saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-05-22 22:01:08 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
onnxruntime/cmake/patches/composable_kernel/Fix_Clang_Build.patch

87 lines
3.2 KiB
Diff
Raw Normal View History

Add composable kernel GEMM baseline for kernel explorer (#12364) * Split GemmBase RocBlasGemm * Add composable kernel GEMM baseline * Make linter happy * Address review comment * Update bert cases with batchsize * Adjust includes to fix IWYU lint * Only builds and links used ck kernels to improve building time * Remove warmup run on SelectImpl * Add comment to utility function * Mute cpplint * Make RocBlasGemm<T>::SelectImpl semantically correct * Add reduced basic test cases for ck gemm * More robust gemm testing * Fix warnings * Fix grammar
2022-08-05 00:32:20 +00:00
diff --git a/CMakeLists.txt b/CMakeLists.txt
[ROCm] Add GemmFastGelu CK implementation (#13759) ### Description <!-- Describe your changes. --> Add GemmFastGelu CK implementation. TODO 1. The performance of CK GemmFastGelu in ORT is not good as using CK directly, still need to investigate the reason and improve the CK in ORT. `GemmFastGeluUnfused float16 NN m=49152 n=3072 k=768 2298.8064 us 100.89 tflops` `withbias DeviceGemmMultipleD_Xdl_CShuffle<256, 256, 128, 32, 8, 8, Default> LoopScheduler: Default, PipelineVersion: v1 float16 NN m=49152 n=3072 k=768 2401.9799 us 96.56 tflops` ### Motivation and Context <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. --> Co-authored-by: peixuanzuo <peixuanzuo@linmif39a000004.zvflicr54joexhdgnhvmxrxygg.phxx.internal.cloudapp.net>
2023-01-05 09:53:30 +00:00
index f861e302..f0b6bcea 100644
Add composable kernel GEMM baseline for kernel explorer (#12364) * Split GemmBase RocBlasGemm * Add composable kernel GEMM baseline * Make linter happy * Address review comment * Update bert cases with batchsize * Adjust includes to fix IWYU lint * Only builds and links used ck kernels to improve building time * Remove warmup run on SelectImpl * Add comment to utility function * Mute cpplint * Make RocBlasGemm<T>::SelectImpl semantically correct * Add reduced basic test cases for ck gemm * More robust gemm testing * Fix warnings * Fix grammar
2022-08-05 00:32:20 +00:00
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
Update ROCm CI to use HIP LANGUAGE (#13214) Update for ROCm CI before reland tunable GEMM #12853. This PR also update composable kernel to use CMakes's HIP language support so that we can mix C/C++ compiler with HIP compiler instead of locking to hip-clang
2022-10-05 08:15:16 +00:00
@@ -1,7 +1,7 @@
cmake_minimum_required(VERSION 3.14)
# Check support for CUDA/HIP in Cmake
-project(composable_kernel)
+project(composable_kernel LANGUAGES CXX HIP)
list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
Update CK and fix performance issue on dev machine (#13531) 1. Update CK to its latest develop branch 2. `-mllvm -amdgpu-early-inline-all=true` is critical to CK's performance, ensure it is properly configured. - The flags are propagated from target `hip-lang::device`'s `INTERFACE_COMPILE_OPTIONS`, we must not manually add the flags. - Instead, we must ensure this target is properly configured by checking _CMAKE_HIP_DEVICE_RUNTIME_TARGET is set. TL,DR `hip-lang::device` sometime will be not be properly configured if our `CMAKE_PREFIX_PATH` is not configured carefully. In the CI docker, the configuration is in good state, but on dev machine it is not, which then silently result poor performance for kernels. We fixed it in this PR and add a guard to avoid unsuccessful future editing and to prevent convoluted debugging process. `_CMAKE_HIP_DEVICE_RUNTIME_TARGET ` is shared in `/opt/rocm/lib/cmake/hip-lang/hip-lang-config.cmake` and it is internal to [CMake](https://gitlab.kitware.com/cmake/cmake/-/merge_requests/6121/diffs), the variable name will not be changed in the foreseeable future.
