From 8917c073fa50cbf50f2a9aac6540a5fc0c68ee0b Mon Sep 17 00:00:00 2001 From: foreman Date: Tue, 28 Nov 2017 16:25:11 -0500 Subject: [PATCH] P4 to Git Change 1487815 by emankov@em-hsa-lightning on 2017/11/28 16:19:33 SWDEV-134396 - cl_amd_assembly_program extension support. Part 2 of 4. Rocm Runtime: Compiler invocation. precheckin: http://ocltc.amd.com:8111/viewModification.html?modId=94552&personal=true&init=1&tab=vcsModificationBuilds reviewed by: German Andryeyev Affected files ... ... //depot/stg/opencl/drivers/opencl/compiler/lib/include/v0_8/aclEnums.h#27 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/roccompiler.cpp#40 edit ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.cpp#77 edit [ROCm/clr commit: a5884cdbf4a0b7b0c5883614fe3837323ae13bf2] --- .../compiler/lib/include/v0_8/aclEnums.h | 3 +- .../runtime/device/rocm/roccompiler.cpp | 23 +- .../rocclr/runtime/device/rocm/rocprogram.cpp | 228 ++++++++++-------- 3 files changed, 144 insertions(+), 110 deletions(-) diff --git a/projects/clr/rocclr/compiler/lib/include/v0_8/aclEnums.h b/projects/clr/rocclr/compiler/lib/include/v0_8/aclEnums.h index 928cd048fc..dd2774fe0c 100644 --- a/projects/clr/rocclr/compiler/lib/include/v0_8/aclEnums.h +++ b/projects/clr/rocclr/compiler/lib/include/v0_8/aclEnums.h @@ -114,7 +114,8 @@ typedef enum _acl_type_enum_0_8 { ACL_TYPE_HEADER = 15, ACL_TYPE_RSLLVMIR_BINARY = 16, ACL_TYPE_SPIRV_BINARY = 17, - ACL_TYPE_LAST = 18 + ACL_TYPE_ASM_TEXT = 18, + ACL_TYPE_LAST = 19 } aclType_0_8; //! Enum of the various loader types that are supported. diff --git a/projects/clr/rocclr/runtime/device/rocm/roccompiler.cpp b/projects/clr/rocclr/runtime/device/rocm/roccompiler.cpp index 6c10254e71..e0a28a1455 100644 --- a/projects/clr/rocclr/runtime/device/rocm/roccompiler.cpp +++ b/projects/clr/rocclr/runtime/device/rocm/roccompiler.cpp @@ -165,13 +165,25 @@ bool HSAILProgram::compileImpl(const std::string& sourceCode, #if defined(WITH_LIGHTNING_COMPILER) bool LightningProgram::compileImpl(const std::string& sourceCode, - const std::vector& headers, - const char** headerIncludeNames, amd::option::Options* options) { + const std::vector& headers, + const char** headerIncludeNames, amd::option::Options* options) { + const char* xLang = options->oVariables->XLang; + if (xLang != nullptr) { + if (strcmp(xLang,"asm") == 0) { + clBinary()->elfOut()->addSection(amd::OclElf::SOURCE, sourceCode.data(), sourceCode.size()); + return true; + } else if (!strcmp(xLang,"cl")) { + buildLog_ += "Unsupported language: \"" + std::string(xLang) + "\".\n"; + return false; + } + } + using namespace amd::opencl_driver; + amd::opencl_driver::DataType inputType(DT_CL); std::unique_ptr C(newCompilerInstance()); std::vector inputs; - Data* input = C->NewBufferReference(DT_CL, sourceCode.c_str(), sourceCode.length()); + Data* input = C->NewBufferReference(inputType, sourceCode.c_str(), sourceCode.length()); if (input == nullptr) { buildLog_ += "Error while creating data from source code"; return false; @@ -193,11 +205,6 @@ bool LightningProgram::compileImpl(const std::string& sourceCode, std::string driverOptions(ostrstr.str()); - const char* xLang = options->oVariables->XLang; - if (xLang != nullptr && strcmp(xLang, "cl")) { - buildLog_ += "Unsupported OpenCL language.