_DOMAIN_NUMA_EXT 0x10 + #define CL_AFFINITY_DOMAIN_NEXT_FISSIONABLE_EXT 0x100 + /* cl_device_partition_property_ext list terminators */ + #define CL_PROPERTIES_LIST_END_EXT ((cl_device_partition_property_ext) 0) + #define CL_PARTITION_BY_COUNTS_LIST_END_EXT ((cl_device_partition_property_ext) 0) + #define CL_PARTITION_BY_NAMES_LIST_END_EXT ((cl_device_partition_property_ext) 0 - 1) + +/********************************* +* cl_qcom_ext_host_ptr extension +*********************************/ + +#define CL_MEM_EXT_HOST_PTR_QCOM (1 << 29) + +#define CL_DEVICE_EXT_MEM_PADDING_IN_BYTES_QCOM 0x40A0 +#define CL_DEVICE_PAGE_SIZE_QCOM 0x40A1 +#define CL_IMAGE_ROW_ALIGNMENT_QCOM 0x40A2 +#define CL_IMAGE_SLICE_ALIGNMENT_QCOM 0x40A3 +#define CL_MEM_HOST_UNCACHED_QCOM 0x40A4 +#define CL_MEM_HOST_WRITEBACK_QCOM 0x40A5 +#define CL_MEM_HOST_WRITETHROUGH_QCOM 0x40A6 +#define CL_MEM_HOST_WRITE_COMBINING_QCOM 0x40A7 + +typedef cl_uint cl_image_pitch_info_qcom; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceImageInfoQCOM(cl_device_id device, + size_t image_width, + size_t image_height, + const cl_image_format *image_format, + cl_image_pitch_info_qcom param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret); + +typedef struct _cl_mem_ext_host_ptr +{ + /* Type of external memory allocation. */ + /* Legal values will be defined in layered extensions. */ + cl_uint allocation_type; + + /* Host cache policy for this external memory allocation. */ + cl_uint host_cache_policy; + +} cl_mem_ext_host_ptr; + +/********************************* +* cl_qcom_ion_host_ptr extension +*********************************/ + +#define CL_MEM_ION_HOST_PTR_QCOM 0x40A8 + +typedef struct _cl_mem_ion_host_ptr +{ + /* Type of external memory allocation. */ + /* Must be CL_MEM_ION_HOST_PTR_QCOM for ION allocations. */ + cl_mem_ext_host_ptr ext_host_ptr; + + /* ION file descriptor */ + int ion_filedesc; + + /* Host pointer to the ION allocated memory */ + void* ion_hostptr; + +} cl_mem_ion_host_ptr; + + // + /************************* + * cl_amd_object_metadata * + **************************/ + #define cl_amd_object_metadata 1 + + typedef size_t cl_key_amd; + + #define CL_INVALID_OBJECT_AMD 0x403A + #define CL_INVALID_KEY_AMD 0x403B + #define CL_PLATFORM_MAX_KEYS_AMD 0x403C + + typedef CL_API_ENTRY cl_key_amd (CL_API_CALL * clCreateKeyAMD_fn)( + cl_platform_id /* platform */, + void (CL_CALLBACK * /* destructor */)( void* /* old_value */), + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int (CL_API_CALL * clObjectGetValueForKeyAMD_fn)( + void * /* object */, + cl_key_amd /* key */, + void ** /* ret_val */) CL_API_SUFFIX__VERSION_1_1; + + typedef CL_API_ENTRY cl_int (CL_API_CALL * clObjectSetValueForKeyAMD_fn)( + void * /* object */, + cl_key_amd /* key */, + void * /* value */) CL_API_SUFFIX__VERSION_1_1; +// +#endif /* CL_VERSION_1_1 */ + +#ifdef CL_VERSION_1_2 + /******************************** + * cl_amd_bus_addressable_memory * + ********************************/ + + /* cl_mem flag - bitfield */ + #define CL_MEM_BUS_ADDRESSABLE_AMD (1<<30) + #define CL_MEM_EXTERNAL_PHYSICAL_AMD (1<<31) + + #define CL_COMMAND_WAIT_SIGNAL_AMD 0x4080 + #define CL_COMMAND_WRITE_SIGNAL_AMD 0x4081 + #define CL_COMMAND_MAKE_BUFFERS_RESIDENT_AMD 0x4082 + + typedef struct _cl_bus_address_amd + { + cl_ulong surface_bus_address; + cl_ulong marker_bus_address; + } cl_bus_address_amd; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL * clEnqueueWaitSignalAMD_fn)( cl_command_queue /*command_queue*/, + cl_mem /*mem_object*/, + cl_uint /*value*/, + cl_uint /*num_events*/, + const cl_event * /*event_wait_list*/, + cl_event * /*event*/) CL_EXT_SUFFIX__VERSION_1_2; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL * clEnqueueWriteSignalAMD_fn)( cl_command_queue /*command_queue*/, + cl_mem /*mem_object*/, + cl_uint /*value*/, + cl_ulong /*offset*/, + cl_uint /*num_events*/, + const cl_event * /*event_list*/, + cl_event * /*event*/) CL_EXT_SUFFIX__VERSION_1_2; + + typedef CL_API_ENTRY cl_int + (CL_API_CALL * clEnqueueMakeBuffersResidentAMD_fn)( cl_command_queue /*command_queue*/, + cl_uint /*num_mem_objs*/, + cl_mem * /*mem_objects*/, + cl_bool /*blocking_make_resident*/, + cl_bus_address_amd * /*bus_addresses*/, + cl_uint /*num_events*/, + const cl_event * /*event_list*/, + cl_event * /*event*/) CL_EXT_SUFFIX__VERSION_1_2; +#endif /* CL_VERSION_1_2 */ + +#ifdef __cplusplus +} +#endif + + +#endif /* __CL_EXT_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl.h new file mode 100644 index 0000000000..04080937a2 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl.h @@ -0,0 +1,162 @@ +/********************************************************************************** + * Copyright (c) 2008 - 2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +#ifndef __OPENCL_CL_GL_H +#define __OPENCL_CL_GL_H + +#ifdef __APPLE__ +#include +#else +#include +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +typedef cl_uint cl_gl_object_type; +typedef cl_uint cl_gl_texture_info; +typedef cl_uint cl_gl_platform_info; +typedef struct __GLsync *cl_GLsync; + +/* cl_gl_object_type = 0x2000 - 0x200F enum values are currently taken */ +#define CL_GL_OBJECT_BUFFER 0x2000 +#define CL_GL_OBJECT_TEXTURE2D 0x2001 +#define CL_GL_OBJECT_TEXTURE3D 0x2002 +#define CL_GL_OBJECT_RENDERBUFFER 0x2003 +#define CL_GL_OBJECT_TEXTURE2D_ARRAY 0x200E +#define CL_GL_OBJECT_TEXTURE1D 0x200F +#define CL_GL_OBJECT_TEXTURE1D_ARRAY 0x2010 +#define CL_GL_OBJECT_TEXTURE_BUFFER 0x2011 + +/* cl_gl_texture_info */ +#define CL_GL_TEXTURE_TARGET 0x2004 +#define CL_GL_MIPMAP_LEVEL 0x2005 +#define CL_GL_NUM_SAMPLES 0x2012 + + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromGLBuffer(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLuint /* bufobj */, + int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromGLTexture(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromGLRenderbuffer(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLuint /* renderbuffer */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetGLObjectInfo(cl_mem /* memobj */, + cl_gl_object_type * /* gl_object_type */, + cl_GLuint * /* gl_object_name */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetGLTextureInfo(cl_mem /* memobj */, + cl_gl_texture_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireGLObjects(cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseGLObjects(cl_command_queue /* command_queue */, + cl_uint /* num_objects */, + const cl_mem * /* mem_objects */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; + + +/* Deprecated OpenCL 1.1 APIs */ +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateFromGLTexture2D(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL +clCreateFromGLTexture3D(cl_context /* context */, + cl_mem_flags /* flags */, + cl_GLenum /* target */, + cl_GLint /* miplevel */, + cl_GLuint /* texture */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +/* cl_khr_gl_sharing extension */ + +#define cl_khr_gl_sharing 1 + +typedef cl_uint cl_gl_context_info; + +/* Additional Error Codes */ +#define CL_INVALID_GL_SHAREGROUP_REFERENCE_KHR -1000 + +/* cl_gl_context_info */ +#define CL_CURRENT_DEVICE_FOR_GL_CONTEXT_KHR 0x2006 +#define CL_DEVICES_FOR_GL_CONTEXT_KHR 0x2007 + +/* Additional cl_context_properties */ +#define CL_GL_CONTEXT_KHR 0x2008 +#define CL_EGL_DISPLAY_KHR 0x2009 +#define CL_GLX_DISPLAY_KHR 0x200A +#define CL_WGL_HDC_KHR 0x200B +#define CL_CGL_SHAREGROUP_KHR 0x200C + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetGLContextInfoKHR(const cl_context_properties * /* properties */, + cl_gl_context_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetGLContextInfoKHR_fn)( + const cl_context_properties * properties, + cl_gl_context_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret); + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_GL_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl_ext.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl_ext.h new file mode 100644 index 0000000000..a46e0a2e0b --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_gl_ext.h @@ -0,0 +1,69 @@ +/********************************************************************************** + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +/* cl_gl_ext.h contains vendor (non-KHR) OpenCL extensions which have */ +/* OpenGL dependencies. */ + +#ifndef __OPENCL_CL_GL_EXT_H +#define __OPENCL_CL_GL_EXT_H + +#ifdef __cplusplus +extern "C" { +#endif + +#ifdef __APPLE__ + #include +#else + #include +#endif + +/* + * For each extension, follow this template + * cl_VEN_extname extension */ +/* #define cl_VEN_extname 1 + * ... define new types, if any + * ... define new tokens, if any + * ... define new APIs, if any + * + * If you need GLtypes here, mirror them with a cl_GLtype, rather than including a GL header + * This allows us to avoid having to decide whether to include GL headers or GLES here. + */ + +/* + * cl_khr_gl_event extension + * See section 9.9 in the OpenCL 1.1 spec for more information + */ +#define CL_COMMAND_GL_FENCE_SYNC_OBJECT_KHR 0x200D + +extern CL_API_ENTRY cl_event CL_API_CALL +clCreateEventFromGLsyncKHR(cl_context /* context */, + cl_GLsync /* cl_GLsync */, + cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1; + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_CL_GL_EXT_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_platform.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_platform.h new file mode 100644 index 0000000000..fd1ece10aa --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/cl_platform.h @@ -0,0 +1,1216 @@ +/********************************************************************************** + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + **********************************************************************************/ + +/* $Revision: 14829 $ on $Date: 2011-05-26 08:22:50 -0700 (Thu, 26 May 2011) $ */ + +#ifndef __CL_PLATFORM_H +#define __CL_PLATFORM_H + +#ifdef __APPLE__ + /* Contains #defines for AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER below */ + #include +#endif + +#if !defined(_WIN32) || !defined(_MSC_VER) +# include +#endif /* !_WIN32 */ +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +#if defined(_WIN32) || defined(__CYGWIN__) + #define CL_API_ENTRY + #define CL_API_CALL __stdcall + #define CL_CALLBACK __stdcall +#else + #define CL_API_ENTRY + #define CL_API_CALL + #define CL_CALLBACK +#endif + +#ifdef __APPLE__ + #define CL_EXTENSION_WEAK_LINK __attribute__((weak_import)) + #define CL_API_SUFFIX__VERSION_1_0 AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_0 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER + #define CL_API_SUFFIX__VERSION_1_1 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define GCL_API_SUFFIX__VERSION_1_1 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_1 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER_BUT_DEPRECATED_IN_MAC_OS_X_VERSION_10_7 + + #ifdef AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define CL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define GCL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_2 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER_BUT_DEPRECATED_IN_MAC_OS_X_VERSION_10_8 + #else + #warning This path should never happen outside of internal operating system development. AvailabilityMacros do not function correctly here! + #define CL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define GCL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_2 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER + #endif +#else + #define CL_EXTENSION_WEAK_LINK + #define CL_API_SUFFIX__VERSION_1_0 + #define CL_EXT_SUFFIX__VERSION_1_0 + #define CL_API_SUFFIX__VERSION_1_1 + #define CL_EXT_SUFFIX__VERSION_1_1 + #define CL_API_SUFFIX__VERSION_1_2 + #define CL_EXT_SUFFIX__VERSION_1_2 + #define CL_API_SUFFIX__VERSION_2_0 + #define CL_EXT_SUFFIX__VERSION_2_0 + + #ifdef __GNUC__ + #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED __attribute__((deprecated)) + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + #endif + + #ifdef CL_USE_DEPRECATED_OPENCL_1_1_APIS + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED __attribute__((deprecated)) + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #endif + + #ifdef CL_USE_DEPRECATED_OPENCL_2_0_APIS + #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED __attribute__((deprecated)) + #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED + #endif + #elif _WIN32 + #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED __declspec(deprecated) + #endif + + #ifdef CL_USE_DEPRECATED_OPENCL_1_1_APIS + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED __declspec(deprecated) + #endif + + #ifdef CL_USE_DEPRECATED_OPENCL_2_0_APIS + #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED + #else + #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED __declspec(deprecated) + #endif + #else + #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED + + #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + + #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED + #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED + #endif +#endif + +#if (defined (_WIN32) && defined(_MSC_VER)) + +/* scalar types */ +typedef signed __int8 cl_char; +typedef unsigned __int8 cl_uchar; +typedef signed __int16 cl_short; +typedef unsigned __int16 cl_ushort; +typedef signed __int32 cl_int; +typedef unsigned __int32 cl_uint; +typedef signed __int64 cl_long; +typedef unsigned __int64 cl_ulong; + +typedef unsigned __int16 cl_half; +#else /* !_WIN32 */ +typedef int8_t cl_char; +typedef uint8_t cl_uchar; +typedef int16_t cl_short; +typedef uint16_t cl_ushort; +typedef int32_t cl_int; +typedef uint32_t cl_uint; +typedef int64_t cl_long; +typedef uint64_t cl_ulong; + +typedef uint16_t cl_half; +#endif /* !_WIN32 */ +typedef float cl_float; +typedef double cl_double; + +#define CL_CHAR_BIT 8 +#define CL_SCHAR_MAX 127 +#define CL_SCHAR_MIN (-127-1) +#define CL_CHAR_MAX CL_SCHAR_MAX +#define CL_CHAR_MIN CL_SCHAR_MIN +#define CL_UCHAR_MAX 255 +#define CL_SHRT_MAX 32767 +#define CL_SHRT_MIN (-32767-1) +#define CL_USHRT_MAX 65535 +#define CL_INT_MAX 2147483647 +#define CL_INT_MIN (-2147483647-1) +#define CL_UINT_MAX 0xffffffffU +#define CL_LONG_MAX ((cl_long) 0x7FFFFFFFFFFFFFFFLL) +#define CL_LONG_MIN ((cl_long) -0x7FFFFFFFFFFFFFFFLL - 1LL) +#define CL_ULONG_MAX ((cl_ulong) 0xFFFFFFFFFFFFFFFFULL) + +#define CL_FLT_DIG 6 +#define CL_FLT_MANT_DIG 24 +#define CL_FLT_MAX_10_EXP +38 +#define CL_FLT_MAX_EXP +128 +#define CL_FLT_MIN_10_EXP -37 +#define CL_FLT_MIN_EXP -125 +#define CL_FLT_RADIX 2 +#define CL_FLT_MAX FLT_MAX +#define CL_FLT_MIN FLT_MIN +#define CL_FLT_EPSILON FLT_EPSILON + +#define CL_DBL_DIG 15 +#define CL_DBL_MANT_DIG 53 +#define CL_DBL_MAX_10_EXP +308 +#define CL_DBL_MAX_EXP +1024 +#define CL_DBL_MIN_10_EXP -307 +#define CL_DBL_MIN_EXP -1021 +#define CL_DBL_RADIX 2 +#define CL_DBL_MAX DBL_MAX +#define CL_DBL_MIN DBL_MIN +#define CL_DBL_EPSILON DBL_EPSILON + +#define CL_M_E 2.718281828459045090796 +#define CL_M_LOG2E 1.442695040888963387005 +#define CL_M_LOG10E 0.434294481903251816668 +#define CL_M_LN2 0.693147180559945286227 +#define CL_M_LN10 2.302585092994045901094 +#define CL_M_PI 3.141592653589793115998 +#define CL_M_PI_2 1.570796326794896557999 +#define CL_M_PI_4 0.785398163397448278999 +#define CL_M_1_PI 0.318309886183790691216 +#define CL_M_2_PI 0.636619772367581382433 +#define CL_M_2_SQRTPI 1.128379167095512558561 +#define CL_M_SQRT2 1.414213562373095145475 +#define CL_M_SQRT1_2 0.707106781186547572737 + +#define CL_M_E_F 2.71828174591064f +#define CL_M_LOG2E_F 1.44269502162933f +#define CL_M_LOG10E_F 0.43429449200630f +#define CL_M_LN2_F 0.69314718246460f +#define CL_M_LN10_F 2.30258512496948f +#define CL_M_PI_F 3.14159274101257f +#define CL_M_PI_2_F 1.57079637050629f +#define CL_M_PI_4_F 0.78539818525314f +#define CL_M_1_PI_F 0.31830987334251f +#define CL_M_2_PI_F 0.63661974668503f +#define CL_M_2_SQRTPI_F 1.12837922573090f +#define CL_M_SQRT2_F 1.41421353816986f +#define CL_M_SQRT1_2_F 0.70710676908493f + +#define CL_NAN (CL_INFINITY - CL_INFINITY) +#define CL_HUGE_VALF ((cl_float) 1e50) +#define CL_HUGE_VAL ((cl_double) 1e500) +#define CL_MAXFLOAT CL_FLT_MAX +#define CL_INFINITY CL_HUGE_VALF + +#include + +/* Mirror types to GL types. Mirror types allow us to avoid deciding which 87s to load based on whether we are using GL or GLES here. */ +typedef unsigned int cl_GLuint; +typedef int cl_GLint; +typedef unsigned int cl_GLenum; + +/* + * Vector types + * + * Note: OpenCL requires that all types be naturally aligned. + * This means that vector types must be naturally aligned. + * For example, a vector of four floats must be aligned to + * a 16 byte boundary (calculated as 4 * the natural 4-byte + * alignment of the float). The alignment qualifiers here + * will only function properly if your compiler supports them + * and if you don't actively work to defeat them. For example, + * in order for a cl_float4 to be 16 byte aligned in a struct, + * the start of the struct must itself be 16-byte aligned. + * + * Maintaining proper alignment is the user's responsibility. + */ + +/* Define basic vector types */ +#if defined( __VEC__ ) + #include /* may be omitted depending on compiler. AltiVec spec provides no way to detect whether the header is required. */ + typedef vector unsigned char __cl_uchar16; + typedef vector signed char __cl_char16; + typedef vector unsigned short __cl_ushort8; + typedef vector signed short __cl_short8; + typedef vector unsigned int __cl_uint4; + typedef vector signed int __cl_int4; + typedef vector float __cl_float4; + #define __CL_UCHAR16__ 1 + #define __CL_CHAR16__ 1 + #define __CL_USHORT8__ 1 + #define __CL_SHORT8__ 1 + #define __CL_UINT4__ 1 + #define __CL_INT4__ 1 + #define __CL_FLOAT4__ 1 +#endif + +#if defined( __SSE__ ) + #if defined( __MINGW64__ ) + #include + #else + #include + #endif + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef float __cl_float4 __attribute__((vector_size(16))); + #else + typedef __m128 __cl_float4; + #endif + #define __CL_FLOAT4__ 1 +#endif + +#if defined( __SSE2__ ) + #if defined( __MINGW64__ ) + #include + #else + #include + #endif + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef cl_uchar __cl_uchar16 __attribute__((vector_size(16))); + typedef cl_char __cl_char16 __attribute__((vector_size(16))); + typedef cl_ushort __cl_ushort8 __attribute__((vector_size(16))); + typedef cl_short __cl_short8 __attribute__((vector_size(16))); + typedef cl_uint __cl_uint4 __attribute__((vector_size(16))); + typedef cl_int __cl_int4 __attribute__((vector_size(16))); + typedef cl_ulong __cl_ulong2 __attribute__((vector_size(16))); + typedef cl_long __cl_long2 __attribute__((vector_size(16))); + typedef cl_double __cl_double2 __attribute__((vector_size(16))); + #else + typedef __m128i __cl_uchar16; + typedef __m128i __cl_char16; + typedef __m128i __cl_ushort8; + typedef __m128i __cl_short8; + typedef __m128i __cl_uint4; + typedef __m128i __cl_int4; + typedef __m128i __cl_ulong2; + typedef __m128i __cl_long2; + typedef __m128d __cl_double2; + #endif + #define __CL_UCHAR16__ 1 + #define __CL_CHAR16__ 1 + #define __CL_USHORT8__ 1 + #define __CL_SHORT8__ 1 + #define __CL_INT4__ 1 + #define __CL_UINT4__ 1 + #define __CL_ULONG2__ 1 + #define __CL_LONG2__ 1 + #define __CL_DOUBLE2__ 1 +#endif + +#if defined( __MMX__ ) + #include + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef cl_uchar __cl_uchar8 __attribute__((vector_size(8))); + typedef cl_char __cl_char8 __attribute__((vector_size(8))); + typedef cl_ushort __cl_ushort4 __attribute__((vector_size(8))); + typedef cl_short __cl_short4 __attribute__((vector_size(8))); + typedef cl_uint __cl_uint2 __attribute__((vector_size(8))); + typedef cl_int __cl_int2 __attribute__((vector_size(8))); + typedef cl_ulong __cl_ulong1 __attribute__((vector_size(8))); + typedef cl_long __cl_long1 __attribute__((vector_size(8))); + typedef cl_float __cl_float2 __attribute__((vector_size(8))); + #else + typedef __m64 __cl_uchar8; + typedef __m64 __cl_char8; + typedef __m64 __cl_ushort4; + typedef __m64 __cl_short4; + typedef __m64 __cl_uint2; + typedef __m64 __cl_int2; + typedef __m64 __cl_ulong1; + typedef __m64 __cl_long1; + typedef __m64 __cl_float2; + #endif + #define __CL_UCHAR8__ 1 + #define __CL_CHAR8__ 1 + #define __CL_USHORT4__ 1 + #define __CL_SHORT4__ 1 + #define __CL_INT2__ 1 + #define __CL_UINT2__ 1 + #define __CL_ULONG1__ 1 + #define __CL_LONG1__ 1 + #define __CL_FLOAT2__ 1 +#endif + +#if defined( __AVX__ ) + #if defined( __MINGW64__ ) + #include + #else + #include + #endif + #if defined( __GNUC__ ) && !defined( __ICC ) + typedef cl_float __cl_float8 __attribute__((vector_size(32))); + typedef cl_double __cl_double4 __attribute__((vector_size(32))); + #else + typedef __m256 __cl_float8; + typedef __m256d __cl_double4; + #endif + #define __CL_FLOAT8__ 1 + #define __CL_DOUBLE4__ 1 +#endif + +/* Define capabilities for anonymous struct members. */ +#if (defined( __GNUC__ ) || defined( __IBMC__ )) && ! defined( __STRICT_ANSI__ ) +#define __CL_HAS_ANON_STRUCT__ 1 +#define __CL_ANON_STRUCT__ __extension__ +#elif defined( _WIN32) && (_MSC_VER >= 1500) + /* Microsoft Developer Studio 2008 supports anonymous structs, but + * complains by default. */ +#define __CL_HAS_ANON_STRUCT__ 1 +#define __CL_ANON_STRUCT__ + /* Disable warning C4201: nonstandard extension used : nameless + * struct/union */ +#pragma warning( push ) +#pragma warning( disable : 4201 ) +#else +#define __CL_HAS_ANON_STRUCT__ 0 +#define __CL_ANON_STRUCT__ +#endif + +/* Define alignment keys */ +#if (defined( __GNUC__ ) || defined( __IBMC__ )) + #define CL_ALIGNED(_x) __attribute__ ((aligned(_x))) +#elif defined( _WIN32) && (_MSC_VER) + /* Alignment keys neutered on windows because MSVC can't swallow function arguments with alignment requirements */ + /* http://msdn.microsoft.com/en-us/library/373ak2y1%28VS.71%29.aspx */ + /* #include */ + /* #define CL_ALIGNED(_x) _CRT_ALIGN(_x) */ + #define CL_ALIGNED(_x) +#else + #warning Need to implement some method to align data here + #define CL_ALIGNED(_x) +#endif + +/* Indicate whether .xyzw, .s0123 and .hi.lo are supported */ +#if __CL_HAS_ANON_STRUCT__ + /* .xyzw and .s0123...{f|F} are supported */ + #define CL_HAS_NAMED_VECTOR_FIELDS 1 + /* .hi and .lo are supported */ + #define CL_HAS_HI_LO_VECTOR_FIELDS 1 +#endif + +/* Define cl_vector types */ + +/* ---- cl_charn ---- */ +typedef union +{ + cl_char CL_ALIGNED(2) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_char x, y; }; + __CL_ANON_STRUCT__ struct{ cl_char s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_char lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2; +#endif +}cl_char2; + +typedef union +{ + cl_char CL_ALIGNED(4) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_char x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_char s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_char2 lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2[2]; +#endif +#if defined( __CL_CHAR4__) + __cl_char4 v4; +#endif +}cl_char4; + +/* cl_char3 is identical in size, alignment and behavior to cl_char4. See section 6.1.5. */ +typedef cl_char4 cl_char3; + +typedef union +{ + cl_char CL_ALIGNED(8) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_char x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_char4 lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2[4]; +#endif +#if defined( __CL_CHAR4__) + __cl_char4 v4[2]; +#endif +#if defined( __CL_CHAR8__ ) + __cl_char8 v8; +#endif +}cl_char8; + +typedef union +{ + cl_char CL_ALIGNED(16) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_char x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_char8 lo, hi; }; +#endif +#if defined( __CL_CHAR2__) + __cl_char2 v2[8]; +#endif +#if defined( __CL_CHAR4__) + __cl_char4 v4[4]; +#endif +#if defined( __CL_CHAR8__ ) + __cl_char8 v8[2]; +#endif +#if defined( __CL_CHAR16__ ) + __cl_char16 v16; +#endif +}cl_char16; + + +/* ---- cl_ucharn ---- */ +typedef union +{ + cl_uchar CL_ALIGNED(2) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uchar x, y; }; + __CL_ANON_STRUCT__ struct{ cl_uchar s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_uchar lo, hi; }; +#endif +#if defined( __cl_uchar2__) + __cl_uchar2 v2; +#endif +}cl_uchar2; + +typedef union +{ + cl_uchar CL_ALIGNED(4) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uchar x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_uchar s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_uchar2 lo, hi; }; +#endif +#if defined( __CL_UCHAR2__) + __cl_uchar2 v2[2]; +#endif +#if defined( __CL_UCHAR4__) + __cl_uchar4 v4; +#endif +}cl_uchar4; + +/* cl_uchar3 is identical in size, alignment and behavior to cl_uchar4. See section 6.1.5. */ +typedef cl_uchar4 cl_uchar3; + +typedef union +{ + cl_uchar CL_ALIGNED(8) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uchar x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_uchar4 lo, hi; }; +#endif +#if defined( __CL_UCHAR2__) + __cl_uchar2 v2[4]; +#endif +#if defined( __CL_UCHAR4__) + __cl_uchar4 v4[2]; +#endif +#if defined( __CL_UCHAR8__ ) + __cl_uchar8 v8; +#endif +}cl_uchar8; + +typedef union +{ + cl_uchar CL_ALIGNED(16) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uchar x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_uchar8 lo, hi; }; +#endif +#if defined( __CL_UCHAR2__) + __cl_uchar2 v2[8]; +#endif +#if defined( __CL_UCHAR4__) + __cl_uchar4 v4[4]; +#endif +#if defined( __CL_UCHAR8__ ) + __cl_uchar8 v8[2]; +#endif +#if defined( __CL_UCHAR16__ ) + __cl_uchar16 v16; +#endif +}cl_uchar16; + + +/* ---- cl_shortn ---- */ +typedef union +{ + cl_short CL_ALIGNED(4) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_short x, y; }; + __CL_ANON_STRUCT__ struct{ cl_short s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_short lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2; +#endif +}cl_short2; + +typedef union +{ + cl_short CL_ALIGNED(8) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_short x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_short s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_short2 lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2[2]; +#endif +#if defined( __CL_SHORT4__) + __cl_short4 v4; +#endif +}cl_short4; + +/* cl_short3 is identical in size, alignment and behavior to cl_short4. See section 6.1.5. */ +typedef cl_short4 cl_short3; + +typedef union +{ + cl_short CL_ALIGNED(16) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_short x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_short4 lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2[4]; +#endif +#if defined( __CL_SHORT4__) + __cl_short4 v4[2]; +#endif +#if defined( __CL_SHORT8__ ) + __cl_short8 v8; +#endif +}cl_short8; + +typedef union +{ + cl_short CL_ALIGNED(32) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_short x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_short8 lo, hi; }; +#endif +#if defined( __CL_SHORT2__) + __cl_short2 v2[8]; +#endif +#if defined( __CL_SHORT4__) + __cl_short4 v4[4]; +#endif +#if defined( __CL_SHORT8__ ) + __cl_short8 v8[2]; +#endif +#if defined( __CL_SHORT16__ ) + __cl_short16 v16; +#endif +}cl_short16; + + +/* ---- cl_ushortn ---- */ +typedef union +{ + cl_ushort CL_ALIGNED(4) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ushort x, y; }; + __CL_ANON_STRUCT__ struct{ cl_ushort s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_ushort lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2; +#endif +}cl_ushort2; + +typedef union +{ + cl_ushort CL_ALIGNED(8) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ushort x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_ushort s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_ushort2 lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2[2]; +#endif +#if defined( __CL_USHORT4__) + __cl_ushort4 v4; +#endif +}cl_ushort4; + +/* cl_ushort3 is identical in size, alignment and behavior to cl_ushort4. See section 6.1.5. */ +typedef cl_ushort4 cl_ushort3; + +typedef union +{ + cl_ushort CL_ALIGNED(16) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ushort x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_ushort4 lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2[4]; +#endif +#if defined( __CL_USHORT4__) + __cl_ushort4 v4[2]; +#endif +#if defined( __CL_USHORT8__ ) + __cl_ushort8 v8; +#endif +}cl_ushort8; + +typedef union +{ + cl_ushort CL_ALIGNED(32) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ushort x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_ushort8 lo, hi; }; +#endif +#if defined( __CL_USHORT2__) + __cl_ushort2 v2[8]; +#endif +#if defined( __CL_USHORT4__) + __cl_ushort4 v4[4]; +#endif +#if defined( __CL_USHORT8__ ) + __cl_ushort8 v8[2]; +#endif +#if defined( __CL_USHORT16__ ) + __cl_ushort16 v16; +#endif +}cl_ushort16; + +/* ---- cl_intn ---- */ +typedef union +{ + cl_int CL_ALIGNED(8) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_int x, y; }; + __CL_ANON_STRUCT__ struct{ cl_int s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_int lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2; +#endif +}cl_int2; + +typedef union +{ + cl_int CL_ALIGNED(16) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_int x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_int s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_int2 lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2[2]; +#endif +#if defined( __CL_INT4__) + __cl_int4 v4; +#endif +}cl_int4; + +/* cl_int3 is identical in size, alignment and behavior to cl_int4. See section 6.1.5. */ +typedef cl_int4 cl_int3; + +typedef union +{ + cl_int CL_ALIGNED(32) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_int x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_int4 lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2[4]; +#endif +#if defined( __CL_INT4__) + __cl_int4 v4[2]; +#endif +#if defined( __CL_INT8__ ) + __cl_int8 v8; +#endif +}cl_int8; + +typedef union +{ + cl_int CL_ALIGNED(64) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_int x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_int8 lo, hi; }; +#endif +#if defined( __CL_INT2__) + __cl_int2 v2[8]; +#endif +#if defined( __CL_INT4__) + __cl_int4 v4[4]; +#endif +#if defined( __CL_INT8__ ) + __cl_int8 v8[2]; +#endif +#if defined( __CL_INT16__ ) + __cl_int16 v16; +#endif +}cl_int16; + + +/* ---- cl_uintn ---- */ +typedef union +{ + cl_uint CL_ALIGNED(8) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uint x, y; }; + __CL_ANON_STRUCT__ struct{ cl_uint s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_uint lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2; +#endif +}cl_uint2; + +typedef union +{ + cl_uint CL_ALIGNED(16) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uint x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_uint s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_uint2 lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2[2]; +#endif +#if defined( __CL_UINT4__) + __cl_uint4 v4; +#endif +}cl_uint4; + +/* cl_uint3 is identical in size, alignment and behavior to cl_uint4. See section 6.1.5. */ +typedef cl_uint4 cl_uint3; + +typedef union +{ + cl_uint CL_ALIGNED(32) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uint x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_uint4 lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2[4]; +#endif +#if defined( __CL_UINT4__) + __cl_uint4 v4[2]; +#endif +#if defined( __CL_UINT8__ ) + __cl_uint8 v8; +#endif +}cl_uint8; + +typedef union +{ + cl_uint CL_ALIGNED(64) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_uint x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_uint8 lo, hi; }; +#endif +#if defined( __CL_UINT2__) + __cl_uint2 v2[8]; +#endif +#if defined( __CL_UINT4__) + __cl_uint4 v4[4]; +#endif +#if defined( __CL_UINT8__ ) + __cl_uint8 v8[2]; +#endif +#if defined( __CL_UINT16__ ) + __cl_uint16 v16; +#endif +}cl_uint16; + +/* ---- cl_longn ---- */ +typedef union +{ + cl_long CL_ALIGNED(16) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_long x, y; }; + __CL_ANON_STRUCT__ struct{ cl_long s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_long lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2; +#endif +}cl_long2; + +typedef union +{ + cl_long CL_ALIGNED(32) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_long x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_long s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_long2 lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2[2]; +#endif +#if defined( __CL_LONG4__) + __cl_long4 v4; +#endif +}cl_long4; + +/* cl_long3 is identical in size, alignment and behavior to cl_long4. See section 6.1.5. */ +typedef cl_long4 cl_long3; + +typedef union +{ + cl_long CL_ALIGNED(64) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_long x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_long4 lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2[4]; +#endif +#if defined( __CL_LONG4__) + __cl_long4 v4[2]; +#endif +#if defined( __CL_LONG8__ ) + __cl_long8 v8; +#endif +}cl_long8; + +typedef union +{ + cl_long CL_ALIGNED(128) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_long x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_long8 lo, hi; }; +#endif +#if defined( __CL_LONG2__) + __cl_long2 v2[8]; +#endif +#if defined( __CL_LONG4__) + __cl_long4 v4[4]; +#endif +#if defined( __CL_LONG8__ ) + __cl_long8 v8[2]; +#endif +#if defined( __CL_LONG16__ ) + __cl_long16 v16; +#endif +}cl_long16; + + +/* ---- cl_ulongn ---- */ +typedef union +{ + cl_ulong CL_ALIGNED(16) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ulong x, y; }; + __CL_ANON_STRUCT__ struct{ cl_ulong s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_ulong lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2; +#endif +}cl_ulong2; + +typedef union +{ + cl_ulong CL_ALIGNED(32) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ulong x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_ulong s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_ulong2 lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2[2]; +#endif +#if defined( __CL_ULONG4__) + __cl_ulong4 v4; +#endif +}cl_ulong4; + +/* cl_ulong3 is identical in size, alignment and behavior to cl_ulong4. See section 6.1.5. */ +typedef cl_ulong4 cl_ulong3; + +typedef union +{ + cl_ulong CL_ALIGNED(64) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ulong x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_ulong4 lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2[4]; +#endif +#if defined( __CL_ULONG4__) + __cl_ulong4 v4[2]; +#endif +#if defined( __CL_ULONG8__ ) + __cl_ulong8 v8; +#endif +}cl_ulong8; + +typedef union +{ + cl_ulong CL_ALIGNED(128) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_ulong x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_ulong8 lo, hi; }; +#endif +#if defined( __CL_ULONG2__) + __cl_ulong2 v2[8]; +#endif +#if defined( __CL_ULONG4__) + __cl_ulong4 v4[4]; +#endif +#if defined( __CL_ULONG8__ ) + __cl_ulong8 v8[2]; +#endif +#if defined( __CL_ULONG16__ ) + __cl_ulong16 v16; +#endif +}cl_ulong16; + + +/* --- cl_floatn ---- */ + +typedef union +{ + cl_float CL_ALIGNED(8) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_float x, y; }; + __CL_ANON_STRUCT__ struct{ cl_float s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_float lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2; +#endif +}cl_float2; + +typedef union +{ + cl_float CL_ALIGNED(16) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_float x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_float s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_float2 lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2[2]; +#endif +#if defined( __CL_FLOAT4__) + __cl_float4 v4; +#endif +}cl_float4; + +/* cl_float3 is identical in size, alignment and behavior to cl_float4. See section 6.1.5. */ +typedef cl_float4 cl_float3; + +typedef union +{ + cl_float CL_ALIGNED(32) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_float x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_float4 lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2[4]; +#endif +#if defined( __CL_FLOAT4__) + __cl_float4 v4[2]; +#endif +#if defined( __CL_FLOAT8__ ) + __cl_float8 v8; +#endif +}cl_float8; + +typedef union +{ + cl_float CL_ALIGNED(64) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_float x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_float8 lo, hi; }; +#endif +#if defined( __CL_FLOAT2__) + __cl_float2 v2[8]; +#endif +#if defined( __CL_FLOAT4__) + __cl_float4 v4[4]; +#endif +#if defined( __CL_FLOAT8__ ) + __cl_float8 v8[2]; +#endif +#if defined( __CL_FLOAT16__ ) + __cl_float16 v16; +#endif +}cl_float16; + +/* --- cl_doublen ---- */ + +typedef union +{ + cl_double CL_ALIGNED(16) s[2]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_double x, y; }; + __CL_ANON_STRUCT__ struct{ cl_double s0, s1; }; + __CL_ANON_STRUCT__ struct{ cl_double lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2; +#endif +}cl_double2; + +typedef union +{ + cl_double CL_ALIGNED(32) s[4]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_double x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_double s0, s1, s2, s3; }; + __CL_ANON_STRUCT__ struct{ cl_double2 lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2[2]; +#endif +#if defined( __CL_DOUBLE4__) + __cl_double4 v4; +#endif +}cl_double4; + +/* cl_double3 is identical in size, alignment and behavior to cl_double4. See section 6.1.5. */ +typedef cl_double4 cl_double3; + +typedef union +{ + cl_double CL_ALIGNED(64) s[8]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_double x, y, z, w; }; + __CL_ANON_STRUCT__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7; }; + __CL_ANON_STRUCT__ struct{ cl_double4 lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2[4]; +#endif +#if defined( __CL_DOUBLE4__) + __cl_double4 v4[2]; +#endif +#if defined( __CL_DOUBLE8__ ) + __cl_double8 v8; +#endif +}cl_double8; + +typedef union +{ + cl_double CL_ALIGNED(128) s[16]; +#if __CL_HAS_ANON_STRUCT__ + __CL_ANON_STRUCT__ struct{ cl_double x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; + __CL_ANON_STRUCT__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; + __CL_ANON_STRUCT__ struct{ cl_double8 lo, hi; }; +#endif +#if defined( __CL_DOUBLE2__) + __cl_double2 v2[8]; +#endif +#if defined( __CL_DOUBLE4__) + __cl_double4 v4[4]; +#endif +#if defined( __CL_DOUBLE8__ ) + __cl_double8 v8[2]; +#endif +#if defined( __CL_DOUBLE16__ ) + __cl_double16 v16; +#endif +}cl_double16; + +/* Macro to facilitate debugging + * Usage: + * Place CL_PROGRAM_STRING_DEBUG_INFO on the line before the first line of your source. + * The first line ends with: CL_PROGRAM_STRING_DEBUG_INFO \" + * Each line thereafter of OpenCL C source must end with: \n\ + * The last line ends in "; + * + * Example: + * + * const char *my_program = CL_PROGRAM_STRING_DEBUG_INFO "\ + * kernel void foo( int a, float * b ) \n\ + * { \n\ + * // my comment \n\ + * *b[ get_global_id(0)] = a; \n\ + * } \n\ + * "; + * + * This should correctly set up the line, (column) and file information for your source + * string so you can do source level debugging. + */ +#define __CL_STRINGIFY( _x ) # _x +#define _CL_STRINGIFY( _x ) __CL_STRINGIFY( _x ) +#define CL_PROGRAM_STRING_DEBUG_INFO "#line " _CL_STRINGIFY(__LINE__) " \"" __FILE__ "\" \n\n" + +#ifdef __cplusplus +} +#endif + +#undef __CL_HAS_ANON_STRUCT__ +#undef __CL_ANON_STRUCT__ +#if defined( _WIN32) && (_MSC_VER >= 1500) +#pragma warning( pop ) +#endif + +#endif /* __CL_PLATFORM_H */ diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/opencl.