mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-04-20 06:49:15 +00:00
4d041837ad
* Adding RapidJson Library * Adding Json Dumps in all CK_Tile Examples Not verified yet * Adding json to cktile Batched Transpose * adding json dumps to layernorm2d_fwd * Adding json dump to flatmm_basic * Adding RapidJson Library * Adding Json Dumps in all CK_Tile Examples Not verified yet * Adding json to cktile Batched Transpose * adding json dumps to layernorm2d_fwd * Adding json dump to flatmm_basic * Adding json in 03_gemm * Add json dump to 16_batched_gemm * Add json dump to gemm_multi_d_fp16 * Add json dump to grouped_gemm * fix fmha_bwd/fwd * Fix clang-format errors exclude include/rapidjson in jenkins as its a third-party library * Saparating function and defination. * Update Documentation of 03_gemm * Refactoring as per code review * Disable fp8 instances on unsupported targets (#2592) * Restrict building of gemm_universal_preshuffle_f8 instances to specific targets in CMakeLists.txt * Add condition to skip gemm_xdl_universal_preshuffle_f8 instances for unsupported targets in CMakeLists.txt * Add conditions to skip unsupported targets for gemm_universal_preshuffle_f8 and gemm_xdl_universal_preshuffle_f8 instances in CMakeLists.txt * Refine conditions to exclude gemm_universal_preshuffle_f8 instances for unsupported targets in CMakeLists.txt --------- Co-authored-by: AviralGoelAMD <aviralgoel@amd.com> * fix clang format * remove duplicate lines of code from library/src/tensor_operation_instance/gpu/CMakeLists.txt * Fixing Readme and unifying jsondumps * adding moe_smoothquant * adding fused_moe * Fixing Readme for batched_gemm * Fixing Readme for grouped_gemm * adding flatmm * adding gemm_multi_d_fp16 * adding elementwise * adding File name when json is dumped * Fixing Reduce after merge * adding batched_transpose * Adding Warptile in Gemm * Fixing Clang Format --------- Co-authored-by: Aviral Goel <aviral.goel@amd.com> Co-authored-by: AviralGoelAMD <aviralgoel@amd.com> Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
241 lines
12 KiB
CMake
241 lines
12 KiB
CMake
include_directories(BEFORE
|
|
${PROJECT_SOURCE_DIR}/include
|
|
${PROJECT_SOURCE_DIR}/library/include
|
|
${PROJECT_SOURCE_DIR}/example/include
|
|
)
|
|
|
|
add_custom_target(examples)
|
|
|
|
|
|
# list of examples that are labelled as REGRESSION_EXAMPLE for make regression (runtime more than 30 seconds)
|
|
# all other tests are labelled as SMOKE_EXAMPLE
|
|
set(REGRESSION_EXAMPLES
|
|
example_sparse_embedding3_forward_layernorm
|
|
)
|
|
|
|
|
|
function(add_example_dependencies EXAMPLE_NAME FILE_NAME)
|
|
if(FILE_NAME)
|
|
add_dependencies(EXAMPLE_NAME FILE_NAME)
|
|
endif()
|
|
endfunction(add_example_dependencies EXAMPLE_NAME)
|
|
|
|
function(add_example_executable EXAMPLE_NAME FILE_NAME)
|
|
message(DEBUG "adding example ${EXAMPLE_NAME}")
|
|
set(result 1)
|
|
if(DEFINED DTYPES)
|
|
foreach(source IN LISTS FILE_NAME)
|
|
get_filename_component(source_name ${source} NAME)
|
|
set(test 0)
|
|