\n"; - } - // FIXME_Nikolay: the program manager should be setting the language // driverOptions.append(" -x cl"); diff --git a/projects/clr/rocclr/runtime/device/rocm/rocprogram.cpp b/projects/clr/rocclr/runtime/device/rocm/rocprogram.cpp index b7b72e3bea..4b90aa457c 100644 --- a/projects/clr/rocclr/runtime/device/rocm/rocprogram.cpp +++ b/projects/clr/rocclr/runtime/device/rocm/rocprogram.cpp @@ -284,6 +284,14 @@ aclType Program::getNextCompilationStageFromBinary(amd::option::Options* options } } } +#if defined(WITH_LIGHTNING_COMPILER) + else { + const char* xLang = options->oVariables->XLang; + if (xLang != nullptr && strcmp(xLang, "asm") == 0) { + continueCompileFrom = ACL_TYPE_ASM_TEXT; + } + } +#endif return continueCompileFrom; } @@ -999,7 +1007,10 @@ bool LightningProgram::linkImpl(const std::vector& inputProgra bool LightningProgram::linkImpl(amd::option::Options* options) { acl_error errorCode; aclType continueCompileFrom = ACL_TYPE_LLVMIR_BINARY; - + using namespace amd::opencl_driver; + std::vector inputs; + std::unique_ptr C(newCompilerInstance()); + bool bLinkLLVMBitcode = true; if (llvmBinary_.empty()) { continueCompileFrom = getNextCompilationStageFromBinary(options); } @@ -1007,6 +1018,19 @@ bool LightningProgram::linkImpl(amd::option::Options* options) { case ACL_TYPE_CG: case ACL_TYPE_LLVMIR_BINARY: { break; + } + case ACL_TYPE_ASM_TEXT: { + char* section; + size_t sz; + clBinary()->elfOut()->getSection(amd::OclElf::SOURCE, §ion, &sz); + Data* input = C->NewBufferReference(DT_ASSEMBLY, section, sz); + if (!input) { + buildLog_ += "Error: Failed to open the assembler text.\n"; + return false; + } + inputs.push_back(input); + bLinkLLVMBitcode = false; + break; } break; case ACL_TYPE_ISA: { @@ -1020,119 +1044,117 @@ bool LightningProgram::linkImpl(amd::option::Options* options) { break; } default: - buildLog_ += "Error while BRIG Codegen phase: the binary is incomplete \n"; + buildLog_ += "Error while Codegen phase: the binary is incomplete \n"; return false; } - using namespace amd::opencl_driver; - std::unique_ptr C(newCompilerInstance()); - // call LinkLLVMBitcode - std::vector inputs; + if (bLinkLLVMBitcode) { + // open the input IR source + Data* input = C->NewBufferReference(DT_LLVM_BC, llvmBinary_.data(), llvmBinary_.size()); - // open the input IR source - Data* input = C->NewBufferReference(DT_LLVM_BC, llvmBinary_.data(), llvmBinary_.size()); + if (!input) { + buildLog_ += "Error: Failed to open the compiled program.\n"; + return false; + } - if (!input) { - buildLog_ += "Error: Failed to open the compiled program.\n"; - return false; - } + inputs.push_back(input); //< must be the first input - inputs.push_back(input); //< must be the first input - - // open the bitcode libraries - Data* opencl_bc = + // open the bitcode libraries + Data* opencl_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)opencl_amdgcn, opencl_amdgcn_size); - Data* ocml_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)ocml_amdgcn, ocml_amdgcn_size); - Data* ockl_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)ockl_amdgcn, ockl_amdgcn_size); - Data* irif_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)irif_amdgcn, irif_amdgcn_size); + Data* ocml_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)ocml_amdgcn, ocml_amdgcn_size); + Data* ockl_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)ockl_amdgcn, ockl_amdgcn_size); + Data* irif_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)irif_amdgcn, irif_amdgcn_size); - if (!opencl_bc || !ocml_bc || !ockl_bc || !irif_bc) { - buildLog_ += "Error: Failed to open the bitcode library.