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/opencl.h new file mode 100644 index 0000000000..0c2e639cd1 --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/opencl.h @@ -0,0 +1,54 @@ +/******************************************************************************* + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ + +#ifndef __OPENCL_H +#define __OPENCL_H + +#ifdef __cplusplus +extern "C" { +#endif + +#ifdef __APPLE__ + +#include +#include +#include +#include + +#else + +#include +#include +#include +#include + +#endif + +#ifdef __cplusplus +} +#endif + +#endif /* __OPENCL_H */ + diff --git a/opencl/api/opencl/khronos/headers/opencl2.0/CL/spir.h b/opencl/api/opencl/khronos/headers/opencl2.0/CL/spir.h new file mode 100644 index 0000000000..5f0dbf6dea --- /dev/null +++ b/opencl/api/opencl/khronos/headers/opencl2.0/CL/spir.h @@ -0,0 +1,118 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. All rights reserved. + * + * NOTICE TO KHRONOS MEMBER: + * + * AMD has assigned the copyright for this object code to Khronos. + * This object code is subject to Khronos ownership rights under U.S. and + * international Copyright laws. + * + * Permission is hereby granted, free of charge, to any Khronos Member + * obtaining a copy of this software and/or associated documentation files + * (the "Materials"), to use, copy, modify and merge the Materials in object + * form only and to publish, distribute and/or sell copies of the Materials + * solely in object code form as part of conformant OpenCL API implementations, + * subject to the following conditions: + * + * Khronos Members shall ensure that their respective ICD implementation, + * that is installed over another Khronos Members' ICD implementation, will + * continue to support all OpenCL devices (hardware and software) supported + * by the replaced ICD implementation. For the purposes of this notice, "ICD" + * shall mean a library that presents an implementation of the OpenCL API for + * the purpose routing API calls to different vendor implementation. + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * KHRONOS AND AMD MAKE NO REPRESENTATION ABOUT THE SUITABILITY OF THIS + * SOURCE CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" WITHOUT EXPRESS OR + * IMPLIED WARRANTY OF ANY KIND. KHRONOS AND AMD DISCLAIM ALL WARRANTIES + * WITH REGARD TO THIS SOURCE CODE, INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY, NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE. + * IN NO EVENT SHALL KHRONOS OR AMD BE LIABLE FOR ANY SPECIAL, INDIRECT, + * INCIDENTAL, OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING + * FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, + * NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH + * THE USE OR PERFORMANCE OF THIS SOURCE CODE. + * + * U.S. Government End Users. This source code is a "commercial item" as + * that term is defined at 48 C.F.R. 2.101 (OCT 1995), consisting of + * "commercial computer software" and "commercial computer software + * documentation" as such terms are used in 48 C.F.R. 12.212 (SEPT 1995) + * and is provided to the U.S. Government only as a commercial end item. + * Consistent with 48 C.F.R.12.212 and 48 C.F.R. 227.7202-1 through + * 227.7202-4 (JUNE 1995), all U.S. Government End Users acquire the + * source code with only those rights set forth herein. + */ +/* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ +#ifndef __SPIR_H +#define __SPIR_H + +#ifdef __cplusplus +extern "C" { +#endif +// CLS is for CL SPIR. +typedef enum { + CLS_PRIVATE_AS = 0, + CLS_GLOBAL_AS = 1, + CLS_CONSTANT_AS = 2, + CLS_LOCAL_AS = 3, + CLS_GLOBAL_HOST_AS = 4, + CLS_CONSTANT_HOST_AS = 5, + CLS_REGION_AS_AMD = 16, +} CLS_ADDRESS_SPACES; + +// Table 13 +typedef enum { + CLS_ADDRESS_MIRRORED_REPEAT = 0, + CLS_ADDRESS_REPEAT = 1, + CLS_ADDRESS_CLAMP_TO_EDGE = 2, + CLS_ADDRESS_CLAMP = 3, + CLS_ADDRESS_NONE = 4 +} CLS_ADDRESSING_MODES; + +// Table 13 +typedef enum { + CLS_FILTER_NEAREST = 0, + CLS_FILTER_LINEAR = 1 +} CLS_FILTER_MODE; + +// Table 13 +typedef enum { + CLS_NORMALIZED_COORDS_TRUE = 0, + CLS_NORMALIZED_COORDS_FALSE = 1 +} CLS_NORMALIZED_COORDS; + +// Section 2.3 +typedef enum { + CLS_READ_ONLY = 0, + CLS_WRITE_ONLY = 1, + CLS_READ_WRITE = 2, + CLS_NONE = 3 +} CLS_ACCESS_QUALIFIERS; + +// Table 14 +typedef enum { + CLS_ARG_CONST = 1, + CLS_ARG_RESTRICT = 2, + CLS_ARG_VOLATILE = 4, + CLS_ARG_NONE = 0, + CLS_ARG_MASK = 0xf, + CLS_ARG_PIPE = 8 +} CLS_ARGTYPE_QUALIFIERS; + +static const char* SPIR_TRIPLE_32BIT = "spir-unknown-unknown"; +static const char* SPIR_TRIPLE_64BIT = "spir64-unknown-unknown"; +static const char* SPIR_DATA_LAYOUT_32BIT = "p:32:32:32-i1:8:8-i8:8:8" + "-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16" + "-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128" + "-v192:256:256-v256:256:256-v512:512:512-v1024:1024:1024"; +static const char* SPIR_DATA_LAYOUT_64BIT = "p:64:64:64-i1:8:8-i8:8:8" + "-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16" + "-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128" + "-v192:256:256-v256:256:256-v512:512:512-v1024:1024:1024"; + +#ifdef __cplusplus +} +#endif +#endif // __SPIR_H diff --git a/opencl/api/opencl/khronos/icd/OpenCL.rc b/opencl/api/opencl/khronos/icd/OpenCL.rc new file mode 100644 index 0000000000..9dbe6a9fdd --- /dev/null +++ b/opencl/api/opencl/khronos/icd/OpenCL.rc @@ -0,0 +1,74 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#include + +#ifdef RC_INVOKED + +VS_VERSION_INFO VERSIONINFO +FILEVERSION 2,0,1,0 +PRODUCTVERSION 2,0,1,0 +FILETYPE VFT_DLL + +BEGIN + BLOCK "StringFileInfo" + BEGIN + BLOCK "040904E4" + BEGIN + VALUE "FileDescription" ,"OpenCL Client DLL" + VALUE "ProductName" ,"Khronos OpenCL ICD" + VALUE "LegalCopyright" ,"Copyright \251 The Khronos Group Inc 2014" + VALUE "FileVersion" ,"2.0.1.0" + + VALUE "CompanyName" ,"Khronos Group" + VALUE "InternalName" ,"OpenCL" + VALUE "OriginalFilename","OpenCL.dll" + END + END + + BLOCK "VarFileInfo" + BEGIN + // extend this line for localized versions + VALUE "Translation", 0x0409, 0x04E4 + END +END + +#endif + diff --git a/opencl/api/opencl/khronos/icd/icd.c b/opencl/api/opencl/khronos/icd/icd.c new file mode 100644 index 0000000000..08465f7baa --- /dev/null +++ b/opencl/api/opencl/khronos/icd/icd.c @@ -0,0 +1,215 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#include "icd.h" +#include "icd_dispatch.h" +#include +#include + +KHRicdState khrIcdState = {0}; + +// entrypoint to initialize the ICD and add all vendors +void khrIcdInitialize(void) +{ + // make sure we don't double-initialize + if (khrIcdState.initialized) + { + return; + } + // khrIcdOsVendorsEnumerate is atomic + khrIcdState.initialized = khrIcdOsVendorsEnumerate(); +} + +void khrIcdVendorAdd(const char *libraryName) +{ + void *library = NULL; + cl_int result = CL_SUCCESS; + pfn_clGetExtensionFunctionAddress p_clGetExtensionFunctionAddress = NULL; + pfn_clIcdGetPlatformIDs p_clIcdGetPlatformIDs = NULL; + cl_uint i = 0; + cl_uint platformCount = 0; + cl_platform_id *platforms = NULL; + + // require that the library name be valid + if (!libraryName) + { + goto Done; + } + KHR_ICD_TRACE("attempting to add vendor %s...\n", libraryName); + + // load its library and query its function pointers + library = khrIcdOsLibraryLoad(libraryName); + if (!library) + { + KHR_ICD_TRACE("failed to load library %s\n", libraryName); + goto Done; + } + + // get the library's clGetExtensionFunctionAddress pointer + p_clGetExtensionFunctionAddress = khrIcdOsLibraryGetFunctionAddress(library, "clGetExtensionFunctionAddress"); + if (!p_clGetExtensionFunctionAddress) + { + KHR_ICD_TRACE("failed to get function address clGetExtensionFunctionAddress\n"); + goto Done; + } + + // use that function to get the clIcdGetPlatformIDsKHR function pointer + p_clIcdGetPlatformIDs = p_clGetExtensionFunctionAddress("clIcdGetPlatformIDsKHR"); + if (!p_clIcdGetPlatformIDs) + { + KHR_ICD_TRACE("failed to get extension function address clIcdGetPlatformIDsKHR\n"); + goto Done; + } + + // query the number of platforms available and allocate space to store them + result = p_clIcdGetPlatformIDs(0, NULL, &platformCount); + if (CL_SUCCESS != result) + { + KHR_ICD_TRACE("failed clIcdGetPlatformIDs\n"); + goto Done; + } + platforms = (cl_platform_id *)malloc(platformCount * sizeof(cl_platform_id) ); + if (!platforms) + { + KHR_ICD_TRACE("failed to allocate memory\n"); + goto Done; + } + memset(platforms, 0, platformCount * sizeof(cl_platform_id) ); + result = p_clIcdGetPlatformIDs(platformCount, platforms, NULL); + if (CL_SUCCESS != result) + { + KHR_ICD_TRACE("failed clIcdGetPlatformIDs\n"); + goto Done; + } + + // for each platform, add it + for (i = 0; i < platformCount; ++i) + { + KHRicdVendor* vendor = NULL; + char *suffix; + size_t suffixSize; + + // call clGetPlatformInfo on the returned platform to get the suffix + if (!platforms[i]) + { + continue; + } + result = platforms[i]->dispatch->clGetPlatformInfo( + platforms[i], + CL_PLATFORM_ICD_SUFFIX_KHR, + 0, + NULL, + &suffixSize); + if (CL_SUCCESS != result) + { + continue; + } + suffix = (char *)malloc(suffixSize); + if (!suffix) + { + continue; + } + result = platforms[i]->dispatch->clGetPlatformInfo( + platforms[i], + CL_PLATFORM_ICD_SUFFIX_KHR, + suffixSize, + suffix, + NULL); + if (CL_SUCCESS != result) + { + free(suffix); + continue; + } + + // allocate a structure for the vendor + vendor = (KHRicdVendor*)malloc(sizeof(*vendor) ); + if (!vendor) + { + free(suffix); + KHR_ICD_TRACE("failed to allocate memory\n"); + continue; + } + memset(vendor, 0, sizeof(*vendor) ); + + // populate vendor data + vendor->library = khrIcdOsLibraryLoad(libraryName); + if (!vendor->library) + { + free(suffix); + free(vendor); + KHR_ICD_TRACE("failed get platform handle to library\n"); + continue; + } + vendor->clGetExtensionFunctionAddress = p_clGetExtensionFunctionAddress; + vendor->platform = platforms[i]; + vendor->suffix = suffix; + + // add this vendor to the list of vendors at the tail + { + KHRicdVendor **prevNextPointer = NULL; + for (prevNextPointer = &khrIcdState.vendors; *prevNextPointer; prevNextPointer = &( (*prevNextPointer)->next) ); + *prevNextPointer = vendor; + } + + KHR_ICD_TRACE("successfully added vendor %s with suffix %s\n", libraryName, suffix); + + } + +Done: + + if (library) + { + khrIcdOsLibraryUnload(library); + } +} + +void khrIcdContextPropertiesGetPlatform(const cl_context_properties *properties, cl_platform_id *outPlatform) +{ + const cl_context_properties *property = (cl_context_properties *)NULL; + *outPlatform = NULL; + for (property = properties; property && property[0]; property += 2) + { + if ((cl_context_properties)CL_CONTEXT_PLATFORM == property[0]) + { + *outPlatform = (cl_platform_id)property[1]; + } + } +} + diff --git a/opencl/api/opencl/khronos/icd/icd.h b/opencl/api/opencl/khronos/icd/icd.h new file mode 100644 index 0000000000..d8b1ef0b4f --- /dev/null +++ b/opencl/api/opencl/khronos/icd/icd.h @@ -0,0 +1,272 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#ifndef _ICD_H_ +#define _ICD_H_ + +#include +#include + +#include "icd_dispatch.h" + +/* + * + * vendor dispatch table structure + * + */ + +struct _cl_object +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_platform_id +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_device_id +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_context +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_command_queue +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_mem +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_program +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_kernel +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_event +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_sampler +{ + KHRicdVendorDispatch *dispatch; +}; + +struct _cl_counter +{ + KHRicdVendorDispatch *dispatch; +}; + +/* + * type definitions + */ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *pfn_clIcdGetPlatformIDs)( + cl_uint num_entries, + cl_platform_id *platforms, + cl_uint *num_platforms) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *pfn_clGetPlatformInfo)( + cl_platform_id platform, + cl_platform_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY void *(CL_API_CALL *pfn_clGetExtensionFunctionAddress)( + const char *function_name) CL_API_SUFFIX__VERSION_1_0; + +typedef struct KHRicdVendorRec KHRicdVendor; +typedef struct KHRicdStateRec KHRicdState; + +/* + * KHRicdVendor + * + * Data for a single ICD vendor platform. + */ +struct KHRicdVendorRec +{ + // the loaded library object (true type varies on Linux versus Windows) + void *library; + + // the extension suffix for this platform + char *suffix; + + // function pointer to the ICD platform IDs extracted from the library + pfn_clGetExtensionFunctionAddress clGetExtensionFunctionAddress; + + // the platform retrieved from clGetIcdPlatformIDsKHR + cl_platform_id platform; + + // next vendor in the list vendors + KHRicdVendor *next; +}; + + +/* + * KHRicdState + * + * The global state of all vendors + * + * TODO: write access to this structure needs to be protected via a mutex + */ + +struct KHRicdStateRec +{ + // has this structure been initialized + cl_bool initialized; + + // the list of vendors which have been loaded + KHRicdVendor *vendors; +}; + +// the global state +extern KHRicdState khrIcdState; + +/* + * khrIcd interface + */ + +// read vendors from system configuration and store the data +// loaded into khrIcdState. this will call the OS-specific +// function khrIcdEnumerateVendors. this is called at every +// dispatch function which may be a valid first call into the +// API (e.g, getPlatformIDs, etc). +void khrIcdInitialize(void); + +// go through the list of vendors (in /etc/OpenCL.conf or through +// the registry) and call khrIcdVendorAdd for each vendor encountered +// n.b, this call is OS-specific +cl_bool khrIcdOsVendorsEnumerate(void); + +// add a vendor's implementation to the list of libraries +void khrIcdVendorAdd(const char *libraryName); + +// dynamically load a library. returns NULL on failure +// n.b, this call is OS-specific +void *khrIcdOsLibraryLoad(const char *libraryName); + +// get a function pointer from a loaded library. returns NULL on failure. +// n.b, this call is OS-specific +void *khrIcdOsLibraryGetFunctionAddress(void *library, const char *functionName); + +// unload a library. +// n.b, this call is OS-specific +void khrIcdOsLibraryUnload(void *library); + +// parse properties and determine the platform to use from them +void khrIcdContextPropertiesGetPlatform( + const cl_context_properties *properties, + cl_platform_id *outPlatform); + +// internal tracing macros +#if 0 + #include + #define KHR_ICD_TRACE(...) \ + do \ + { \ + fprintf(stderr, "KHR ICD trace at %s:%d: ", __FILE__, __LINE__); \ + fprintf(stderr, __VA_ARGS__); \ + } while (0) + + #define KHR_ICD_ASSERT(x) \ + do \ + { \ + if (!(x)) \ + { \ + fprintf(stderr, "KHR ICD assert at %s:%d: %s failed", __FILE__, __LINE__, #x); \ + } \ + } while (0) +#else + #define KHR_ICD_TRACE(...) + #define KHR_ICD_ASSERT(x) +#endif + +// if handle is NULL then return +#define KHR_ICD_VALIDATE_HANDLE(handle) \ + do \ + { \ + if (!handle) \ + { \ + return; \ + } \ + } while (0) + +// if handle is NULL then return invalid_handle_error_code +#define KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(handle,invalid_handle_error_code) \ + do \ + { \ + if (!handle) \ + { \ + return invalid_handle_error_code; \ + } \ + } while (0) + +// if handle is NULL then set errcode_ret to invalid_handle_error and return NULL +// (NULL being an invalid handle) +#define KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(handle,invalid_handle_error) \ + do \ + { \ + if (!handle) \ + { \ + if (errcode_ret) \ + { \ + *errcode_ret = invalid_handle_error; \ + } \ + return NULL; \ + } \ + } while (0) + + +#endif + diff --git a/opencl/api/opencl/khronos/icd/icd_dispatch.c b/opencl/api/opencl/khronos/icd/icd_dispatch.c new file mode 100644 index 0000000000..8340d1e9e7 --- /dev/null +++ b/opencl/api/opencl/khronos/icd/icd_dispatch.c @@ -0,0 +1,2524 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#include "icd_dispatch.h" +#include "icd.h" +#include +#include + +// Platform APIs +CL_API_ENTRY cl_int CL_API_CALL +clGetPlatformIDs(cl_uint num_entries, + cl_platform_id * platforms, + cl_uint * num_platforms) CL_API_SUFFIX__VERSION_1_0 +{ + KHRicdVendor* vendor = NULL; + cl_uint i; + + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + + if (!num_entries && platforms) + { + return CL_INVALID_VALUE; + } + if (!platforms && !num_platforms) + { + return CL_INVALID_VALUE; + } + // set num_platforms to 0 and set all platform pointers to NULL + if (num_platforms) + { + *num_platforms = 0; + } + for (i = 0; i < num_entries && platforms; ++i) + { + platforms[i] = NULL; + } + // return error if we have no platforms + if (!khrIcdState.vendors) + { + return CL_PLATFORM_NOT_FOUND_KHR; + } + // otherwise enumerate all platforms + for (vendor = khrIcdState.vendors; vendor; vendor = vendor->next) + { + if (num_entries && platforms) + { + *(platforms++) = vendor->platform; + --num_entries; + } + if (num_platforms) + { + ++(*num_platforms); + } + } + return CL_SUCCESS; +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetPlatformInfo(cl_platform_id platform, + cl_platform_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetPlatformInfo( + platform, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Device APIs +CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDs(cl_platform_id platform, + cl_device_type device_type, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_0 +{ + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetDeviceIDs( + platform, + device_type, + num_entries, + devices, + num_devices); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceInfo( + cl_device_id device, + cl_device_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(device, CL_INVALID_DEVICE); + return device->dispatch->clGetDeviceInfo( + device, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clCreateSubDevices(cl_device_id in_device, + const cl_device_partition_property * properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(in_device, CL_INVALID_DEVICE); + return in_device->dispatch->clCreateSubDevices( + in_device, + properties, + num_entries, + out_devices, + num_devices); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainDevice(cl_device_id device) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(device, CL_INVALID_DEVICE); + return device->dispatch->clRetainDevice(device); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseDevice(cl_device_id device) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(device, CL_INVALID_DEVICE); + return device->dispatch->clReleaseDevice(device); +} + +// Context APIs +CL_API_ENTRY cl_context CL_API_CALL +clCreateContext(const cl_context_properties * properties, + cl_uint num_devices, + const cl_device_id * devices, + void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *), + void * user_data, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + if (!num_devices || !devices) + { + if (errcode_ret) + { + *errcode_ret = CL_INVALID_VALUE; + } + return NULL; + } + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(devices[0], CL_INVALID_DEVICE); + return devices[0]->dispatch->clCreateContext( + properties, + num_devices, + devices, + pfn_notify, + user_data, + errcode_ret); +} + +CL_API_ENTRY cl_context CL_API_CALL +clCreateContextFromType(const cl_context_properties * properties, + cl_device_type device_type, + void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *), + void * user_data, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + cl_platform_id platform = NULL; + + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + + // determine the platform to use from the properties specified + khrIcdContextPropertiesGetPlatform(properties, &platform); + + // validate the platform handle and dispatch + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clCreateContextFromType( + properties, + device_type, + pfn_notify, + user_data, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainContext(cl_context context) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(context, CL_INVALID_CONTEXT); + return context->dispatch->clRetainContext(context); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseContext(cl_context context) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(context, CL_INVALID_CONTEXT); + return context->dispatch->clReleaseContext(context); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetContextInfo(cl_context context, + cl_context_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(context, CL_INVALID_CONTEXT); + return context->dispatch->clGetContextInfo( + context, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Command Queue APIs +CL_API_ENTRY cl_command_queue CL_API_CALL +clCreateCommandQueue(cl_context context, + cl_device_id device, + cl_command_queue_properties properties, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateCommandQueue( + context, + device, + properties, + errcode_ret); +} + +CL_API_ENTRY cl_command_queue CL_API_CALL +clCreateCommandQueueWithProperties(cl_context context, + cl_device_id device, + const cl_queue_properties * properties, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateCommandQueueWithProperties( + context, + device, + properties, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainCommandQueue(cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clRetainCommandQueue(command_queue); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseCommandQueue(cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clReleaseCommandQueue(command_queue); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetCommandQueueInfo(cl_command_queue command_queue, + cl_command_queue_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clGetCommandQueueInfo( + command_queue, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Memory Object APIs +CL_API_ENTRY cl_mem CL_API_CALL +clCreateBuffer(cl_context context, + cl_mem_flags flags, + size_t size, + void * host_ptr, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateBuffer( + context, + flags, + size, + host_ptr, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateImage(cl_context context, + cl_mem_flags flags, + const cl_image_format * image_format, + const cl_image_desc * image_desc, + void * host_ptr, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateImage( + context, + flags, + image_format, + image_desc, + host_ptr, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainMemObject(cl_mem memobj) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(memobj, CL_INVALID_MEM_OBJECT); + return memobj->dispatch->clRetainMemObject(memobj); +} + + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseMemObject(cl_mem memobj) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(memobj, CL_INVALID_MEM_OBJECT); + return memobj->dispatch->clReleaseMemObject(memobj); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetSupportedImageFormats(cl_context context, + cl_mem_flags flags, + cl_mem_object_type image_type, + cl_uint num_entries, + cl_image_format * image_formats, + cl_uint * num_image_formats) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(context, CL_INVALID_CONTEXT); + return context->dispatch->clGetSupportedImageFormats( + context, + flags, + image_type, + num_entries, + image_formats, + num_image_formats); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetMemObjectInfo(cl_mem memobj, + cl_mem_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(memobj, CL_INVALID_MEM_OBJECT); + return memobj->dispatch->clGetMemObjectInfo( + memobj, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetImageInfo(cl_mem image, + cl_image_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(image, CL_INVALID_MEM_OBJECT); + return image->dispatch->clGetImageInfo( + image, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Sampler APIs +CL_API_ENTRY cl_sampler CL_API_CALL +clCreateSampler(cl_context context, + cl_bool normalized_coords, + cl_addressing_mode addressing_mode, + cl_filter_mode filter_mode, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateSampler( + context, + normalized_coords, + addressing_mode, + filter_mode, + errcode_ret); +} + +CL_API_ENTRY cl_sampler CL_API_CALL +clCreateSamplerWithProperties(cl_context context, + const cl_sampler_properties * properties, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateSamplerWithProperties( + context, + properties, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainSampler(cl_sampler sampler) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(sampler, CL_INVALID_SAMPLER); + return sampler->dispatch->clRetainSampler(sampler); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseSampler(cl_sampler sampler) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(sampler, CL_INVALID_SAMPLER); + return sampler->dispatch->clReleaseSampler(sampler); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetSamplerInfo(cl_sampler sampler, + cl_sampler_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(sampler, CL_INVALID_SAMPLER); + return sampler->dispatch->clGetSamplerInfo( + sampler, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Program Object APIs +CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithSource(cl_context context, + cl_uint count, + const char ** strings, + const size_t * lengths, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateProgramWithSource( + context, + count, + strings, + lengths, + errcode_ret); +} + +CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithBinary(cl_context context, + cl_uint num_devices, + const cl_device_id * device_list, + const size_t * lengths, + const unsigned char ** binaries, + cl_int * binary_status, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateProgramWithBinary( + context, + num_devices, + device_list, + lengths, + binaries, + binary_status, + errcode_ret); +} + +CL_API_ENTRY cl_program CL_API_CALL +clCreateProgramWithBuiltInKernels(cl_context context, + cl_uint num_devices, + const cl_device_id * device_list, + const char * kernel_names, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateProgramWithBuiltInKernels( + context, + num_devices, + device_list, + kernel_names, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainProgram(cl_program program) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clRetainProgram(program); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseProgram(cl_program program) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clReleaseProgram(program); +} + +CL_API_ENTRY cl_int CL_API_CALL +clBuildProgram(cl_program program, + cl_uint num_devices, + const cl_device_id * device_list, + const char * options, + void (CL_CALLBACK *pfn_notify)(cl_program program, void * user_data), + void * user_data) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clBuildProgram( + program, + num_devices, + device_list, + options, + pfn_notify, + user_data); +} + +CL_API_ENTRY cl_int CL_API_CALL +clCompileProgram(cl_program program, + cl_uint num_devices, + const cl_device_id * device_list, + const char * options, + cl_uint num_input_headers, + const cl_program * input_headers, + const char ** header_include_names, + void (CL_CALLBACK * pfn_notify)(cl_program program, void * user_data), + void * user_data) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clCompileProgram( + program, + num_devices, + device_list, + options, + num_input_headers, + input_headers, + header_include_names, + pfn_notify, + user_data); +} + +CL_API_ENTRY cl_program CL_API_CALL +clLinkProgram(cl_context context, + cl_uint num_devices, + const cl_device_id * device_list, + const char * options, + cl_uint num_input_programs, + const cl_program * input_programs, + void (CL_CALLBACK * pfn_notify)(cl_program program, void * user_data), + void * user_data, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clLinkProgram( + context, + num_devices, + device_list, + options, + num_input_programs, + input_programs, + pfn_notify, + user_data, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clUnloadPlatformCompiler(cl_platform_id platform) CL_API_SUFFIX__VERSION_1_2 +{ + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clUnloadPlatformCompiler(platform); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetProgramInfo(cl_program program, + cl_program_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clGetProgramInfo( + program, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetProgramBuildInfo(cl_program program, + cl_device_id device, + cl_program_build_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clGetProgramBuildInfo( + program, + device, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Kernel Object APIs +CL_API_ENTRY cl_kernel CL_API_CALL +clCreateKernel(cl_program program, + const char * kernel_name, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(program, CL_INVALID_PROGRAM); + return program->dispatch->clCreateKernel( + program, + kernel_name, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clCreateKernelsInProgram(cl_program program, + cl_uint num_kernels, + cl_kernel * kernels, + cl_uint * num_kernels_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(program, CL_INVALID_PROGRAM); + return program->dispatch->clCreateKernelsInProgram( + program, + num_kernels, + kernels, + num_kernels_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainKernel(cl_kernel kernel) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clRetainKernel(kernel); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseKernel(cl_kernel kernel) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clReleaseKernel(kernel); +} + +CL_API_ENTRY cl_int CL_API_CALL +clSetKernelArg(cl_kernel kernel, + cl_uint arg_index, + size_t arg_size, + const void * arg_value) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clSetKernelArg( + kernel, + arg_index, + arg_size, + arg_value); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetKernelInfo(cl_kernel kernel, + cl_kernel_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clGetKernelInfo( + kernel, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetKernelArgInfo(cl_kernel kernel, + cl_uint arg_indx, + cl_kernel_arg_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clGetKernelArgInfo( + kernel, + arg_indx, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetKernelWorkGroupInfo(cl_kernel kernel, + cl_device_id device, + cl_kernel_work_group_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clGetKernelWorkGroupInfo( + kernel, + device, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Event Object APIs +CL_API_ENTRY cl_int CL_API_CALL +clWaitForEvents(cl_uint num_events, + const cl_event * event_list) CL_API_SUFFIX__VERSION_1_0 +{ + if (!num_events || !event_list) + { + return CL_INVALID_VALUE; + } + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event_list[0], CL_INVALID_EVENT); + return event_list[0]->dispatch->clWaitForEvents( + num_events, + event_list); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetEventInfo(cl_event event, + cl_event_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event, CL_INVALID_EVENT); + return event->dispatch->clGetEventInfo( + event, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainEvent(cl_event event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event, CL_INVALID_EVENT); + return event->dispatch->clRetainEvent(event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseEvent(cl_event event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event, CL_INVALID_EVENT); + return event->dispatch->clReleaseEvent(event); +} + +// Profiling APIs +CL_API_ENTRY cl_int CL_API_CALL +clGetEventProfilingInfo(cl_event event, + cl_profiling_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event, CL_INVALID_EVENT); + return event->dispatch->clGetEventProfilingInfo( + event, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +// Flush and Finish APIs +CL_API_ENTRY cl_int CL_API_CALL +clFlush(cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clFlush(command_queue); +} + +CL_API_ENTRY cl_int CL_API_CALL +clFinish(cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clFinish(command_queue); +} + +// Enqueued Commands APIs +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadBuffer(cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + size_t offset, + size_t cb, + void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReadBuffer( + command_queue, + buffer, + blocking_read, + offset, + cb, + ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadBufferRect( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + const size_t * buffer_origin, + const size_t * host_origin, + const size_t * region, + size_t buffer_row_pitch, + size_t buffer_slice_pitch, + size_t host_row_pitch, + size_t host_slice_pitch, + void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReadBufferRect( + command_queue, + buffer, + blocking_read, + buffer_origin, + host_origin, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteBuffer(cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_write, + size_t offset, + size_t cb, + const void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueWriteBuffer( + command_queue, + buffer, + blocking_write, + offset, + cb, + ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteBufferRect( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + const size_t * buffer_origin, + const size_t * host_origin, + const size_t * region, + size_t buffer_row_pitch, + size_t buffer_slice_pitch, + size_t host_row_pitch, + size_t host_slice_pitch, + const void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueWriteBufferRect( + command_queue, + buffer, + blocking_read, + buffer_origin, + host_origin, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueFillBuffer(cl_command_queue command_queue, + cl_mem buffer, + const void * pattern, + size_t pattern_size, + size_t offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueFillBuffer( + command_queue, + buffer, + pattern, + pattern_size, + offset, + cb, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBuffer(cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_buffer, + size_t src_offset, + size_t dst_offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueCopyBuffer( + command_queue, + src_buffer, + dst_buffer, + src_offset, + dst_offset, + cb, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBufferRect( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_buffer, + const size_t * src_origin, + const size_t * dst_origin, + const size_t * region, + size_t src_row_pitch, + size_t src_slice_pitch, + size_t dst_row_pitch, + size_t dst_slice_pitch, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueCopyBufferRect( + command_queue, + src_buffer, + dst_buffer, + src_origin, + dst_origin, + region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReadImage(cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_read, + const size_t * origin, + const size_t * region, + size_t row_pitch, + size_t slice_pitch, + void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReadImage( + command_queue, + image, + blocking_read, + origin, + region, + row_pitch, + slice_pitch, + ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWriteImage(cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_write, + const size_t * origin, + const size_t * region, + size_t input_row_pitch, + size_t input_slice_pitch, + const void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueWriteImage( + command_queue, + image, + blocking_write, + origin, + region, + input_row_pitch, + input_slice_pitch, + ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueFillImage(cl_command_queue command_queue, + cl_mem image, + const void * fill_color, + const size_t origin[3], + const size_t region[3], + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueFillImage( + command_queue, + image, + fill_color, + origin, + region, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyImage(cl_command_queue command_queue, + cl_mem src_image, + cl_mem dst_image, + const size_t * src_origin, + const size_t * dst_origin, + const size_t * region, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueCopyImage( + command_queue, + src_image, + dst_image, + src_origin, + dst_origin, + region, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyImageToBuffer(cl_command_queue command_queue, + cl_mem src_image, + cl_mem dst_buffer, + const size_t * src_origin, + const size_t * region, + size_t dst_offset, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueCopyImageToBuffer( + command_queue, + src_image, + dst_buffer, + src_origin, + region, + dst_offset, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueCopyBufferToImage(cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_image, + size_t src_offset, + const size_t * dst_origin, + const size_t * region, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueCopyBufferToImage( + command_queue, + src_buffer, + dst_image, + src_offset, + dst_origin, + region, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY void * CL_API_CALL +clEnqueueMapBuffer(cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_map, + cl_map_flags map_flags, + size_t offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueMapBuffer( + command_queue, + buffer, + blocking_map, + map_flags, + offset, + cb, + num_events_in_wait_list, + event_wait_list, + event, + errcode_ret); +} + +CL_API_ENTRY void * CL_API_CALL +clEnqueueMapImage(cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_map, + cl_map_flags map_flags, + const size_t * origin, + const size_t * region, + size_t * image_row_pitch, + size_t * image_slice_pitch, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueMapImage( + command_queue, + image, + blocking_map, + map_flags, + origin, + region, + image_row_pitch, + image_slice_pitch, + num_events_in_wait_list, + event_wait_list, + event, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueUnmapMemObject(cl_command_queue command_queue, + cl_mem memobj, + void * mapped_ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueUnmapMemObject( + command_queue, + memobj, + mapped_ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMigrateMemObjects(cl_command_queue command_queue, + cl_uint num_mem_objects, + const cl_mem * mem_objects, + cl_mem_migration_flags flags, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueMigrateMemObjects( + command_queue, + num_mem_objects, + mem_objects, + flags, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueNDRangeKernel(cl_command_queue command_queue, + cl_kernel kernel, + cl_uint work_dim, + const size_t * global_work_offset, + const size_t * global_work_size, + const size_t * local_work_size, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueNDRangeKernel( + command_queue, + kernel, + work_dim, + global_work_offset, + global_work_size, + local_work_size, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueTask(cl_command_queue command_queue, + cl_kernel kernel, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueTask( + command_queue, + kernel, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueNativeKernel(cl_command_queue command_queue, + void (CL_CALLBACK * user_func)(void *), + void * args, + size_t cb_args, + cl_uint num_mem_objects, + const cl_mem * mem_list, + const void ** args_mem_loc, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueNativeKernel( + command_queue, + user_func, + args, + cb_args, + num_mem_objects, + mem_list, + args_mem_loc, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMarkerWithWaitList(cl_command_queue command_queue, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueMarkerWithWaitList( + command_queue, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueBarrierWithWaitList(cl_command_queue command_queue, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueBarrierWithWaitList( + command_queue, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY void * CL_API_CALL +clGetExtensionFunctionAddressForPlatform(cl_platform_id platform, + const char * function_name) CL_API_SUFFIX__VERSION_1_2 +{ + // make sure the ICD is initialized + khrIcdInitialize(); + + // return any ICD-aware extensions + #define CL_COMMON_EXTENSION_ENTRYPOINT_ADD(name) if (!strcmp(function_name, #name) ) return (void *)&name + + // Are these core or ext? This is unclear, but they appear to be + // independent from cl_khr_gl_sharing. + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLBuffer); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLTexture); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLTexture2D); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLTexture3D); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLRenderbuffer); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetGLObjectInfo); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetGLTextureInfo); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireGLObjects); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseGLObjects); + + // cl_khr_gl_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetGLContextInfoKHR); + + // cl_khr_gl_event + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateEventFromGLsyncKHR); + +#if defined(_WIN32) + // cl_khr_d3d10_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetDeviceIDsFromD3D10KHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D10BufferKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D10Texture2DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D10Texture3DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireD3D10ObjectsKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseD3D10ObjectsKHR); + // cl_khr_d3d11_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetDeviceIDsFromD3D11KHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D11BufferKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D11Texture2DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D11Texture3DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireD3D11ObjectsKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseD3D11ObjectsKHR); + // cl_khr_dx9_media_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetDeviceIDsFromDX9MediaAdapterKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromDX9MediaSurfaceKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireDX9MediaSurfacesKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseDX9MediaSurfacesKHR); +#endif + + // cl_ext_device_fission + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateSubDevicesEXT); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clRetainDeviceEXT); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clReleaseDeviceEXT); + + /* cl_khr_egl_image */ + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromEGLImageKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireEGLObjectsKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseEGLObjectsKHR); + + /* cl_khr_egl_event */ + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateEventFromEGLSyncKHR); + + // fall back to vendor extension detection + + // FIXME Now that we have a platform id here, we need to validate that it isn't NULL, so shouldn't we have an errcode_ret + // KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetExtensionFunctionAddressForPlatform( + platform, + function_name); +} + +// Deprecated APIs +CL_API_ENTRY cl_int