diff --git a/projects/clr/hipamd/include/hip/hcc_detail/hip_hcc.h b/projects/clr/hipamd/include/hip/hcc_detail/hip_hcc.h index 645e980376..889e04eb9f 100644 --- a/projects/clr/hipamd/include/hip/hcc_detail/hip_hcc.h +++ b/projects/clr/hipamd/include/hip/hcc_detail/hip_hcc.h @@ -28,6 +28,17 @@ THE SOFTWARE. #if __cplusplus #ifdef __HCC__ #include + + +/** + *------------------------------------------------------------------------------------------------- + *------------------------------------------------------------------------------------------------- + * @defgroup HCC-specific features + * @warning These APIs provide access to special features of HCC compiler and are not available through the CUDA path. + * @{ + */ + + /** * @brief Return hc::accelerator associated with the specified deviceId * @return #hipSuccess, #hipErrorInvalidDevice @@ -45,6 +56,30 @@ hipError_t hipHccGetAcceleratorView(hipStream_t stream, hc::accelerator_view **a #endif // #ifdef __HCC__ +/** + * @brief launches kernel f with launch parameters and shared memory on stream with arguments passed to kernelparams or extra + * + * @param [in[ f Kernel to launch. + * @param [in] gridDimX X grid dimension specified in work-items + * @param [in] gridDimY Y grid dimension specified in work-items + * @param [in] gridDimZ Z grid dimension specified in work-items + * @param [in] blockDimX X block dimensions specified in work-items + * @param [in] blockDimY Y grid dimension specified in work-items + * @param [in] blockDimZ Z grid dimension specified in work-items + * @param [in] sharedMemBytes Amount of dynamic shared memory to allocate for this kernel. The kernel can access this with HIP_DYNAMIC_SHARED. + * @param [in] stream Stream where the kernel should be dispatched. May be 0, in which case th default stream is used with associated synchronization rules. + * @param [in] kernelParams + * @param [in] extra Pointer to kernel arguments. These are passed directly to the kernel and must be in the memory layout and alignment expected by the kernel. + * @param [in] startEvent If non-null, specified event will be updated to track the start time of the kernel launch. The event must be created before calling this API. + * @param [in] stopEvent If non-null, specified event will be updated to track the stop time of the kernel launch. The event must be created before calling this API. + * + * @returns hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue + + * If startNanos or stopNanos is specified, this API will record and return the start and stop timestamps for the command. The timestamps are collected on the GPU device + * and converted into ns resolution. Typically programs will specify both pointers. Collecting performance timestamps may have a small overhead (approx 1us). + * + * @warning kernellParams argument is not yet implemented in HIP. Please use extra instead. Please refer to hip_porting_driver_api.md for sample usage. + */ hipError_t hipHccModuleLaunchKernel(hipFunction_t f, uint32_t globalWorkSizeX, uint32_t globalWorkSizeY, @@ -55,8 +90,15 @@ hipError_t hipHccModuleLaunchKernel(hipFunction_t f, size_t sharedMemBytes, hipStream_t hStream, void **kernelParams, - void **extra); + void **extra, + uint64_t *startNanos=nullptr, + uint64_t *stopNanos=nullptr + ); +// doxygen end HCC-specific features +/** + * @} + */ #endif // #if __cplusplus #endif // diff --git a/projects/clr/hipamd/include/hip/hcc_detail/hip_runtime_api.h b/projects/clr/hipamd/include/hip/hcc_detail/hip_runtime_api.h index 0daca7a53b..f9bfb5a310 100644 --- a/projects/clr/hipamd/include/hip/hcc_detail/hip_runtime_api.h +++ b/projects/clr/hipamd/include/hip/hcc_detail/hip_runtime_api.h @@ -1913,19 +1913,18 @@ hipError_t hipModuleLoadData(hipModule_t *module, const void *image); /** * @brief launches kernel f with launch parameters and shared memory on stream with arguments passed to kernelparams or extra * - * @param [in[ f - * @param [in] gridDimX - * @param [in] gridDimY - * @param [in] gridDimZ - * @param [in] blockDimX - * @param [in] blockDimY - * @param [in] blockDimZ - * @param [in] sharedMemBytes - * @param [in] stream - * @param [in] kernelParams - * @param [in] extraa + * @param [in[ f Kernel to launch. + * @param [in] gridDimX X grid dimension specified as multiple of blockDimX. + * @param [in] gridDimY Y grid dimension specified as multiple of blockDimY. + * @param [in] gridDimZ Z grid dimension specified as multiple of blockDimZ. + * @param [in] blockDimX X block dimensions specified in work-items + * @param [in] blockDimY Y grid dimension specified in work-items + * @param [in] blockDimZ Z grid dimension specified in work-items + * @param [in] sharedMemBytes Amount of dynamic shared memory to allocate for this kernel. The kernel can access this with HIP_DYNAMIC_SHARED. + * @param [in] stream Stream where the kernel should be dispatched. May be 0, in which case th default stream is used with associated synchronization rules. + * @param [in] kernelParams + * @param [in] extra Pointer to kernel arguments. These are passed directly to the kernel and must be in the memory layout and alignment expected by the kernel. * - * The function takes the above arguments and run the kernel in hipFunction_t f. with launch parameters specified in gridDimX, gridDimY, gridDimZ, blockDimX, blockDimY and blockDimmZ. The amount of shared memory is specificed and can be used with HIP_DYNAMIC_SHARED. The arguemt extra is used to pass in the arguments for the kernel. * @returns hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue * * @warning kernellParams argument is not yet implemented in HIP. Please use extra instead. Please refer to hip_porting_driver_api.md for sample usage.