\n"; - return false; - } + if (!opencl_bc || !ocml_bc || !ockl_bc || !irif_bc) { + buildLog_ += "Error: Failed to open the bitcode library.\n"; + return false; + } - inputs.push_back(opencl_bc); // depends on oclm & ockl - inputs.push_back(ockl_bc); // depends on irif - inputs.push_back(ocml_bc); // depends on irif - inputs.push_back(irif_bc); + inputs.push_back(opencl_bc); // depends on oclm & ockl + inputs.push_back(ockl_bc); // depends on irif + inputs.push_back(ocml_bc); // depends on irif + inputs.push_back(irif_bc); - // open the control functions - auto isa_version = get_oclc_isa_version(dev().deviceInfo().gfxipVersion_); - if (!isa_version.first) { - buildLog_ += "Error: Linking for this device is not supported\n"; - return false; - } + // open the control functions + auto isa_version = get_oclc_isa_version(dev().deviceInfo().gfxipVersion_); + if (!isa_version.first) { + buildLog_ += "Error: Linking for this device is not supported\n"; + return false; + } - Data* isa_version_bc = + Data* isa_version_bc = C->NewBufferReference(DT_LLVM_BC, (const char*)isa_version.first, isa_version.second); - if (!isa_version_bc) { - buildLog_ += "Error: Failed to open the control functions.\n"; - return false; - } - - inputs.push_back(isa_version_bc); - - auto correctly_rounded_sqrt = - get_oclc_correctly_rounded_sqrt(options->oVariables->FP32RoundDivideSqrt); - Data* correctly_rounded_sqrt_bc = C->NewBufferReference(DT_LLVM_BC, correctly_rounded_sqrt.first, - correctly_rounded_sqrt.second); - - auto daz_opt = get_oclc_daz_opt(options->oVariables->DenormsAreZero || - AMD_GPU_FORCE_SINGLE_FP_DENORM == 0 || - (dev().deviceInfo().gfxipVersion_ < 900 && - AMD_GPU_FORCE_SINGLE_FP_DENORM < 0)); - Data* daz_opt_bc = C->NewBufferReference(DT_LLVM_BC, daz_opt.first, daz_opt.second); - - auto finite_only = get_oclc_finite_only(options->oVariables->FiniteMathOnly || - options->oVariables->FastRelaxedMath); - Data* finite_only_bc = C->NewBufferReference(DT_LLVM_BC, finite_only.first, finite_only.second); - - auto unsafe_math = get_oclc_unsafe_math(options->oVariables->UnsafeMathOpt || - options->oVariables->FastRelaxedMath); - Data* unsafe_math_bc = C->NewBufferReference(DT_LLVM_BC, unsafe_math.first, unsafe_math.second); - - if (!correctly_rounded_sqrt_bc || !daz_opt_bc || !finite_only_bc || !unsafe_math_bc) { - buildLog_ += "Error: Failed to open the control functions.\n"; - return false; - } - - inputs.push_back(correctly_rounded_sqrt_bc); - inputs.push_back(daz_opt_bc); - inputs.push_back(finite_only_bc); - inputs.push_back(unsafe_math_bc); - - // open the linked output - std::vector linkOptions; - Buffer* linked_bc = C->NewBuffer(DT_LLVM_BC); - - if (!linked_bc) { - buildLog_ += "Error: Failed to open the linked program.\n"; - return false; - } - - // NOTE: The linkOptions parameter is also used to identy cached code object. This parameter - // should not contain any dyanamically generated filename. - bool ret = - dev().cacheCompilation()->linkLLVMBitcode(C.get(), inputs, linked_bc, linkOptions, buildLog_); - buildLog_ += C->Output(); - if (!ret) { - buildLog_ += "Error: Linking bitcode failed: linking source & IR libraries.\n"; - return false; - } - - if (options->isDumpFlagSet(amd::option::DUMP_BC_LINKED)) { - std::ofstream f(options->getDumpFileName("_linked.bc").c_str(), std::ios::trunc); - if (f.is_open()) { - f.write(linked_bc->Buf().data(), linked_bc->Size()); - } else { - buildLog_ += "Warning: opening the file to dump the linked IR failed.