CL_API_CALL +clSetCommandQueueProperty(cl_command_queue command_queue, + cl_command_queue_properties properties, + cl_bool enable, + cl_command_queue_properties * old_properties) /*CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED*/ +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clSetCommandQueueProperty( + command_queue, + properties, + enable, + old_properties); +} + +CL_API_ENTRY cl_int CL_API_CALL +clCreateSubDevicesEXT( + cl_device_id in_device, + const cl_device_partition_property_ext * partition_properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(in_device, CL_INVALID_DEVICE); + return in_device->dispatch->clCreateSubDevicesEXT( + in_device, + partition_properties, + num_entries, + out_devices, + num_devices); +} + +CL_API_ENTRY cl_int CL_API_CALL +clRetainDeviceEXT(cl_device_id device) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(device, CL_INVALID_DEVICE); + return device->dispatch->clRetainDeviceEXT(device); +} + +CL_API_ENTRY cl_int CL_API_CALL +clReleaseDeviceEXT(cl_device_id device) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(device, CL_INVALID_DEVICE); + return device->dispatch->clReleaseDeviceEXT(device); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateImage2D(cl_context context, + cl_mem_flags flags, + const cl_image_format * image_format, + size_t image_width, + size_t image_height, + size_t image_row_pitch, + void * host_ptr, + cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateImage2D( + context, + flags, + image_format, + image_width, + image_height, + image_row_pitch, + host_ptr, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateImage3D(cl_context context, + cl_mem_flags flags, + const cl_image_format * image_format, + size_t image_width, + size_t image_height, + size_t image_depth, + size_t image_row_pitch, + size_t image_slice_pitch, + void * host_ptr, + cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateImage3D( + context, + flags, + image_format, + image_width, + image_height, + image_depth, + image_row_pitch, + image_slice_pitch, + host_ptr, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clUnloadCompiler(void) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + return CL_SUCCESS; +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueMarker(cl_command_queue command_queue, + cl_event * event) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueMarker( + command_queue, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueWaitForEvents(cl_command_queue command_queue, + cl_uint num_events, + const cl_event * event_list) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueWaitForEvents( + command_queue, + num_events, + event_list); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueBarrier(cl_command_queue command_queue) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueBarrier(command_queue); +} + +CL_API_ENTRY void * CL_API_CALL +clGetExtensionFunctionAddress(const char *function_name) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ + size_t function_name_length = strlen(function_name); + KHRicdVendor* vendor = NULL; + + // make sure the ICD is initialized + khrIcdInitialize(); + + // return any ICD-aware extensions + #define CL_COMMON_EXTENSION_ENTRYPOINT_ADD(name) if (!strcmp(function_name, #name) ) return (void *)&name + + // Are these core or ext? This is unclear, but they appear to be + // independent from cl_khr_gl_sharing. + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLBuffer); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLTexture); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLTexture2D); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLTexture3D); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromGLRenderbuffer); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetGLObjectInfo); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetGLTextureInfo); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireGLObjects); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseGLObjects); + + // cl_khr_gl_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetGLContextInfoKHR); + + // cl_khr_gl_event + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateEventFromGLsyncKHR); + +#if defined(_WIN32) + // cl_khr_d3d10_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetDeviceIDsFromD3D10KHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D10BufferKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D10Texture2DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D10Texture3DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireD3D10ObjectsKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseD3D10ObjectsKHR); + // cl_khr_d3d11_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetDeviceIDsFromD3D11KHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D11BufferKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D11Texture2DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromD3D11Texture3DKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireD3D11ObjectsKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseD3D11ObjectsKHR); + // cl_khr_dx9_media_sharing + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clGetDeviceIDsFromDX9MediaAdapterKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromDX9MediaSurfaceKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireDX9MediaSurfacesKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseDX9MediaSurfacesKHR); +#endif + + // cl_ext_device_fission + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateSubDevicesEXT); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clRetainDeviceEXT); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clReleaseDeviceEXT); + + /* cl_khr_egl_image */ + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateFromEGLImageKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueAcquireEGLObjectsKHR); + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clEnqueueReleaseEGLObjectsKHR); + + /* cl_khr_egl_event */ + CL_COMMON_EXTENSION_ENTRYPOINT_ADD(clCreateEventFromEGLSyncKHR); + + // fall back to vendor extension detection + for (vendor = khrIcdState.vendors; vendor; vendor = vendor->next) + { + size_t vendor_suffix_length = strlen(vendor->suffix); + if (vendor_suffix_length <= function_name_length && vendor_suffix_length > 0) + { + const char *function_suffix = function_name+function_name_length-vendor_suffix_length; + if (!strcmp(function_suffix, vendor->suffix) ) + { + return vendor->clGetExtensionFunctionAddress(function_name); + } + } + } + return NULL; +} + +// GL and other APIs +CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLBuffer( + cl_context context, + cl_mem_flags flags, + GLuint bufobj, + int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromGLBuffer( + context, + flags, + bufobj, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLTexture( + cl_context context, + cl_mem_flags flags, + cl_GLenum target, + cl_GLint miplevel, + cl_GLuint texture, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromGLTexture( + context, + flags, + target, + miplevel, + texture, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLTexture2D( + cl_context context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texture, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromGLTexture2D( + context, + flags, + target, + miplevel, + texture, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLTexture3D( + cl_context context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texture, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromGLTexture3D( + context, + flags, + target, + miplevel, + texture, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLRenderbuffer( + cl_context context, + cl_mem_flags flags, + GLuint renderbuffer, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromGLRenderbuffer( + context, + flags, + renderbuffer, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL clGetGLObjectInfo( + cl_mem memobj, + cl_gl_object_type * gl_object_type, + GLuint * gl_object_name) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(memobj, CL_INVALID_MEM_OBJECT); + return memobj->dispatch->clGetGLObjectInfo( + memobj, + gl_object_type, + gl_object_name); +} + +CL_API_ENTRY cl_int CL_API_CALL clGetGLTextureInfo( + cl_mem memobj, + cl_gl_texture_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(memobj, CL_INVALID_MEM_OBJECT); + return memobj->dispatch->clGetGLTextureInfo( + memobj, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL clEnqueueAcquireGLObjects( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueAcquireGLObjects( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL clEnqueueReleaseGLObjects( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReleaseGLObjects( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL clGetGLContextInfoKHR( + const cl_context_properties *properties, + cl_gl_context_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret) CL_API_SUFFIX__VERSION_1_0 +{ + cl_platform_id platform = NULL; + + // initialize the platforms (in case they have not been already) + khrIcdInitialize(); + + // determine the platform to use from the properties specified + khrIcdContextPropertiesGetPlatform(properties, &platform); + + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetGLContextInfoKHR( + properties, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_event CL_API_CALL clCreateEventFromGLsyncKHR( + cl_context context, + cl_GLsync sync, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateEventFromGLsyncKHR( + context, + sync, + errcode_ret); +} + +#if defined(_WIN32) +/* + * + * cl_d3d10_sharing_khr + * + */ + +CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromD3D10KHR( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void *d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id *devices, + cl_uint *num_devices) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetDeviceIDsFromD3D10KHR( + platform, + d3d_device_source, + d3d_object, + d3d_device_set, + num_entries, + devices, + num_devices); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10BufferKHR( + cl_context context, + cl_mem_flags flags, + ID3D10Buffer *resource, + cl_int *errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromD3D10BufferKHR( + context, + flags, + resource, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10Texture2DKHR( + cl_context context, + cl_mem_flags flags, + ID3D10Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromD3D10Texture2DKHR( + context, + flags, + resource, + subresource, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10Texture3DKHR( + cl_context context, + cl_mem_flags flags, + ID3D10Texture3D *resource, + UINT subresource, + cl_int *errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromD3D10Texture3DKHR( + context, + flags, + resource, + subresource, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireD3D10ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueAcquireD3D10ObjectsKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseD3D10ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReleaseD3D10ObjectsKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +/* + * + * cl_d3d11_sharing_khr + * + */ + +CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromD3D11KHR( + cl_platform_id platform, + cl_d3d11_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d11_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetDeviceIDsFromD3D11KHR( + platform, + d3d_device_source, + d3d_object, + d3d_device_set, + num_entries, + devices, + num_devices); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11BufferKHR( + cl_context context, + cl_mem_flags flags, + ID3D11Buffer * resource, + cl_int * errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromD3D11BufferKHR( + context, + flags, + resource, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11Texture2DKHR( + cl_context context, + cl_mem_flags flags, + ID3D11Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromD3D11Texture2DKHR( + context, + flags, + resource, + subresource, + errcode_ret); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11Texture3DKHR( + cl_context context, + cl_mem_flags flags, + ID3D11Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromD3D11Texture3DKHR( + context, + flags, + resource, + subresource, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireD3D11ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueAcquireD3D11ObjectsKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseD3D11ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReleaseD3D11ObjectsKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +/* + * + * cl_khr_dx9_media_sharing + * + */ + +CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromDX9MediaAdapterKHR( + cl_platform_id platform, + cl_uint num_media_adapters, + cl_dx9_media_adapter_type_khr * media_adapters_type, + void * media_adapters, + cl_dx9_media_adapter_set_khr media_adapter_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(platform, CL_INVALID_PLATFORM); + return platform->dispatch->clGetDeviceIDsFromDX9MediaAdapterKHR( + platform, + num_media_adapters, + media_adapters_type, + media_adapters, + media_adapter_set, + num_entries, + devices, + num_devices); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromDX9MediaSurfaceKHR( + cl_context context, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + void * surface_info, + cl_uint plane, + cl_int * errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromDX9MediaSurfaceKHR( + context, + flags, + adapter_type, + surface_info, + plane, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireDX9MediaSurfacesKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueAcquireDX9MediaSurfacesKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseDX9MediaSurfacesKHR( + cl_command_queue command_queue, + cl_uint num_objects, + cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReleaseDX9MediaSurfacesKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +#endif + +CL_API_ENTRY cl_int CL_API_CALL +clSetEventCallback( + cl_event event, + cl_int command_exec_callback_type, + void (CL_CALLBACK *pfn_notify)(cl_event, cl_int, void *), + void *user_data) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event, CL_INVALID_EVENT); + return event->dispatch->clSetEventCallback( + event, + command_exec_callback_type, + pfn_notify, + user_data); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateSubBuffer( + cl_mem buffer, + cl_mem_flags flags, + cl_buffer_create_type buffer_create_type, + const void * buffer_create_info, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(buffer, CL_INVALID_MEM_OBJECT); + return buffer->dispatch->clCreateSubBuffer( + buffer, + flags, + buffer_create_type, + buffer_create_info, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clSetMemObjectDestructorCallback( + cl_mem memobj, + void (CL_CALLBACK * pfn_notify)( cl_mem, void*), + void * user_data ) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(memobj, CL_INVALID_MEM_OBJECT); + return memobj->dispatch->clSetMemObjectDestructorCallback( + memobj, + pfn_notify, + user_data); +} + +CL_API_ENTRY cl_event CL_API_CALL +clCreateUserEvent( + cl_context context, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateUserEvent( + context, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clSetUserEventStatus( + cl_event event, + cl_int execution_status) CL_API_SUFFIX__VERSION_1_1 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(event, CL_INVALID_EVENT); + return event->dispatch->clSetUserEventStatus( + event, + execution_status); +} + +CL_API_ENTRY void * CL_API_CALL +clSVMAlloc(cl_context context, + cl_svm_mem_flags flags, + size_t size, + cl_uint alignment) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(context, NULL); + return context->dispatch->clSVMAlloc( + context, + flags, + size, + alignment); +} + +CL_API_ENTRY void CL_API_CALL +clSVMFree(cl_context context, + void * svm_pointer) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE(context); + context->dispatch->clSVMFree( + context, + svm_pointer); +} + +CL_API_ENTRY cl_int CL_API_CALL +clSetKernelArgSVMPointer(cl_kernel kernel, + cl_uint arg_index, + const void * arg_value) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clSetKernelArgSVMPointer( + kernel, + arg_index, + arg_value); +} + +CL_API_ENTRY cl_int CL_API_CALL +clSetKernelExecInfo(cl_kernel kernel, + cl_kernel_exec_info param_name, + size_t param_value_size, + const void * param_value) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clSetKernelExecInfo( + kernel, + param_name, + param_value_size, + param_value); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMFree(cl_command_queue command_queue, + cl_uint num_svm_pointers, + void * svm_pointers[], + void (CL_CALLBACK * pfn_free_func)(cl_command_queue queue, + cl_uint num_svm_pointers, + void * svm_pointers[], + void * user_data), + void * user_data, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueSVMFree( + command_queue, + num_svm_pointers, + svm_pointers, + pfn_free_func, + user_data, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMMemcpy(cl_command_queue command_queue, + cl_bool blocking_copy, + void * dst_ptr, + const void * src_ptr, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueSVMMemcpy( + command_queue, + blocking_copy, + dst_ptr, + src_ptr, + size, + num_events_in_wait_list, + event_wait_list, + event); +} + + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMMemFill(cl_command_queue command_queue, + void * svm_ptr, + const void * pattern, + size_t pattern_size, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueSVMMemFill( + command_queue, + svm_ptr, + pattern, + pattern_size, + size, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMMap(cl_command_queue command_queue, + cl_bool blocking_map, + cl_map_flags flags, + void * svm_ptr, + size_t size, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueSVMMap( + command_queue, + blocking_map, + flags, + svm_ptr, + size, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueSVMUnmap(cl_command_queue command_queue, + void * svm_ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueSVMUnmap( + command_queue, + svm_ptr, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreatePipe(cl_context context, + cl_mem_flags flags, + cl_uint pipe_packet_size, + cl_uint pipe_max_packets, + const cl_pipe_properties * properties, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreatePipe( + context, + flags, + pipe_packet_size, + pipe_max_packets, + properties, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetPipeInfo(cl_mem pipe, + cl_image_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(pipe, CL_INVALID_MEM_OBJECT); + return pipe->dispatch->clGetPipeInfo( + pipe, + param_name, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clGetKernelSubGroupInfoKHR( + cl_kernel kernel, + cl_device_id device, + cl_kernel_sub_group_info param_name, + size_t input_value_size, + const void * input_value, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(kernel, CL_INVALID_KERNEL); + return kernel->dispatch->clGetKernelSubGroupInfoKHR( + kernel, + device, + param_name, + input_value_size, + input_value, + param_value_size, + param_value, + param_value_size_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clTerminateContextKHR( + cl_context context) CL_API_SUFFIX__VERSION_2_0 +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(context, CL_INVALID_CONTEXT); + return context->dispatch->clTerminateContextKHR(context); +} + +CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromEGLImageKHR( + cl_context context, + CLeglDisplayKHR display, + CLeglImageKHR image, + cl_mem_flags flags, + const cl_egl_image_properties_khr *properties, + cl_int *errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateFromEGLImageKHR( + context, + display, + image, + flags, + properties, + errcode_ret); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireEGLObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueAcquireEGLObjectsKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseEGLObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_ERROR(command_queue, CL_INVALID_COMMAND_QUEUE); + return command_queue->dispatch->clEnqueueReleaseEGLObjectsKHR( + command_queue, + num_objects, + mem_objects, + num_events_in_wait_list, + event_wait_list, + event); +} + +/* cl_khr_egl_event */ +CL_API_ENTRY cl_event CL_API_CALL +clCreateEventFromEGLSyncKHR( + cl_context context, + CLeglSyncKHR sync, + CLeglDisplayKHR display, + cl_int *errcode_ret) +{ + KHR_ICD_VALIDATE_HANDLE_RETURN_HANDLE(context, CL_INVALID_CONTEXT); + return context->dispatch->clCreateEventFromEGLSyncKHR( + context, + sync, + display, + errcode_ret); +} + diff --git a/opencl/api/opencl/khronos/icd/icd_dispatch.h b/opencl/api/opencl/khronos/icd/icd_dispatch.h new file mode 100644 index 0000000000..023f4427e8 --- /dev/null +++ b/opencl/api/opencl/khronos/icd/icd_dispatch.h @@ -0,0 +1,1411 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#ifndef _ICD_DISPATCH_H_ +#define _ICD_DISPATCH_H_ + +#ifndef CL_USE_DEPRECATED_OPENCL_1_0_APIS +#define CL_USE_DEPRECATED_OPENCL_1_0_APIS +#endif + +#ifndef CL_USE_DEPRECATED_OPENCL_1_1_APIS +#define CL_USE_DEPRECATED_OPENCL_1_1_APIS +#endif + +// cl.h +#include + +// cl_gl.h and required files +#ifdef _WIN32 +#include +#include +#include +#include +#include +#include +#endif +#include +#include +#include +#include +#include + +/* + * + * function pointer typedefs + * + */ + +// Platform APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetPlatformIDs)( + cl_uint num_entries, + cl_platform_id * platforms, + cl_uint * num_platforms) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetPlatformInfo)( + cl_platform_id platform, + cl_platform_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +// Device APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetDeviceIDs)( + cl_platform_id platform, + cl_device_type device_type, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetDeviceInfo)( + cl_device_id device, + cl_device_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clCreateSubDevices)( + cl_device_id in_device, + const cl_device_partition_property * partition_properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices); + +typedef CL_API_ENTRY cl_int (CL_API_CALL * KHRpfn_clRetainDevice)( + cl_device_id device) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL * KHRpfn_clReleaseDevice)( + cl_device_id device) CL_API_SUFFIX__VERSION_1_2; + +// Context APIs +typedef CL_API_ENTRY cl_context (CL_API_CALL *KHRpfn_clCreateContext)( + const cl_context_properties * properties, + cl_uint num_devices, + const cl_device_id * devices, + void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *), + void * user_data, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_context (CL_API_CALL *KHRpfn_clCreateContextFromType)( + const cl_context_properties * properties, + cl_device_type device_type, + void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *), + void * user_data, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainContext)( + cl_context context) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseContext)( + cl_context context) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetContextInfo)( + cl_context context, + cl_context_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +// Command Queue APIs +typedef CL_API_ENTRY cl_command_queue (CL_API_CALL *KHRpfn_clCreateCommandQueue)( + cl_context context, + cl_device_id device, + cl_command_queue_properties properties, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_command_queue (CL_API_CALL *KHRpfn_clCreateCommandQueueWithProperties)( + cl_context /* context */, + cl_device_id /* device */, + const cl_queue_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainCommandQueue)( + cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseCommandQueue)( + cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetCommandQueueInfo)( + cl_command_queue command_queue, + cl_command_queue_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +// Memory Object APIs +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateBuffer)( + cl_context context, + cl_mem_flags flags, + size_t size, + void * host_ptr, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateImage)( + cl_context context, + cl_mem_flags flags, + const cl_image_format * image_format, + const cl_image_desc * image_desc, + void * host_ptr, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainMemObject)(cl_mem memobj) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseMemObject)(cl_mem memobj) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetSupportedImageFormats)( + cl_context context, + cl_mem_flags flags, + cl_mem_object_type image_type, + cl_uint num_entries, + cl_image_format * image_formats, + cl_uint * num_image_formats) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetMemObjectInfo)( + cl_mem memobj, + cl_mem_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetImageInfo)( + cl_mem image, + cl_image_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreatePipe)( + cl_context /* context */, + cl_mem_flags /* flags */, + cl_uint /* pipe_packet_size */, + cl_uint /* pipe_max_packets */, + const cl_pipe_properties * /* properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetPipeInfo)( + cl_mem /* pipe */, + cl_pipe_info /* param_name */, + size_t /* param_value_size */, + void * /* param_value */, + size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY void * (CL_API_CALL *KHRpfn_clSVMAlloc)( + cl_context /* context */, + cl_svm_mem_flags /* flags */, + size_t /* size */, + unsigned int /* alignment */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY void (CL_API_CALL *KHRpfn_clSVMFree)( + cl_context /* context */, + void * /* svm_pointer */) CL_API_SUFFIX__VERSION_2_0; + +// Sampler APIs +typedef CL_API_ENTRY cl_sampler (CL_API_CALL *KHRpfn_clCreateSampler)( + cl_context context, + cl_bool normalized_coords, + cl_addressing_mode addressing_mode, + cl_filter_mode filter_mode, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainSampler)(cl_sampler sampler) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseSampler)(cl_sampler sampler) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetSamplerInfo)( + cl_sampler sampler, + cl_sampler_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_sampler (CL_API_CALL *KHRpfn_clCreateSamplerWithProperties)( + cl_context /* context */, + const cl_sampler_properties * /* sampler_properties */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_2_0; + +// Program Object APIs +typedef CL_API_ENTRY cl_program (CL_API_CALL *KHRpfn_clCreateProgramWithSource)( + cl_context context, + cl_uint count, + const char ** strings, + const size_t * lengths, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_program (CL_API_CALL *KHRpfn_clCreateProgramWithBinary)( + cl_context context, + cl_uint num_devices, + const cl_device_id * device_list, + const size_t * lengths, + const unsigned char ** binaries, + cl_int * binary_status, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_program (CL_API_CALL *KHRpfn_clCreateProgramWithBuiltInKernels)( + cl_context context, + cl_uint num_devices, + const cl_device_id * device_list, + const char * kernel_names, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainProgram)(cl_program program) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseProgram)(cl_program program) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clBuildProgram)( + cl_program program, + cl_uint num_devices, + const cl_device_id * device_list, + const char * options, + void (CL_CALLBACK *pfn_notify)(cl_program program, void * user_data), + void * user_data) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clCompileProgram)( + cl_program program, + cl_uint num_devices, + const cl_device_id * device_list, + const char * options, + cl_uint num_input_headers, + const cl_program * input_headers, + const char ** header_include_names, + void (CL_CALLBACK * pfn_notify)(cl_program program, void * user_data), + void * user_data) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_program (CL_API_CALL *KHRpfn_clLinkProgram)( + cl_context context, + cl_uint num_devices, + const cl_device_id * device_list, + const char * options, + cl_uint num_input_programs, + const cl_program * input_programs, + void (CL_CALLBACK * pfn_notify)(cl_program program, void * user_data), + void * user_data, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clUnloadPlatformCompiler)( + cl_platform_id platform) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetProgramInfo)( + cl_program program, + cl_program_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetProgramBuildInfo)( + cl_program program, + cl_device_id device, + cl_program_build_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +// Kernel Object APIs +typedef CL_API_ENTRY cl_kernel (CL_API_CALL *KHRpfn_clCreateKernel)( + cl_program program, + const char * kernel_name, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clCreateKernelsInProgram)( + cl_program program, + cl_uint num_kernels, + cl_kernel * kernels, + cl_uint * num_kernels_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainKernel)(cl_kernel kernel) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseKernel)(cl_kernel kernel) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetKernelArg)( + cl_kernel kernel, + cl_uint arg_index, + size_t arg_size, + const void * arg_value) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetKernelInfo)( + cl_kernel kernel, + cl_kernel_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetKernelArgInfo)( + cl_kernel kernel, + cl_uint arg_indx, + cl_kernel_arg_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetKernelWorkGroupInfo)( + cl_kernel kernel, + cl_device_id device, + cl_kernel_work_group_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetKernelArgSVMPointer)( + cl_kernel /* kernel */, + cl_uint /* arg_index */, + const void * /* arg_value */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetKernelExecInfo)( + cl_kernel /* kernel */, + cl_kernel_exec_info /* param_name */, + size_t /* param_value_size */, + const void * /* param_value */) CL_API_SUFFIX__VERSION_2_0; + +// Event Object APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clWaitForEvents)( + cl_uint num_events, + const cl_event * event_list) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetEventInfo)( + cl_event event, + cl_event_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clRetainEvent)(cl_event event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clReleaseEvent)(cl_event event) CL_API_SUFFIX__VERSION_1_0; + +// Profiling APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetEventProfilingInfo)( + cl_event event, + cl_profiling_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +// Flush and Finish APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clFlush)(cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clFinish)(cl_command_queue command_queue) CL_API_SUFFIX__VERSION_1_0; + +// Enqueued Commands APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReadBuffer)( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + size_t offset, + size_t cb, + void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReadBufferRect)( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + const size_t * buffer_origin, + const size_t * host_origin, + const size_t * region, + size_t buffer_row_pitch, + size_t buffer_slice_pitch, + size_t host_row_pitch, + size_t host_slice_pitch, + void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueWriteBuffer)( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_write, + size_t offset, + size_t cb, + const void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueWriteBufferRect)( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_read, + const size_t * buffer_origin, + const size_t * host_origin, + const size_t * region, + size_t buffer_row_pitch, + size_t buffer_slice_pitch, + size_t host_row_pitch, + size_t host_slice_pitch, + const void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueFillBuffer)( + cl_command_queue command_queue, + cl_mem buffer, + const void * pattern, + size_t pattern_size, + size_t offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueCopyBuffer)( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_buffer, + size_t src_offset, + size_t dst_offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueCopyBufferRect)( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_buffer, + const size_t * src_origin, + const size_t * dst_origin, + const size_t * region, + size_t src_row_pitch, + size_t src_slice_pitch, + size_t dst_row_pitch, + size_t dst_slice_pitch, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReadImage)( + cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_read, + const size_t * origin, + const size_t * region, + size_t row_pitch, + size_t slice_pitch, + void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueWriteImage)( + cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_write, + const size_t * origin, + const size_t * region, + size_t input_row_pitch, + size_t input_slice_pitch, + const void * ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueFillImage)( + cl_command_queue command_queue, + cl_mem image, + const void * fill_color, + const size_t origin[3], + const size_t region[3], + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueCopyImage)( + cl_command_queue command_queue, + cl_mem src_image, + cl_mem dst_image, + const size_t * src_origin, + const size_t * dst_origin, + const size_t * region, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueCopyImageToBuffer)( + cl_command_queue command_queue, + cl_mem src_image, + cl_mem dst_buffer, + const size_t * src_origin, + const size_t * region, + size_t dst_offset, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueCopyBufferToImage)( + cl_command_queue command_queue, + cl_mem src_buffer, + cl_mem dst_image, + size_t src_offset, + const size_t * dst_origin, + const size_t * region, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY void * (CL_API_CALL *KHRpfn_clEnqueueMapBuffer)( + cl_command_queue command_queue, + cl_mem buffer, + cl_bool blocking_map, + cl_map_flags map_flags, + size_t offset, + size_t cb, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY void * (CL_API_CALL *KHRpfn_clEnqueueMapImage)( + cl_command_queue command_queue, + cl_mem image, + cl_bool blocking_map, + cl_map_flags map_flags, + const size_t * origin, + const size_t * region, + size_t * image_row_pitch, + size_t * image_slice_pitch, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueUnmapMemObject)( + cl_command_queue command_queue, + cl_mem memobj, + void * mapped_ptr, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueMigrateMemObjects)( + cl_command_queue command_queue, + cl_uint num_mem_objects, + const cl_mem * mem_objects, + cl_mem_migration_flags flags, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueNDRangeKernel)( + cl_command_queue command_queue, + cl_kernel kernel, + cl_uint work_dim, + const size_t * global_work_offset, + const size_t * global_work_size, + const size_t * local_work_size, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueTask)( + cl_command_queue command_queue, + cl_kernel kernel, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueNativeKernel)( + cl_command_queue command_queue, + void (CL_CALLBACK * user_func)(void *), + void * args, + size_t cb_args, + cl_uint num_mem_objects, + const cl_mem * mem_list, + const void ** args_mem_loc, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueMarkerWithWaitList)( + cl_command_queue command_queue, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueBarrierWithWaitList)( + cl_command_queue command_queue, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY void * (CL_API_CALL *KHRpfn_clGetExtensionFunctionAddressForPlatform)( + cl_platform_id platform, + const char * function_name) CL_API_SUFFIX__VERSION_1_2; + +// Shared Virtual Memory APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueSVMFree)( + cl_command_queue /* command_queue */, + cl_uint /* num_svm_pointers */, + void ** /* svm_pointers */, + void (CL_CALLBACK *pfn_free_func)( + cl_command_queue /* queue */, + cl_uint /* num_svm_pointers */, + void ** /* svm_pointers[] */, + void * /* user_data */), + void * /* user_data */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueSVMMemcpy)( + cl_command_queue /* command_queue */, + cl_bool /* blocking_copy */, + void * /* dst_ptr */, + const void * /* src_ptr */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueSVMMemFill)( + cl_command_queue /* command_queue */, + void * /* svm_ptr */, + const void * /* pattern */, + size_t /* pattern_size */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueSVMMap)( + cl_command_queue /* command_queue */, + cl_bool /* blocking_map */, + cl_map_flags /* map_flags */, + void * /* svm_ptr */, + size_t /* size */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueSVMUnmap)( + cl_command_queue /* command_queue */, + void * /* svm_ptr */, + cl_uint /* num_events_in_wait_list */, + const cl_event * /* event_wait_list */, + cl_event * /* event */) CL_API_SUFFIX__VERSION_2_0; + +// Deprecated APIs +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetCommandQueueProperty)( + cl_command_queue command_queue, + cl_command_queue_properties properties, + cl_bool enable, + cl_command_queue_properties * old_properties) /*CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED*/; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateImage2D)( + cl_context context, + cl_mem_flags flags, + const cl_image_format * image_format, + size_t image_width, + size_t image_height, + size_t image_row_pitch, + void * host_ptr, + cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateImage3D)( + cl_context context, + cl_mem_flags flags, + const cl_image_format * image_format, + size_t image_width, + size_t image_height, + size_t image_depth, + size_t image_row_pitch, + size_t image_slice_pitch, + void * host_ptr, + cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clUnloadCompiler)(void) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueMarker)( + cl_command_queue command_queue, + cl_event * event) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueWaitForEvents)( + cl_command_queue command_queue, + cl_uint num_events, + const cl_event * event_list) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueBarrier)(cl_command_queue command_queue) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +typedef CL_API_ENTRY void * (CL_API_CALL *KHRpfn_clGetExtensionFunctionAddress)(const char *function_name) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; + +// GL and other APIs +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromGLBuffer)( + cl_context context, + cl_mem_flags flags, + GLuint bufobj, + int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromGLTexture)( + cl_context context, + cl_mem_flags flags, + cl_GLenum target, + cl_GLint miplevel, + cl_GLuint texture, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromGLTexture2D)( + cl_context context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texture, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromGLTexture3D)( + cl_context context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texture, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromGLRenderbuffer)( + cl_context context, + cl_mem_flags flags, + GLuint renderbuffer, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetGLObjectInfo)( + cl_mem memobj, + cl_gl_object_type * gl_object_type, + GLuint * gl_object_name) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetGLTextureInfo)( + cl_mem memobj, + cl_gl_texture_info param_name, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueAcquireGLObjects)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReleaseGLObjects)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +/* cl_khr_gl_sharing */ +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetGLContextInfoKHR)( + const cl_context_properties *properties, + cl_gl_context_info param_name, + size_t param_value_size, + void *param_value, + size_t *param_value_size_ret); + +/* cl_khr_gl_event */ +typedef CL_API_ENTRY cl_event (CL_API_CALL *KHRpfn_clCreateEventFromGLsyncKHR)( + cl_context context, + cl_GLsync sync, + cl_int *errcode_ret); + + +#if defined(_WIN32) + +/* cl_khr_d3d10_sharing */ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetDeviceIDsFromD3D10KHR)( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromD3D10BufferKHR)( + cl_context context, + cl_mem_flags flags, + ID3D10Buffer * resource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromD3D10Texture2DKHR)( + cl_context context, + cl_mem_flags flags, + ID3D10Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromD3D10Texture3DKHR)( + cl_context context, + cl_mem_flags flags, + ID3D10Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueAcquireD3D10ObjectsKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReleaseD3D10ObjectsKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_0; + +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromD3D10KHR( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void *d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id *devices, + cl_uint *num_devices); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10BufferKHR( + cl_context context, + cl_mem_flags flags, + ID3D10Buffer *resource, + cl_int *errcode_ret); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10Texture2DKHR( + cl_context context, + cl_mem_flags