if((source_name MATCHES "_fp16" OR source_name MATCHES "_f16") AND NOT "fp16" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_fp32" OR source_name MATCHES "_f32") AND NOT "fp32" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_fp64" OR source_name MATCHES "_f64") AND NOT "fp64" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_fp8" OR source_name MATCHES "_f8") AND NOT "fp8" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_bf8" OR source_name MATCHES "_bf8") AND NOT "bf8" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_bf16" OR source_name MATCHES "_b16") AND NOT "bf16" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_int8" OR source_name MATCHES "_i8") AND NOT "int8" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if(test EQUAL 1)
|
|
message(DEBUG "removing example source file ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
endforeach()
|
|
endif()
|
|
|
|
set(EX_TARGETS ${SUPPORTED_GPU_TARGETS})
|
|
|
|
foreach(source IN LISTS FILE_NAME)
|
|
get_filename_component(source_name ${source} NAME)
|
|
#Do not build any DL examples if DL_KERNELS not set
|
|
if(NOT DEFINED DL_KERNELS AND source_name MATCHES "_dl")
|
|
message(DEBUG "removing dl example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any DPP examples if DPP_KERNELS not set
|
|
if(NOT DEFINED DPP_KERNELS AND source_name MATCHES "_dpp")
|
|
message(DEBUG "removing dpp example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any XDL examples if gfx9 targets are not on the list
|
|
if(NOT EX_TARGETS MATCHES "gfx9" AND source_name MATCHES "_xdl")
|
|
message(DEBUG "removing xdl example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any WMMA examples if gfx11 targets are not on the list
|
|
if(NOT EX_TARGETS MATCHES "gfx11" AND NOT EX_TARGETS MATCHES "gfx12" AND source_name MATCHES "_wmma")
|
|
message(DEBUG "removing wmma example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any microscaling examples if gfx950 target is not on the list
|
|
if(NOT EX_TARGETS MATCHES "gfx950" AND source_name MATCHES "_mx")
|
|
message(DEBUG "removing microscaling example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any FP8 examples if CK_ENABLE_FP8 not set
|
|
if(NOT DEFINED CK_ENABLE_FP8 AND source_name MATCHES "_fp8")
|
|
message(DEBUG "removing fp8 example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any BF8 examples if CK_ENABLE_BF8 not set
|
|
if(NOT DEFINED CK_ENABLE_BF8 AND source_name MATCHES "_bf8")
|
|
message(DEBUG "removing bf8 example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
# Build fp8 gemm_multiply_multiply and moe only on gfx94/95
|
|
if(NOT EX_TARGETS MATCHES "gfx94" AND NOT EX_TARGETS MATCHES "gfx95")
|
|
if(source_name MATCHES "fp8" AND source_name MATCHES "(gemm_multiply_multiply|moe)")
|
|
message(DEBUG "Skipping ${source} example for current target")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
endif()
|
|
endforeach()
|
|
#only continue if there are some source files left on the list
|
|
set(source_name_list "")
|
|
foreach(source IN LISTS FILE_NAME)
|
|
get_filename_component(source_name ${source} NAME)
|
|
list(APPEND source_name_list ${source_name})
|
|
endforeach()
|
|
if(FILE_NAME)
|
|
if(source_name_list MATCHES "_xdl" AND NOT source_name_list MATCHES "_pk_i4")
|
|
list(REMOVE_ITEM EX_TARGETS gfx900 gfx906 gfx906:xnack- gfx1030 gfx1100 gfx1101 gfx1102 gfx1103 gfx1150 gfx1151 gfx1152 gfx1200 gfx1201 gfx10-3-generic gfx11-generic gfx12-generic)
|
|
elseif(source_name_list MATCHES "_wmma")
|
|
