From c8b3f3d587b1464c8a776880c5f2cccf4f9c3447 Mon Sep 17 00:00:00 2001 From: apoorva Date: Thu, 19 Jun 2025 12:35:33 +0000 Subject: [PATCH] Restored the Cmake file that was reverted by mistake during rebase. --- profiler/src/CMakeLists.txt | 54 +++++++++++++++++-------------------- 1 file changed, 25 insertions(+), 29 deletions(-) diff --git a/profiler/src/CMakeLists.txt b/profiler/src/CMakeLists.txt index 7c381a4093..e75a10ad19 100644 --- a/profiler/src/CMakeLists.txt +++ b/profiler/src/CMakeLists.txt @@ -40,21 +40,19 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9") list(APPEND PROFILER_OPS profile_contraction_scale.cpp) endif() if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES) - list(APPEND PROFILER_SOURCES profile_gemm_reduce.cpp) - list(APPEND PROFILER_SOURCES profile_batched_gemm_gemm.cpp) - list(APPEND PROFILER_SOURCES profile_batched_gemm_add_relu_gemm_add.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_add_add_fastgelu.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_add_fastgelu.cpp) - list(APPEND PROFILER_SOURCES profile_grouped_gemm.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_streamk.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_fastgelu.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_add_relu.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_add_silu.cpp) - list(APPEND PROFILER_SOURCES profile_gemm_add_relu_add_layernorm.cpp) - list(APPEND PROFILER_SOURCES profile_grouped_gemm_fixed_nk.cpp) - list(APPEND PROFILER_SOURCES profile_grouped_gemm_fastgelu.cpp) - list(APPEND PROFILER_SOURCES profile_grouped_gemm_tile_loop.cpp) - list(APPEND PROFILER_SOURCES profile_grouped_gemm_multiply_tile_loop.cpp) + list(APPEND PROFILER_OPS profile_gemm_reduce.cpp) + list(APPEND PROFILER_OPS profile_batched_gemm_gemm.cpp) + list(APPEND PROFILER_OPS profile_batched_gemm_add_relu_gemm_add.cpp) + list(APPEND PROFILER_OPS profile_gemm_add.cpp) + list(APPEND PROFILER_OPS profile_grouped_gemm.cpp) + list(APPEND PROFILER_OPS profile_gemm_streamk.cpp) + list(APPEND PROFILER_OPS profile_gemm_add_relu.cpp) + list(APPEND PROFILER_OPS profile_gemm_add_silu.cpp) + list(APPEND PROFILER_OPS profile_gemm_add_relu_add_layernorm.cpp) + list(APPEND PROFILER_OPS profile_grouped_gemm_fixed_nk.cpp) + list(APPEND PROFILER_OPS profile_grouped_gemm_fastgelu.cpp) + list(APPEND PROFILER_OPS profile_grouped_gemm_tile_loop.cpp) + list(APPEND PROFILER_OPS profile_grouped_gemm_multiply_tile_loop.cpp) endif() list(APPEND PROFILER_OPS profile_gemm_multiply_add.cpp) if(SUPPORTED_GPU_TARGETS MATCHES "gfx9[45]") @@ -151,19 +149,17 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9") list(APPEND DEVICE_INSTANCES device_contraction_scale_instance) endif() if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_add_fastgelu_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_fastgelu_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_gemm_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_add_relu_gemm_add_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_streamk_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_fastgelu_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_silu_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_add_layernorm_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fixed_nk_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fastgelu_instance) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_tile_loop_instance) + list(APPEND DEVICE_INSTANCES device_gemm_add_instance) + list(APPEND DEVICE_INSTANCES device_batched_gemm_gemm_instance) + list(APPEND DEVICE_INSTANCES device_batched_gemm_add_relu_gemm_add_instance) + list(APPEND DEVICE_INSTANCES device_grouped_gemm_instance) + list(APPEND DEVICE_INSTANCES device_gemm_streamk_instance) + list(APPEND DEVICE_INSTANCES device_gemm_add_relu_instance) + list(APPEND DEVICE_INSTANCES device_gemm_add_silu_instance) + list(APPEND DEVICE_INSTANCES device_gemm_add_relu_add_layernorm_instance) + list(APPEND DEVICE_INSTANCES device_grouped_gemm_fixed_nk_instance) + list(APPEND DEVICE_INSTANCES device_grouped_gemm_fastgelu_instance) + list(APPEND DEVICE_INSTANCES device_grouped_gemm_tile_loop_instance) endif() list(APPEND DEVICE_INSTANCES device_batched_gemm_instance) list(APPEND DEVICE_INSTANCES device_batched_gemm_reduce_instance) @@ -210,7 +206,7 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx1[1 list(APPEND DEVICE_INSTANCES device_grouped_conv3d_bwd_weight_instance) list(APPEND DEVICE_INSTANCES device_gemm_add_fastgelu_instance) if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES) - target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_instance) + list(APPEND DEVICE_INSTANCES device_gemm_add_instance) list(APPEND DEVICE_INSTANCES device_gemm_fastgelu_instance) list(APPEND DEVICE_INSTANCES device_gemm_add_add_fastgelu_instance) endif()