flags, + ID3D10Texture2D * resource, + UINT subresource, + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D10Texture3DKHR( + cl_context context, + cl_mem_flags flags, + ID3D10Texture3D *resource, + UINT subresource, + cl_int *errcode_ret); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireD3D10ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseD3D10ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event); + +/* cl_khr_d3d11_sharing */ +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetDeviceIDsFromD3D11KHR)( + cl_platform_id platform, + cl_d3d11_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d11_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromD3D11BufferKHR)( + cl_context context, + cl_mem_flags flags, + ID3D11Buffer * resource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromD3D11Texture2DKHR)( + cl_context context, + cl_mem_flags flags, + ID3D11Texture2D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromD3D11Texture3DKHR)( + cl_context context, + cl_mem_flags flags, + ID3D11Texture3D * resource, + UINT subresource, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueAcquireD3D11ObjectsKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReleaseD3D11ObjectsKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +/* cl_khr_dx9_media_sharing */ +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clGetDeviceIDsFromDX9MediaAdapterKHR)( + cl_platform_id platform, + cl_uint num_media_adapters, + cl_dx9_media_adapter_type_khr * media_adapters_type, + void * media_adapters, + cl_dx9_media_adapter_set_khr media_adapter_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromDX9MediaSurfaceKHR)( + cl_context context, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + void * surface_info, + cl_uint plane, + cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueAcquireDX9MediaSurfacesKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReleaseDX9MediaSurfacesKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event) CL_API_SUFFIX__VERSION_1_2; + +/* cl_khr_d3d11_sharing */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromD3D11KHR( + cl_platform_id platform, + cl_d3d11_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d11_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11BufferKHR( + cl_context context, + cl_mem_flags flags, + ID3D11Buffer * resource, + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11Texture2DKHR( + cl_context context, + cl_mem_flags flags, + ID3D11Texture2D * resource, + UINT subresource, + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromD3D11Texture3DKHR( + cl_context context, + cl_mem_flags flags, + ID3D11Texture3D * resource, + UINT subresource, + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireD3D11ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseD3D11ObjectsKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event); + +/* cl_khr_dx9_media_sharing */ +extern CL_API_ENTRY cl_int CL_API_CALL +clGetDeviceIDsFromDX9MediaAdapterKHR( + cl_platform_id platform, + cl_uint num_media_adapters, + cl_dx9_media_adapter_type_khr * media_adapter_type, + void * media_adapters, + cl_dx9_media_adapter_set_khr media_adapter_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint * num_devices); + +extern CL_API_ENTRY cl_mem CL_API_CALL +clCreateFromDX9MediaSurfaceKHR( + cl_context context, + cl_mem_flags flags, + cl_dx9_media_adapter_type_khr adapter_type, + void * surface_info, + cl_uint plane, + cl_int * errcode_ret); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueAcquireDX9MediaSurfacesKHR( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event); + +extern CL_API_ENTRY cl_int CL_API_CALL +clEnqueueReleaseDX9MediaSurfacesKHR( + cl_command_queue command_queue, + cl_uint num_objects, + cl_mem * mem_objects, + cl_uint num_events_in_wait_list, + const cl_event * event_wait_list, + cl_event * event); + +#else + +/* cl_khr_d3d10_sharing */ +typedef void *KHRpfn_clGetDeviceIDsFromD3D10KHR; +typedef void *KHRpfn_clCreateFromD3D10BufferKHR; +typedef void *KHRpfn_clCreateFromD3D10Texture2DKHR; +typedef void *KHRpfn_clCreateFromD3D10Texture3DKHR; +typedef void *KHRpfn_clEnqueueAcquireD3D10ObjectsKHR; +typedef void *KHRpfn_clEnqueueReleaseD3D10ObjectsKHR; + +/* cl_khr_d3d11_sharing */ +typedef void *KHRpfn_clGetDeviceIDsFromD3D11KHR; +typedef void *KHRpfn_clCreateFromD3D11BufferKHR; +typedef void *KHRpfn_clCreateFromD3D11Texture2DKHR; +typedef void *KHRpfn_clCreateFromD3D11Texture3DKHR; +typedef void *KHRpfn_clEnqueueAcquireD3D11ObjectsKHR; +typedef void *KHRpfn_clEnqueueReleaseD3D11ObjectsKHR; + +/* cl_khr_dx9_media_sharing */ +typedef void *KHRpfn_clCreateFromDX9MediaSurfaceKHR; +typedef void *KHRpfn_clEnqueueAcquireDX9MediaSurfacesKHR; +typedef void *KHRpfn_clEnqueueReleaseDX9MediaSurfacesKHR; +typedef void *KHRpfn_clGetDeviceIDsFromDX9MediaAdapterKHR; + +#endif + +/* OpenCL 1.1 */ + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetEventCallback)( + cl_event /* event */, + cl_int /* command_exec_callback_type */, + void (CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *), + void * /* user_data */) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateSubBuffer)( + cl_mem /* buffer */, + cl_mem_flags /* flags */, + cl_buffer_create_type /* buffer_create_type */, + const void * /* buffer_create_info */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetMemObjectDestructorCallback)( + cl_mem /* memobj */, + void (CL_CALLBACK * /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), + void * /*user_data */ ) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_event (CL_API_CALL *KHRpfn_clCreateUserEvent)( + cl_context /* context */, + cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clSetUserEventStatus)( + cl_event /* event */, + cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1; + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clCreateSubDevicesEXT)( + cl_device_id in_device, + const cl_device_partition_property_ext * partition_properties, + cl_uint num_entries, + cl_device_id * out_devices, + cl_uint * num_devices); + +typedef CL_API_ENTRY cl_int (CL_API_CALL * KHRpfn_clRetainDeviceEXT)( + cl_device_id device) CL_API_SUFFIX__VERSION_1_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL * KHRpfn_clReleaseDeviceEXT)( + cl_device_id device) CL_API_SUFFIX__VERSION_1_0; + +/* cl_khr_egl_image */ +typedef CL_API_ENTRY cl_mem (CL_API_CALL *KHRpfn_clCreateFromEGLImageKHR)( + cl_context context, + CLeglDisplayKHR display, + CLeglImageKHR image, + cl_mem_flags flags, + const cl_egl_image_properties_khr *properties, + cl_int *errcode_ret); + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueAcquireEGLObjectsKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event); + +typedef CL_API_ENTRY cl_int (CL_API_CALL *KHRpfn_clEnqueueReleaseEGLObjectsKHR)( + cl_command_queue command_queue, + cl_uint num_objects, + const cl_mem *mem_objects, + cl_uint num_events_in_wait_list, + const cl_event *event_wait_list, + cl_event *event); + +/* cl_khr_egl_event */ +typedef CL_API_ENTRY cl_event (CL_API_CALL *KHRpfn_clCreateEventFromEGLSyncKHR)( + cl_context context, + CLeglSyncKHR sync, + CLeglDisplayKHR display, + cl_int *errcode_ret); + +typedef CL_API_ENTRY cl_int (CL_API_CALL * KHRpfn_clGetKernelSubGroupInfoKHR)( + cl_kernel kernel, + cl_device_id device, + cl_kernel_sub_group_info param_name, + size_t input_value_size, + const void * input_value, + size_t param_value_size, + void * param_value, + size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_2_0; + +typedef CL_API_ENTRY cl_int (CL_API_CALL * KHRpfn_clTerminateContextKHR)( + cl_context context) CL_API_SUFFIX__VERSION_2_0; + + +/* + * + * vendor dispatch table structure + * + * note that the types in the structure KHRicdVendorDispatch mirror the function + * names listed in the string table khrIcdVendorDispatchFunctionNames + * + */ + +typedef struct KHRicdVendorDispatchRec KHRicdVendorDispatch; + +struct KHRicdVendorDispatchRec +{ + /* OpenCL 1.0 */ + KHRpfn_clGetPlatformIDs clGetPlatformIDs; + KHRpfn_clGetPlatformInfo clGetPlatformInfo; + KHRpfn_clGetDeviceIDs clGetDeviceIDs; + KHRpfn_clGetDeviceInfo clGetDeviceInfo; + KHRpfn_clCreateContext clCreateContext; + KHRpfn_clCreateContextFromType clCreateContextFromType; + KHRpfn_clRetainContext clRetainContext; + KHRpfn_clReleaseContext clReleaseContext; + KHRpfn_clGetContextInfo clGetContextInfo; + KHRpfn_clCreateCommandQueue clCreateCommandQueue; + KHRpfn_clRetainCommandQueue clRetainCommandQueue; + KHRpfn_clReleaseCommandQueue clReleaseCommandQueue; + KHRpfn_clGetCommandQueueInfo clGetCommandQueueInfo; + KHRpfn_clSetCommandQueueProperty clSetCommandQueueProperty; + KHRpfn_clCreateBuffer clCreateBuffer; + KHRpfn_clCreateImage2D clCreateImage2D; + KHRpfn_clCreateImage3D clCreateImage3D; + KHRpfn_clRetainMemObject clRetainMemObject; + KHRpfn_clReleaseMemObject clReleaseMemObject; + KHRpfn_clGetSupportedImageFormats clGetSupportedImageFormats; + KHRpfn_clGetMemObjectInfo clGetMemObjectInfo; + KHRpfn_clGetImageInfo clGetImageInfo; + KHRpfn_clCreateSampler clCreateSampler; + KHRpfn_clRetainSampler clRetainSampler; + KHRpfn_clReleaseSampler clReleaseSampler; + KHRpfn_clGetSamplerInfo clGetSamplerInfo; + KHRpfn_clCreateProgramWithSource clCreateProgramWithSource; + KHRpfn_clCreateProgramWithBinary clCreateProgramWithBinary; + KHRpfn_clRetainProgram clRetainProgram; + KHRpfn_clReleaseProgram clReleaseProgram; + KHRpfn_clBuildProgram clBuildProgram; + KHRpfn_clUnloadCompiler clUnloadCompiler; + KHRpfn_clGetProgramInfo clGetProgramInfo; + KHRpfn_clGetProgramBuildInfo clGetProgramBuildInfo; + KHRpfn_clCreateKernel clCreateKernel; + KHRpfn_clCreateKernelsInProgram clCreateKernelsInProgram; + KHRpfn_clRetainKernel clRetainKernel; + KHRpfn_clReleaseKernel clReleaseKernel; + KHRpfn_clSetKernelArg clSetKernelArg; + KHRpfn_clGetKernelInfo clGetKernelInfo; + KHRpfn_clGetKernelWorkGroupInfo clGetKernelWorkGroupInfo; + KHRpfn_clWaitForEvents clWaitForEvents; + KHRpfn_clGetEventInfo clGetEventInfo; + KHRpfn_clRetainEvent clRetainEvent; + KHRpfn_clReleaseEvent clReleaseEvent; + KHRpfn_clGetEventProfilingInfo clGetEventProfilingInfo; + KHRpfn_clFlush clFlush; + KHRpfn_clFinish clFinish; + KHRpfn_clEnqueueReadBuffer clEnqueueReadBuffer; + KHRpfn_clEnqueueWriteBuffer clEnqueueWriteBuffer; + KHRpfn_clEnqueueCopyBuffer clEnqueueCopyBuffer; + KHRpfn_clEnqueueReadImage clEnqueueReadImage; + KHRpfn_clEnqueueWriteImage clEnqueueWriteImage; + KHRpfn_clEnqueueCopyImage clEnqueueCopyImage; + KHRpfn_clEnqueueCopyImageToBuffer clEnqueueCopyImageToBuffer; + KHRpfn_clEnqueueCopyBufferToImage clEnqueueCopyBufferToImage; + KHRpfn_clEnqueueMapBuffer clEnqueueMapBuffer; + KHRpfn_clEnqueueMapImage clEnqueueMapImage; + KHRpfn_clEnqueueUnmapMemObject clEnqueueUnmapMemObject; + KHRpfn_clEnqueueNDRangeKernel clEnqueueNDRangeKernel; + KHRpfn_clEnqueueTask clEnqueueTask; + KHRpfn_clEnqueueNativeKernel clEnqueueNativeKernel; + KHRpfn_clEnqueueMarker clEnqueueMarker; + KHRpfn_clEnqueueWaitForEvents clEnqueueWaitForEvents; + KHRpfn_clEnqueueBarrier clEnqueueBarrier; + KHRpfn_clGetExtensionFunctionAddress clGetExtensionFunctionAddress; + KHRpfn_clCreateFromGLBuffer clCreateFromGLBuffer; + KHRpfn_clCreateFromGLTexture2D clCreateFromGLTexture2D; + KHRpfn_clCreateFromGLTexture3D clCreateFromGLTexture3D; + KHRpfn_clCreateFromGLRenderbuffer clCreateFromGLRenderbuffer; + KHRpfn_clGetGLObjectInfo clGetGLObjectInfo; + KHRpfn_clGetGLTextureInfo clGetGLTextureInfo; + KHRpfn_clEnqueueAcquireGLObjects clEnqueueAcquireGLObjects; + KHRpfn_clEnqueueReleaseGLObjects clEnqueueReleaseGLObjects; + KHRpfn_clGetGLContextInfoKHR clGetGLContextInfoKHR; + + /* cl_khr_d3d10_sharing */ + KHRpfn_clGetDeviceIDsFromD3D10KHR clGetDeviceIDsFromD3D10KHR; + KHRpfn_clCreateFromD3D10BufferKHR clCreateFromD3D10BufferKHR; + KHRpfn_clCreateFromD3D10Texture2DKHR clCreateFromD3D10Texture2DKHR; + KHRpfn_clCreateFromD3D10Texture3DKHR clCreateFromD3D10Texture3DKHR; + KHRpfn_clEnqueueAcquireD3D10ObjectsKHR clEnqueueAcquireD3D10ObjectsKHR; + KHRpfn_clEnqueueReleaseD3D10ObjectsKHR clEnqueueReleaseD3D10ObjectsKHR; + + /* OpenCL 1.1 */ + KHRpfn_clSetEventCallback clSetEventCallback; + KHRpfn_clCreateSubBuffer clCreateSubBuffer; + KHRpfn_clSetMemObjectDestructorCallback clSetMemObjectDestructorCallback; + KHRpfn_clCreateUserEvent clCreateUserEvent; + KHRpfn_clSetUserEventStatus clSetUserEventStatus; + KHRpfn_clEnqueueReadBufferRect clEnqueueReadBufferRect; + KHRpfn_clEnqueueWriteBufferRect clEnqueueWriteBufferRect; + KHRpfn_clEnqueueCopyBufferRect clEnqueueCopyBufferRect; + + /* cl_ext_device_fission */ + KHRpfn_clCreateSubDevicesEXT clCreateSubDevicesEXT; + KHRpfn_clRetainDeviceEXT clRetainDeviceEXT; + KHRpfn_clReleaseDeviceEXT clReleaseDeviceEXT; + + /* cl_khr_gl_event */ + KHRpfn_clCreateEventFromGLsyncKHR clCreateEventFromGLsyncKHR; + + /* OpenCL 1.2 */ + KHRpfn_clCreateSubDevices clCreateSubDevices; + KHRpfn_clRetainDevice clRetainDevice; + KHRpfn_clReleaseDevice clReleaseDevice; + KHRpfn_clCreateImage clCreateImage; + KHRpfn_clCreateProgramWithBuiltInKernels clCreateProgramWithBuiltInKernels; + KHRpfn_clCompileProgram clCompileProgram; + KHRpfn_clLinkProgram clLinkProgram; + KHRpfn_clUnloadPlatformCompiler clUnloadPlatformCompiler; + KHRpfn_clGetKernelArgInfo clGetKernelArgInfo; + KHRpfn_clEnqueueFillBuffer clEnqueueFillBuffer; + KHRpfn_clEnqueueFillImage clEnqueueFillImage; + KHRpfn_clEnqueueMigrateMemObjects clEnqueueMigrateMemObjects; + KHRpfn_clEnqueueMarkerWithWaitList clEnqueueMarkerWithWaitList; + KHRpfn_clEnqueueBarrierWithWaitList clEnqueueBarrierWithWaitList; + KHRpfn_clGetExtensionFunctionAddressForPlatform clGetExtensionFunctionAddressForPlatform; + KHRpfn_clCreateFromGLTexture clCreateFromGLTexture; + + /* cl_khr_d3d11_sharing */ + KHRpfn_clGetDeviceIDsFromD3D11KHR clGetDeviceIDsFromD3D11KHR; + KHRpfn_clCreateFromD3D11BufferKHR clCreateFromD3D11BufferKHR; + KHRpfn_clCreateFromD3D11Texture2DKHR clCreateFromD3D11Texture2DKHR; + KHRpfn_clCreateFromD3D11Texture3DKHR clCreateFromD3D11Texture3DKHR; + KHRpfn_clCreateFromDX9MediaSurfaceKHR clCreateFromDX9MediaSurfaceKHR; + KHRpfn_clEnqueueAcquireD3D11ObjectsKHR clEnqueueAcquireD3D11ObjectsKHR; + KHRpfn_clEnqueueReleaseD3D11ObjectsKHR clEnqueueReleaseD3D11ObjectsKHR; + + /* cl_khr_dx9_media_sharing */ + KHRpfn_clGetDeviceIDsFromDX9MediaAdapterKHR clGetDeviceIDsFromDX9MediaAdapterKHR; + KHRpfn_clEnqueueAcquireDX9MediaSurfacesKHR clEnqueueAcquireDX9MediaSurfacesKHR; + KHRpfn_clEnqueueReleaseDX9MediaSurfacesKHR clEnqueueReleaseDX9MediaSurfacesKHR; + + /* cl_khr_egl_image */ + KHRpfn_clCreateFromEGLImageKHR clCreateFromEGLImageKHR; + KHRpfn_clEnqueueAcquireEGLObjectsKHR clEnqueueAcquireEGLObjectsKHR; + KHRpfn_clEnqueueReleaseEGLObjectsKHR clEnqueueReleaseEGLObjectsKHR; + + /* cl_khr_egl_event */ + KHRpfn_clCreateEventFromEGLSyncKHR clCreateEventFromEGLSyncKHR; + + /* OpenCL 2.0 */ + KHRpfn_clCreateCommandQueueWithProperties clCreateCommandQueueWithProperties; + KHRpfn_clCreatePipe clCreatePipe; + KHRpfn_clGetPipeInfo clGetPipeInfo; + KHRpfn_clSVMAlloc clSVMAlloc; + KHRpfn_clSVMFree clSVMFree; + KHRpfn_clEnqueueSVMFree clEnqueueSVMFree; + KHRpfn_clEnqueueSVMMemcpy clEnqueueSVMMemcpy; + KHRpfn_clEnqueueSVMMemFill clEnqueueSVMMemFill; + KHRpfn_clEnqueueSVMMap clEnqueueSVMMap; + KHRpfn_clEnqueueSVMUnmap clEnqueueSVMUnmap; + KHRpfn_clCreateSamplerWithProperties clCreateSamplerWithProperties; + KHRpfn_clSetKernelArgSVMPointer clSetKernelArgSVMPointer; + KHRpfn_clSetKernelExecInfo clSetKernelExecInfo; + + KHRpfn_clGetKernelSubGroupInfoKHR clGetKernelSubGroupInfoKHR; + KHRpfn_clTerminateContextKHR clTerminateContextKHR; +}; + +#endif // _ICD_DISPATCH_H_ + diff --git a/opencl/api/opencl/khronos/icd/icd_linux.c b/opencl/api/opencl/khronos/icd/icd_linux.c new file mode 100644 index 0000000000..cbee92da76 --- /dev/null +++ b/opencl/api/opencl/khronos/icd/icd_linux.c @@ -0,0 +1,191 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#include "icd.h" +#include +#include +#include +#include +#include +#include +#include + +/* + * + * Vendor enumeration functions + * + */ +static pthread_once_t InitOnce = PTHREAD_ONCE_INIT; + +// go through the list of vendors in the two configuration files +void khrIcdOsVendorsEnumerateCallBack(void) +{ + DIR *dir = NULL; + struct dirent *dirEntry = NULL; + char *vendorPath; + + vendorPath = getenv("OPENCL_VENDOR_PATH"); + if (vendorPath == NULL) + { + vendorPath = "/etc/OpenCL/vendors"; + } + + // open the directory + dir = opendir(vendorPath); + if (NULL == dir) + { + KHR_ICD_TRACE("Failed to open path %s\n", vendorPath); + goto Cleanup; + } + + // attempt to load all files in the directory + for (dirEntry = readdir(dir); dirEntry; dirEntry = readdir(dir) ) + { + switch(dirEntry->d_type) + { + case DT_UNKNOWN: + case DT_REG: + case DT_LNK: + { + const char* extension = ".icd"; + FILE *fin = NULL; + char* fileName = NULL; + char* buffer = NULL; + long bufferSize = 0; + + // make sure the file name ends in .icd + if (strlen(extension) > strlen(dirEntry->d_name) ) + { + break; + } + if (strcmp(dirEntry->d_name + strlen(dirEntry->d_name) - strlen(extension), extension) ) + { + break; + } + + // allocate space for the full path of the vendor library name + fileName = malloc(strlen(dirEntry->d_name) + 1 + strlen(vendorPath) + 1); + if (!fileName) + { + KHR_ICD_TRACE("Failed allocate space for ICD file path\n"); + break; + } + sprintf(fileName, "%s/%s", vendorPath, dirEntry->d_name); + + // open the file and read its contents + fin = fopen(fileName, "r"); + if (!fin) + { + free(fileName); + break; + } + fseek(fin, 0, SEEK_END); + bufferSize = ftell(fin); + + buffer = malloc(bufferSize+1); + if (!buffer) + { + free(fileName); + fclose(fin); + break; + } + memset(buffer, 0, bufferSize+1); + fseek(fin, 0, SEEK_SET); + if (bufferSize != (long)fread(buffer, 1, bufferSize, fin) ) + { + free(fileName); + free(buffer); + fclose(fin); + break; + } + // ignore a newline at the end of the file + if (buffer[bufferSize-1] == '\n') buffer[bufferSize-1] = '\0'; + + // load the string read from the file + khrIcdVendorAdd(buffer); + + free(fileName); + free(buffer); + fclose(fin); + } + break; + default: + break; + } + } + +Cleanup: + + // free resources and exit + if (dir) + { + closedir(dir); + } +} + +/* + * + * Dynamic library loading functions + * + */ + +// dynamically load a library. returns NULL on failure +void *khrIcdOsLibraryLoad(const char *libraryName) +{ + return dlopen (libraryName, RTLD_NOW); +} + +// get a function pointer from a loaded library. returns NULL on failure. +void *khrIcdOsLibraryGetFunctionAddress(void *library, const char *functionName) +{ + return dlsym(library, functionName); +} + +// unload a library +void khrIcdOsLibraryUnload(void *library) +{ + dlclose(library); +} + +cl_bool khrIcdOsVendorsEnumerate(void) +{ + pthread_once(&InitOnce, khrIcdOsVendorsEnumerateCallBack); + return CL_TRUE; +} diff --git a/opencl/api/opencl/khronos/icd/icd_windows.c b/opencl/api/opencl/khronos/icd/icd_windows.c new file mode 100644 index 0000000000..74020a14c0 --- /dev/null +++ b/opencl/api/opencl/khronos/icd/icd_windows.c @@ -0,0 +1,163 @@ +/* + * Copyright (c) 2012 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software source and associated documentation files (the "Materials"), + * to use, copy, modify and compile the Materials to create a binary under the + * following terms and conditions: + * + * 1. The Materials shall NOT be distributed to any third party; + * + * 2. The binary may be distributed without restriction, including without + * limitation the rights to use, copy, merge, publish, distribute, sublicense, + * and/or sell copies, and to permit persons to whom the binary is furnished to + * do so; + * + * 3. All modifications to the Materials used to create a binary that is + * distributed to third parties shall be provided to Khronos with an + * unrestricted license to use for the purposes of implementing bug fixes and + * enhancements to the Materials; + * + * 4. If the binary is used as part of an OpenCL(TM) implementation, whether + * binary is distributed together with or separately to that implementation, + * then recipient must become an OpenCL Adopter and follow the published OpenCL + * conformance process for that implementation, details at: + * http://www.khronos.org/conformance/; + * + * 5. The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE MATERIALS OR THE USE OR OTHER DEALINGS IN + * THE MATERIALS. + * + * OpenCL is a trademark of Apple Inc. used under license by Khronos. + */ + +#include "icd.h" +#include +#include +#include +#include + +/* + * + * Vendor enumeration functions + * + */ +static INIT_ONCE InitOnce = INIT_ONCE_STATIC_INIT; + +// go through the list of vendors in the registry and call khrIcdVendorAdd +// for each vendor encountered +BOOL CALLBACK khrIcdOsVendorsEnumerateCallBack( + _Inout_ PINIT_ONCE InitOnce, + _Inout_opt_ PVOID Parameter, + _Out_opt_ PVOID *Context) +{ + LONG result; + const char* platformsName = "SOFTWARE\\Khronos\\OpenCL\\Vendors"; + HKEY platformsKey = NULL; + DWORD dwIndex; + + KHR_ICD_TRACE("Opening key HKLM\\%s...\n", platformsName); + result = RegOpenKeyExA( + HKEY_LOCAL_MACHINE, + platformsName, + 0, + KEY_READ, + &platformsKey); + if (ERROR_SUCCESS != result) + { + KHR_ICD_TRACE("Failed to open platforms key %s, continuing\n", platformsName); + return FALSE; + } + + // for each value + for (dwIndex = 0;; ++dwIndex) + { + char cszLibraryName[1024] = {0}; + DWORD dwLibraryNameSize = sizeof(cszLibraryName); + DWORD dwLibraryNameType = 0; + DWORD dwValue = 0; + DWORD dwValueSize = sizeof(dwValue); + + // read the value name + KHR_ICD_TRACE("Reading value %d...\n", dwIndex); + result = RegEnumValueA( + platformsKey, + dwIndex, + cszLibraryName, + &dwLibraryNameSize, + NULL, + &dwLibraryNameType, + (LPBYTE)&dwValue, + &dwValueSize); + // if RegEnumKeyEx fails, we are done with the enumeration + if (ERROR_SUCCESS != result) + { + KHR_ICD_TRACE("Failed to read value %d, done reading key.\n", dwIndex); + break; + } + KHR_ICD_TRACE("Value %s found...\n", cszLibraryName); + + // Require that the value be a DWORD and equal zero + if (REG_DWORD != dwLibraryNameType) + { + KHR_ICD_TRACE("Value not a DWORD, skipping\n"); + continue; + } + if (dwValue) + { + KHR_ICD_TRACE("Value not zero, skipping\n"); + continue; + } + + // add the library + khrIcdVendorAdd(cszLibraryName); + } + + result = RegCloseKey(platformsKey); + if (ERROR_SUCCESS != result) + { + KHR_ICD_TRACE("Failed to close platforms key %s, ignoring\n", platformsName); + } + return TRUE; +} + +/* + * + * Dynamic library loading functions + * + */ + +// dynamically load a library. returns NULL on failure +void *khrIcdOsLibraryLoad(const char *libraryName) +{ + return (void *)LoadLibraryA(libraryName); +} + +// get a function pointer from a loaded library. returns NULL on failure. +void *khrIcdOsLibraryGetFunctionAddress(void *library, const char *functionName) +{ + if (!library || !functionName) + { + return NULL; + } + return GetProcAddress( (HMODULE)library, functionName); +} + +// unload a library. +void khrIcdOsLibraryUnload(void *library) +{ + FreeLibrary( (HMODULE)library); +} + +cl_bool khrIcdOsVendorsEnumerate(void) +{ + InitOnceExecuteOnce(&InitOnce, khrIcdOsVendorsEnumerateCallBack, NULL, NULL); + return CL_TRUE; +} diff --git a/opencl/init.txt b/opencl/init.txt deleted file mode 100644 index cef216937d..0000000000 --- a/opencl/init.txt +++ /dev/null @@ -1 +0,0 @@ -Fri Jul 4 09:40:50 EDT 2014 diff --git a/opencl/tools/clinfo/cl.hpp b/opencl/tools/clinfo/cl.hpp new file mode 100644 index 0000000000..e885099fc1 --- /dev/null +++ b/opencl/tools/clinfo/cl.hpp @@ -0,0 +1,12474 @@ +/******************************************************************************* + * Copyright (c) 2008-2013 The Khronos Group Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and/or associated documentation files (the + * "Materials"), to deal in the Materials without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Materials, and to + * permit persons to whom the Materials are furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Materials. + * + * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. + ******************************************************************************/ + +/*! \file + * + * \brief C++ bindings for OpenCL 1.0 (rev 48), OpenCL 1.1 (rev 33) and + * OpenCL 1.2 (rev 15) + * \author Benedict R. Gaster, Laurent Morichetti and Lee Howes + * + * Additions and fixes from: + * Brian Cole, March 3rd 2010 and April 2012 + * Matt Gruenke, April 2012. + * Bruce Merry, February 2013. + * Tom Deakin and Simon McIntosh-Smith, July 2013 + * + * \version 1.2.6 + * \date August 2013 + * + * Optional extension support + * + * cl + * cl_ext_device_fission + * #define USE_CL_DEVICE_FISSION + */ + +/*! \mainpage + * \section intro Introduction + * For many large applications C++ is the language of choice and so it seems + * reasonable to define C++ bindings for OpenCL. + * + * + * The interface is contained with a single C++ header file \em cl.hpp and all + * definitions are contained within the namespace \em cl. There is no additional + * requirement to include \em cl.h and to use either the C++ or original C + * bindings it is enough to simply include \em cl.hpp. + * + * The bindings themselves are lightweight and correspond closely to the + * underlying C API. Using the C++ bindings introduces no additional execution + * overhead. + * + * For detail documentation on the bindings see: + * + * The OpenCL C++ Wrapper API 1.2 (revision 09) + * http://www.khronos.org/registry/cl/specs/opencl-cplusplus-1.2.pdf + * + * \section example Example + * + * The following example shows a general use case for the C++ + * bindings, including support for the optional exception feature and + * also the supplied vector and string classes, see following sections for + * decriptions of these features. + * + * \code + * #define __CL_ENABLE_EXCEPTIONS + * + * #if defined(__APPLE__) || defined(__MACOSX) + * #include + * #else + * #include + * #endif + * #include + * #include + * #include + * + * const char * helloStr = "__kernel void " + * "hello(void) " + * "{ " + * " " + * "} "; + * + * int + * main(void) + * { + * cl_int err = CL_SUCCESS; + * try { + * + * std::vector platforms; + * cl::Platform::get(&platforms); + * if (platforms.size() == 0) { + * std::cout << "Platform size 0\n"; + * return -1; + * } + * + * cl_context_properties properties[] = + * { CL_CONTEXT_PLATFORM, (cl_context_properties)(platforms[0])(), 0}; + * cl::Context context(CL_DEVICE_TYPE_CPU, properties); + * + * std::vector devices = context.getInfo(); + * + * cl::Program::Sources source(1, + * std::make_pair(helloStr,strlen(helloStr))); + * cl::Program program_ = cl::Program(context, source); + * program_.build(devices); + * + * cl::Kernel kernel(program_, "hello", &err); + * + * cl::Event event; + * cl::CommandQueue queue(context, devices[0], 0, &err); + * queue.enqueueNDRangeKernel( + * kernel, + * cl::NullRange, + * cl::NDRange(4,4), + * cl::NullRange, + * NULL, + * &event); + * + * event.wait(); + * } + * catch (cl::Error err) { + * std::cerr + * << "ERROR: " + * << err.what() + * << "(" + * << err.err() + * << ")" + * << std::endl; + * } + * + * return EXIT_SUCCESS; + * } + * + * \endcode + * + */ +#ifndef CL_HPP_ +#define CL_HPP_ + +#ifdef _WIN32 + +#include +#include +#include +#include + +#if defined(__CL_ENABLE_EXCEPTIONS) +#include +#endif // #if defined(__CL_ENABLE_EXCEPTIONS) + +#pragma push_macro("max") +#undef max +#if defined(USE_DX_INTEROP) +#include +#include +#endif +#endif // _WIN32 + +// +#if defined(USE_CL_DEVICE_FISSION) +#include +#endif + +#if defined(__APPLE__) || defined(__MACOSX) +#include +#include +#include +#else +#include +#include +#endif // !__APPLE__ + +// To avoid accidentally taking ownership of core OpenCL types +// such as cl_kernel constructors are made explicit +// under OpenCL 1.2 +#if defined(CL_VERSION_1_2) && !defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +#define __CL_EXPLICIT_CONSTRUCTORS explicit +#else // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +#define __CL_EXPLICIT_CONSTRUCTORS +#endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + +// Define deprecated prefixes and suffixes to ensure compilation +// in case they are not pre-defined +#if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) +#define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED +#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) +#if !defined(CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED) +#define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) + +#if !defined(CL_CALLBACK) +#define CL_CALLBACK +#endif //CL_CALLBACK + +#include +#include + +#if !defined(__NO_STD_VECTOR) +#include +#endif + +#if !defined(__NO_STD_STRING) +#include +#endif + +#if defined(linux) || defined(__APPLE__) || defined(__MACOSX) +#include + +#include +#include +#endif // linux + +#include + + +/*! \namespace cl + * + * \brief The OpenCL C++ bindings are defined within this namespace. + * + */ +namespace cl { + +class Memory; + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +#define __INIT_CL_EXT_FCN_PTR(name) \ + if(!pfn_##name) { \ + pfn_##name = (PFN_##name) \ + clGetExtensionFunctionAddress(#name); \ + if(!pfn_##name) { \ + } \ + } +#endif // #if defined(CL_VERSION_1_1) + +#if defined(CL_VERSION_1_2) +#define __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, name) \ + if(!pfn_##name) { \ + pfn_##name = (PFN_##name) \ + clGetExtensionFunctionAddressForPlatform(platform, #name); \ + if(!pfn_##name) { \ + } \ + } +#endif // #if defined(CL_VERSION_1_1) + +class Program; +class Device; +class Context; +class CommandQueue; +class Memory; +class Buffer; + +#if defined(__CL_ENABLE_EXCEPTIONS) +/*! \brief Exception class + * + * This may be thrown by API functions when __CL_ENABLE_EXCEPTIONS is defined. + */ +class Error : public std::exception +{ +private: + cl_int err_; + const char * errStr_; +public: + /*! \brief Create a new CL error exception for a given error code + * and corresponding message. + * + * \param err error code value. + * + * \param errStr a descriptive string that must remain in scope until + * handling of the exception has concluded. If set, it + * will be returned by what(). + */ + Error(cl_int err, const char * errStr = NULL) : err_(err), errStr_(errStr) + {} + + ~Error() throw() {} + + /*! \brief Get error string associated with exception + * + * \return A memory pointer to the error message string. + */ + virtual const char * what() const throw () + { + if (errStr_ == NULL) { + return "empty"; + } + else { + return errStr_; + } + } + + /*! \brief Get error code associated with exception + * + * \return The error code. + */ + cl_int err(void) const { return err_; } +}; + +#define __ERR_STR(x) #x +#else +#define __ERR_STR(x) NULL +#endif // __CL_ENABLE_EXCEPTIONS + + +namespace detail +{ +#if defined(__CL_ENABLE_EXCEPTIONS) +static inline cl_int errHandler ( + cl_int err, + const char * errStr = NULL) +{ + if (err != CL_SUCCESS) { + throw Error(err, errStr); + } + return err; +} +#else +static inline cl_int errHandler (cl_int err, const char * errStr = NULL) +{ + (void) errStr; // suppress unused variable warning + return err; +} +#endif // __CL_ENABLE_EXCEPTIONS +} + + + +//! \cond DOXYGEN_DETAIL +#if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) +#define __GET_DEVICE_INFO_ERR __ERR_STR(clGetDeviceInfo) +#define __GET_PLATFORM_INFO_ERR __ERR_STR(clGetPlatformInfo) +#define __GET_DEVICE_IDS_ERR __ERR_STR(clGetDeviceIDs) +#define __GET_PLATFORM_IDS_ERR __ERR_STR(clGetPlatformIDs) +#define __GET_CONTEXT_INFO_ERR __ERR_STR(clGetContextInfo) +#define __GET_EVENT_INFO_ERR __ERR_STR(clGetEventInfo) +#define __GET_EVENT_PROFILE_INFO_ERR __ERR_STR(clGetEventProfileInfo) +#define __GET_MEM_OBJECT_INFO_ERR __ERR_STR(clGetMemObjectInfo) +#define __GET_IMAGE_INFO_ERR __ERR_STR(clGetImageInfo) +#define __GET_SAMPLER_INFO_ERR __ERR_STR(clGetSamplerInfo) +#define __GET_KERNEL_INFO_ERR __ERR_STR(clGetKernelInfo) +#if defined(CL_VERSION_1_2) +#define __GET_KERNEL_ARG_INFO_ERR __ERR_STR(clGetKernelArgInfo) +#endif // #if defined(CL_VERSION_1_2) +#define __GET_KERNEL_WORK_GROUP_INFO_ERR __ERR_STR(clGetKernelWorkGroupInfo) +#define __GET_PROGRAM_INFO_ERR __ERR_STR(clGetProgramInfo) +#define __GET_PROGRAM_BUILD_INFO_ERR __ERR_STR(clGetProgramBuildInfo) +#define __GET_COMMAND_QUEUE_INFO_ERR __ERR_STR(clGetCommandQueueInfo) + +#define __CREATE_CONTEXT_ERR __ERR_STR(clCreateContext) +#define __CREATE_CONTEXT_FROM_TYPE_ERR __ERR_STR(clCreateContextFromType) +#define __GET_SUPPORTED_IMAGE_FORMATS_ERR __ERR_STR(clGetSupportedImageFormats) + +#define __CREATE_BUFFER_ERR __ERR_STR(clCreateBuffer) +#define __COPY_ERR __ERR_STR(cl::copy) +#define __CREATE_SUBBUFFER_ERR __ERR_STR(clCreateSubBuffer) +#define __CREATE_GL_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) +#define __CREATE_GL_RENDER_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) +#define __GET_GL_OBJECT_INFO_ERR __ERR_STR(clGetGLObjectInfo) +#if defined(CL_VERSION_1_2) +#define __CREATE_IMAGE_ERR __ERR_STR(clCreateImage) +#define __CREATE_GL_TEXTURE_ERR __ERR_STR(clCreateFromGLTexture) +#define __IMAGE_DIMENSION_ERR __ERR_STR(Incorrect image dimensions) +#endif // #if defined(CL_VERSION_1_2) +#define __CREATE_SAMPLER_ERR __ERR_STR(clCreateSampler) +#define __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR __ERR_STR(clSetMemObjectDestructorCallback) + +#define __CREATE_USER_EVENT_ERR __ERR_STR(clCreateUserEvent) +#define __SET_USER_EVENT_STATUS_ERR __ERR_STR(clSetUserEventStatus) +#define __SET_EVENT_CALLBACK_ERR __ERR_STR(clSetEventCallback) +#define __WAIT_FOR_EVENTS_ERR __ERR_STR(clWaitForEvents) + +#define __CREATE_KERNEL_ERR __ERR_STR(clCreateKernel) +#define __SET_KERNEL_ARGS_ERR __ERR_STR(clSetKernelArg) +#define __CREATE_PROGRAM_WITH_SOURCE_ERR __ERR_STR(clCreateProgramWithSource) +#define __CREATE_PROGRAM_WITH_BINARY_ERR __ERR_STR(clCreateProgramWithBinary) +#if defined(CL_VERSION_1_2) +#define __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR __ERR_STR(clCreateProgramWithBuiltInKernels) +#endif // #if defined(CL_VERSION_1_2) +#define __BUILD_PROGRAM_ERR __ERR_STR(clBuildProgram) +#if defined(CL_VERSION_1_2) +#define __COMPILE_PROGRAM_ERR __ERR_STR(clCompileProgram) + +#endif // #if defined(CL_VERSION_1_2) +#define __CREATE_KERNELS_IN_PROGRAM_ERR __ERR_STR(clCreateKernelsInProgram) + +#define __CREATE_COMMAND_QUEUE_ERR __ERR_STR(clCreateCommandQueue) +#define __SET_COMMAND_QUEUE_PROPERTY_ERR __ERR_STR(clSetCommandQueueProperty) +#define __ENQUEUE_READ_BUFFER_ERR __ERR_STR(clEnqueueReadBuffer) +#define __ENQUEUE_READ_BUFFER_RECT_ERR __ERR_STR(clEnqueueReadBufferRect) +#define __ENQUEUE_WRITE_BUFFER_ERR __ERR_STR(clEnqueueWriteBuffer) +#define __ENQUEUE_WRITE_BUFFER_RECT_ERR __ERR_STR(clEnqueueWriteBufferRect) +#define __ENQEUE_COPY_BUFFER_ERR __ERR_STR(clEnqueueCopyBuffer) +#define __ENQEUE_COPY_BUFFER_RECT_ERR __ERR_STR(clEnqueueCopyBufferRect) +#define __ENQUEUE_FILL_BUFFER_ERR __ERR_STR(clEnqueueFillBuffer) +#define __ENQUEUE_READ_IMAGE_ERR __ERR_STR(clEnqueueReadImage) +#define __ENQUEUE_WRITE_IMAGE_ERR __ERR_STR(clEnqueueWriteImage) +#define __ENQUEUE_COPY_IMAGE_ERR __ERR_STR(clEnqueueCopyImage) +#define __ENQUEUE_FILL_IMAGE_ERR __ERR_STR(clEnqueueFillImage) +#define __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR __ERR_STR(clEnqueueCopyImageToBuffer) +#define __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR __ERR_STR(clEnqueueCopyBufferToImage) +#define __ENQUEUE_MAP_BUFFER_ERR __ERR_STR(clEnqueueMapBuffer) +#define __ENQUEUE_MAP_IMAGE_ERR __ERR_STR(clEnqueueMapImage) +#define __ENQUEUE_UNMAP_MEM_OBJECT_ERR __ERR_STR(clEnqueueUnMapMemObject) +#define __ENQUEUE_NDRANGE_KERNEL_ERR __ERR_STR(clEnqueueNDRangeKernel) +#define __ENQUEUE_TASK_ERR __ERR_STR(clEnqueueTask) +#define __ENQUEUE_NATIVE_KERNEL __ERR_STR(clEnqueueNativeKernel) +#if defined(CL_VERSION_1_2) +#define __ENQUEUE_MIGRATE_MEM_OBJECTS_ERR __ERR_STR(clEnqueueMigrateMemObjects) +#endif // #if defined(CL_VERSION_1_2) + +#define __ENQUEUE_ACQUIRE_GL_ERR __ERR_STR(clEnqueueAcquireGLObjects) +#define __ENQUEUE_RELEASE_GL_ERR __ERR_STR(clEnqueueReleaseGLObjects) + + +#define __RETAIN_ERR __ERR_STR(Retain Object) +#define __RELEASE_ERR __ERR_STR(Release Object) +#define __FLUSH_ERR __ERR_STR(clFlush) +#define __FINISH_ERR __ERR_STR(clFinish) +#define __VECTOR_CAPACITY_ERR __ERR_STR(Vector capacity error) + +/** + * CL 1.2 version that uses device fission. + */ +#if defined(CL_VERSION_1_2) +#define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevices) +#else +#define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevicesEXT) +#endif // #if defined(CL_VERSION_1_2) + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +#define __ENQUEUE_MARKER_ERR __ERR_STR(clEnqueueMarker) +#define __ENQUEUE_WAIT_FOR_EVENTS_ERR __ERR_STR(clEnqueueWaitForEvents) +#define __ENQUEUE_BARRIER_ERR __ERR_STR(clEnqueueBarrier) +#define __UNLOAD_COMPILER_ERR __ERR_STR(clUnloadCompiler) +#define __CREATE_GL_TEXTURE_2D_ERR __ERR_STR(clCreateFromGLTexture2D) +#define __CREATE_GL_TEXTURE_3D_ERR __ERR_STR(clCreateFromGLTexture3D) +#define __CREATE_IMAGE2D_ERR __ERR_STR(clCreateImage2D) +#define __CREATE_IMAGE3D_ERR __ERR_STR(clCreateImage3D) +#endif // #if defined(CL_VERSION_1_1) + +#endif // __CL_USER_OVERRIDE_ERROR_STRINGS +//! \endcond + +/** + * CL 1.2 marker and barrier commands + */ +#if defined(CL_VERSION_1_2) +#define __ENQUEUE_MARKER_WAIT_LIST_ERR __ERR_STR(clEnqueueMarkerWithWaitList) +#define __ENQUEUE_BARRIER_WAIT_LIST_ERR __ERR_STR(clEnqueueBarrierWithWaitList) +#endif // #if defined(CL_VERSION_1_2) + +#if !defined(__USE_DEV_STRING) && !defined(__NO_STD_STRING) +typedef std::string STRING_CLASS; +#elif !defined(__USE_DEV_STRING) + +/*! \class string + * \brief Simple string class, that provides a limited subset of std::string + * functionality but avoids many of the issues that come with that class. + + * \note Deprecated. Please use std::string as default or + * re-define the string class to match the std::string + * interface by defining STRING_CLASS + */ +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED string CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ +private: + ::size_t size_; + char * str_; +public: + //! \brief Constructs an empty string, allocating no memory. + string(void) : size_(0), str_(NULL) + { + } + + /*! \brief Constructs a string populated from an arbitrary value of + * specified size. + * + * An extra '\0' is added, in case none was contained in str. + * + * \param str the initial value of the string instance. Note that '\0' + * characters receive no special treatment. If NULL, + * the string is left empty, with a size of 0. + * + * \param size the number of characters to copy from str. + */ + string(const char * str, ::size_t size) : + size_(size), + str_(NULL) + { + if( size > 0 ) { + str_ = new char[size_+1]; + if (str_ != NULL) { + memcpy(str_, str, size_ * sizeof(char)); + str_[size_] = '\0'; + } + else { + size_ = 0; + } + } + } + + /*! \brief Constructs a string populated from a null-terminated value. + * + * \param str the null-terminated initial value of the string instance. + * If NULL, the string is left empty, with a size of 0. + */ + string(const char * str) : + size_(0), + str_(NULL) + { + if( str ) { + size_= ::strlen(str); + } + if( size_ > 0 ) { + str_ = new char[size_ + 1]; + if (str_ != NULL) { + memcpy(str_, str, (size_ + 1) * sizeof(char)); + } + } + } + + void resize( ::size_t n ) + { + if( size_ == n ) { + return; + } + if (n == 0) { + if( str_ ) { + delete [] str_; + } + str_ = NULL; + size_ = 0; + } + else { + char *newString = new char[n + 1]; + int copySize = n; + if( size_ < n ) { + copySize = size_; + } + size_ = n; + + if(str_) { + memcpy(newString, str_, (copySize + 1) * sizeof(char)); + } + if( copySize < size_ ) { + memset(newString + copySize, 0, size_ - copySize); + } + newString[size_] = '\0'; + + delete [] str_; + str_ = newString; + } + } + + const char& operator[] ( ::size_t pos ) const + { + return str_[pos]; + } + + char& operator[] ( ::size_t pos ) + { + return str_[pos]; + } + + /*! \brief Copies the value of another string to this one. + * + * \param rhs the string to copy. + * + * \returns a reference to the modified instance. + */ + string& operator=(const string& rhs) + { + if (this == &rhs) { + return *this; + } + + if( str_ != NULL ) { + delete [] str_; + str_ = NULL; + size_ = 0; + } + + if (rhs.size_ == 0 || rhs.str_ == NULL) { + str_ = NULL; + size_ = 0; + } + else { + str_ = new char[rhs.size_ + 1]; + size_ = rhs.size_; + + if (str_ != NULL) { + memcpy(str_, rhs.str_, (size_ + 1) * sizeof(char)); + } + else { + size_ = 0; + } + } + + return *this; + } + + /*! \brief Constructs a string by copying the value of another instance. + * + * \param rhs the string to copy. + */ + string(const string& rhs) : + size_(0), + str_(NULL) + { + *this = rhs; + } + + //! \brief Destructor - frees memory used to hold the current value. + ~string() + { + delete[] str_; + str_ = NULL; + } + + //! \brief Queries the length of the string, excluding any added '\0's. + ::size_t size(void) const { return size_; } + + //! \brief Queries the length of the string, excluding any added '\0's. + ::size_t length(void) const { return size(); } + + /*! \brief Returns a pointer to the private copy held by this instance, + * or "" if empty/unset. + */ + const char * c_str(void) const { return (str_) ? str_ : "";} +}; +typedef cl::string STRING_CLASS; +#endif // #elif !defined(__USE_DEV_STRING) + +#if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) +#define VECTOR_CLASS std::vector +#elif !defined(__USE_DEV_VECTOR) +#define VECTOR_CLASS cl::vector + +#if !defined(__MAX_DEFAULT_VECTOR_SIZE) +#define __MAX_DEFAULT_VECTOR_SIZE 10 +#endif + +/*! \class vector + * \brief Fixed sized vector implementation that mirroring + * + * \note Deprecated. Please use std::vector as default or + * re-define the vector class to match the std::vector + * interface by defining VECTOR_CLASS + + * \note Not recommended for use with custom objects as + * current implementation will construct N elements + * + * std::vector functionality. + * \brief Fixed sized vector compatible with std::vector. + * + * \note + * This differs from std::vector<> not just in memory allocation, + * but also in terms of when members are constructed, destroyed, + * and assigned instead of being copy constructed. + * + * \param T type of element contained in the vector. + * + * \param N maximum size of the vector. + */ +template +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED vector CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED +{ +private: + T data_[N]; + unsigned int size_; + +public: + //! \brief Constructs an empty vector with no memory allocated. + vector() : + size_(static_cast(0)) + {} + + //! \brief Deallocates the vector's memory and destroys all of its elements. + ~vector() + { + clear(); + } + + //! \brief Returns the number of elements currently contained. + unsigned int size(void) const + { + return size_; + } + + /*! \brief Empties the vector of all elements. + * \note + * This does not deallocate memory but will invoke destructors + * on contained elements. + */ + void clear() + { + while(!empty()) { + pop_back(); + } + } + + /*! \brief Appends an element after the last valid element. + * Calling this on a vector that has reached capacity will throw an + * exception if exceptions are enabled. + */ + void push_back (const T& x) + { + if (size() < N) { + new (&data_[size_]) T(x); + size_++; + } else { + detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); + } + } + + /*! \brief Removes the last valid element from the vector. + * Calling this on an empty vector will throw an exception + * if exceptions are enabled. + */ + void pop_back(void) + { + if (size_ != 0) { + --size_; + data_[size_].