2022-11-03 11:32:30 +00:00
@@ -41,27 +41,6 @@ set(CMAKE_CXX_STANDARD_REQUIRED ON)
Add composable kernel GEMM baseline for kernel explorer (#12364) * Split GemmBase RocBlasGemm * Add composable kernel GEMM baseline * Make linter happy * Address review comment * Update bert cases with batchsize * Adjust includes to fix IWYU lint * Only builds and links used ck kernels to improve building time * Remove warmup run on SelectImpl * Add comment to utility function * Mute cpplint * Make RocBlasGemm<T>::SelectImpl semantically correct * Add reduced basic test cases for ck gemm * More robust gemm testing * Fix warnings * Fix grammar
2022-08-05 00:32:20 +00:00
set(CMAKE_CXX_EXTENSIONS OFF)
message("CMAKE_CXX_COMPILER_ID: ${CMAKE_CXX_COMPILER_ID}")
Update ROCm CI to use HIP LANGUAGE (#13214) Update for ROCm CI before reland tunable GEMM #12853. This PR also update composable kernel to use CMakes's HIP language support so that we can mix C/C++ compiler with HIP compiler instead of locking to hip-clang
2022-10-05 08:15:16 +00:00
-## OpenMP
-if(CMAKE_CXX_COMPILER_ID MATCHES "Clang")
- # workaround issue hipcc in rocm3.5 cannot find openmp
- set(OpenMP_CXX "${CMAKE_CXX_COMPILER}")
- set(OpenMP_CXX_FLAGS "-fopenmp=libomp -Wno-unused-command-line-argument")
- set(OpenMP_CXX_LIB_NAMES "libomp" "libgomp" "libiomp5")
- set(OpenMP_libomp_LIBRARY ${OpenMP_CXX_LIB_NAMES})
- set(OpenMP_libgomp_LIBRARY ${OpenMP_CXX_LIB_NAMES})
- set(OpenMP_libiomp5_LIBRARY ${OpenMP_CXX_LIB_NAMES})
-else()
- find_package(OpenMP REQUIRED)
-endif()
-
-message("OpenMP_CXX_LIB_NAMES: ${OpenMP_CXX_LIB_NAMES}")
-message("OpenMP_gomp_LIBRARY: ${OpenMP_gomp_LIBRARY}")
-message("OpenMP_pthread_LIBRARY: ${OpenMP_pthread_LIBRARY}")
-message("OpenMP_CXX_FLAGS: ${OpenMP_CXX_FLAGS}")
-
-link_libraries(${OpenMP_gomp_LIBRARY})
-link_libraries(${OpenMP_pthread_LIBRARY})
-
## HIP
find_package(HIP REQUIRED)
# Override HIP version in config.h, if necessary.
Update CK and fix performance issue on dev machine (#13531) 1. Update CK to its latest develop branch 2. `-mllvm -amdgpu-early-inline-all=true` is critical to CK's performance, ensure it is properly configured. - The flags are propagated from target `hip-lang::device`'s `INTERFACE_COMPILE_OPTIONS`, we must not manually add the flags. - Instead, we must ensure this target is properly configured by checking _CMAKE_HIP_DEVICE_RUNTIME_TARGET is set. TL,DR `hip-lang::device` sometime will be not be properly configured if our `CMAKE_PREFIX_PATH` is not configured carefully. In the CI docker, the configuration is in good state, but on dev machine it is not, which then silently result poor performance for kernels. We fixed it in this PR and add a guard to avoid unsuccessful future editing and to prevent convoluted debugging process. `_CMAKE_HIP_DEVICE_RUNTIME_TARGET ` is shared in `/opt/rocm/lib/cmake/hip-lang/hip-lang-config.cmake` and it is internal to [CMake](https://gitlab.kitware.com/cmake/cmake/-/merge_requests/6121/diffs), the variable name will not be changed in the foreseeable future.