\n"; + if (!isa_version_bc) { + buildLog_ += "Error: Failed to open the control functions.\n"; + return false; } - } - inputs.clear(); - inputs.push_back(linked_bc); + inputs.push_back(isa_version_bc); + + auto correctly_rounded_sqrt = + get_oclc_correctly_rounded_sqrt(options->oVariables->FP32RoundDivideSqrt); + Data* correctly_rounded_sqrt_bc = C->NewBufferReference(DT_LLVM_BC, correctly_rounded_sqrt.first, + correctly_rounded_sqrt.second); + + auto daz_opt = get_oclc_daz_opt(options->oVariables->DenormsAreZero || + AMD_GPU_FORCE_SINGLE_FP_DENORM == 0 || + (dev().deviceInfo().gfxipVersion_ < 900 && + AMD_GPU_FORCE_SINGLE_FP_DENORM < 0)); + Data* daz_opt_bc = C->NewBufferReference(DT_LLVM_BC, daz_opt.first, daz_opt.second); + + auto finite_only = get_oclc_finite_only(options->oVariables->FiniteMathOnly || + options->oVariables->FastRelaxedMath); + Data* finite_only_bc = C->NewBufferReference(DT_LLVM_BC, finite_only.first, finite_only.second); + + auto unsafe_math = get_oclc_unsafe_math(options->oVariables->UnsafeMathOpt || + options->oVariables->FastRelaxedMath); + Data* unsafe_math_bc = C->NewBufferReference(DT_LLVM_BC, unsafe_math.first, unsafe_math.second); + + if (!correctly_rounded_sqrt_bc || !daz_opt_bc || !finite_only_bc || !unsafe_math_bc) { + buildLog_ += "Error: Failed to open the control functions.\n"; + return false; + } + + inputs.push_back(correctly_rounded_sqrt_bc); + inputs.push_back(daz_opt_bc); + inputs.push_back(finite_only_bc); + inputs.push_back(unsafe_math_bc); + + // open the linked output + std::vector linkOptions; + Buffer* linked_bc = C->NewBuffer(DT_LLVM_BC); + + if (!linked_bc) { + buildLog_ += "Error: Failed to open the linked program.\n"; + return false; + } + + // NOTE: The linkOptions parameter is also used to identy cached code object. This parameter + // should not contain any dyanamically generated filename. + bool ret = + dev().cacheCompilation()->linkLLVMBitcode(C.get(), inputs, linked_bc, linkOptions, buildLog_); + buildLog_ += C->Output(); + if (!ret) { + buildLog_ += "Error: Linking bitcode failed: linking source & IR libraries.\n"; + return false; + } + + if (options->isDumpFlagSet(amd::option::DUMP_BC_LINKED)) { + std::ofstream f(options->getDumpFileName("_linked.bc").c_str(), std::ios::trunc); + if (f.is_open()) { + f.write(linked_bc->Buf().data(), linked_bc->Size()); + } + else { + buildLog_ += "Warning: opening the file to dump the linked IR failed.\n"; + } + } + + inputs.clear(); + inputs.push_back(linked_bc); + } Buffer* out_exec = C->NewBuffer(DT_EXECUTABLE); if (!out_exec) { @@ -1165,13 +1187,17 @@ bool LightningProgram::linkImpl(amd::option::Options* options) { std::istream_iterator sit(strstr), end; std::vector params(sit, end); - // NOTE: The params is also used to identy cached code object. This paramete + // NOTE: The params is also used to identy cached code object. This parameter // should not contain any dyanamically generated filename. - ret = dev().cacheCompilation()->compileAndLinkExecutable(C.get(), inputs, out_exec, params, + bool ret = dev().cacheCompilation()->compileAndLinkExecutable(C.get(), inputs, out_exec, params, buildLog_); buildLog_ += C->Output(); if (!ret) { - buildLog_ += "Error: Creating the executable failed: Compiling LLVM IRs to executable\n"; + if (continueCompileFrom == ACL_TYPE_ASM_TEXT) { + buildLog_ += "Error: Creating the executable from ISA assembly text failed.\n"; + } else { + buildLog_ += "Error: Creating the executable from LLVM IRs failed.\n"; + } return false; }