~T(); + } else { + detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); + } + } + + /*! \brief Constructs with a value copied from another. + * + * \param vec the vector to copy. + */ + vector(const vector& vec) : + size_(vec.size_) + { + if (size_ != 0) { + assign(vec.begin(), vec.end()); + } + } + + /*! \brief Constructs with a specified number of initial elements. + * + * \param size number of initial elements. + * + * \param val value of initial elements. + */ + vector(unsigned int size, const T& val = T()) : + size_(0) + { + for (unsigned int i = 0; i < size; i++) { + push_back(val); + } + } + + /*! \brief Overwrites the current content with that copied from another + * instance. + * + * \param rhs vector to copy. + * + * \returns a reference to this. + */ + vector& operator=(const vector& rhs) + { + if (this == &rhs) { + return *this; + } + + if (rhs.size_ != 0) { + assign(rhs.begin(), rhs.end()); + } else { + clear(); + } + + return *this; + } + + /*! \brief Tests equality against another instance. + * + * \param vec the vector against which to compare. + */ + bool operator==(vector &vec) + { + if (size() != vec.size()) { + return false; + } + + for( unsigned int i = 0; i < size(); ++i ) { + if( operator[](i) != vec[i] ) { + return false; + } + } + return true; + } + + //! \brief Conversion operator to T*. + operator T* () { return data_; } + + //! \brief Conversion operator to const T*. + operator const T* () const { return data_; } + + //! \brief Tests whether this instance has any elements. + bool empty (void) const + { + return size_==0; + } + + //! \brief Returns the maximum number of elements this instance can hold. + unsigned int max_size (void) const + { + return N; + } + + //! \brief Returns the maximum number of elements this instance can hold. + unsigned int capacity () const + { + return N; + } + + /*! \brief Returns a reference to a given element. + * + * \param index which element to access. * + * \note + * The caller is responsible for ensuring index is >= 0 and < size(). + */ + T& operator[](int index) + { + return data_[index]; + } + + /*! \brief Returns a const reference to a given element. + * + * \param index which element to access. + * + * \note + * The caller is responsible for ensuring index is >= 0 and < size(). + */ + const T& operator[](int index) const + { + return data_[index]; + } + + /*! \brief Assigns elements of the vector based on a source iterator range. + * + * \param start Beginning iterator of source range + * \param end Enditerator of source range + * + * \note + * Will throw an exception if exceptions are enabled and size exceeded. + */ + template + void assign(I start, I end) + { + clear(); + while(start != end) { + push_back(*start); + start++; + } + } + + /*! \class iterator + * \brief Const iterator class for vectors + */ + class iterator + { + private: + const vector *vec_; + int index_; + + /** + * Internal iterator constructor to capture reference + * to the vector it iterates over rather than taking + * the vector by copy. + */ + iterator (const vector &vec, int index) : + vec_(&vec) + { + if( !vec.empty() ) { + index_ = index; + } else { + index_ = -1; + } + } + + public: + iterator(void) : + index_(-1), + vec_(NULL) + { + } + + iterator(const iterator& rhs) : + vec_(rhs.vec_), + index_(rhs.index_) + { + } + + ~iterator(void) {} + + static iterator begin(const cl::vector &vec) + { + iterator i(vec, 0); + + return i; + } + + static iterator end(const cl::vector &vec) + { + iterator i(vec, vec.size()); + + return i; + } + + bool operator==(iterator i) + { + return ((vec_ == i.vec_) && + (index_ == i.index_)); + } + + bool operator!=(iterator i) + { + return (!(*this==i)); + } + + iterator& operator++() + { + ++index_; + return *this; + } + + iterator operator++(int) + { + iterator retVal(*this); + ++index_; + return retVal; + } + + iterator& operator--() + { + --index_; + return *this; + } + + iterator operator--(int) + { + iterator retVal(*this); + --index_; + return retVal; + } + + const T& operator *() const + { + return (*vec_)[index_]; + } + }; + + iterator begin(void) + { + return iterator::begin(*this); + } + + iterator begin(void) const + { + return iterator::begin(*this); + } + + iterator end(void) + { + return iterator::end(*this); + } + + iterator end(void) const + { + return iterator::end(*this); + } + + T& front(void) + { + return data_[0]; + } + + T& back(void) + { + return data_[size_]; + } + + const T& front(void) const + { + return data_[0]; + } + + const T& back(void) const + { + return data_[size_-1]; + } +}; +#endif // #if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) + + + + + +namespace detail { +#define __DEFAULT_NOT_INITIALIZED 1 +#define __DEFAULT_BEING_INITIALIZED 2 +#define __DEFAULT_INITIALIZED 4 + + /* + * Compare and exchange primitives are needed for handling of defaults + */ + inline int compare_exchange(volatile int * dest, int exchange, int comparand) + { +#ifdef _WIN32 + return (int)(InterlockedCompareExchange( + (volatile long*)dest, + (long)exchange, + (long)comparand)); +#elif defined(__APPLE__) || defined(__MACOSX) + return OSAtomicOr32Orig((uint32_t)exchange, (volatile uint32_t*)dest); +#else // !_WIN32 || defined(__APPLE__) || defined(__MACOSX) + return (__sync_val_compare_and_swap( + dest, + comparand, + exchange)); +#endif // !_WIN32 + } + + inline void fence() { _mm_mfence(); } +}; // namespace detail + + +/*! \brief class used to interface between C++ and + * OpenCL C calls that require arrays of size_t values, whose + * size is known statically. + */ +template +class size_t +{ +private: + ::size_t data_[N]; + +public: + //! \brief Initialize size_t to all 0s + size_t() + { + for( int i = 0; i < N; ++i ) { + data_[i] = 0; + } + } + + ::size_t& operator[](int index) + { + return data_[index]; + } + + const ::size_t& operator[](int index) const + { + return data_[index]; + } + + //! \brief Conversion operator to T*. + operator ::size_t* () { return data_; } + + //! \brief Conversion operator to const T*. + operator const ::size_t* () const { return data_; } +}; + +namespace detail { + +// Generic getInfoHelper. The final parameter is used to guide overload +// resolution: the actual parameter passed is an int, which makes this +// a worse conversion sequence than a specialization that declares the +// parameter as an int. +template +inline cl_int getInfoHelper(Functor f, cl_uint name, T* param, long) +{ + return f(name, sizeof(T), param, NULL); +} + +// Specialized getInfoHelper for VECTOR_CLASS params +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + T* value = (T*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + param->assign(&value[0], &value[required/sizeof(T)]); + return CL_SUCCESS; +} + +/* Specialization for reference-counted types. This depends on the + * existence of Wrapper::cl_type, and none of the other types having the + * cl_type member. Note that simplify specifying the parameter as Wrapper + * does not work, because when using a derived type (e.g. Context) the generic + * template will provide a better match. + */ +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int, typename T::cl_type = 0) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + typename T::cl_type * value = (typename T::cl_type *) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + ::size_t elements = required / sizeof(typename T::cl_type); + param->assign(&value[0], &value[elements]); + for (::size_t i = 0; i < elements; i++) + { + if (value[i] != NULL) + { + err = (*param)[i].retain(); + if (err != CL_SUCCESS) { + return err; + } + } + } + return CL_SUCCESS; +} + +// Specialized for getInfo +template +inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int) +{ + cl_int err = f(name, param->size() * sizeof(char *), &(*param)[0], NULL); + + if (err != CL_SUCCESS) { + return err; + } + + return CL_SUCCESS; +} + +// Specialized GetInfoHelper for STRING_CLASS params +template +inline cl_int getInfoHelper(Func f, cl_uint name, STRING_CLASS* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + char* value = (char*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + *param = value; + return CL_SUCCESS; +} + +// Specialized GetInfoHelper for cl::size_t params +template +inline cl_int getInfoHelper(Func f, cl_uint name, size_t* param, long) +{ + ::size_t required; + cl_int err = f(name, 0, NULL, &required); + if (err != CL_SUCCESS) { + return err; + } + + ::size_t* value = (::size_t*) alloca(required); + err = f(name, required, value, NULL); + if (err != CL_SUCCESS) { + return err; + } + + for(int i = 0; i < N; ++i) { + (*param)[i] = value[i]; + } + + return CL_SUCCESS; +} + +template struct ReferenceHandler; + +/* Specialization for reference-counted types. This depends on the + * existence of Wrapper::cl_type, and none of the other types having the + * cl_type member. Note that simplify specifying the parameter as Wrapper + * does not work, because when using a derived type (e.g. Context) the generic + * template will provide a better match. + */ +template +inline cl_int getInfoHelper(Func f, cl_uint name, T* param, int, typename T::cl_type = 0) +{ + typename T::cl_type value; + cl_int err = f(name, sizeof(value), &value, NULL); + if (err != CL_SUCCESS) { + return err; + } + *param = value; + if (value != NULL) + { + err = param->retain(); + if (err != CL_SUCCESS) { + return err; + } + } + return CL_SUCCESS; +} + +#define __PARAM_NAME_INFO_1_0(F) \ + F(cl_platform_info, CL_PLATFORM_PROFILE, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_VERSION, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_NAME, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_VENDOR, STRING_CLASS) \ + F(cl_platform_info, CL_PLATFORM_EXTENSIONS, STRING_CLASS) \ + \ + F(cl_device_info, CL_DEVICE_TYPE, cl_device_type) \ + F(cl_device_info, CL_DEVICE_VENDOR_ID, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_COMPUTE_UNITS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_GROUP_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_SIZES, VECTOR_CLASS< ::size_t>) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_CLOCK_FREQUENCY, cl_uint) \ + F(cl_device_info, CL_DEVICE_ADDRESS_BITS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_READ_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_WRITE_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_MEM_ALLOC_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_WIDTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_HEIGHT, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_WIDTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_HEIGHT, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_DEPTH, ::size_t) \ + F(cl_device_info, CL_DEVICE_IMAGE_SUPPORT, cl_bool) \ + F(cl_device_info, CL_DEVICE_MAX_PARAMETER_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_MAX_SAMPLERS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MEM_BASE_ADDR_ALIGN, cl_uint) \ + F(cl_device_info, CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE, cl_uint) \ + F(cl_device_info, CL_DEVICE_SINGLE_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, cl_device_mem_cache_type) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE, cl_uint)\ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_GLOBAL_MEM_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_MAX_CONSTANT_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_LOCAL_MEM_TYPE, cl_device_local_mem_type) \ + F(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE, cl_ulong) \ + F(cl_device_info, CL_DEVICE_ERROR_CORRECTION_SUPPORT, cl_bool) \ + F(cl_device_info, CL_DEVICE_PROFILING_TIMER_RESOLUTION, ::size_t) \ + F(cl_device_info, CL_DEVICE_ENDIAN_LITTLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_AVAILABLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_COMPILER_AVAILABLE, cl_bool) \ + F(cl_device_info, CL_DEVICE_EXECUTION_CAPABILITIES, cl_device_exec_capabilities) \ + F(cl_device_info, CL_DEVICE_QUEUE_ON_HOST_PROPERTIES, cl_command_queue_properties) \ + F(cl_device_info, CL_DEVICE_PLATFORM, cl_platform_id) \ + F(cl_device_info, CL_DEVICE_NAME, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_VENDOR, STRING_CLASS) \ + F(cl_device_info, CL_DRIVER_VERSION, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_PROFILE, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_VERSION, STRING_CLASS) \ + F(cl_device_info, CL_DEVICE_EXTENSIONS, STRING_CLASS) \ + \ + F(cl_context_info, CL_CONTEXT_REFERENCE_COUNT, cl_uint) \ + F(cl_context_info, CL_CONTEXT_DEVICES, VECTOR_CLASS) \ + F(cl_context_info, CL_CONTEXT_PROPERTIES, VECTOR_CLASS) \ + \ + F(cl_event_info, CL_EVENT_COMMAND_QUEUE, cl::CommandQueue) \ + F(cl_event_info, CL_EVENT_COMMAND_TYPE, cl_command_type) \ + F(cl_event_info, CL_EVENT_REFERENCE_COUNT, cl_uint) \ + F(cl_event_info, CL_EVENT_COMMAND_EXECUTION_STATUS, cl_uint) \ + \ + F(cl_profiling_info, CL_PROFILING_COMMAND_QUEUED, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_SUBMIT, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_START, cl_ulong) \ + F(cl_profiling_info, CL_PROFILING_COMMAND_END, cl_ulong) \ + \ + F(cl_mem_info, CL_MEM_TYPE, cl_mem_object_type) \ + F(cl_mem_info, CL_MEM_FLAGS, cl_mem_flags) \ + F(cl_mem_info, CL_MEM_SIZE, ::size_t) \ + F(cl_mem_info, CL_MEM_HOST_PTR, void*) \ + F(cl_mem_info, CL_MEM_MAP_COUNT, cl_uint) \ + F(cl_mem_info, CL_MEM_REFERENCE_COUNT, cl_uint) \ + F(cl_mem_info, CL_MEM_CONTEXT, cl::Context) \ + \ + F(cl_image_info, CL_IMAGE_FORMAT, cl_image_format) \ + F(cl_image_info, CL_IMAGE_ELEMENT_SIZE, ::size_t) \ + F(cl_image_info, CL_IMAGE_ROW_PITCH, ::size_t) \ + F(cl_image_info, CL_IMAGE_SLICE_PITCH, ::size_t) \ + F(cl_image_info, CL_IMAGE_WIDTH, ::size_t) \ + F(cl_image_info, CL_IMAGE_HEIGHT, ::size_t) \ + F(cl_image_info, CL_IMAGE_DEPTH, ::size_t) \ + \ + F(cl_sampler_info, CL_SAMPLER_REFERENCE_COUNT, cl_uint) \ + F(cl_sampler_info, CL_SAMPLER_CONTEXT, cl::Context) \ + F(cl_sampler_info, CL_SAMPLER_NORMALIZED_COORDS, cl_addressing_mode) \ + F(cl_sampler_info, CL_SAMPLER_ADDRESSING_MODE, cl_filter_mode) \ + F(cl_sampler_info, CL_SAMPLER_FILTER_MODE, cl_bool) \ + \ + F(cl_program_info, CL_PROGRAM_REFERENCE_COUNT, cl_uint) \ + F(cl_program_info, CL_PROGRAM_CONTEXT, cl::Context) \ + F(cl_program_info, CL_PROGRAM_NUM_DEVICES, cl_uint) \ + F(cl_program_info, CL_PROGRAM_DEVICES, VECTOR_CLASS) \ + F(cl_program_info, CL_PROGRAM_SOURCE, STRING_CLASS) \ + F(cl_program_info, CL_PROGRAM_BINARY_SIZES, VECTOR_CLASS< ::size_t>) \ + F(cl_program_info, CL_PROGRAM_BINARIES, VECTOR_CLASS) \ + \ + F(cl_program_build_info, CL_PROGRAM_BUILD_STATUS, cl_build_status) \ + F(cl_program_build_info, CL_PROGRAM_BUILD_OPTIONS, STRING_CLASS) \ + F(cl_program_build_info, CL_PROGRAM_BUILD_LOG, STRING_CLASS) \ + \ + F(cl_kernel_info, CL_KERNEL_FUNCTION_NAME, STRING_CLASS) \ + F(cl_kernel_info, CL_KERNEL_NUM_ARGS, cl_uint) \ + F(cl_kernel_info, CL_KERNEL_REFERENCE_COUNT, cl_uint) \ + F(cl_kernel_info, CL_KERNEL_CONTEXT, cl::Context) \ + F(cl_kernel_info, CL_KERNEL_PROGRAM, cl::Program) \ + \ + F(cl_kernel_work_group_info, CL_KERNEL_WORK_GROUP_SIZE, ::size_t) \ + F(cl_kernel_work_group_info, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, cl::size_t<3>) \ + F(cl_kernel_work_group_info, CL_KERNEL_LOCAL_MEM_SIZE, cl_ulong) \ + \ + F(cl_command_queue_info, CL_QUEUE_CONTEXT, cl::Context) \ + F(cl_command_queue_info, CL_QUEUE_DEVICE, cl::Device) \ + F(cl_command_queue_info, CL_QUEUE_REFERENCE_COUNT, cl_uint) \ + F(cl_command_queue_info, CL_QUEUE_PROPERTIES, cl_command_queue_properties) + +#if defined(CL_VERSION_1_1) +#define __PARAM_NAME_INFO_1_1(F) \ + F(cl_context_info, CL_CONTEXT_NUM_DEVICES, cl_uint)\ + F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_INT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE, cl_uint) \ + F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF, cl_uint) \ + F(cl_device_info, CL_DEVICE_DOUBLE_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_HALF_FP_CONFIG, cl_device_fp_config) \ + F(cl_device_info, CL_DEVICE_HOST_UNIFIED_MEMORY, cl_bool) \ + F(cl_device_info, CL_DEVICE_OPENCL_C_VERSION, STRING_CLASS) \ + \ + F(cl_mem_info, CL_MEM_ASSOCIATED_MEMOBJECT, cl::Memory) \ + F(cl_mem_info, CL_MEM_OFFSET, ::size_t) \ + \ + F(cl_kernel_work_group_info, CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE, ::size_t) \ + F(cl_kernel_work_group_info, CL_KERNEL_PRIVATE_MEM_SIZE, cl_ulong) \ + \ + F(cl_event_info, CL_EVENT_CONTEXT, cl::Context) +#endif // CL_VERSION_1_1 + + +#if defined(CL_VERSION_1_2) +#define __PARAM_NAME_INFO_1_2(F) \ + F(cl_image_info, CL_IMAGE_BUFFER, cl::Buffer) \ + \ + F(cl_program_info, CL_PROGRAM_NUM_KERNELS, ::size_t) \ + F(cl_program_info, CL_PROGRAM_KERNEL_NAMES, STRING_CLASS) \ + \ + F(cl_program_build_info, CL_PROGRAM_BINARY_TYPE, cl_program_binary_type) \ + \ + F(cl_kernel_info, CL_KERNEL_ATTRIBUTES, STRING_CLASS) \ + \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_ADDRESS_QUALIFIER, cl_kernel_arg_address_qualifier) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_ACCESS_QUALIFIER, cl_kernel_arg_access_qualifier) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_TYPE_NAME, STRING_CLASS) \ + F(cl_kernel_arg_info, CL_KERNEL_ARG_NAME, STRING_CLASS) \ + \ + F(cl_device_info, CL_DEVICE_PARENT_DEVICE, cl_device_id) \ + F(cl_device_info, CL_DEVICE_PARTITION_PROPERTIES, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_PARTITION_TYPE, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_REFERENCE_COUNT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, ::size_t) \ + F(cl_device_info, CL_DEVICE_PARTITION_AFFINITY_DOMAIN, cl_device_affinity_domain) \ + F(cl_device_info, CL_DEVICE_BUILT_IN_KERNELS, STRING_CLASS) +#endif // #if defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_2_0) +#define __PARAM_NAME_INFO_2_0(F) \ + F(cl_device_info, CL_DEVICE_MAX_PIPE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_PIPE_MAX_ACTIVE_RESERVATIONS, cl_uint) \ + F(cl_device_info, CL_DEVICE_PIPE_MAX_PACKET_SIZE, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_GLOBAL_VARIABLE_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_GLOBAL_VARIABLE_PREFERRED_TOTAL_SIZE, ::size_t) \ + F(cl_device_info, CL_DEVICE_QUEUE_ON_DEVICE_PREFERRED_SIZE, cl_uint) \ + F(cl_device_info, CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_ON_DEVICE_QUEUES, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_ON_DEVICE_EVENTS, cl_uint) \ + F(cl_device_info, CL_DEVICE_MAX_READ_WRITE_IMAGE_ARGS, cl_uint) \ + F(cl_device_info, CL_DEVICE_SVM_CAPABILITIES, cl_device_svm_capabilities) \ + F(cl_device_info, CL_DEVICE_PREFERRED_PLATFORM_ATOMIC_ALIGNMENT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_GLOBAL_ATOMIC_ALIGNMENT, cl_uint) \ + F(cl_device_info, CL_DEVICE_PREFERRED_LOCAL_ATOMIC_ALIGNMENT, cl_uint) \ + F(cl_device_info, CL_DEVICE_QUEUE_ON_DEVICE_PROPERTIES, cl_command_queue_properties) +#endif // #if defined(CL_VERSION_2_0) + +#if defined(USE_CL_DEVICE_FISSION) +#define __PARAM_NAME_DEVICE_FISSION(F) \ + F(cl_device_info, CL_DEVICE_PARENT_DEVICE_EXT, cl_device_id) \ + F(cl_device_info, CL_DEVICE_PARTITION_TYPES_EXT, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, VECTOR_CLASS) \ + F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \ + F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, VECTOR_CLASS) +#endif // USE_CL_DEVICE_FISSION + +template +struct param_traits {}; + +#define __CL_DECLARE_PARAM_TRAITS(token, param_name, T) \ +struct token; \ +template<> \ +struct param_traits \ +{ \ + enum { value = param_name }; \ + typedef T param_type; \ +}; + +__PARAM_NAME_INFO_1_0(__CL_DECLARE_PARAM_TRAITS) +#if defined(CL_VERSION_1_1) +__PARAM_NAME_INFO_1_1(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_1_1 +#if defined(CL_VERSION_1_2) +__PARAM_NAME_INFO_1_2(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_1_2 +#if defined(CL_VERSION_2_0) +__PARAM_NAME_INFO_2_0(__CL_DECLARE_PARAM_TRAITS) +#endif // CL_VERSION_2_0 + +#if defined(USE_CL_DEVICE_FISSION) +__PARAM_NAME_DEVICE_FISSION(__CL_DECLARE_PARAM_TRAITS); +#endif // USE_CL_DEVICE_FISSION + +#ifdef CL_PLATFORM_ICD_SUFFIX_KHR +__CL_DECLARE_PARAM_TRAITS(cl_platform_info, CL_PLATFORM_ICD_SUFFIX_KHR, STRING_CLASS) +#endif + +#ifdef CL_DEVICE_PROFILING_TIMER_OFFSET_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_PROFILING_TIMER_OFFSET_AMD, cl_ulong) +#endif + +#ifdef CL_DEVICE_GLOBAL_FREE_MEMORY_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_FREE_MEMORY_AMD, VECTOR_CLASS< ::size_t>) +#endif +#ifdef CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_SIMD_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_WAVEFRONT_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WAVEFRONT_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD, cl_uint) +#endif +#ifdef CL_DEVICE_LOCAL_MEM_BANKS_AMD +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_BANKS_AMD, cl_uint) +#endif + +#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV, cl_uint) +#endif +#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV, cl_uint) +#endif +#ifdef CL_DEVICE_REGISTERS_PER_BLOCK_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_REGISTERS_PER_BLOCK_NV, cl_uint) +#endif +#ifdef CL_DEVICE_WARP_SIZE_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WARP_SIZE_NV, cl_uint) +#endif +#ifdef CL_DEVICE_GPU_OVERLAP_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GPU_OVERLAP_NV, cl_bool) +#endif +#ifdef CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV, cl_bool) +#endif +#ifdef CL_DEVICE_INTEGRATED_MEMORY_NV +__CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_INTEGRATED_MEMORY_NV, cl_bool) +#endif + +// Convenience functions + +template +inline cl_int +getInfo(Func f, cl_uint name, T* param) +{ + return getInfoHelper(f, name, param, 0); +} + +template +struct GetInfoFunctor0 +{ + Func f_; const Arg0& arg0_; + cl_int operator ()( + cl_uint param, ::size_t size, void* value, ::size_t* size_ret) + { return f_(arg0_, param, size, value, size_ret); } +}; + +template +struct GetInfoFunctor1 +{ + Func f_; const Arg0& arg0_; const Arg1& arg1_; + cl_int operator ()( + cl_uint param, ::size_t size, void* value, ::size_t* size_ret) + { return f_(arg0_, arg1_, param, size, value, size_ret); } +}; + +template +inline cl_int +getInfo(Func f, const Arg0& arg0, cl_uint name, T* param) +{ + GetInfoFunctor0 f0 = { f, arg0 }; + return getInfoHelper(f0, name, param, 0); +} + +template +inline cl_int +getInfo(Func f, const Arg0& arg0, const Arg1& arg1, cl_uint name, T* param) +{ + GetInfoFunctor1 f0 = { f, arg0, arg1 }; + return getInfoHelper(f0, name, param, 0); +} + +template +struct ReferenceHandler +{ }; + +#if defined(CL_VERSION_1_2) +/** + * OpenCL 1.2 devices do have retain/release. + */ +template <> +struct ReferenceHandler +{ + /** + * Retain the device. + * \param device A valid device created using createSubDevices + * \return + * CL_SUCCESS if the function executed successfully. + * CL_INVALID_DEVICE if device was not a valid subdevice + * CL_OUT_OF_RESOURCES + * CL_OUT_OF_HOST_MEMORY + */ + static cl_int retain(cl_device_id device) + { return ::clRetainDevice(device); } + /** + * Retain the device. + * \param device A valid device created using createSubDevices + * \return + * CL_SUCCESS if the function executed successfully. + * CL_INVALID_DEVICE if device was not a valid subdevice + * CL_OUT_OF_RESOURCES + * CL_OUT_OF_HOST_MEMORY + */ + static cl_int release(cl_device_id device) + { return ::clReleaseDevice(device); } +}; +#else // #if defined(CL_VERSION_1_2) +/** + * OpenCL 1.1 devices do not have retain/release. + */ +template <> +struct ReferenceHandler +{ + // cl_device_id does not have retain(). + static cl_int retain(cl_device_id) + { return CL_SUCCESS; } + // cl_device_id does not have release(). + static cl_int release(cl_device_id) + { return CL_SUCCESS; } +}; +#endif // #if defined(CL_VERSION_1_2) + +template <> +struct ReferenceHandler +{ + // cl_platform_id does not have retain(). + static cl_int retain(cl_platform_id) + { return CL_SUCCESS; } + // cl_platform_id does not have release(). + static cl_int release(cl_platform_id) + { return CL_SUCCESS; } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_context context) + { return ::clRetainContext(context); } + static cl_int release(cl_context context) + { return ::clReleaseContext(context); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_command_queue queue) + { return ::clRetainCommandQueue(queue); } + static cl_int release(cl_command_queue queue) + { return ::clReleaseCommandQueue(queue); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_mem memory) + { return ::clRetainMemObject(memory); } + static cl_int release(cl_mem memory) + { return ::clReleaseMemObject(memory); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_sampler sampler) + { return ::clRetainSampler(sampler); } + static cl_int release(cl_sampler sampler) + { return ::clReleaseSampler(sampler); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_program program) + { return ::clRetainProgram(program); } + static cl_int release(cl_program program) + { return ::clReleaseProgram(program); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_kernel kernel) + { return ::clRetainKernel(kernel); } + static cl_int release(cl_kernel kernel) + { return ::clReleaseKernel(kernel); } +}; + +template <> +struct ReferenceHandler +{ + static cl_int retain(cl_event event) + { return ::clRetainEvent(event); } + static cl_int release(cl_event event) + { return ::clReleaseEvent(event); } +}; + + +// Extracts version number with major in the upper 16 bits, minor in the lower 16 +static cl_uint getVersion(const char *versionInfo) +{ + int highVersion = 0; + int lowVersion = 0; + int index = 7; + while(versionInfo[index] != '.' ) { + highVersion *= 10; + highVersion += versionInfo[index]-'0'; + ++index; + } + ++index; + while(versionInfo[index] != ' ' ) { + lowVersion *= 10; + lowVersion += versionInfo[index]-'0'; + ++index; + } + return (highVersion << 16) | lowVersion; +} + +static cl_uint getPlatformVersion(cl_platform_id platform) +{ + ::size_t size = 0; + clGetPlatformInfo(platform, CL_PLATFORM_VERSION, 0, NULL, &size); + char *versionInfo = (char *) alloca(size); + clGetPlatformInfo(platform, CL_PLATFORM_VERSION, size, &versionInfo[0], &size); + return getVersion(versionInfo); +} + +static cl_uint getDevicePlatformVersion(cl_device_id device) +{ + cl_platform_id platform; + clGetDeviceInfo(device, CL_DEVICE_PLATFORM, sizeof(platform), &platform, NULL); + return getPlatformVersion(platform); +} + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) +static cl_uint getContextPlatformVersion(cl_context context) +{ + // The platform cannot be queried directly, so we first have to grab a + // device and obtain its context + ::size_t size = 0; + clGetContextInfo(context, CL_CONTEXT_DEVICES, 0, NULL, &size); + if (size == 0) + return 0; + cl_device_id *devices = (cl_device_id *) alloca(size); + clGetContextInfo(context, CL_CONTEXT_DEVICES, size, devices, NULL); + return getDevicePlatformVersion(devices[0]); +} +#endif // #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + +template +class Wrapper +{ +public: + typedef T cl_type; + +protected: + cl_type object_; + +public: + Wrapper() : object_(NULL) { } + + Wrapper(const cl_type &obj) : object_(obj) { } + + ~Wrapper() + { + if (object_ != NULL) { release(); } + } + + Wrapper(const Wrapper& rhs) + { + object_ = rhs.object_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + } + + Wrapper& operator = (const Wrapper& rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs.object_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + return *this; + } + + Wrapper& operator = (const cl_type &rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs; + return *this; + } + + cl_type operator ()() const { return object_; } + + cl_type& operator ()() { return object_; } + +protected: + template + friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); + + cl_int retain() const + { + return ReferenceHandler::retain(object_); + } + + cl_int release() const + { + return ReferenceHandler::release(object_); + } +}; + +template <> +class Wrapper +{ +public: + typedef cl_device_id cl_type; + +protected: + cl_type object_; + bool referenceCountable_; + + static bool isReferenceCountable(cl_device_id device) + { + bool retVal = false; + if (device != NULL) { + int version = getDevicePlatformVersion(device); + if(version > ((1 << 16) + 1)) { + retVal = true; + } + } + return retVal; + } + +public: + Wrapper() : object_(NULL), referenceCountable_(false) + { + } + + Wrapper(const cl_type &obj) : object_(obj), referenceCountable_(false) + { + referenceCountable_ = isReferenceCountable(obj); + } + + ~Wrapper() + { + if (object_ != NULL) { release(); } + } + + Wrapper(const Wrapper& rhs) + { + object_ = rhs.object_; + referenceCountable_ = isReferenceCountable(object_); + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + } + + Wrapper& operator = (const Wrapper& rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs.object_; + referenceCountable_ = rhs.referenceCountable_; + if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } + return *this; + } + + Wrapper& operator = (const cl_type &rhs) + { + if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } + object_ = rhs; + referenceCountable_ = isReferenceCountable(object_); + return *this; + } + + cl_type operator ()() const { return object_; } + + cl_type& operator ()() { return object_; } + +protected: + template + friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); + + template + friend inline cl_int getInfoHelper(Func, cl_uint, VECTOR_CLASS*, int, typename U::cl_type); + + cl_int retain() const + { + if( referenceCountable_ ) { + return ReferenceHandler::retain(object_); + } + else { + return CL_SUCCESS; + } + } + + cl_int release() const + { + if( referenceCountable_ ) { + return ReferenceHandler::release(object_); + } + else { + return CL_SUCCESS; + } + } +}; + +} // namespace detail +//! \endcond + +/*! \stuct ImageFormat + * \brief Adds constructors and member functions for cl_image_format. + * + * \see cl_image_format + */ +struct ImageFormat : public cl_image_format +{ + //! \brief Default constructor - performs no initialization. + ImageFormat(){} + + //! \brief Initializing constructor. + ImageFormat(cl_channel_order order, cl_channel_type type) + { + image_channel_order = order; + image_channel_data_type = type; + } + + //! \brief Assignment operator. + ImageFormat& operator = (const ImageFormat& rhs) + { + if (this != &rhs) { + this->image_channel_data_type = rhs.image_channel_data_type; + this->image_channel_order = rhs.image_channel_order; + } + return *this; + } +}; + +/*! \brief Class interface for cl_device_id. + * + * \note Copies of these objects are inexpensive, since they don't 'own' + * any underlying resources or data structures. + * + * \see cl_device_id + */ +class Device : public detail::Wrapper +{ +public: + //! \brief Default constructor - initializes to NULL. + Device() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device(const Device& device) : detail::Wrapper(device) { } + + /*! \brief Constructor from cl_device_id. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device(const cl_device_id &device) : detail::Wrapper(device) { } + + /*! \brief Returns the first device on the default context. + * + * \see Context::getDefault() + */ + static Device getDefault(cl_int * err = NULL); + + /*! \brief Assignment operator from Device. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device& operator = (const Device& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_device_id. + * + * This simply copies the device ID value, which is an inexpensive operation. + */ + Device& operator = (const cl_device_id& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetDeviceInfo(). + template + cl_int getInfo(cl_device_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetDeviceInfo, object_, name, param), + __GET_DEVICE_INFO_ERR); + } + + //! \brief Wrapper for clGetDeviceInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_device_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /** + * CL 1.2 version + */ +#if defined(CL_VERSION_1_2) + //! \brief Wrapper for clCreateSubDevicesEXT(). + cl_int createSubDevices( + const cl_device_partition_property * properties, + VECTOR_CLASS* devices) + { + cl_uint n = 0; + cl_int err = clCreateSubDevices(object_, properties, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = clCreateSubDevices(object_, properties, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif // #if defined(CL_VERSION_1_2) + +/** + * CL 1.1 version that uses device fission. + */ +#if defined(CL_VERSION_1_1) +#if defined(USE_CL_DEVICE_FISSION) + cl_int createSubDevices( + const cl_device_partition_property_ext * properties, + VECTOR_CLASS* devices) + { + typedef CL_API_ENTRY cl_int + ( CL_API_CALL * PFN_clCreateSubDevicesEXT)( + cl_device_id /*in_device*/, + const cl_device_partition_property_ext * /* properties */, + cl_uint /*num_entries*/, + cl_device_id * /*out_devices*/, + cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; + + static PFN_clCreateSubDevicesEXT pfn_clCreateSubDevicesEXT = NULL; + __INIT_CL_EXT_FCN_PTR(clCreateSubDevicesEXT); + + cl_uint n = 0; + cl_int err = pfn_clCreateSubDevicesEXT(object_, properties, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = pfn_clCreateSubDevicesEXT(object_, properties, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_SUB_DEVICES); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif // #if defined(USE_CL_DEVICE_FISSION) +#endif // #if defined(CL_VERSION_1_1) +}; + +/*! \brief Class interface for cl_platform_id. + * + * \note Copies of these objects are inexpensive, since they don't 'own' + * any underlying resources or data structures. + * + * \see cl_platform_id + */ +class Platform : public detail::Wrapper +{ +public: + //! \brief Default constructor - initializes to NULL. + Platform() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform(const Platform& platform) : detail::Wrapper(platform) { } + + /*! \brief Constructor from cl_platform_id. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform(const cl_platform_id &platform) : detail::Wrapper(platform) { } + + /*! \brief Assignment operator from Platform. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform& operator = (const Platform& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_platform_id. + * + * This simply copies the platform ID value, which is an inexpensive operation. + */ + Platform& operator = (const cl_platform_id& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetPlatformInfo(). + cl_int getInfo(cl_platform_info name, STRING_CLASS* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetPlatformInfo, object_, name, param), + __GET_PLATFORM_INFO_ERR); + } + + //! \brief Wrapper for clGetPlatformInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_platform_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Gets a list of devices for this platform. + * + * Wraps clGetDeviceIDs(). + */ + cl_int getDevices( + cl_device_type type, + VECTOR_CLASS* devices) const + { + cl_uint n = 0; + if( devices == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); + } + cl_int err = ::clGetDeviceIDs(object_, type, 0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = ::clGetDeviceIDs(object_, type, n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } + +#if defined(USE_DX_INTEROP) + /*! \brief Get the list of available D3D10 devices. + * + * \param d3d_device_source. + * + * \param d3d_object. + * + * \param d3d_device_set. + * + * \param devices returns a vector of OpenCL D3D10 devices found. The cl::Device + * values returned in devices can be used to identify a specific OpenCL + * device. If \a devices argument is NULL, this argument is ignored. + * + * \return One of the following values: + * - CL_SUCCESS if the function is executed successfully. + * + * The application can query specific capabilities of the OpenCL device(s) + * returned by cl::getDevices. This can be used by the application to + * determine which device(s) to use. + * + * \note In the case that exceptions are enabled and a return value + * other than CL_SUCCESS is generated, then cl::Error exception is + * generated. + */ + cl_int getDevices( + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + VECTOR_CLASS* devices) const + { + typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clGetDeviceIDsFromD3D10KHR)( + cl_platform_id platform, + cl_d3d10_device_source_khr d3d_device_source, + void * d3d_object, + cl_d3d10_device_set_khr d3d_device_set, + cl_uint num_entries, + cl_device_id * devices, + cl_uint* num_devices); + + if( devices == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); + } + + static PFN_clGetDeviceIDsFromD3D10KHR pfn_clGetDeviceIDsFromD3D10KHR = NULL; + __INIT_CL_EXT_FCN_PTR_PLATFORM(object_, clGetDeviceIDsFromD3D10KHR); + + cl_uint n = 0; + cl_int err = pfn_clGetDeviceIDsFromD3D10KHR( + object_, + d3d_device_source, + d3d_object, + d3d_device_set, + 0, + NULL, + &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); + err = pfn_clGetDeviceIDsFromD3D10KHR( + object_, + d3d_device_source, + d3d_object, + d3d_device_set, + n, + ids, + NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_DEVICE_IDS_ERR); + } + + devices->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } +#endif + + /*! \brief Gets a list of available platforms. + * + * Wraps clGetPlatformIDs(). + */ + static cl_int get( + VECTOR_CLASS* platforms) + { + cl_uint n = 0; + + if( platforms == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); + } + + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + platforms->assign(&ids[0], &ids[n]); + return CL_SUCCESS; + } + + /*! \brief Gets the first available platform. + * + * Wraps clGetPlatformIDs(), returning the first result. + */ + static cl_int get( + Platform * platform) + { + cl_uint n = 0; + + if( platform == NULL ) { + return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); + } + + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + *platform = ids[0]; + return CL_SUCCESS; + } + + /*! \brief Gets the first available platform, returning it by value. + * + * Wraps clGetPlatformIDs(), returning the first result. + */ + static Platform get( + cl_int * errResult = NULL) + { + Platform platform; + cl_uint n = 0; + cl_int err = ::clGetPlatformIDs(0, NULL, &n); + if (err != CL_SUCCESS) { + detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + if (errResult != NULL) { + *errResult = err; + } + } + + cl_platform_id* ids = (cl_platform_id*) alloca( + n * sizeof(cl_platform_id)); + err = ::clGetPlatformIDs(n, ids, NULL); + + if (err != CL_SUCCESS) { + detail::errHandler(err, __GET_PLATFORM_IDS_ERR); + } + + if (errResult != NULL) { + *errResult = err; + } + + return ids[0]; + } + + static Platform getDefault( + cl_int *errResult = NULL ) + { + return get(errResult); + } + + +#if defined(CL_VERSION_1_2) + //! \brief Wrapper for clUnloadCompiler(). + cl_int + unloadCompiler() + { + return ::clUnloadPlatformCompiler(object_); + } +#endif // #if defined(CL_VERSION_1_2) +}; // class Platform + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) +/** + * Unload the OpenCL compiler. + * \note Deprecated for OpenCL 1.2. Use Platform::unloadCompiler instead. + */ +inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int +UnloadCompiler() CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; +inline cl_int +UnloadCompiler() +{ + return ::clUnloadCompiler(); +} +#endif // #if defined(CL_VERSION_1_1) + +/*! \brief Class interface for cl_context. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_context as the original. For details, see + * clRetainContext() and clReleaseContext(). + * + * \see cl_context + */ +class Context + : public detail::Wrapper +{ +private: + static volatile int default_initialized_; + static Context default_; + static volatile cl_int default_error_; +public: + /*! \brief Destructor. + * + * This calls clReleaseContext() on the value held by this instance. + */ + ~Context() { } + + /*! \brief Constructs a context including a list of specified devices. + * + * Wraps clCreateContext(). + */ + Context( + const VECTOR_CLASS& devices, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + object_ = ::clCreateContext( + properties, (cl_uint) numDevices, + deviceIDs, + notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_ERR); + if (err != NULL) { + *err = error; + } + } + + Context( + const Device& device, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + + cl_device_id deviceID = device(); + + object_ = ::clCreateContext( + properties, 1, + &deviceID, + notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Constructs a context including all or a subset of devices of a specified type. + * + * Wraps clCreateContextFromType(). + */ + Context( + cl_device_type type, + cl_context_properties* properties = NULL, + void (CL_CALLBACK * notifyFptr)( + const char *, + const void *, + ::size_t, + void *) = NULL, + void* data = NULL, + cl_int* err = NULL) + { + cl_int error; + +#if !defined(__APPLE__) || !defined(__MACOS) + cl_context_properties prop[4] = {CL_CONTEXT_PLATFORM, 0, 0, 0 }; + + if (properties == NULL) { + // Get a valid platform ID as we cannot send in a blank one + VECTOR_CLASS platforms; + error = Platform::get(&platforms); + if (error != CL_SUCCESS) { + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + } + return; + } + + // Check the platforms we found for a device of our specified type + cl_context_properties platform_id = 0; + for (unsigned int i = 0; i < platforms.size(); i++) { + + VECTOR_CLASS devices; + +#if defined(__CL_ENABLE_EXCEPTIONS) + try { +#endif + + error = platforms[i].getDevices(type, &devices); + +#if defined(__CL_ENABLE_EXCEPTIONS) + } catch (Error) {} + // Catch if exceptions are enabled as we don't want to exit if first platform has no devices of type + // We do error checking next anyway, and can throw there if needed +#endif + + // Only squash CL_SUCCESS and CL_DEVICE_NOT_FOUND + if (error != CL_SUCCESS && error != CL_DEVICE_NOT_FOUND) { + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + } + } + + if (devices.size() > 0) { + platform_id = (cl_context_properties)platforms[i](); + break; + } + } + + if (platform_id == 0) { + detail::errHandler(CL_DEVICE_NOT_FOUND, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = CL_DEVICE_NOT_FOUND; + } + return; + } + + prop[1] = platform_id; + properties = &prop[0]; + } +#endif + object_ = ::clCreateContextFromType( + properties, type, notifyFptr, data, &error); + + detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Returns a singleton context including all devices of CL_DEVICE_TYPE_DEFAULT. + * + * \note All calls to this function return the same cl_context as the first. + */ + static Context getDefault(cl_int * err = NULL) + { + int state = detail::compare_exchange( + &default_initialized_, + __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); + + if (state & __DEFAULT_INITIALIZED) { + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + if (state & __DEFAULT_BEING_INITIALIZED) { + // Assume writes will propagate eventually... + while(default_initialized_ != __DEFAULT_INITIALIZED) { + detail::fence(); + } + + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + cl_int error; + default_ = Context( + CL_DEVICE_TYPE_DEFAULT, + NULL, + NULL, + NULL, + &error); + + detail::fence(); + + default_error_ = error; + // Assume writes will propagate eventually... + default_initialized_ = __DEFAULT_INITIALIZED; + + detail::fence(); + + if (err != NULL) { + *err = default_error_; + } + return default_; + + } + + //! \brief Default constructor - initializes to NULL. + Context() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This calls clRetainContext() on the parameter's cl_context. + */ + Context(const Context& context) : detail::Wrapper(context) { } + + /*! \brief Constructor from cl_context - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_context + * into the new Context object. + */ + __CL_EXPLICIT_CONSTRUCTORS Context(const cl_context& context) : detail::Wrapper(context) { } + + /*! \brief Assignment operator from Context. + * + * This calls clRetainContext() on the parameter and clReleaseContext() on + * the previous value held by this instance. + */ + Context& operator = (const Context& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_context - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseContext() on the value previously held by this instance. + */ + Context& operator = (const cl_context& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetContextInfo(). + template + cl_int getInfo(cl_context_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetContextInfo, object_, name, param), + __GET_CONTEXT_INFO_ERR); + } + + //! \brief Wrapper for clGetContextInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_context_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Gets a list of supported image formats. + * + * Wraps clGetSupportedImageFormats(). + */ + cl_int getSupportedImageFormats( + cl_mem_flags flags, + cl_mem_object_type type, + VECTOR_CLASS* formats) const + { + cl_uint numEntries; + cl_int err = ::clGetSupportedImageFormats( + object_, + flags, + type, + 0, + NULL, + &numEntries); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); + } + + ImageFormat* value = (ImageFormat*) + alloca(numEntries * sizeof(ImageFormat)); + err = ::clGetSupportedImageFormats( + object_, + flags, + type, + numEntries, + (cl_image_format*) value, + NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); + } + + formats->assign(&value[0], &value[numEntries]); + return CL_SUCCESS; + } +}; + +inline Device Device::getDefault(cl_int * err) +{ + cl_int error; + Device device; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + device = context.getInfo()[0]; + if (err != NULL) { + *err = CL_SUCCESS; + } + } + + return device; +} + + +#ifdef _WIN32 +__declspec(selectany) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__declspec(selectany) Context Context::default_; +__declspec(selectany) volatile cl_int Context::default_error_ = CL_SUCCESS; +#else +__attribute__((weak)) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__attribute__((weak)) Context Context::default_; +__attribute__((weak)) volatile cl_int Context::default_error_ = CL_SUCCESS; +#endif + +/*! \brief Class interface for cl_event. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_event as the original. For details, see + * clRetainEvent() and clReleaseEvent(). + * + * \see cl_event + */ +class Event : public detail::Wrapper +{ +public: + /*! \brief Destructor. + * + * This calls clReleaseEvent() on the value held by this instance. + */ + ~Event() { } + + //! \brief Default constructor - initializes to NULL. + Event() : detail::Wrapper() { } + + /*! \brief Copy constructor. + * + * This calls clRetainEvent() on the parameter's cl_event. + */ + Event(const Event& event) : detail::Wrapper(event) { } + + /*! \brief Constructor from cl_event - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_event + * into the new Event object. + */ + Event(const cl_event& event) : detail::Wrapper(event) { } + + /*! \brief Assignment operator from cl_event - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseEvent() on the value previously held by this instance. + */ + Event& operator = (const Event& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_event. + * + * This calls clRetainEvent() on the parameter and clReleaseEvent() on + * the previous value held by this instance. + */ + Event& operator = (const cl_event& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetEventInfo(). + template + cl_int getInfo(cl_event_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetEventInfo, object_, name, param), + __GET_EVENT_INFO_ERR); + } + + //! \brief Wrapper for clGetEventInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_event_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + //! \brief Wrapper for clGetEventProfilingInfo(). + template + cl_int getProfilingInfo(cl_profiling_info name, T* param) const + { + return detail::errHandler(detail::getInfo( + &::clGetEventProfilingInfo, object_, name, param), + __GET_EVENT_PROFILE_INFO_ERR); + } + + //! \brief Wrapper for clGetEventProfilingInfo() that returns by value. + template typename + detail::param_traits::param_type + getProfilingInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_profiling_info, name>::param_type param; + cl_int result = getProfilingInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + /*! \brief Blocks the calling thread until this event completes. + * + * Wraps clWaitForEvents(). + */ + cl_int wait() const + { + return detail::errHandler( + ::clWaitForEvents(1, &object_), + __WAIT_FOR_EVENTS_ERR); + } + +#if defined(CL_VERSION_1_1) + /*! \brief Registers a user callback function for a specific command execution status. + * + * Wraps clSetEventCallback(). + */ + cl_int setCallback( + cl_int type, + void (CL_CALLBACK * pfn_notify)(cl_event, cl_int, void *), + void * user_data = NULL) + { + return detail::errHandler( + ::clSetEventCallback( + object_, + type, + pfn_notify, + user_data), + __SET_EVENT_CALLBACK_ERR); + } +#endif + + /*! \brief Blocks the calling thread until every event specified is complete. + * + * Wraps clWaitForEvents(). + */ + static cl_int + waitForEvents(const VECTOR_CLASS& events) + { + return detail::errHandler( + ::clWaitForEvents( + (cl_uint) events.size(), (cl_event*)&events.front()), + __WAIT_FOR_EVENTS_ERR); + } +}; + +#if defined(CL_VERSION_1_1) +/*! \brief Class interface for user events (a subset of cl_event's). + * + * See Event for details about copy semantics, etc. + */ +class UserEvent : public Event +{ +public: + /*! \brief Constructs a user event on a given context. + * + * Wraps clCreateUserEvent(). + */ + UserEvent( + const Context& context, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateUserEvent( + context(), + &error); + + detail::errHandler(error, __CREATE_USER_EVENT_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + UserEvent() : Event() { } + + //! \brief Copy constructor - performs shallow copy. + UserEvent(const UserEvent& event) : Event(event) { } + + //! \brief Assignment Operator - performs shallow copy. + UserEvent& operator = (const UserEvent& rhs) + { + if (this != &rhs) { + Event::operator=(rhs); + } + return *this; + } + + /*! \brief Sets the execution status of a user event object. + * + * Wraps clSetUserEventStatus(). + */ + cl_int setStatus(cl_int status) + { + return detail::errHandler( + ::clSetUserEventStatus(object_,status), + __SET_USER_EVENT_STATUS_ERR); + } +}; +#endif + +/*! \brief Blocks the calling thread until every event specified is complete. + * + * Wraps clWaitForEvents(). + */ +inline static cl_int +WaitForEvents(const VECTOR_CLASS& events) +{ + return detail::errHandler( + ::clWaitForEvents( + (cl_uint) events.size(), (cl_event*)&events.front()), + __WAIT_FOR_EVENTS_ERR); +} + +/*! \brief Class interface for cl_mem. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_mem as the original. For details, see + * clRetainMemObject() and clReleaseMemObject(). + * + * \see cl_mem + */ +class Memory : public detail::Wrapper +{ +public: + + /*! \brief Destructor. + * + * This calls clReleaseMemObject() on the value held by this instance. + */ + ~Memory() {} + + //! \brief Default constructor - initializes to NULL. + Memory() : detail::Wrapper() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainMemObject() on the parameter's cl_mem. + */ + Memory(const Memory& memory) : detail::Wrapper(memory) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_mem + * into the new Memory object. + */ + __CL_EXPLICIT_CONSTRUCTORS Memory(const cl_mem& memory) : detail::Wrapper(memory) { } + + /*! \brief Assignment operator from Memory. + * + * This calls clRetainMemObject() on the parameter and clReleaseMemObject() + * on the previous value held by this instance. + */ + Memory& operator = (const Memory& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_mem - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseMemObject() on the value previously held by this instance. + */ + Memory& operator = (const cl_mem& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetMemObjectInfo(). + template + cl_int getInfo(cl_mem_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetMemObjectInfo, object_, name, param), + __GET_MEM_OBJECT_INFO_ERR); + } + + //! \brief Wrapper for clGetMemObjectInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_mem_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + +#if defined(CL_VERSION_1_1) + /*! \brief Registers a callback function to be called when the memory object + * is no longer needed. + * + * Wraps clSetMemObjectDestructorCallback(). + * + * Repeated calls to this function, for a given cl_mem value, will append + * to the list of functions called (in reverse order) when memory object's + * resources are freed and the memory object is deleted. + * + * \note + * The registered callbacks are associated with the underlying cl_mem + * value - not the Memory class instance. + */ + cl_int setDestructorCallback( + void (CL_CALLBACK * pfn_notify)(cl_mem, void *), + void * user_data = NULL) + { + return detail::errHandler( + ::clSetMemObjectDestructorCallback( + object_, + pfn_notify, + user_data), + __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR); + } +#endif + +}; + +// Pre-declare copy functions +class Buffer; +template< typename IteratorType > +cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ); +template< typename IteratorType > +cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ); +template< typename IteratorType > +cl_int copy( const CommandQueue &queue, IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ); +template< typename IteratorType > +cl_int copy( const CommandQueue &queue, const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ); + + +/*! \brief Class interface for Buffer Memory Objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Buffer : public Memory +{ +public: + + /*! \brief Constructs a Buffer in a specified context. + * + * Wraps clCreateBuffer(). + * + * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was + * specified. Note alignment & exclusivity requirements. + */ + Buffer( + const Context& context, + cl_mem_flags flags, + ::size_t size, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Constructs a Buffer in the default context. + * + * Wraps clCreateBuffer(). + * + * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was + * specified. Note alignment & exclusivity requirements. + * + * \see Context::getDefault() + */ + Buffer( + cl_mem_flags flags, + ::size_t size, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + + Context context = Context::getDefault(err); + + object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! + * \brief Construct a Buffer from a host container via iterators. + * IteratorType must be random access. + * If useHostPtr is specified iterators must represent contiguous data. + */ + template< typename IteratorType > + Buffer( + IteratorType startIterator, + IteratorType endIterator, + bool readOnly, + bool useHostPtr = false, + cl_int* err = NULL) + { + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + cl_mem_flags flags = 0; + if( readOnly ) { + flags |= CL_MEM_READ_ONLY; + } + else { + flags |= CL_MEM_READ_WRITE; + } + if( useHostPtr ) { + flags |= CL_MEM_USE_HOST_PTR; + } + + ::size_t size = sizeof(DataType)*(endIterator - startIterator); + + Context context = Context::getDefault(err); + + if( useHostPtr ) { + object_ = ::clCreateBuffer(context(), flags, size, static_cast(&*startIterator), &error); + } else { + object_ = ::clCreateBuffer(context(), flags, size, 0, &error); + } + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + if( !useHostPtr ) { + error = cl::copy(startIterator, endIterator, *this); + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + } + + /*! + * \brief Construct a Buffer from a host container via iterators using a specified context. + * IteratorType must be random access. + * If useHostPtr is specified iterators must represent contiguous data. + */ + template< typename IteratorType > + Buffer(const Context &context, IteratorType startIterator, IteratorType endIterator, + bool readOnly, bool useHostPtr = false, cl_int* err = NULL); + + //! \brief Default constructor - initializes to NULL. + Buffer() : Memory() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Buffer(const Buffer& buffer) : Memory(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Buffer(const cl_mem& buffer) : Memory(buffer) { } + + /*! \brief Assignment from Buffer - performs shallow copy. + * + * See Memory for further details. + */ + Buffer& operator = (const Buffer& rhs) + { + if (this != &rhs) { + Memory::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Buffer& operator = (const cl_mem& rhs) + { + Memory::operator=(rhs); + return *this; + } + +#if defined(CL_VERSION_1_1) + /*! \brief Creates a new buffer object from this. + * + * Wraps clCreateSubBuffer(). + */ + Buffer createSubBuffer( + cl_mem_flags flags, + cl_buffer_create_type buffer_create_type, + const void * buffer_create_info, + cl_int * err = NULL) + { + Buffer result; + cl_int error; + result.object_ = ::clCreateSubBuffer( + object_, + flags, + buffer_create_type, + buffer_create_info, + &error); + + detail::errHandler(error, __CREATE_SUBBUFFER_ERR); + if (err != NULL) { + *err = error; + } + + return result; + } +#endif +}; + +#if defined (USE_DX_INTEROP) +/*! \brief Class interface for creating OpenCL buffers from ID3D10Buffer's. + * + * This is provided to facilitate interoperability with Direct3D. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferD3D10 : public Buffer +{ +public: + typedef CL_API_ENTRY cl_mem (CL_API_CALL *PFN_clCreateFromD3D10BufferKHR)( + cl_context context, cl_mem_flags flags, ID3D10Buffer* buffer, + cl_int* errcode_ret); + + /*! \brief Constructs a BufferD3D10, in a specified context, from a + * given ID3D10Buffer. + * + * Wraps clCreateFromD3D10BufferKHR(). + */ + BufferD3D10( + const Context& context, + cl_mem_flags flags, + ID3D10Buffer* bufobj, + cl_int * err = NULL) + { + static PFN_clCreateFromD3D10BufferKHR pfn_clCreateFromD3D10BufferKHR = NULL; + +#if defined(CL_VERSION_1_2) + vector props = context.getInfo(); + cl_platform platform = -1; + for( int i = 0; i < props.size(); ++i ) { + if( props[i] == CL_CONTEXT_PLATFORM ) { + platform = props[i+1]; + } + } + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clCreateFromD3D10BufferKHR); +#endif +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clCreateFromD3D10BufferKHR); +#endif + + cl_int error; + object_ = pfn_clCreateFromD3D10BufferKHR( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferD3D10() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10(const BufferD3D10& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferD3D10(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferD3D10 - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10& operator = (const BufferD3D10& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferD3D10& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } +}; +#endif + +/*! \brief Class interface for GL Buffer Memory Objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferGL : public Buffer +{ +public: + /*! \brief Constructs a BufferGL in a specified context, from a given + * GL buffer. + * + * Wraps clCreateFromGLBuffer(). + */ + BufferGL( + const Context& context, + cl_mem_flags flags, + GLuint bufobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLBuffer( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferGL() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL(const BufferGL& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferGL(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferGL - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL& operator = (const BufferGL& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferGL& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetGLObjectInfo(). + cl_int getObjectInfo( + cl_gl_object_type *type, + GLuint * gl_object_name) + { + return detail::errHandler( + ::clGetGLObjectInfo(object_,type,gl_object_name), + __GET_GL_OBJECT_INFO_ERR); + } +}; + +/*! \brief Class interface for GL Render Buffer Memory Objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class BufferRenderGL : public Buffer +{ +public: + /*! \brief Constructs a BufferRenderGL in a specified context, from a given + * GL Renderbuffer. + * + * Wraps clCreateFromGLRenderbuffer(). + */ + BufferRenderGL( + const Context& context, + cl_mem_flags flags, + GLuint bufobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLRenderbuffer( + context(), + flags, + bufobj, + &error); + + detail::errHandler(error, __CREATE_GL_RENDER_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + BufferRenderGL() : Buffer() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL(const BufferGL& buffer) : Buffer(buffer) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS BufferRenderGL(const cl_mem& buffer) : Buffer(buffer) { } + + /*! \brief Assignment from BufferGL - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL& operator = (const BufferRenderGL& rhs) + { + if (this != &rhs) { + Buffer::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + BufferRenderGL& operator = (const cl_mem& rhs) + { + Buffer::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetGLObjectInfo(). + cl_int getObjectInfo( + cl_gl_object_type *type, + GLuint * gl_object_name) + { + return detail::errHandler( + ::clGetGLObjectInfo(object_,type,gl_object_name), + __GET_GL_OBJECT_INFO_ERR); + } +}; + +/*! \brief C++ base class for Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image : public Memory +{ +protected: + //! \brief Default constructor - initializes to NULL. + Image() : Memory() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image(const Image& image) : Memory(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image(const cl_mem& image) : Memory(image) { } + + /*! \brief Assignment from Image - performs shallow copy. + * + * See Memory for further details. + */ + Image& operator = (const Image& rhs) + { + if (this != &rhs) { + Memory::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image& operator = (const cl_mem& rhs) + { + Memory::operator=(rhs); + return *this; + } + +public: + //! \brief Wrapper for clGetImageInfo(). + template + cl_int getImageInfo(cl_image_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetImageInfo, object_, name, param), + __GET_IMAGE_INFO_ERR); + } + + //! \brief Wrapper for clGetImageInfo() that returns by value. + template typename + detail::param_traits::param_type + getImageInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_image_info, name>::param_type param; + cl_int result = getImageInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +}; + +#if defined(CL_VERSION_1_2) +/*! \brief Class interface for 1D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image1D : public Image +{ +public: + /*! \brief Constructs a 1D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image1D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc = + { + CL_MEM_OBJECT_IMAGE1D, + width, + 0, 0, 0, 0, 0, 0, 0, 0 + }; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + Image1D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image1D(const Image1D& image1D) : Image(image1D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image1D(const cl_mem& image1D) : Image(image1D) { } + + /*! \brief Assignment from Image1D - performs shallow copy. + * + * See Memory for further details. + */ + Image1D& operator = (const Image1D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image1D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +/*! \class Image1DBuffer + * \brief Image interface for 1D buffer images. + */ +class Image1DBuffer : public Image +{ +public: + Image1DBuffer( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + const Buffer &buffer, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc = + { + CL_MEM_OBJECT_IMAGE1D_BUFFER, + width, + 0, 0, 0, 0, 0, 0, 0, + buffer() + }; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + NULL, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image1DBuffer() { } + + Image1DBuffer(const Image1DBuffer& image1D) : Image(image1D) { } + + __CL_EXPLICIT_CONSTRUCTORS Image1DBuffer(const cl_mem& image1D) : Image(image1D) { } + + Image1DBuffer& operator = (const Image1DBuffer& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image1DBuffer& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +/*! \class Image1DArray + * \brief Image interface for arrays of 1D images. + */ +class Image1DArray : public Image +{ +public: + Image1DArray( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t arraySize, + ::size_t width, + ::size_t rowPitch, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc = + { + CL_MEM_OBJECT_IMAGE1D_ARRAY, + width, + 0, 0, // height, depth (unused) + arraySize, + rowPitch, + 0, 0, 0, 0 + }; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image1DArray() { } + + Image1DArray(const Image1DArray& imageArray) : Image(imageArray) { } + + __CL_EXPLICIT_CONSTRUCTORS Image1DArray(const cl_mem& imageArray) : Image(imageArray) { } + + Image1DArray& operator = (const Image1DArray& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image1DArray& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + + +/*! \brief Class interface for 2D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image2D : public Image +{ +public: + /*! \brief Constructs a 1D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image2D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + ::size_t height, + ::size_t row_pitch = 0, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + bool useCreateImage; + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + // Run-time decision based on the actual platform + { + cl_uint version = detail::getContextPlatformVersion(context()); + useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above + } +#elif defined(CL_VERSION_1_2) + useCreateImage = true; +#else + useCreateImage = false; +#endif + +#if defined(CL_VERSION_1_2) + if (useCreateImage) + { + cl_image_desc desc = + { + CL_MEM_OBJECT_IMAGE2D, + width, + height, + 0, 0, // depth, array size (unused) + row_pitch, + 0, 0, 0, 0 + }; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) +#if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + if (!useCreateImage) + { + object_ = ::clCreateImage2D( + context(), flags,&format, width, height, row_pitch, host_ptr, &error); + + detail::errHandler(error, __CREATE_IMAGE2D_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + } + + //! \brief Default constructor - initializes to NULL. + Image2D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image2D(const Image2D& image2D) : Image(image2D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image2D(const cl_mem& image2D) : Image(image2D) { } + + /*! \brief Assignment from Image2D - performs shallow copy. + * + * See Memory for further details. + */ + Image2D& operator = (const Image2D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image2D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + + +#if !defined(CL_VERSION_1_2) +/*! \brief Class interface for GL 2D Image Memory objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + * \note Deprecated for OpenCL 1.2. Please use ImageGL instead. + */ +class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED Image2DGL CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED : public Image2D +{ +public: + /*! \brief Constructs an Image2DGL in a specified context, from a given + * GL Texture. + * + * Wraps clCreateFromGLTexture2D(). + */ + Image2DGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture2D( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_2D_ERR); + if (err != NULL) { + *err = error; + } + + } + + //! \brief Default constructor - initializes to NULL. + Image2DGL() : Image2D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL(const Image2DGL& image) : Image2D(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image2DGL(const cl_mem& image) : Image2D(image) { } + + /*! \brief Assignment from Image2DGL - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL& operator = (const Image2DGL& rhs) + { + if (this != &rhs) { + Image2D::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image2DGL& operator = (const cl_mem& rhs) + { + Image2D::operator=(rhs); + return *this; + } +}; +#endif // #if !defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_1_2) +/*! \class Image2DArray + * \brief Image interface for arrays of 2D images. + */ +class Image2DArray : public Image +{ +public: + Image2DArray( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t arraySize, + ::size_t width, + ::size_t height, + ::size_t rowPitch, + ::size_t slicePitch, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + cl_image_desc desc = + { + CL_MEM_OBJECT_IMAGE2D_ARRAY, + width, + height, + 0, // depth (unused) + arraySize, + rowPitch, + slicePitch, + 0, 0, 0 + }; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } + + Image2DArray() { } + + Image2DArray(const Image2DArray& imageArray) : Image(imageArray) { } + + __CL_EXPLICIT_CONSTRUCTORS Image2DArray(const cl_mem& imageArray) : Image(imageArray) { } + + Image2DArray& operator = (const Image2DArray& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + Image2DArray& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + +/*! \brief Class interface for 3D Image Memory objects. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image3D : public Image +{ +public: + /*! \brief Constructs a 3D Image in a specified context. + * + * Wraps clCreateImage(). + */ + Image3D( + const Context& context, + cl_mem_flags flags, + ImageFormat format, + ::size_t width, + ::size_t height, + ::size_t depth, + ::size_t row_pitch = 0, + ::size_t slice_pitch = 0, + void* host_ptr = NULL, + cl_int* err = NULL) + { + cl_int error; + bool useCreateImage; + +#if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + // Run-time decision based on the actual platform + { + cl_uint version = detail::getContextPlatformVersion(context()); + useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above + } +#elif defined(CL_VERSION_1_2) + useCreateImage = true; +#else + useCreateImage = false; +#endif + +#if defined(CL_VERSION_1_2) + if (useCreateImage) + { + cl_image_desc desc = + { + CL_MEM_OBJECT_IMAGE3D, + width, + height, + depth, + 0, // array size (unused) + row_pitch, + slice_pitch, + 0, 0, 0 + }; + object_ = ::clCreateImage( + context(), + flags, + &format, + &desc, + host_ptr, + &error); + + detail::errHandler(error, __CREATE_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) +#if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + if (!useCreateImage) + { + object_ = ::clCreateImage3D( + context(), flags, &format, width, height, depth, row_pitch, + slice_pitch, host_ptr, &error); + + detail::errHandler(error, __CREATE_IMAGE3D_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) + } + + //! \brief Default constructor - initializes to NULL. + Image3D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image3D(const Image3D& image3D) : Image(image3D) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image3D(const cl_mem& image3D) : Image(image3D) { } + + /*! \brief Assignment from Image3D - performs shallow copy. + * + * See Memory for further details. + */ + Image3D& operator = (const Image3D& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image3D& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; + +#if !defined(CL_VERSION_1_2) +/*! \brief Class interface for GL 3D Image Memory objects. + * + * This is provided to facilitate interoperability with OpenGL. + * + * See Memory for details about copy semantics, etc. + * + * \see Memory + */ +class Image3DGL : public Image3D +{ +public: + /*! \brief Constructs an Image3DGL in a specified context, from a given + * GL Texture. + * + * Wraps clCreateFromGLTexture3D(). + */ + Image3DGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture3D( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_3D_ERR); + if (err != NULL) { + *err = error; + } + } + + //! \brief Default constructor - initializes to NULL. + Image3DGL() : Image3D() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL(const Image3DGL& image) : Image3D(image) { } + + /*! \brief Constructor from cl_mem - takes ownership. + * + * See Memory for further details. + */ + __CL_EXPLICIT_CONSTRUCTORS Image3DGL(const cl_mem& image) : Image3D(image) { } + + /*! \brief Assignment from Image3DGL - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL& operator = (const Image3DGL& rhs) + { + if (this != &rhs) { + Image3D::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment from cl_mem - performs shallow copy. + * + * See Memory for further details. + */ + Image3DGL& operator = (const cl_mem& rhs) + { + Image3D::operator=(rhs); + return *this; + } +}; +#endif // #if !defined(CL_VERSION_1_2) + +#if defined(CL_VERSION_1_2) +/*! \class ImageGL + * \brief general image interface for GL interop. + * We abstract the 2D and 3D GL images into a single instance here + * that wraps all GL sourced images on the grounds that setup information + * was performed by OpenCL anyway. + */ +class ImageGL : public Image +{ +public: + ImageGL( + const Context& context, + cl_mem_flags flags, + GLenum target, + GLint miplevel, + GLuint texobj, + cl_int * err = NULL) + { + cl_int error; + object_ = ::clCreateFromGLTexture( + context(), + flags, + target, + miplevel, + texobj, + &error); + + detail::errHandler(error, __CREATE_GL_TEXTURE_ERR); + if (err != NULL) { + *err = error; + } + } + + ImageGL() : Image() { } + + ImageGL(const ImageGL& image) : Image(image) { } + + __CL_EXPLICIT_CONSTRUCTORS ImageGL(const cl_mem& image) : Image(image) { } + + ImageGL& operator = (const ImageGL& rhs) + { + if (this != &rhs) { + Image::operator=(rhs); + } + return *this; + } + + ImageGL& operator = (const cl_mem& rhs) + { + Image::operator=(rhs); + return *this; + } +}; +#endif // #if defined(CL_VERSION_1_2) + +/*! \brief Class interface for cl_sampler. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_sampler as the original. For details, see + * clRetainSampler() and clReleaseSampler(). + * + * \see cl_sampler + */ +class Sampler : public detail::Wrapper +{ +public: + /*! \brief Destructor. + * + * This calls clReleaseSampler() on the value held by this instance. + */ + ~Sampler() { } + + //! \brief Default constructor - initializes to NULL. + Sampler() { } + + /*! \brief Constructs a Sampler in a specified context. + * + * Wraps clCreateSampler(). + */ + Sampler( + const Context& context, + cl_bool normalized_coords, + cl_addressing_mode addressing_mode, + cl_filter_mode filter_mode, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateSampler( + context(), + normalized_coords, + addressing_mode, + filter_mode, + &error); + + detail::errHandler(error, __CREATE_SAMPLER_ERR); + if (err != NULL) { + *err = error; + } + } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainSampler() on the parameter's cl_sampler. + */ + Sampler(const Sampler& sampler) : detail::Wrapper(sampler) { } + + /*! \brief Constructor from cl_sampler - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_sampler + * into the new Sampler object. + */ + Sampler(const cl_sampler& sampler) : detail::Wrapper(sampler) { } + + /*! \brief Assignment operator from Sampler. + * + * This calls clRetainSampler() on the parameter and clReleaseSampler() + * on the previous value held by this instance. + */ + Sampler& operator = (const Sampler& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_sampler - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseSampler() on the value previously held by this instance. + */ + Sampler& operator = (const cl_sampler& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + //! \brief Wrapper for clGetSamplerInfo(). + template + cl_int getInfo(cl_sampler_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetSamplerInfo, object_, name, param), + __GET_SAMPLER_INFO_ERR); + } + + //! \brief Wrapper for clGetSamplerInfo() that returns by value. + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_sampler_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +}; + +class Program; +class CommandQueue; +class Kernel; + +//! \brief Class interface for specifying NDRange values. +class NDRange +{ +private: + size_t<3> sizes_; + cl_uint dimensions_; + +public: + //! \brief Default constructor - resulting range has zero dimensions. + NDRange() + : dimensions_(0) + { } + + //! \brief Constructs one-dimensional range. + NDRange(::size_t size0) + : dimensions_(1) + { + sizes_[0] = size0; + } + + //! \brief Constructs two-dimensional range. + NDRange(::size_t size0, ::size_t size1) + : dimensions_(2) + { + sizes_[0] = size0; + sizes_[1] = size1; + } + + //! \brief Constructs three-dimensional range. + NDRange(::size_t size0, ::size_t size1, ::size_t size2) + : dimensions_(3) + { + sizes_[0] = size0; + sizes_[1] = size1; + sizes_[2] = size2; + } + + /*! \brief Conversion operator to const ::size_t *. + * + * \returns a pointer to the size of the first dimension. + */ + operator const ::size_t*() const { + return (const ::size_t*) sizes_; + } + + //! \brief Queries the number of dimensions in the range. + ::size_t dimensions() const { return dimensions_; } +}; + +//! \brief A zero-dimensional range. +static const NDRange NullRange; + +//! \brief Local address wrapper for use with Kernel::setArg +struct LocalSpaceArg +{ + ::size_t size_; +}; + +namespace detail { + +template +struct KernelArgumentHandler +{ + static ::size_t size(const T&) { return sizeof(T); } + static T* ptr(T& value) { return &value; } +}; + +template <> +struct KernelArgumentHandler +{ + static ::size_t size(const LocalSpaceArg& value) { return value.size_; } + static void* ptr(LocalSpaceArg&) { return NULL; } +}; + +} +//! \endcond + +/*! __local + * \brief Helper function for generating LocalSpaceArg objects. + * Deprecated. Replaced with Local. + */ +inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED LocalSpaceArg +__local(::size_t size) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; +inline LocalSpaceArg +__local(::size_t size) +{ + LocalSpaceArg ret = { size }; + return ret; +} + +/*! Local + * \brief Helper function for generating LocalSpaceArg objects. + */ +inline LocalSpaceArg +Local(::size_t size) +{ + LocalSpaceArg ret = { size }; + return ret; +} + +//class KernelFunctor; + +/*! \brief Class interface for cl_kernel. + * + * \note Copies of these objects are shallow, meaning that the copy will refer + * to the same underlying cl_kernel as the original. For details, see + * clRetainKernel() and clReleaseKernel(). + * + * \see cl_kernel + */ +class Kernel : public detail::Wrapper +{ +public: + inline Kernel(const Program& program, const char* name, cl_int* err = NULL); + + /*! \brief Destructor. + * + * This calls clReleaseKernel() on the value held by this instance. + */ + ~Kernel() { } + + //! \brief Default constructor - initializes to NULL. + Kernel() { } + + /*! \brief Copy constructor - performs shallow copy. + * + * This calls clRetainKernel() on the parameter's cl_kernel. + */ + Kernel(const Kernel& kernel) : detail::Wrapper(kernel) { } + + /*! \brief Constructor from cl_kernel - takes ownership. + * + * This effectively transfers ownership of a refcount on the cl_kernel + * into the new Kernel object. + */ + __CL_EXPLICIT_CONSTRUCTORS Kernel(const cl_kernel& kernel) : detail::Wrapper(kernel) { } + + /*! \brief Assignment operator from Kernel. + * + * This calls clRetainKernel() on the parameter and clReleaseKernel() + * on the previous value held by this instance. + */ + Kernel& operator = (const Kernel& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + /*! \brief Assignment operator from cl_kernel - takes ownership. + * + * This effectively transfers ownership of a refcount on the rhs and calls + * clReleaseKernel() on the value previously held by this instance. + */ + Kernel& operator = (const cl_kernel& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + template + cl_int getInfo(cl_kernel_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetKernelInfo, object_, name, param), + __GET_KERNEL_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + +#if defined(CL_VERSION_1_2) + template + cl_int getArgInfo(cl_uint argIndex, cl_kernel_arg_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetKernelArgInfo, object_, argIndex, name, param), + __GET_KERNEL_ARG_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getArgInfo(cl_uint