2022-11-03 11:32:30 +00:00
@@ -83,8 +62,6 @@ if( DEFINED CK_OVERRIDE_HIP_VERSION_PATCH )
message(STATUS "CK_HIP_VERSION_PATCH overriden with ${CK_OVERRIDE_HIP_VERSION_PATCH}")
endif()
message(STATUS "Build with HIP ${HIP_VERSION}")
-link_libraries(hip::device)
-add_compile_definitions(__HIP_PLATFORM_HCC__=1)
## tidy
include(EnableCompilerWarnings)
[ROCm] Add GemmFastGelu CK implementation (#13759) ### Description <!-- Describe your changes. --> Add GemmFastGelu CK implementation. TODO 1. The performance of CK GemmFastGelu in ORT is not good as using CK directly, still need to investigate the reason and improve the CK in ORT. `GemmFastGeluUnfused float16 NN m=49152 n=3072 k=768 2298.8064 us 100.89 tflops` `withbias DeviceGemmMultipleD_Xdl_CShuffle<256, 256, 128, 32, 8, 8, Default> LoopScheduler: Default, PipelineVersion: v1 float16 NN m=49152 n=3072 k=768 2401.9799 us 96.56 tflops` ### Motivation and Context <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. --> Co-authored-by: peixuanzuo <peixuanzuo@linmif39a000004.zvflicr54joexhdgnhvmxrxygg.phxx.internal.cloudapp.net>
2023-01-05 09:53:30 +00:00
@@ -273,9 +250,6 @@ rocm_package_setup_component(profiler
Add composable kernel GEMM baseline for kernel explorer (#12364) * Split GemmBase RocBlasGemm * Add composable kernel GEMM baseline * Make linter happy * Address review comment * Update bert cases with batchsize * Adjust includes to fix IWYU lint * Only builds and links used ck kernels to improve building time * Remove warmup run on SelectImpl * Add comment to utility function * Mute cpplint * Make RocBlasGemm<T>::SelectImpl semantically correct * Add reduced basic test cases for ck gemm * More robust gemm testing * Fix warnings * Fix grammar
2022-08-05 00:32:20 +00:00
)
add_subdirectory(library)
-add_subdirectory(example)
-add_subdirectory(test)
-add_subdirectory(profiler)
#Create an interface target for the include only files and call it "composablekernels"
include(CMakePackageConfigHelpers)
[ROCm] Add GemmFastGelu CK implementation (#13759) ### Description <!-- Describe your changes. --> Add GemmFastGelu CK implementation. TODO 1. The performance of CK GemmFastGelu in ORT is not good as using CK directly, still need to investigate the reason and improve the CK in ORT. `GemmFastGeluUnfused float16 NN m=49152 n=3072 k=768 2298.8064 us 100.89 tflops` `withbias DeviceGemmMultipleD_Xdl_CShuffle<256, 256, 128, 32, 8, 8, Default> LoopScheduler: Default, PipelineVersion: v1 float16 NN m=49152 n=3072 k=768 2401.9799 us 96.56 tflops` ### Motivation and Context <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. --> Co-authored-by: peixuanzuo <peixuanzuo@linmif39a000004.zvflicr54joexhdgnhvmxrxygg.phxx.internal.cloudapp.net>
2023-01-05 09:53:30 +00:00
@@ -301,11 +275,3 @@ rocm_install(FILES
Add composable kernel GEMM baseline for kernel explorer (#12364) * Split GemmBase RocBlasGemm * Add composable kernel GEMM baseline * Make linter happy * Address review comment * Update bert cases with batchsize * Adjust includes to fix IWYU lint * Only builds and links used ck kernels to improve building time * Remove warmup run on SelectImpl * Add comment to utility function * Mute cpplint * Make RocBlasGemm<T>::SelectImpl semantically correct * Add reduced basic test cases for ck gemm * More robust gemm testing * Fix warnings * Fix grammar
2022-08-05 00:32:20 +00:00
