ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-05 20:55:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fixed cmake errors related to gemm_bilinear. Previously, if the above flags are set, cmake build fails: GPU_TARGETS="gfx1100;gfx1201" -D DTYPES="fp16;bf16;fp8"

Browse Source
This commit is contained in:
Anca Hamuraru
2025-04-30 13:02:57 +00:00
parent edd92fc546
commit 5bdc993dbf
2 changed files with 29 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
library/src/tensor_operation_instance/gpu/CMakeLists.txt
View File

@@ -279,6 +279,15 @@ FOREACH(subdir_path ${dir_list})
message("Found gemm_multiply_multiply_f8 instances, but gfx94/gfx95 not on the target list. Skipping.")
set(add_inst 0)
endif()
if ("${cmake_instance}" MATCHES "gemm_bilinear")
set(add_inst 0)
if((SUPPORTED_GPU_TARGETS MATCHES "gfx9") AND (DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES))
set(add_inst 1)
endif()
if((SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12") AND (DTYPES MATCHES "i8" OR NOT DEFINED DTYPES))
set(add_inst 1)
endif()
endif()
if((add_inst EQUAL 1))
get_filename_component(target_dir ${subdir_path} NAME)
add_subdirectory(${target_dir})

22
profiler/src/CMakeLists.txt
View File

@@ -71,10 +71,19 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12" OR SUPPORTED_GPU_TARGETS MATCHES "gfx9")
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
list(APPEND PROFILER_SOURCES profile_gemm_bilinear.cpp)
endif()
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
if(DTYPES MATCHES "i8" OR NOT DEFINED DTYPES)
list(APPEND PROFILER_SOURCES profile_gemm_bilinear.cpp)
endif()
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12" OR SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_SOURCES profile_gemm_universal.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_conv_fwd.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_conv_bwd_data.cpp)
@@ -165,10 +174,19 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_convinvscale_instance)
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bilinear_instance)
endif()
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
if(DTYPES MATCHES "i8" OR NOT DEFINED DTYPES)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_bilinear_instance)
endif()
endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12")
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_universal_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv3d_fwd_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_conv2d_bwd_data_instance)