argIndex, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_arg_info, name>::param_type param; + cl_int result = getArgInfo(argIndex, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } +#endif // #if defined(CL_VERSION_1_2) + + template + cl_int getWorkGroupInfo( + const Device& device, cl_kernel_work_group_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetKernelWorkGroupInfo, object_, device(), name, param), + __GET_KERNEL_WORK_GROUP_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getWorkGroupInfo(const Device& device, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_kernel_work_group_info, name>::param_type param; + cl_int result = getWorkGroupInfo(device, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + template + cl_int setArg(cl_uint index, T value) + { + return detail::errHandler( + ::clSetKernelArg( + object_, + index, + detail::KernelArgumentHandler::size(value), + detail::KernelArgumentHandler::ptr(value)), + __SET_KERNEL_ARGS_ERR); + } + + cl_int setArg(cl_uint index, ::size_t size, void* argPtr) + { + return detail::errHandler( + ::clSetKernelArg(object_, index, size, argPtr), + __SET_KERNEL_ARGS_ERR); + } +}; + +/*! \class Program + * \brief Program interface that implements cl_program. + */ +class Program : public detail::Wrapper +{ +public: + typedef VECTOR_CLASS > Binaries; + typedef VECTOR_CLASS > Sources; + + Program( + const STRING_CLASS& source, + bool build = false, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + Context context = Context::getDefault(err); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS && build) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const Context& context, + const STRING_CLASS& source, + bool build = false, + cl_int* err = NULL) + { + cl_int error; + + const char * strings = source.c_str(); + const ::size_t length = source.size(); + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)1, &strings, &length, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + + if (error == CL_SUCCESS && build) { + + error = ::clBuildProgram( + object_, + 0, + NULL, + "", + NULL, + NULL); + + detail::errHandler(error, __BUILD_PROGRAM_ERR); + } + + if (err != NULL) { + *err = error; + } + } + + Program( + const Context& context, + const Sources& sources, + cl_int* err = NULL) + { + cl_int error; + + const ::size_t n = (::size_t)sources.size(); + ::size_t* lengths = (::size_t*) alloca(n * sizeof(::size_t)); + const char** strings = (const char**) alloca(n * sizeof(const char*)); + + for (::size_t i = 0; i < n; ++i) { + strings[i] = sources[(int)i].first; + lengths[i] = sources[(int)i].second; + } + + object_ = ::clCreateProgramWithSource( + context(), (cl_uint)n, strings, lengths, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); + if (err != NULL) { + *err = error; + } + } + + /** + * Construct a program object from a list of devices and a per-device list of binaries. + * \param context A valid OpenCL context in which to construct the program. + * \param devices A vector of OpenCL device objects for which the program will be created. + * \param binaries A vector of pairs of a pointer to a binary object and its length. + * \param binaryStatus An optional vector that on completion will be resized to + * match the size of binaries and filled with values to specify if each binary + * was successfully loaded. + * Set to CL_SUCCESS if the binary was successfully loaded. + * Set to CL_INVALID_VALUE if the length is 0 or the binary pointer is NULL. + * Set to CL_INVALID_BINARY if the binary provided is not valid for the matching device. + * \param err if non-NULL will be set to CL_SUCCESS on successful operation or one of the following errors: + * CL_INVALID_CONTEXT if context is not a valid context. + * CL_INVALID_VALUE if the length of devices is zero; or if the length of binaries does not match the length of devices; + * or if any entry in binaries is NULL or has length 0. + * CL_INVALID_DEVICE if OpenCL devices listed in devices are not in the list of devices associated with context. + * CL_INVALID_BINARY if an invalid program binary was encountered for any device. binaryStatus will return specific status for each device. + * CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the OpenCL implementation on the host. + */ + Program( + const Context& context, + const VECTOR_CLASS& devices, + const Binaries& binaries, + VECTOR_CLASS* binaryStatus = NULL, + cl_int* err = NULL) + { + cl_int error; + + const ::size_t numDevices = devices.size(); + + // Catch size mismatch early and return + if(binaries.size() != numDevices) { + error = CL_INVALID_VALUE; + detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); + if (err != NULL) { + *err = error; + } + return; + } + + ::size_t* lengths = (::size_t*) alloca(numDevices * sizeof(::size_t)); + const unsigned char** images = (const unsigned char**) alloca(numDevices * sizeof(const unsigned char**)); + + for (::size_t i = 0; i < numDevices; ++i) { + images[i] = (const unsigned char*)binaries[i].first; + lengths[i] = binaries[(int)i].second; + } + + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + if(binaryStatus) { + binaryStatus->resize(numDevices); + } + + object_ = ::clCreateProgramWithBinary( + context(), (cl_uint) devices.size(), + deviceIDs, + lengths, images, binaryStatus != NULL + ? &binaryStatus->front() + : NULL, &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); + if (err != NULL) { + *err = error; + } + } + + +#if defined(CL_VERSION_1_2) + /** + * Create program using builtin kernels. + * \param kernelNames Semi-colon separated list of builtin kernel names + */ + Program( + const Context& context, + const VECTOR_CLASS& devices, + const STRING_CLASS& kernelNames, + cl_int* err = NULL) + { + cl_int error; + + + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + object_ = ::clCreateProgramWithBuiltInKernels( + context(), + (cl_uint) devices.size(), + deviceIDs, + kernelNames.c_str(), + &error); + + detail::errHandler(error, __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR); + if (err != NULL) { + *err = error; + } + } +#endif // #if defined(CL_VERSION_1_2) + + Program() { } + + Program(const Program& program) : detail::Wrapper(program) { } + + __CL_EXPLICIT_CONSTRUCTORS Program(const cl_program& program) : detail::Wrapper(program) { } + + Program& operator = (const Program& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + Program& operator = (const cl_program& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + cl_int build( + const VECTOR_CLASS& devices, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + ::size_t numDevices = devices.size(); + cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); + for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { + deviceIDs[deviceIndex] = (devices[deviceIndex])(); + } + + return detail::errHandler( + ::clBuildProgram( + object_, + (cl_uint) + devices.size(), + deviceIDs, + options, + notifyFptr, + data), + __BUILD_PROGRAM_ERR); + } + + cl_int build( + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + return detail::errHandler( + ::clBuildProgram( + object_, + 0, + NULL, + options, + notifyFptr, + data), + __BUILD_PROGRAM_ERR); + } + +#if defined(CL_VERSION_1_2) + cl_int compile( + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL) const + { + return detail::errHandler( + ::clCompileProgram( + object_, + 0, + NULL, + options, + 0, + NULL, + NULL, + notifyFptr, + data), + __COMPILE_PROGRAM_ERR); + } +#endif + + template + cl_int getInfo(cl_program_info name, T* param) const + { + return detail::errHandler( + detail::getInfo(&::clGetProgramInfo, object_, name, param), + __GET_PROGRAM_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_program_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + template + cl_int getBuildInfo( + const Device& device, cl_program_build_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetProgramBuildInfo, object_, device(), name, param), + __GET_PROGRAM_BUILD_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getBuildInfo(const Device& device, cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_program_build_info, name>::param_type param; + cl_int result = getBuildInfo(device, name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + cl_int createKernels(VECTOR_CLASS* kernels) + { + cl_uint numKernels; + cl_int err = ::clCreateKernelsInProgram(object_, 0, NULL, &numKernels); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); + } + + Kernel* value = (Kernel*) alloca(numKernels * sizeof(Kernel)); + err = ::clCreateKernelsInProgram( + object_, numKernels, (cl_kernel*) value, NULL); + if (err != CL_SUCCESS) { + return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); + } + + kernels->assign(&value[0], &value[numKernels]); + return CL_SUCCESS; + } +}; + +#if defined(CL_VERSION_1_2) +inline Program linkProgram( + Program input1, + Program input2, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL, + cl_int* err = NULL) +{ + cl_int err_local = CL_SUCCESS; + + cl_program programs[2] = { input1(), input2() }; + + Context ctx = input1.getInfo(); + + cl_program prog = ::clLinkProgram( + ctx(), + 0, + NULL, + options, + 2, + programs, + notifyFptr, + data, + &err_local); + + detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); + if (err != NULL) { + *err = err_local; + } + + return Program(prog); +} + +inline Program linkProgram( + VECTOR_CLASS inputPrograms, + const char* options = NULL, + void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, + void* data = NULL, + cl_int* err = NULL) +{ + cl_int err_local = CL_SUCCESS; + + cl_program * programs = (cl_program*) alloca(inputPrograms.size() * sizeof(cl_program)); + + if (programs != NULL) { + for (unsigned int i = 0; i < inputPrograms.size(); i++) { + programs[i] = inputPrograms[i](); + } + } + + cl_program prog = ::clLinkProgram( + Context::getDefault()(), + 0, + NULL, + options, + (cl_uint)inputPrograms.size(), + programs, + notifyFptr, + data, + &err_local); + + detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); + if (err != NULL) { + *err = err_local; + } + + return Program(prog); +} +#endif + +template<> +inline VECTOR_CLASS cl::Program::getInfo(cl_int* err) const +{ + VECTOR_CLASS< ::size_t> sizes = getInfo(); + VECTOR_CLASS binaries; + for (VECTOR_CLASS< ::size_t>::iterator s = sizes.begin(); s != sizes.end(); ++s) + { + char *ptr = NULL; + if (*s != 0) + ptr = new char[*s]; + binaries.push_back(ptr); + } + + cl_int result = getInfo(CL_PROGRAM_BINARIES, &binaries); + if (err != NULL) { + *err = result; + } + return binaries; +} + +inline Kernel::Kernel(const Program& program, const char* name, cl_int* err) +{ + cl_int error; + + object_ = ::clCreateKernel(program(), name, &error); + detail::errHandler(error, __CREATE_KERNEL_ERR); + + if (err != NULL) { + *err = error; + } + +} + +/*! \class CommandQueue + * \brief CommandQueue interface for cl_command_queue. + */ +class CommandQueue : public detail::Wrapper +{ +private: + static volatile int default_initialized_; + static CommandQueue default_; + static volatile cl_int default_error_; +public: + CommandQueue( + cl_command_queue_properties properties, + cl_int* err = NULL) + { + cl_int error; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + Device device = context.getInfo()[0]; + + object_ = ::clCreateCommandQueue( + context(), device(), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + } + /*! + * \brief Constructs a CommandQueue for an implementation defined device in the given context + */ + explicit CommandQueue( + const Context& context, + cl_command_queue_properties properties = 0, + cl_int* err = NULL) + { + cl_int error; + VECTOR_CLASS devices; + error = context.getInfo(CL_CONTEXT_DEVICES, &devices); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) + { + if (err != NULL) { + *err = error; + } + return; + } + + object_ = ::clCreateCommandQueue(context(), devices[0](), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (err != NULL) { + *err = error; + } + + } + + CommandQueue( + const Context& context, + const Device& device, + cl_command_queue_properties properties = 0, + cl_int* err = NULL) + { + cl_int error; + object_ = ::clCreateCommandQueue( + context(), device(), properties, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + + static CommandQueue getDefault(cl_int * err = NULL) + { + int state = detail::compare_exchange( + &default_initialized_, + __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); + + if (state & __DEFAULT_INITIALIZED) { + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + if (state & __DEFAULT_BEING_INITIALIZED) { + // Assume writes will propagate eventually... + while(default_initialized_ != __DEFAULT_INITIALIZED) { + detail::fence(); + } + + if (err != NULL) { + *err = default_error_; + } + return default_; + } + + cl_int error; + + Context context = Context::getDefault(&error); + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + + if (error != CL_SUCCESS) { + if (err != NULL) { + *err = error; + } + } + else { + Device device = context.getInfo()[0]; + + default_ = CommandQueue(context, device, 0, &error); + + detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); + if (err != NULL) { + *err = error; + } + } + + detail::fence(); + + default_error_ = error; + // Assume writes will propagate eventually... + default_initialized_ = __DEFAULT_INITIALIZED; + + detail::fence(); + + if (err != NULL) { + *err = default_error_; + } + return default_; + + } + + CommandQueue() { } + + CommandQueue(const CommandQueue& commandQueue) : detail::Wrapper(commandQueue) { } + + CommandQueue(const cl_command_queue& commandQueue) : detail::Wrapper(commandQueue) { } + + CommandQueue& operator = (const CommandQueue& rhs) + { + if (this != &rhs) { + detail::Wrapper::operator=(rhs); + } + return *this; + } + + CommandQueue& operator = (const cl_command_queue& rhs) + { + detail::Wrapper::operator=(rhs); + return *this; + } + + template + cl_int getInfo(cl_command_queue_info name, T* param) const + { + return detail::errHandler( + detail::getInfo( + &::clGetCommandQueueInfo, object_, name, param), + __GET_COMMAND_QUEUE_INFO_ERR); + } + + template typename + detail::param_traits::param_type + getInfo(cl_int* err = NULL) const + { + typename detail::param_traits< + detail::cl_command_queue_info, name>::param_type param; + cl_int result = getInfo(name, ¶m); + if (err != NULL) { + *err = result; + } + return param; + } + + cl_int enqueueReadBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadBuffer( + object_, buffer(), blocking, offset, size, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + const void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteBuffer( + object_, buffer(), blocking, offset, size, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBuffer( + const Buffer& src, + const Buffer& dst, + ::size_t src_offset, + ::size_t dst_offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBuffer( + object_, src(), dst(), src_offset, dst_offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQEUE_COPY_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReadBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadBufferRect( + object_, + buffer(), + blocking, + (const ::size_t *)buffer_offset, + (const ::size_t *)host_offset, + (const ::size_t *)region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteBufferRect( + object_, + buffer(), + blocking, + (const ::size_t *)buffer_offset, + (const ::size_t *)host_offset, + (const ::size_t *)region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBufferRect( + const Buffer& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + ::size_t src_row_pitch, + ::size_t src_slice_pitch, + ::size_t dst_row_pitch, + ::size_t dst_slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBufferRect( + object_, + src(), + dst(), + (const ::size_t *)src_origin, + (const ::size_t *)dst_origin, + (const ::size_t *)region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQEUE_COPY_BUFFER_RECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueue a command to fill a buffer object with a pattern + * of a given size. The pattern is specified a as vector. + * \tparam PatternType The datatype of the pattern field. + * The pattern type must be an accepted OpenCL data type. + */ + template + cl_int enqueueFillBuffer( + const Buffer& buffer, + PatternType pattern, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillBuffer( + object_, + buffer(), + static_cast(&pattern), + sizeof(PatternType), + offset, + size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueReadImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReadImage( + object_, image(), blocking, (const ::size_t *) origin, + (const ::size_t *) region, row_pitch, slice_pitch, ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_READ_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueWriteImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueWriteImage( + object_, image(), blocking, (const ::size_t *) origin, + (const ::size_t *) region, row_pitch, slice_pitch, ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_WRITE_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyImage( + const Image& src, + const Image& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyImage( + object_, src(), dst(), (const ::size_t *) src_origin, + (const ::size_t *)dst_origin, (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA floating-point color value if + * the image channel data type is not an unnormalized signed or + * unsigned data type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_float4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA signed integer color value if + * the image channel data type is an unnormalized signed integer + * type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_int4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueue a command to fill an image object with a specified color. + * \param fillColor is the color to use to fill the image. + * This is a four component RGBA unsigned integer color value if + * the image channel data type is an unnormalized unsigned integer + * type. + */ + cl_int enqueueFillImage( + const Image& image, + cl_uint4 fillColor, + const size_t<3>& origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueFillImage( + object_, + image(), + static_cast(&fillColor), + (const ::size_t *) origin, + (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_FILL_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueCopyImageToBuffer( + const Image& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& region, + ::size_t dst_offset, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyImageToBuffer( + object_, src(), dst(), (const ::size_t *) src_origin, + (const ::size_t *) region, dst_offset, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueCopyBufferToImage( + const Buffer& src, + const Image& dst, + ::size_t src_offset, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueCopyBufferToImage( + object_, src(), dst(), src_offset, + (const ::size_t *) dst_origin, (const ::size_t *) region, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + void* enqueueMapBuffer( + const Buffer& buffer, + cl_bool blocking, + cl_map_flags flags, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) const + { + cl_int error; + void * result = ::clEnqueueMapBuffer( + object_, buffer(), blocking, flags, offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + return result; + } + + void* enqueueMapImage( + const Image& buffer, + cl_bool blocking, + cl_map_flags flags, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t * row_pitch, + ::size_t * slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) const + { + cl_int error; + void * result = ::clEnqueueMapImage( + object_, buffer(), blocking, flags, + (const ::size_t *) origin, (const ::size_t *) region, + row_pitch, slice_pitch, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_IMAGE_ERR); + if (err != NULL) { + *err = error; + } + return result; + } + + cl_int enqueueUnmapMemObject( + const Memory& memory, + void* mapped_ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueUnmapMemObject( + object_, memory(), mapped_ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined(CL_VERSION_1_2) + /** + * Enqueues a marker command which waits for either a list of events to complete, + * or all previously enqueued commands to complete. + * + * Enqueues a marker command which waits for either a list of events to complete, + * or if the list is empty it waits for all commands previously enqueued in command_queue + * to complete before it completes. This command returns an event which can be waited on, + * i.e. this event can be waited on to insure that all events either in the event_wait_list + * or all previously enqueued commands, queued before this command to command_queue, + * have completed. + */ + cl_int enqueueMarkerWithWaitList( + const VECTOR_CLASS *events = 0, + Event *event = 0) + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueMarkerWithWaitList( + object_, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_MARKER_WAIT_LIST_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * A synchronization point that enqueues a barrier operation. + * + * Enqueues a barrier command which waits for either a list of events to complete, + * or if the list is empty it waits for all commands previously enqueued in command_queue + * to complete before it completes. This command blocks command execution, that is, any + * following commands enqueued after it do not execute until it completes. This command + * returns an event which can be waited on, i.e. this event can be waited on to insure that + * all events either in the event_wait_list or all previously enqueued commands, queued + * before this command to command_queue, have completed. + */ + cl_int enqueueBarrierWithWaitList( + const VECTOR_CLASS *events = 0, + Event *event = 0) + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueBarrierWithWaitList( + object_, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_BARRIER_WAIT_LIST_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + /** + * Enqueues a command to indicate with which device a set of memory objects + * should be associated. + */ + cl_int enqueueMigrateMemObjects( + const VECTOR_CLASS &memObjects, + cl_mem_migration_flags flags, + const VECTOR_CLASS* events = NULL, + Event* event = NULL + ) + { + cl_event tmp; + + cl_mem* localMemObjects = static_cast(alloca(memObjects.size() * sizeof(cl_mem))); + for( int i = 0; i < (int)memObjects.size(); ++i ) { + localMemObjects[i] = memObjects[i](); + } + + + cl_int err = detail::errHandler( + ::clEnqueueMigrateMemObjects( + object_, + (cl_uint)memObjects.size(), + static_cast(localMemObjects), + flags, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif // #if defined(CL_VERSION_1_2) + + cl_int enqueueNDRangeKernel( + const Kernel& kernel, + const NDRange& offset, + const NDRange& global, + const NDRange& local = NullRange, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueNDRangeKernel( + object_, kernel(), (cl_uint) global.dimensions(), + offset.dimensions() != 0 ? (const ::size_t*) offset : NULL, + (const ::size_t*) global, + local.dimensions() != 0 ? (const ::size_t*) local : NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_NDRANGE_KERNEL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueTask( + const Kernel& kernel, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueTask( + object_, kernel(), + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_TASK_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueNativeKernel( + void (CL_CALLBACK *userFptr)(void *), + std::pair args, + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* mem_locs = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_mem * mems = (mem_objects != NULL && mem_objects->size() > 0) + ? (cl_mem*) alloca(mem_objects->size() * sizeof(cl_mem)) + : NULL; + + if (mems != NULL) { + for (unsigned int i = 0; i < mem_objects->size(); i++) { + mems[i] = ((*mem_objects)[i])(); + } + } + + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueNativeKernel( + object_, userFptr, args.first, args.second, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + mems, + (mem_locs != NULL) ? (const void **) &mem_locs->front() : NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_NATIVE_KERNEL); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueMarker(Event* event = NULL) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueMarker(object_, (cl_event*) event), + __ENQUEUE_MARKER_ERR); + } + + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueWaitForEvents(const VECTOR_CLASS& events) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueWaitForEvents( + object_, + (cl_uint) events.size(), + (const cl_event*) &events.front()), + __ENQUEUE_WAIT_FOR_EVENTS_ERR); + } +#endif // #if defined(CL_VERSION_1_1) + + cl_int enqueueAcquireGLObjects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueAcquireGLObjects( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_ACQUIRE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReleaseGLObjects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueReleaseGLObjects( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_RELEASE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + +#if defined (USE_DX_INTEROP) +typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueAcquireD3D10ObjectsKHR)( + cl_command_queue command_queue, cl_uint num_objects, + const cl_mem* mem_objects, cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, cl_event* event); +typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)( + cl_command_queue command_queue, cl_uint num_objects, + const cl_mem* mem_objects, cl_uint num_events_in_wait_list, + const cl_event* event_wait_list, cl_event* event); + + cl_int enqueueAcquireD3D10Objects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL; +#if defined(CL_VERSION_1_2) + cl_context context = getInfo(); + cl::Device device(getInfo()); + cl_platform_id platform = device.getInfo(); + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueAcquireD3D10ObjectsKHR); +#endif +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clEnqueueAcquireD3D10ObjectsKHR); +#endif + + cl_event tmp; + cl_int err = detail::errHandler( + pfn_clEnqueueAcquireD3D10ObjectsKHR( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_ACQUIRE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } + + cl_int enqueueReleaseD3D10Objects( + const VECTOR_CLASS* mem_objects = NULL, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) const + { + static PFN_clEnqueueReleaseD3D10ObjectsKHR pfn_clEnqueueReleaseD3D10ObjectsKHR = NULL; +#if defined(CL_VERSION_1_2) + cl_context context = getInfo(); + cl::Device device(getInfo()); + cl_platform_id platform = device.getInfo(); + __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueReleaseD3D10ObjectsKHR); +#endif // #if defined(CL_VERSION_1_2) +#if defined(CL_VERSION_1_1) + __INIT_CL_EXT_FCN_PTR(clEnqueueReleaseD3D10ObjectsKHR); +#endif // #if defined(CL_VERSION_1_1) + + cl_event tmp; + cl_int err = detail::errHandler( + pfn_clEnqueueReleaseD3D10ObjectsKHR( + object_, + (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, + (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_RELEASE_GL_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; + } +#endif + +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) + CL_EXT_PREFIX__VERSION_1_1_DEPRECATED + cl_int enqueueBarrier() const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED + { + return detail::errHandler( + ::clEnqueueBarrier(object_), + __ENQUEUE_BARRIER_ERR); + } +#endif // #if defined(CL_VERSION_1_1) + + cl_int flush() const + { + return detail::errHandler(::clFlush(object_), __FLUSH_ERR); + } + + cl_int finish() const + { + return detail::errHandler(::clFinish(object_), __FINISH_ERR); + } +}; + +#ifdef _WIN32 +__declspec(selectany) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__declspec(selectany) CommandQueue CommandQueue::default_; +__declspec(selectany) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; +#else +__attribute__((weak)) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; +__attribute__((weak)) CommandQueue CommandQueue::default_; +__attribute__((weak)) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; +#endif + +template< typename IteratorType > +Buffer::Buffer( + const Context &context, + IteratorType startIterator, + IteratorType endIterator, + bool readOnly, + bool useHostPtr, + cl_int* err) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + cl_mem_flags flags = 0; + if( readOnly ) { + flags |= CL_MEM_READ_ONLY; + } + else { + flags |= CL_MEM_READ_WRITE; + } + if( useHostPtr ) { + flags |= CL_MEM_USE_HOST_PTR; + } + + ::size_t size = sizeof(DataType)*(endIterator - startIterator); + + if( useHostPtr ) { + object_ = ::clCreateBuffer(context(), flags, size, static_cast(&*startIterator), &error); + } else { + object_ = ::clCreateBuffer(context(), flags, size, 0, &error); + } + + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + if( !useHostPtr ) { + CommandQueue queue(context, 0, &error); + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + error = cl::copy(queue, startIterator, endIterator, *this); + detail::errHandler(error, __CREATE_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + } +} + +inline cl_int enqueueReadBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadBuffer(buffer, blocking, offset, size, ptr, events, event); +} + +inline cl_int enqueueWriteBuffer( + const Buffer& buffer, + cl_bool blocking, + ::size_t offset, + ::size_t size, + const void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteBuffer(buffer, blocking, offset, size, ptr, events, event); +} + +inline void* enqueueMapBuffer( + const Buffer& buffer, + cl_bool blocking, + cl_map_flags flags, + ::size_t offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL, + cl_int* err = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + + void * result = ::clEnqueueMapBuffer( + queue(), buffer(), blocking, flags, offset, size, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (cl_event*) event, + &error); + + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (err != NULL) { + *err = error; + } + return result; +} + +inline cl_int enqueueUnmapMemObject( + const Memory& memory, + void* mapped_ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); + if (error != CL_SUCCESS) { + return error; + } + + cl_event tmp; + cl_int err = detail::errHandler( + ::clEnqueueUnmapMemObject( + queue(), memory(), mapped_ptr, + (events != NULL) ? (cl_uint) events->size() : 0, + (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, + (event != NULL) ? &tmp : NULL), + __ENQUEUE_UNMAP_MEM_OBJECT_ERR); + + if (event != NULL && err == CL_SUCCESS) + *event = tmp; + + return err; +} + +inline cl_int enqueueCopyBuffer( + const Buffer& src, + const Buffer& dst, + ::size_t src_offset, + ::size_t dst_offset, + ::size_t size, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBuffer(src, dst, src_offset, dst_offset, size, events, event); +} + +/** + * Blocking copy operation between iterators and a buffer. + * Host to Device. + * Uses default command queue. + */ +template< typename IteratorType > +inline cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + if (error != CL_SUCCESS) + return error; + + return cl::copy(queue, startIterator, endIterator, buffer); +} + +/** + * Blocking copy operation between iterators and a buffer. + * Device to Host. + * Uses default command queue. + */ +template< typename IteratorType > +inline cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + if (error != CL_SUCCESS) + return error; + + return cl::copy(queue, buffer, startIterator, endIterator); +} + +/** + * Blocking copy operation between iterators and a buffer. + * Host to Device. + * Uses specified queue. + */ +template< typename IteratorType > +inline cl_int copy( const CommandQueue &queue, IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + ::size_t length = endIterator-startIterator; + ::size_t byteLength = length*sizeof(DataType); + + DataType *pointer = + static_cast(queue.enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_WRITE, 0, byteLength, 0, 0, &error)); + // if exceptions enabled, enqueueMapBuffer will throw + if( error != CL_SUCCESS ) { + return error; + } +#if defined(_MSC_VER) + std::copy( + startIterator, + endIterator, + stdext::checked_array_iterator( + pointer, length)); +#else + std::copy(startIterator, endIterator, pointer); +#endif + Event endEvent; + error = queue.enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); + // if exceptions enabled, enqueueUnmapMemObject will throw + if( error != CL_SUCCESS ) { + return error; + } + endEvent.wait(); + return CL_SUCCESS; +} + +/** + * Blocking copy operation between iterators and a buffer. + * Device to Host. + * Uses specified queue. + */ +template< typename IteratorType > +inline cl_int copy( const CommandQueue &queue, const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ) +{ + typedef typename std::iterator_traits::value_type DataType; + cl_int error; + + ::size_t length = endIterator-startIterator; + ::size_t byteLength = length*sizeof(DataType); + + DataType *pointer = + static_cast(queue.enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_READ, 0, byteLength, 0, 0, &error)); + // if exceptions enabled, enqueueMapBuffer will throw + if( error != CL_SUCCESS ) { + return error; + } + std::copy(pointer, pointer + length, startIterator); + Event endEvent; + error = queue.enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); + // if exceptions enabled, enqueueUnmapMemObject will throw + if( error != CL_SUCCESS ) { + return error; + } + endEvent.wait(); + return CL_SUCCESS; +} + +#if defined(CL_VERSION_1_1) +inline cl_int enqueueReadBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadBufferRect( + buffer, + blocking, + buffer_offset, + host_offset, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueWriteBufferRect( + const Buffer& buffer, + cl_bool blocking, + const size_t<3>& buffer_offset, + const size_t<3>& host_offset, + const size_t<3>& region, + ::size_t buffer_row_pitch, + ::size_t buffer_slice_pitch, + ::size_t host_row_pitch, + ::size_t host_slice_pitch, + void *ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteBufferRect( + buffer, + blocking, + buffer_offset, + host_offset, + region, + buffer_row_pitch, + buffer_slice_pitch, + host_row_pitch, + host_slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueCopyBufferRect( + const Buffer& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + ::size_t src_row_pitch, + ::size_t src_slice_pitch, + ::size_t dst_row_pitch, + ::size_t dst_slice_pitch, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBufferRect( + src, + dst, + src_origin, + dst_origin, + region, + src_row_pitch, + src_slice_pitch, + dst_row_pitch, + dst_slice_pitch, + events, + event); +} +#endif + +inline cl_int enqueueReadImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueReadImage( + image, + blocking, + origin, + region, + row_pitch, + slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueWriteImage( + const Image& image, + cl_bool blocking, + const size_t<3>& origin, + const size_t<3>& region, + ::size_t row_pitch, + ::size_t slice_pitch, + void* ptr, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueWriteImage( + image, + blocking, + origin, + region, + row_pitch, + slice_pitch, + ptr, + events, + event); +} + +inline cl_int enqueueCopyImage( + const Image& src, + const Image& dst, + const size_t<3>& src_origin, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyImage( + src, + dst, + src_origin, + dst_origin, + region, + events, + event); +} + +inline cl_int enqueueCopyImageToBuffer( + const Image& src, + const Buffer& dst, + const size_t<3>& src_origin, + const size_t<3>& region, + ::size_t dst_offset, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyImageToBuffer( + src, + dst, + src_origin, + region, + dst_offset, + events, + event); +} + +inline cl_int enqueueCopyBufferToImage( + const Buffer& src, + const Image& dst, + ::size_t src_offset, + const size_t<3>& dst_origin, + const size_t<3>& region, + const VECTOR_CLASS* events = NULL, + Event* event = NULL) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.enqueueCopyBufferToImage( + src, + dst, + src_offset, + dst_origin, + region, + events, + event); +} + + +inline cl_int flush(void) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + return queue.flush(); +} + +inline cl_int finish(void) +{ + cl_int error; + CommandQueue queue = CommandQueue::getDefault(&error); + + if (error != CL_SUCCESS) { + return error; + } + + + return queue.finish(); +} + +// Kernel Functor support +// New interface as of September 2011 +// Requires the C++11 std::tr1::function (note do not support TR1) +// Visual Studio 2010 and GCC 4.2 + +struct EnqueueArgs +{ + CommandQueue queue_; + const NDRange offset_; + const NDRange global_; + const NDRange local_; + VECTOR_CLASS events_; + + EnqueueArgs(NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange) + { + + } + + EnqueueArgs(NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local) + { + + } + + EnqueueArgs(NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local) + { + + } + + EnqueueArgs(Event e, NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange) + { + events_.push_back(e); + } + + EnqueueArgs(Event e, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(Event e, NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange global) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(NullRange), + events_(events) + { + + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(NullRange), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : + queue_(CommandQueue::getDefault()), + offset_(offset), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local) + { + + } + + EnqueueArgs(CommandQueue &queue, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local) + { + + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, Event e, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local) + { + events_.push_back(e); + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(NullRange), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global, NDRange local) : + queue_(queue), + offset_(NullRange), + global_(global), + local_(local), + events_(events) + { + + } + + EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : + queue_(queue), + offset_(offset), + global_(global), + local_(local), + events_(events) + { + + } +}; + +namespace detail { + +class NullType {}; + +template +struct SetArg +{ + static void set (Kernel kernel, T0 arg) + { + kernel.setArg(index, arg); + } +}; + +template +struct SetArg +{ + static void set (Kernel, NullType) + { + } +}; + +template < + typename T0, typename T1, typename T2, typename T3, + typename T4, typename T5, typename T6, typename T7, + typename T8, typename T9, typename T10, typename T11, + typename T12, typename T13, typename T14, typename T15, + typename T16, typename T17, typename T18, typename T19, + typename T20, typename T21, typename T22, typename T23, + typename T24, typename T25, typename T26, typename T27, + typename T28, typename T29, typename T30, typename T31 +> +class KernelFunctorGlobal +{ +private: + Kernel kernel_; + +public: + KernelFunctorGlobal( + Kernel kernel) : + kernel_(kernel) + {} + + KernelFunctorGlobal( + const Program& program, + const STRING_CLASS name, + cl_int * err = NULL) : + kernel_(program, name.c_str(), err) + {} + + Event operator() ( + const EnqueueArgs& args, + T0 t0, + T1 t1 = NullType(), + T2 t2 = NullType(), + T3 t3 = NullType(), + T4 t4 = NullType(), + T5 t5 = NullType(), + T6 t6 = NullType(), + T7 t7 = NullType(), + T8 t8 = NullType(), + T9 t9 = NullType(), + T10 t10 = NullType(), + T11 t11 = NullType(), + T12 t12 = NullType(), + T13 t13 = NullType(), + T14 t14 = NullType(), + T15 t15 = NullType(), + T16 t16 = NullType(), + T17 t17 = NullType(), + T18 t18 = NullType(), + T19 t19 = NullType(), + T20 t20 = NullType(), + T21 t21 = NullType(), + T22 t22 = NullType(), + T23 t23 = NullType(), + T24 t24 = NullType(), + T25 t25 = NullType(), + T26 t26 = NullType(), + T27 t27 = NullType(), + T28 t28 = NullType(), + T29 t29 = NullType(), + T30 t30 = NullType(), + T31 t31 = NullType() + ) + { + Event event; + SetArg<0, T0>::set(kernel_, t0); + SetArg<1, T1>::set(kernel_, t1); + SetArg<2, T2>::set(kernel_, t2); + SetArg<3, T3>::set(kernel_, t3); + SetArg<4, T4>::set(kernel_, t4); + SetArg<5, T5>::set(kernel_, t5); + SetArg<6, T6>::set(kernel_, t6); + SetArg<7, T7>::set(kernel_, t7); + SetArg<8, T8>::set(kernel_, t8); + SetArg<9, T9>::set(kernel_, t9); + SetArg<10, T10>::set(kernel_, t10); + SetArg<11, T11>::set(kernel_, t11); + SetArg<12, T12>::set(kernel_, t12); + SetArg<13, T13>::set(kernel_, t13); + SetArg<14, T14>::set(kernel_, t14); + SetArg<15, T15>::set(kernel_, t15); + SetArg<16, T16>::set(kernel_, t16); + SetArg<17, T17>::set(kernel_, t17); + SetArg<18, T18>::set(kernel_, t18); + SetArg<19, T19>::set(kernel_, t19); + SetArg<20, T20>::set(kernel_, t20); + SetArg<21, T21>::set(kernel_, t21); + SetArg<22, T22>::set(kernel_, t22); + SetArg<23, T23>::set(kernel_, t23); + SetArg<24, T24>::set(kernel_, t24); + SetArg<25, T25>::set(kernel_, t25); + SetArg<26, T26>::set(kernel_, t26); + SetArg<27, T27>::set(kernel_, t27); + SetArg<28, T28>::set(kernel_, t28); + SetArg<29, T29>::set(kernel_, t29); + SetArg<30, T30>::set(kernel_, t30); + SetArg<31, T31>::set(kernel_, t31); + + args.queue_.enqueueNDRangeKernel( + kernel_, + args.offset_, + args.global_, + args.local_, + &args.events_, + &event); + + return