set(CPACK_RESOURCE_FILE_LICENSE "${CMAKE_CURRENT_SOURCE_DIR}/LICENSE")
set(CPACK_RPM_PACKAGE_LICENSE "MIT")
-
-rocm_create_package(
- NAME composablekernel
- DESCRIPTION "High Performance Composable Kernel for AMD GPUs"
- MAINTAINER "MIOpen Kernels Dev Team <dl.MIOpen@amd.com>"
- LDCONFIG
- HEADER_ONLY
-)
Update ROCm CI to use HIP LANGUAGE (#13214) Update for ROCm CI before reland tunable GEMM #12853. This PR also update composable kernel to use CMakes's HIP language support so that we can mix C/C++ compiler with HIP compiler instead of locking to hip-clang
2022-10-05 08:15:16 +00:00
diff --git a/library/src/tensor_operation_instance/gpu/CMakeLists.txt b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
Update CK and fix performance issue on dev machine (#13531) 1. Update CK to its latest develop branch 2. `-mllvm -amdgpu-early-inline-all=true` is critical to CK's performance, ensure it is properly configured. - The flags are propagated from target `hip-lang::device`'s `INTERFACE_COMPILE_OPTIONS`, we must not manually add the flags. - Instead, we must ensure this target is properly configured by checking _CMAKE_HIP_DEVICE_RUNTIME_TARGET is set. TL,DR `hip-lang::device` sometime will be not be properly configured if our `CMAKE_PREFIX_PATH` is not configured carefully. In the CI docker, the configuration is in good state, but on dev machine it is not, which then silently result poor performance for kernels. We fixed it in this PR and add a guard to avoid unsuccessful future editing and to prevent convoluted debugging process. `_CMAKE_HIP_DEVICE_RUNTIME_TARGET ` is shared in `/opt/rocm/lib/cmake/hip-lang/hip-lang-config.cmake` and it is internal to [CMake](https://gitlab.kitware.com/cmake/cmake/-/merge_requests/6121/diffs), the variable name will not be changed in the foreseeable future.
2022-11-03 11:32:30 +00:00
index c206c4dc..e45fac9d 100644
Update ROCm CI to use HIP LANGUAGE (#13214) Update for ROCm CI before reland tunable GEMM #12853. This PR also update composable kernel to use CMakes's HIP language support so that we can mix C/C++ compiler with HIP compiler instead of locking to hip-clang
2022-10-05 08:15:16 +00:00
--- a/library/src/tensor_operation_instance/gpu/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
Update CK and fix performance issue on dev machine (#13531) 1. Update CK to its latest develop branch 2. `-mllvm -amdgpu-early-inline-all=true` is critical to CK's performance, ensure it is properly configured. - The flags are propagated from target `hip-lang::device`'s `INTERFACE_COMPILE_OPTIONS`, we must not manually add the flags. - Instead, we must ensure this target is properly configured by checking _CMAKE_HIP_DEVICE_RUNTIME_TARGET is set. TL,DR `hip-lang::device` sometime will be not be properly configured if our `CMAKE_PREFIX_PATH` is not configured carefully. In the CI docker, the configuration is in good state, but on dev machine it is not, which then silently result poor performance for kernels. We fixed it in this PR and add a guard to avoid unsuccessful future editing and to prevent convoluted debugging process. `_CMAKE_HIP_DEVICE_RUNTIME_TARGET ` is shared in `/opt/rocm/lib/cmake/hip-lang/hip-lang-config.cmake` and it is internal to [CMake](https://gitlab.kitware.com/cmake/cmake/-/merge_requests/6121/diffs), the variable name will not be changed in the foreseeable future.
2022-11-03 11:32:30 +00:00
@@ -1,7 +1,9 @@
Update ROCm CI to use HIP LANGUAGE (#13214) Update for ROCm CI before reland tunable GEMM #12853. This PR also update composable kernel to use CMakes's HIP language support so that we can mix C/C++ compiler with HIP compiler instead of locking to hip-clang
2022-10-05 08:15:16 +00:00
function(add_instance_library INSTANCE_NAME)
message("adding instance ${INSTANCE_NAME}")
+ set_source_files_properties(${ARGN} PROPERTIES LANGUAGE HIP)
add_library(${INSTANCE_NAME} OBJECT ${ARGN})
target_compile_features(${INSTANCE_NAME} PUBLIC)
Update CK and fix performance issue on dev machine (#13531) 1. Update CK to its latest develop branch 2. `-mllvm -amdgpu-early-inline-all=true` is critical to CK's performance, ensure it is properly configured. - The flags are propagated from target `hip-lang::device`'s `INTERFACE_COMPILE_OPTIONS`, we must not manually add the flags. - Instead, we must ensure this target is properly configured by checking _CMAKE_HIP_DEVICE_RUNTIME_TARGET is set. TL,DR `hip-lang::device` sometime will be not be properly configured if our `CMAKE_PREFIX_PATH` is not configured carefully. In the CI docker, the configuration is in good state, but on dev machine it is not, which then silently result poor performance for kernels. We fixed it in this PR and add a guard to avoid unsuccessful future editing and to prevent convoluted debugging process. `_CMAKE_HIP_DEVICE_RUNTIME_TARGET ` is shared in `/opt/rocm/lib/cmake/hip-lang/hip-lang-config.cmake` and it is internal to [CMake](https://gitlab.kitware.com/cmake/cmake/-/merge_requests/6121/diffs), the variable name will not be changed in the foreseeable future.
2022-11-03 11:32:30 +00:00
+ target_compile_definitions(${INSTANCE_NAME} PRIVATE "__HIP_PLATFORM_AMD__=1" "__HIP_PLATFORM_HCC__=1")
Update ROCm CI to use HIP LANGUAGE (#13214) Update for ROCm CI before reland tunable GEMM #12853. This PR also update composable kernel to use CMakes's HIP language support so that we can mix C/C++ compiler with HIP compiler instead of locking to hip-clang
2022-10-05 08:15:16 +00:00
set_target_properties(${INSTANCE_NAME} PROPERTIES POSITION_INDEPENDENT_CODE ON)
Update CK and fix performance issue on dev machine (#13531) 1. Update CK to its latest develop branch 2. `-mllvm -amdgpu-early-inline-all=true` is critical to CK's performance, ensure it is properly configured. - The flags are propagated from target `hip-lang::device`'s `INTERFACE_COMPILE_OPTIONS`, we must not manually add the flags. - Instead, we must ensure this target is properly configured by checking _CMAKE_HIP_DEVICE_RUNTIME_TARGET is set. TL,DR `hip-lang::device` sometime will be not be properly configured if our `CMAKE_PREFIX_PATH` is not configured carefully. In the CI docker, the configuration is in good state, but on dev machine it is not, which then silently result poor performance for kernels. We fixed it in this PR and add a guard to avoid unsuccessful future editing and to prevent convoluted debugging process. `_CMAKE_HIP_DEVICE_RUNTIME_TARGET ` is shared in `/opt/rocm/lib/cmake/hip-lang/hip-lang-config.cmake` and it is internal to [CMake](https://gitlab.kitware.com/cmake/cmake/-/merge_requests/6121/diffs), the variable name will not be changed in the foreseeable future.
2022-11-03 11:32:30 +00:00
clang_tidy_check(${INSTANCE_NAME})
endfunction(add_instance_library INSTANCE_NAME)
Reference in a new issue Copy permalink