event; + } + +}; + +//------------------------------------------------------------------------------------------------------ + + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29, + typename T30, + typename T31> +struct functionImplementation_ +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + T31> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 32)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + T31); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29, + T30 arg30, + T31 arg31) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29, + arg30, + arg31); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29, + typename T30> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 31)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + T30); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29, + T30 arg30) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29, + arg30); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28, + typename T29> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 30)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + T29); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28, + T29 arg29) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28, + arg29); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27, + typename T28> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 29)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + T28); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27, + T28 arg28) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27, + arg28); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26, + typename T27> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 28)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + T27); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26, + T27 arg27) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26, + arg27); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25, + typename T26> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 27)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + T26); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25, + T26 arg26) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25, + arg26); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24, + typename T25> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 26)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + T25); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24, + T25 arg25) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24, + arg25); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23, + typename T24> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 25)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + T24); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23, + T24 arg24) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23, + arg24); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22, + typename T23> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 24)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + T23); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22, + T23 arg23) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22, + arg23); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21, + typename T22> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 23)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + T22); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21, + T22 arg22) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21, + arg22); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20, + typename T21> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 22)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + T21); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20, + T21 arg21) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20, + arg21); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19, + typename T20> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 21)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + T20); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19, + T20 arg20) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19, + arg20); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18, + typename T19> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 20)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + T19); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18, + T19 arg19) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18, + arg19); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17, + typename T18> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 19)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + T18); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17, + T18 arg18) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17, + arg18); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16, + typename T17> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 18)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + T17); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16, + T17 arg17) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16, + arg17); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15, + typename T16> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 17)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + T16); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15, + T16 arg16) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15, + arg16); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14, + typename T15> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 16)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + T15); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14, + T15 arg15) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14, + arg15); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13, + typename T14> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 15)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + T14); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13, + T14 arg14) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13, + arg14); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12, + typename T13> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 14)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + T13); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12, + T13 arg13) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12, + arg13); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11, + typename T12> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 13)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + T12); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11, + T12 arg12) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11, + arg12); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10, + typename T11> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 12)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + T11); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10, + T11 arg11) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10, + arg11); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9, + typename T10> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 11)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + T10); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9, + T10 arg10) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9, + arg10); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8, + typename T9> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 10)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + T9); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8, + T9 arg9) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8, + arg9); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7, + typename T8> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 9)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + T8); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7, + T8 arg8) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7, + arg8); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6, + typename T7> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 8)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6, + T7); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6, + T7 arg7) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6, + arg7); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5, + typename T6> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + T6, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + T6, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 7)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5, + T6); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5, + T6 arg6) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5, + arg6); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4, + typename T5> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + T5, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + T5, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 6)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4, + T5); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4, + T5 arg5) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4, + arg5); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3, + typename T4> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + T4, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + T4, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 5)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3, + T4); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3, + T4 arg4) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3, + arg4); + } + + +}; + +template< + typename T0, + typename T1, + typename T2, + typename T3> +struct functionImplementation_ +< T0, + T1, + T2, + T3, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + T3, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 4)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2, + T3); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2, + T3 arg3) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2, + arg3); + } + + +}; + +template< + typename T0, + typename T1, + typename T2> +struct functionImplementation_ +< T0, + T1, + T2, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + T2, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 3)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1, + T2); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1, + T2 arg2) + { + return functor_( + enqueueArgs, + arg0, + arg1, + arg2); + } + + +}; + +template< + typename T0, + typename T1> +struct functionImplementation_ +< T0, + T1, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + T1, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 2)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0, + T1); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0, + T1 arg1) + { + return functor_( + enqueueArgs, + arg0, + arg1); + } + + +}; + +template< + typename T0> +struct functionImplementation_ +< T0, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> +{ + typedef detail::KernelFunctorGlobal< + T0, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType, + NullType> FunctorType; + + FunctorType functor_; + + functionImplementation_(const FunctorType &functor) : + functor_(functor) + { + + #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 1)) + // Fail variadic expansion for dev11 + static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); + #endif + + } + + //! \brief Return type of the functor + typedef Event result_type; + + //! \brief Function signature of kernel functor with no event dependency. + typedef Event type_( + const EnqueueArgs&, + T0); + + Event operator()( + const EnqueueArgs& enqueueArgs, + T0 arg0) + { + return functor_( + enqueueArgs, + arg0); + } + + +}; + + + + + +} // namespace detail + +//---------------------------------------------------------------------------------------------- + +template < + typename T0, typename T1 = detail::NullType, typename T2 = detail::NullType, + typename T3 = detail::NullType, typename T4 = detail::NullType, + typename T5 = detail::NullType, typename T6 = detail::NullType, + typename T7 = detail::NullType, typename T8 = detail::NullType, + typename T9 = detail::NullType, typename T10 = detail::NullType, + typename T11 = detail::NullType, typename T12 = detail::NullType, + typename T13 = detail::NullType, typename T14 = detail::NullType, + typename T15 = detail::NullType, typename T16 = detail::NullType, + typename T17 = detail::NullType, typename T18 = detail::NullType, + typename T19 = detail::NullType, typename T20 = detail::NullType, + typename T21 = detail::NullType, typename T22 = detail::NullType, + typename T23 = detail::NullType, typename T24 = detail::NullType, + typename T25 = detail::NullType, typename T26 = detail::NullType, + typename T27 = detail::NullType, typename T28 = detail::NullType, + typename T29 = detail::NullType, typename T30 = detail::NullType, + typename T31 = detail::NullType +> +struct make_kernel : + public detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + > +{ +public: + typedef detail::KernelFunctorGlobal< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + > FunctorType; + + make_kernel( + const Program& program, + const STRING_CLASS name, + cl_int * err = NULL) : + detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + >( + FunctorType(program, name, err)) + {} + + make_kernel( + const Kernel kernel) : + detail::functionImplementation_< + T0, T1, T2, T3, + T4, T5, T6, T7, + T8, T9, T10, T11, + T12, T13, T14, T15, + T16, T17, T18, T19, + T20, T21, T22, T23, + T24, T25, T26, T27, + T28, T29, T30, T31 + >( + FunctorType(kernel)) + {} +}; + + +//---------------------------------------------------------------------------------------------------------------------- + +#undef __ERR_STR +#if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) +#undef __GET_DEVICE_INFO_ERR +#undef __GET_PLATFORM_INFO_ERR +#undef __GET_DEVICE_IDS_ERR +#undef __GET_CONTEXT_INFO_ERR +#undef __GET_EVENT_INFO_ERR +#undef __GET_EVENT_PROFILE_INFO_ERR +#undef __GET_MEM_OBJECT_INFO_ERR +#undef __GET_IMAGE_INFO_ERR +#undef __GET_SAMPLER_INFO_ERR +#undef __GET_KERNEL_INFO_ERR +#undef __GET_KERNEL_ARG_INFO_ERR +#undef __GET_KERNEL_WORK_GROUP_INFO_ERR +#undef __GET_PROGRAM_INFO_ERR +#undef __GET_PROGRAM_BUILD_INFO_ERR +#undef __GET_COMMAND_QUEUE_INFO_ERR + +#undef __CREATE_CONTEXT_ERR +#undef __CREATE_CONTEXT_FROM_TYPE_ERR +#undef __GET_SUPPORTED_IMAGE_FORMATS_ERR + +#undef __CREATE_BUFFER_ERR +#undef __CREATE_SUBBUFFER_ERR +#undef __CREATE_IMAGE2D_ERR +#undef __CREATE_IMAGE3D_ERR +#undef __CREATE_SAMPLER_ERR +#undef __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR + +#undef __CREATE_USER_EVENT_ERR +#undef __SET_USER_EVENT_STATUS_ERR +#undef __SET_EVENT_CALLBACK_ERR +#undef __SET_PRINTF_CALLBACK_ERR + +#undef __WAIT_FOR_EVENTS_ERR + +#undef __CREATE_KERNEL_ERR +#undef __SET_KERNEL_ARGS_ERR +#undef __CREATE_PROGRAM_WITH_SOURCE_ERR +#undef __CREATE_PROGRAM_WITH_BINARY_ERR +#undef __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR +#undef __BUILD_PROGRAM_ERR +#undef __CREATE_KERNELS_IN_PROGRAM_ERR + +#undef __CREATE_COMMAND_QUEUE_ERR +#undef __SET_COMMAND_QUEUE_PROPERTY_ERR +#undef __ENQUEUE_READ_BUFFER_ERR +#undef __ENQUEUE_WRITE_BUFFER_ERR +#undef __ENQUEUE_READ_BUFFER_RECT_ERR +#undef __ENQUEUE_WRITE_BUFFER_RECT_ERR +#undef __ENQEUE_COPY_BUFFER_ERR +#undef __ENQEUE_COPY_BUFFER_RECT_ERR +#undef __ENQUEUE_READ_IMAGE_ERR +#undef __ENQUEUE_WRITE_IMAGE_ERR +#undef __ENQUEUE_COPY_IMAGE_ERR +#undef __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR +#undef __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR +#undef __ENQUEUE_MAP_BUFFER_ERR +#undef __ENQUEUE_MAP_IMAGE_ERR +#undef __ENQUEUE_UNMAP_MEM_OBJECT_ERR +#undef __ENQUEUE_NDRANGE_KERNEL_ERR +#undef __ENQUEUE_TASK_ERR +#undef __ENQUEUE_NATIVE_KERNEL + +#undef __CL_EXPLICIT_CONSTRUCTORS + +#undef __UNLOAD_COMPILER_ERR +#endif //__CL_USER_OVERRIDE_ERROR_STRINGS + +#undef __CL_FUNCTION_TYPE + +// Extensions +/** + * Deprecated APIs for 1.2 + */ +#if defined(CL_VERSION_1_1) +#undef __INIT_CL_EXT_FCN_PTR +#endif // #if defined(CL_VERSION_1_1) +#undef __CREATE_SUB_DEVICES + +#if defined(USE_CL_DEVICE_FISSION) +#undef __PARAM_NAME_DEVICE_FISSION +#endif // USE_CL_DEVICE_FISSION + +#undef __DEFAULT_NOT_INITIALIZED +#undef __DEFAULT_BEING_INITIALIZED +#undef __DEFAULT_INITIALIZED + +} // namespace cl + +#ifdef _WIN32 +#pragma pop_macro("max") +#endif // _WIN32 + +#endif // CL_HPP_ diff --git a/opencl/tools/clinfo/clinfo.cpp b/opencl/tools/clinfo/clinfo.cpp new file mode 100644 index 0000000000..6a24e93742 --- /dev/null +++ b/opencl/tools/clinfo/clinfo.cpp @@ -0,0 +1,837 @@ +/* ============================================================ + +Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved. + +Redistribution and use of this material is permitted under the following +conditions: + +Redistributions must retain the above copyright notice and all terms of this +license. + +In no event shall anyone redistributing or accessing or using this material +commence or participate in any arbitration or legal action relating to this +material against Advanced Micro Devices, Inc. or any copyright holders or +contributors. The foregoing shall survive any expiration or termination of +this license or any agreement or access or use related to this material. + +ANY BREACH OF ANY TERM OF THIS LICENSE SHALL RESULT IN THE IMMEDIATE REVOCATION +OF ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE THIS MATERIAL. + +THIS MATERIAL IS PROVIDED BY ADVANCED MICRO DEVICES, INC. AND ANY COPYRIGHT +HOLDERS AND CONTRIBUTORS "AS IS" IN ITS CURRENT CONDITION AND WITHOUT ANY +REPRESENTATIONS, GUARANTEE, OR WARRANTY OF ANY KIND OR IN ANY WAY RELATED TO +SUPPORT, INDEMNITY, ERROR FREE OR UNINTERRUPTED OPERA TION, OR THAT IT IS FREE +FROM DEFECTS OR VIRUSES. ALL OBLIGATIONS ARE HEREBY DISCLAIMED - WHETHER +EXPRESS, IMPLIED, OR STATUTORY - INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED +WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, +ACCURACY, COMPLETENESS, OPERABILITY, QUALITY OF SERVICE, OR NON-INFRINGEMENT. +IN NO EVENT SHALL ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, PUNITIVE, +EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT +OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, REVENUE, DATA, OR PROFITS; OR +BUSINESS INTERRUPTION) HOWEVER CAUSED OR BASED ON ANY THEORY OF LIABILITY +ARISING IN ANY WAY RELATED TO THIS MATERIAL, EVEN IF ADVISED OF THE POSSIBILITY +OF SUCH DAMAGE. THE ENTIRE AND AGGREGATE LIABILITY OF ADVANCED MICRO DEVICES, +INC. AND ANY COPYRIGHT HOLDERS AND CONTRIBUTORS SHALL NOT EXCEED TEN DOLLARS +(US $10.00). ANYONE REDISTRIBUTING OR ACCESSING OR USING THIS MATERIAL ACCEPTS +THIS ALLOCATION OF RISK AND AGREES TO RELEASE ADVANCED MICRO DEVICES, INC. AND +ANY COPYRIGHT HOLDERS AND CONTRIBUTORS FROM ANY AND ALL LIABILITIES, +OBLIGATIONS, CLAIMS, OR DEMANDS IN EXCESS OF TEN DOLLARS (US $10.00). THE +FOREGOING ARE ESSENTIAL TERMS OF THIS LICENSE AND, IF ANY OF THESE TERMS ARE +CONSTRUED AS UNENFORCEABLE, FAIL IN ESSENTIAL PURPOSE, OR BECOME VOID OR +DETRIMENTAL TO ADVANCED MICRO DEVICES, INC. OR ANY COPYRIGHT HOLDERS OR +CONTRIBUTORS FOR ANY REASON, THEN ALL RIGHTS TO REDISTRIBUTE, ACCESS OR USE +THIS MATERIAL SHALL TERMINATE IMMEDIATELY. MOREOVER, THE FOREGOING SHALL +SURVIVE ANY EXPIRATION OR TERMINATION OF THIS LICENSE OR ANY AGREEMENT OR +ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE IS HEREBY PROVIDED, AND BY REDISTRIBUTING OR ACCESSING OR USING THIS +MATERIAL SUCH NOTICE IS ACKNOWLEDGED, THAT THIS MATERIAL MAY BE SUBJECT TO +RESTRICTIONS UNDER THE LAWS AND REGULATIONS OF THE UNITED STATES OR OTHER +COUNTRIES, WHICH INCLUDE BUT ARE NOT LIMITED TO, U.S. EXPORT CONTROL LAWS SUCH +AS THE EXPORT ADMINISTRATION REGULATIONS AND NATIONAL SECURITY CONTROLS AS +DEFINED THEREUNDER, AS WELL AS STATE DEPARTMENT CONTROLS UNDER THE U.S. +MUNITIONS LIST. THIS MATERIAL MAY NOT BE USED, RELEASED, TRANSFERRED, IMPORTED, +EXPORTED AND/OR RE-EXPORTED IN ANY MANNER PROHIBITED UNDER ANY APPLICABLE LAWS, +INCLUDING U.S. EXPORT CONTROL LAWS REGARDING SPECIFICALLY DESIGNATED PERSONS, +COUNTRIES AND NATIONALS OF COUNTRIES SUBJECT TO NATIONAL SECURITY CONTROLS. +MOREOVER, THE FOREGOING SHALL SURVIVE ANY EXPIRATION OR TERMINATION OF ANY +LICENSE OR AGREEMENT OR ACCESS OR USE RELATED TO THIS MATERIAL. + +NOTICE REGARDING THE U.S. GOVERNMENT AND DOD AGENCIES: This material is +provided with "RESTRICTED RIGHTS" and/or "LIMITED RIGHTS" as applicable to +computer software and technical data, respectively. Use, duplication, +distribution or disclosure by the U.S. Government and/or DOD agencies is +subject to the full extent of restrictions in all applicable regulations, +including those found at FAR52.227 and DFARS252.227 et seq. and any successor +regulations thereof. Use of this material by the U.S. Government and/or DOD +agencies is acknowledgment of the proprietary rights of any copyright holders +and contributors, including those of Advanced Micro Devices, Inc., as well as +the provisions of FAR52.227-14 through 23 regarding privately developed and/or +commercial computer software. + +This license forms the entire agreement regarding the subject matter hereof and +supersedes all proposals and prior discussions and writings between the parties +with respect thereto. This license does not affect any ownership, rights, title, +or interest in, or relating to, this material. No terms of this license can be +modified or waived, and no breach of this license can be excused, unless done +so in a writing signed by all affected parties. Each term of this license is +separately enforceable. If any term of this license is determined to be or +becomes unenforceable or illegal, such term shall be reformed to the minimum +extent necessary in order for this license to remain in effect in accordance +with its terms as modified by such reformation. This license shall be governed +by and construed in accordance with the laws of the State of Texas without +regard to rules on conflicts of law of any state or jurisdiction or the United +Nations Convention on the International Sale of Goods. All disputes arising out +of this license shall be subject to the jurisdiction of the federal and state +courts in Austin, Texas, and all defenses are hereby waived concerning personal +jurisdiction and venue of these courts. + +============================================================ */ + +#include +#include +#include +#include +#include +#include +#include +#if !defined(_WIN32) +#include +#endif + +#define __CL_ENABLE_EXCEPTIONS +#define __MAX_DEFAULT_VECTOR_SIZE 50 + +#ifdef _MSC_VER +#pragma warning(disable: 4290) +#endif +#define CL_USE_DEPRECATED_OPENCL_1_1_APIS +#define CL_USE_DEPRECATED_OPENCL_2_0_APIS +#include "cl.hpp" + +bool verbose = false; + +inline +void +checkErr(cl_int err, const char * name) +{ + if (err != CL_SUCCESS) { + std::cerr << "ERROR: " << name << " (" << err << ")" << std::endl; + exit(1); + } +} + + +int +main(int argc, char** argv) +{ + /* Error flag */ + cl_int status = 0; + cl_int err; + + //parse input + for(int i = 1; i < argc; i++){ + if ((strcmp(argv[i], "-v") == 0) || + (strcmp(argv[i], "--verbose") == 0)){ + verbose = true; + } else if ((strcmp(argv[i], "-h") == 0) || + (strcmp(argv[i], "--help") == 0)){ + std::cout << "Usage is: " << argv[0] << " [-v|--verbose]" << std::endl; + return 0; + } + } + + // Platform info + std::vector platforms; + err = cl::Platform::get(&platforms); + + checkErr( + err && (platforms.size() == 0 ? -1 : CL_SUCCESS), + "cl::Platform::get()"); + + try { + // Iteratate over platforms + std::cout << "Number of platforms:\t\t\t\t " + << platforms.size() + << std::endl; + for (std::vector::iterator i = platforms.begin(); + i != platforms.end(); + ++i) { + const cl::Platform& platform = *i; + + std::cout << " Platform Profile:\t\t\t\t " + << platform.getInfo().c_str() + << std::endl; + std::cout << " Platform Version:\t\t\t\t " + << platform.getInfo().c_str() + << std::endl; + std::cout << " Platform Name:\t\t\t\t " + << platform.getInfo().c_str() + << std::endl; + std::cout << " Platform Vendor:\t\t\t\t " + << platform.getInfo().c_str() << std::endl; + if (platform.getInfo().size() > 0) { + std::cout << " Platform Extensions:\t\t\t\t " + << platform.getInfo().c_str() + << std::endl; + } + } + + std::cout << std::endl << std:: endl; + // Now Iteratate over each platform and its devices + for (std::vector::iterator p = platforms.begin(); + p != platforms.end(); + ++p) { + const cl::Platform& platform = *p; + std::cout << " Platform Name:\t\t\t\t " + << platform.getInfo().c_str() + << std::endl; + + std::vector devices; + platform.getDevices(CL_DEVICE_TYPE_ALL, &devices); + + // Get OpenCL version + std::string platformVersionStr = platform.getInfo(); + std::string openclVerstionStr(platformVersionStr.c_str()); + size_t vStart = openclVerstionStr.find(" ", 0); + size_t vEnd = openclVerstionStr.find(" ", vStart + 1); + std::string vStrVal = openclVerstionStr.substr(vStart + 1, vEnd - vStart - 1); + + std::cout << "Number of devices:\t\t\t\t " << devices.size() << std::endl; + for (std::vector::iterator i = devices.begin(); + i != devices.end(); + ++i) { + const cl::Device& device = *i; + /* Get device name */ + std::string deviceName = device.getInfo(); + cl_device_type dtype = device.getInfo(); + + /* Get CAL driver version in int */ + std::string driverVersion = device.getInfo(); + std::string calVersion(driverVersion.c_str()); + calVersion = calVersion.substr(calVersion.find_last_of(".") + 1); + int version = atoi(calVersion.c_str()); + + std::cout << " Device Type:\t\t\t\t\t " ; + switch (dtype) { + case CL_DEVICE_TYPE_ACCELERATOR: + std::cout << "CL_DEVICE_TYPE_ACCRLERATOR" << std::endl; + break; + case CL_DEVICE_TYPE_CPU: + std::cout << "CL_DEVICE_TYPE_CPU" << std::endl; + break; + case CL_DEVICE_TYPE_DEFAULT: + std::cout << "CL_DEVICE_TYPE_DEFAULT" << std::endl; + break; + case CL_DEVICE_TYPE_GPU: + std::cout << "CL_DEVICE_TYPE_GPU" << std::endl; + break; + } + + std::cout << " Vendor ID:\t\t\t\t\t " + << std::hex + << device.getInfo() + << "h" + << std::dec + << std::endl; + + bool isAMDPlatform = (strcmp(platform.getInfo().c_str(), "AMD Accelerated Parallel Processing") == 0) ? true : false; + if (isAMDPlatform) + { + std::string boardName; + device.getInfo(CL_DEVICE_BOARD_NAME_AMD, &boardName); + std::cout << " Board name:\t\t\t\t\t " + << boardName.c_str() + << std::endl; + + cl_device_topology_amd topology; + err = device.getInfo(CL_DEVICE_TOPOLOGY_AMD, &topology); + if (topology.raw.type == CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD) { + std::cout << " Device Topology:\t\t\t\t " + << "PCI[ B#" << (int)topology.pcie.bus + << ", D#" << (int)topology.pcie.device + << ", F#" << (int)topology.pcie.function + << " ]" << std::endl; + } + } + + std::cout << " Max compute units:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max work items dimensions:\t\t\t " + << device.getInfo() + << std::endl; + + std::vector< ::size_t> witems = + device.getInfo(); + for (unsigned int x = 0; + x < device.getInfo(); + x++) { + std::cout << " Max work items[" + << x << "]:\t\t\t\t " + << witems[x] + << std::endl; + } + + std::cout << " Max work group size:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred vector width char:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred vector width short:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred vector width int:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred vector width long:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred vector width float:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred vector width double:\t\t " + << device.getInfo() + << std::endl; + +#ifdef CL_VERSION_1_1 + if(vStrVal.compare("1.0") > 0) + { + std::cout << " Native vector width char:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Native vector width short:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Native vector width int:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Native vector width long:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Native vector width float:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Native vector width double:\t\t\t " + << device.getInfo() + << std::endl; + } +#endif // CL_VERSION_1_1 + std::cout << " Max clock frequency:\t\t\t\t " + << device.getInfo() + << "Mhz" + << std::endl; + + std::cout << " Address bits:\t\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max memory allocation:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Image support:\t\t\t\t " + << (device.getInfo() ? "Yes" : "No") + << std::endl; + + if (device.getInfo()) + { + + std::cout << " Max number of images read arguments:\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max number of images write arguments:\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max image 2D width:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max image 2D height:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max image 3D width:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max image 3D height:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max image 3D depth:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max samplers within kernel:\t\t\t " + << device.getInfo() + << std::endl; + + if (verbose) + { + std::cout << " Image formats supported:" << std::endl; + std::vector formats; + + cl_context_properties cps[3] = { CL_CONTEXT_PLATFORM, (cl_context_properties)(*p)(), 0 }; + std::vector device; + device.push_back(*i); + cl::Context context(device, cps, NULL, NULL, &err); + + std::map channelOrder; + channelOrder[CL_R] = "CL_R"; + channelOrder[CL_A] = "CL_A"; + channelOrder[CL_RG] = "CL_RG"; + channelOrder[CL_RA] = "CL_RA"; + channelOrder[CL_RGB] = "CL_RGB"; + channelOrder[CL_RGBA] = "CL_RGBA"; + channelOrder[CL_BGRA] = "CL_BGRA"; + channelOrder[CL_ARGB] = "CL_ARGB"; + channelOrder[CL_INTENSITY] = "CL_INTENSITY"; + channelOrder[CL_LUMINANCE] = "CL_LUMINANCE"; + channelOrder[CL_Rx] = "CL_Rx"; + channelOrder[CL_RGx] = "CL_RGx"; + channelOrder[CL_RGBx] = "CL_RGBx"; + + std::map > channelType; + channelType[CL_SNORM_INT8] = std::make_pair("snorm", "int8"); + channelType[CL_SNORM_INT16] = std::make_pair("snorm", "int16"); + channelType[CL_UNORM_INT8] = std::make_pair("unorm", "int8"); + channelType[CL_UNORM_INT16] = std::make_pair("unorm", "int16"); + channelType[CL_UNORM_SHORT_565] = std::make_pair("unorm", "short_565"); + channelType[CL_UNORM_SHORT_555] = std::make_pair("unorm", "short_555"); + channelType[CL_UNORM_INT_101010] = std::make_pair("unorm", "int_101010"); + channelType[CL_SIGNED_INT8] = std::make_pair("signed", "int8"); + channelType[CL_SIGNED_INT16] = std::make_pair("signed", "int16"); + channelType[CL_SIGNED_INT32] = std::make_pair("signed", "int32"); + channelType[CL_UNSIGNED_INT8] = std::make_pair("unsigned", "int8"); + channelType[CL_UNSIGNED_INT16] = std::make_pair("unsigned", "int16"); + channelType[CL_UNSIGNED_INT32] = std::make_pair("unsigned", "int32"); + channelType[CL_HALF_FLOAT] = std::make_pair("half_float", ""); + channelType[CL_FLOAT] = std::make_pair("float", ""); + + std::vector > imageDimensions; + imageDimensions.push_back(std::make_pair(CL_MEM_OBJECT_IMAGE2D, std::string("2D "))); + imageDimensions.push_back(std::make_pair(CL_MEM_OBJECT_IMAGE3D, std::string("3D "))); + for(std::vector >::iterator id = imageDimensions.begin(); + id != imageDimensions.end(); + id++){ + + struct imageAccessStruct { + std::string name; + int access; + std::vector formats; + } imageAccess[] = {{std::string("Read-Write/Read-Only/Write-Only"), CL_MEM_READ_WRITE, std::vector()}, + {std::string("Read-Only"), CL_MEM_READ_ONLY, std::vector()}, + {std::string("Write-Only"), CL_MEM_WRITE_ONLY, std::vector()}}; + + for(size_t ia=0; ia < sizeof(imageAccess)/sizeof(imageAccessStruct); ia++){ + context.getSupportedImageFormats(imageAccess[ia].access, (*id).first, &(imageAccess[ia].formats)); + bool printTopHeader = true; + for (std::map::iterator o = channelOrder.begin(); + o != channelOrder.end(); + o++) + { + bool printHeader = true; + + for (std::vector::iterator it = imageAccess[ia].formats.begin(); + it != imageAccess[ia].formats.end(); + ++it) + { + if ( (*o).first == (int)(*it).image_channel_order) + { + bool printedAlready = false; + //see if this was already print in RW/RO/WO + if (ia !=0) + { + for (std::vector::iterator searchIt = imageAccess[0].formats.begin(); + searchIt != imageAccess[0].formats.end(); + searchIt++) + { + if ( ((*searchIt).image_channel_data_type == (*it).image_channel_data_type) && + ((*searchIt).image_channel_order == (*it).image_channel_order)) + { + printedAlready = true; + break; + } + } + } + if (printedAlready) + { + continue; + } + if (printTopHeader) + { + std::cout << " " << (*id).second << imageAccess[ia].name << std::endl; + printTopHeader = false; + } + if (printHeader) + { + std::cout << " " << (*o).second << ": "; + printHeader = false; + } + std::cout << channelType[(*it).image_channel_data_type].first; + if (channelType[(*it).image_channel_data_type].second != "") + { + std::cout << "-" + << channelType[(*it).image_channel_data_type].second; + } + if (it != (imageAccess[ia].formats.end() - 1)) + { + std::cout << " "; + } + } + } + if (printHeader == false) + { + std::cout << std::endl; + } + } + } + } + } + } + + std::cout << " Max size of kernel argument:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Alignment (bits) of base address:\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Minimum alignment (bytes) for any datatype:\t " + << device.getInfo() + << std::endl; + + std::cout << " Single precision floating point capability" << std::endl; + std::cout << " Denorms:\t\t\t\t\t " + << (device.getInfo() & + CL_FP_DENORM ? "Yes" : "No") + << std::endl; + std::cout << " Quiet NaNs:\t\t\t\t\t " + << (device.getInfo() & + CL_FP_INF_NAN ? "Yes" : "No") + << std::endl; + std::cout << " Round to nearest even:\t\t\t " + << (device.getInfo() & + CL_FP_ROUND_TO_NEAREST ? "Yes" : "No") + << std::endl; + std::cout << " Round to zero:\t\t\t\t " + << (device.getInfo() & + CL_FP_ROUND_TO_ZERO ? "Yes" : "No") + << std::endl; + std::cout << " Round to +ve and infinity:\t\t\t " + << (device.getInfo() & + CL_FP_ROUND_TO_INF ? "Yes" : "No") + << std::endl; + std::cout << " IEEE754-2008 fused multiply-add:\t\t " + << (device.getInfo() & + CL_FP_FMA ? "Yes" : "No") + << std::endl; + + std::cout << " Cache type:\t\t\t\t\t " ; + switch (device.getInfo()) { + case CL_NONE: + std::cout << "None" << std::endl; + break; + case CL_READ_ONLY_CACHE: + std::cout << "Read only" << std::endl; + break; + case CL_READ_WRITE_CACHE: + std::cout << "Read/Write" << std::endl; + break; + } + + std::cout << " Cache line size:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Cache size:\t\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Global memory size:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Constant buffer size:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max number of constant args:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Local memory type:\t\t\t\t " ; + switch (device.getInfo()) { + case CL_LOCAL: + std::cout << "Scratchpad" << std::endl; + break; + case CL_GLOBAL: + std::cout << "Global" << std::endl; + break; + } + + + std::cout << " Local memory size:\t\t\t\t " + << device.getInfo() + << std::endl; + +#if defined(CL_VERSION_2_0) + if(vStrVal.compare("2") > 0) + { + std::cout << " Max pipe arguments:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max pipe active reservations:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max pipe packet size:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max global variable size:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max global variable preferred total size:\t " + << device.getInfo() + << std::endl; + + std::cout << " Max read/write image args:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max on device events:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Queue on device max size:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Max on device queues:\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Queue on device preferred size:\t\t " + << device.getInfo() + << std::endl; + + std::cout << " SVM capabilities:\t\t\t\t " << std::endl; + std::cout << " Coarse grain buffer:\t\t\t " + << (device.getInfo() & + CL_DEVICE_SVM_COARSE_GRAIN_BUFFER ? "Yes" : "No") + << std::endl; + std::cout << " Fine grain buffer:\t\t\t\t " + << (device.getInfo() & + CL_DEVICE_SVM_FINE_GRAIN_BUFFER ? "Yes" : "No") + << std::endl; + std::cout << " Fine grain system:\t\t\t\t " + << (device.getInfo() & + CL_DEVICE_SVM_FINE_GRAIN_SYSTEM ? "Yes" : "No") + << std::endl; + std::cout << " Atomics:\t\t\t\t\t " + << (device.getInfo() & + CL_DEVICE_SVM_ATOMICS ? "Yes" : "No") + << std::endl; + + std::cout << " Preferred platform atomic alignment:\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred global atomic alignment:\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Preferred local atomic alignment:\t\t " + << device.getInfo() + << std::endl; + } +#endif // CL_VERSION_2_0 + +#if defined(CL_VERSION_1_1) && !defined(ATI_ARCH_ARM) + if(vStrVal.compare("1.0") > 0) + { + cl_context_properties cps[3] = { CL_CONTEXT_PLATFORM, (cl_context_properties)(*p)(), 0 }; + + std::vector device; + device.push_back(*i); + + cl::Context context(device, cps, NULL, NULL, &err); + if (err != CL_SUCCESS) { + std::cerr << "Context::Context() failed (" << err << ")\n"; + return 1; + } + std::string kernelStr("__kernel void hello(){ size_t i = get_global_id(0); size_t j = get_global_id(1);}"); + cl::Program::Sources sources(1, std::make_pair(kernelStr.data(), kernelStr.size())); + + cl::Program program = cl::Program(context, sources, &err); + if (err != CL_SUCCESS) { + std::cerr << "Program::Program() failed (" << err << ")\n"; + return 1; + } + + err = program.build(device); + if (err != CL_SUCCESS) { + + if(err == CL_BUILD_PROGRAM_FAILURE) + { + std::string str = program.getBuildInfo((*i)); + + std::cout << " \n\t\t\tBUILD LOG\n"; + std::cout << " ************************************************\n"; + std::cout << str.c_str() << std::endl; + std::cout << " ************************************************\n"; + } + + std::cerr << "Program::build() failed (" << err << ")\n"; + return 1; + } + + cl::Kernel kernel(program, "hello", &err); + if (err != CL_SUCCESS) { + std::cerr << "Kernel::Kernel() failed (" << err << ")\n"; + return 1; + } + + std::cout << " Kernel Preferred work group size multiple:\t " + << kernel.getWorkGroupInfo((*i), &err) + << std::endl; + } + +#endif // CL_VERSION_1_1 + + std::cout << " Error correction support:\t\t\t " + << device.getInfo() + << std::endl; +#ifdef CL_VERSION_1_1 + if(vStrVal.compare("1.0") > 0) + { + std::cout << " Unified memory for Host and Device:\t\t " + << device.getInfo() + << std::endl; + } +#endif // CL_VERSION_1_1 + std::cout << " Profiling timer resolution:\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Device endianess:\t\t\t\t " + << (device.getInfo() ? "Little" : "Big") + << std::endl; + + std::cout << " Available:\t\t\t\t\t " + << (device.getInfo() ? "Yes" : "No") + << std::endl; + + std::cout << " Compiler available:\t\t\t\t " + << (device.getInfo() ? "Yes" : "No") + << std::endl; + + std::cout << " Execution capabilities:\t\t\t\t " << std::endl; + std::cout << " Execute OpenCL kernels:\t\t\t " + << (device.getInfo() & + CL_EXEC_KERNEL ? "Yes" : "No") + << std::endl; + std::cout << " Execute native function:\t\t\t " + << (device.getInfo() & + CL_EXEC_NATIVE_KERNEL ? "Yes" : "No") + << std::endl; + + std::cout << " Queue on Host properties:\t\t\t\t " << std::endl; + std::cout << " Out-of-Order:\t\t\t\t " + << (device.getInfo() & + CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE ? "Yes" : "No") + << std::endl; + std::cout << " Profiling :\t\t\t\t\t " + << (device.getInfo() & + CL_QUEUE_PROFILING_ENABLE ? "Yes" : "No") + << std::endl; + +#ifdef CL_VERSION_2_0 + if(vStrVal.compare("2") > 0) + { + std::cout << " Queue on Device properties:\t\t\t\t " << std::endl; + std::cout << " Out-of-Order:\t\t\t\t " + << (device.getInfo() & + CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE ? "Yes" : "No") + << std::endl; + std::cout << " Profiling :\t\t\t\t\t " + << (device.getInfo() & + CL_QUEUE_PROFILING_ENABLE ? "Yes" : "No") + << std::endl; + } +#endif + + std::cout << " Platform ID:\t\t\t\t\t " + << device.getInfo() + << std::endl; + + std::cout << " Name:\t\t\t\t\t\t " + << device.getInfo().c_str() + << std::endl; + + std::cout << " Vendor:\t\t\t\t\t " + << device.getInfo().c_str() + << std::endl; +#ifdef CL_VERSION_1_1 + if(vStrVal.compare("1.0") > 0) + { + std::cout << " Device OpenCL C version:\t\t\t " + << device.getInfo().c_str() + << std::endl; + } +#endif // CL_VERSION_1_1 + std::cout << " Driver version:\t\t\t\t " + << device.getInfo().c_str() + << std::endl; + + std::cout << " Profile:\t\t\t\t\t " + << device.getInfo().c_str() + << std::endl; + + std::cout << " Version:\t\t\t\t\t " + << device.getInfo().c_str() + << std::endl; + + + std::cout << " Extensions:\t\t\t\t\t " + << device.getInfo().c_str() + << std::endl; + + std::cout << std::endl << std::endl; + } + } + } + catch (cl::Error err) + { + std::cerr + << "ERROR: " + << err.what() + << "(" + << err.err() + << ")" + << std::endl; + } + + return status; +}