list(REMOVE_ITEM EX_TARGETS gfx900 gfx906 gfx906:xnack- gfx908:xnack+ gfx908:xnack- gfx90a:xnack+ gfx90a:xnack- gfx908 gfx90a gfx942 gfx1030 gfx950)
|
|
elseif(source_name_list MATCHES "_mx") #only build mx example for gfx950
|
|
list(REMOVE_ITEM EX_TARGETS gfx900 gfx906 gfx906:xnack- gfx908:xnack+ gfx908:xnack- gfx90a:xnack+ gfx90a:xnack- gfx908 gfx90a gfx942 gfx1030 gfx1100 gfx1101 gfx1102 gfx1103 gfx1150 gfx1151 gfx1152 gfx1200 gfx1201 gfx10-3-generic gfx11-generic gfx12-generic)
|
|
elseif(source_name_list MATCHES "_pk_i4") #only build these examples for gfx942 and gfx950
|
|
message(DEBUG "trimming targets for ${FILE_NAME}")
|
|
list(REMOVE_ITEM EX_TARGETS gfx900 gfx906 gfx906:xnack- gfx908:xnack+ gfx908:xnack- gfx90a:xnack+ gfx90a:xnack- gfx908 gfx90a gfx1030 gfx1100 gfx1101 gfx1102 gfx1103 gfx1150 gfx1151 gfx1152 gfx1200 gfx1201 gfx10-3-generic gfx11-generic gfx12-generic)
|
|
endif()
|
|
set_source_files_properties(${FILE_NAME} PROPERTIES LANGUAGE HIP)
|
|
add_executable(${EXAMPLE_NAME} ${FILE_NAME})
|
|
target_link_libraries(${EXAMPLE_NAME} PRIVATE utility)
|
|
target_link_libraries(${EXAMPLE_NAME} PRIVATE getopt::getopt)
|
|
add_test(NAME ${EXAMPLE_NAME} COMMAND $<TARGET_FILE:${EXAMPLE_NAME}> ${ARGN})
|
|
set_property(TARGET ${EXAMPLE_NAME} PROPERTY HIP_ARCHITECTURES ${EX_TARGETS})
|
|
add_dependencies(examples ${EXAMPLE_NAME})
|
|
add_dependencies(check ${EXAMPLE_NAME})
|
|
rocm_install(TARGETS ${EXAMPLE_NAME} COMPONENT examples)
|
|
set(result 0)
|
|
endif()
|
|
message(DEBUG "add_example returns ${result}")
|
|
if(result EQUAL 0 AND NOT "${EXAMPLE_NAME}" IN_LIST REGRESSION_EXAMPLES)
|
|
set_tests_properties(${EXAMPLE_NAME} PROPERTIES LABELS "SMOKE_TEST")
|
|
add_dependencies(smoke ${EXAMPLE_NAME})
|
|
elseif(result EQUAL 0 AND "${EXAMPLE_NAME}" IN_LIST REGRESSION_EXAMPLES)
|
|
set_tests_properties(${EXAMPLE_NAME} PROPERTIES LABELS "REGRESSION_TEST")
|
|
add_dependencies(regression ${EXAMPLE_NAME})
|
|
endif()
|
|
set(result ${result} PARENT_SCOPE)
|
|
endfunction(add_example_executable EXAMPLE_NAME)
|
|
|
|
function(add_example_dependencies EXAMPLE_NAME FILE_NAME)
|
|
if(result EQUAL 0)
|
|
add_dependencies(${EXAMPLE_NAME} ${FILE_NAME})
|
|
endif()
|
|
endfunction(add_example_dependencies EXAMPLE_NAME)
|
|
|
|
function(add_example_executable_no_testing EXAMPLE_NAME FILE_NAME)
|
|
message(DEBUG "adding example ${EXAMPLE_NAME}")
|
|
set(result 1)
|
|
if(DEFINED DTYPES)
|
|
foreach(source IN LISTS FILE_NAME)
|
|
get_filename_component(source_name ${source} NAME)
|
|
set(test 0)
|
|
if((source_name MATCHES "_fp16" OR source_name MATCHES "_f16") AND NOT "fp16" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_fp32" OR source_name MATCHES "_f32") AND NOT "fp32" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_fp64" OR source_name MATCHES "_f64") AND NOT "fp64" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_fp8" OR source_name MATCHES "_f8") AND NOT "fp8" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_bf8" OR source_name MATCHES "_bf8") AND NOT "bf8" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_bf16" OR source_name MATCHES "_b16") AND NOT "bf16" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if((source_name MATCHES "_int8" OR source_name MATCHES "_i8") AND NOT "int8" IN_LIST DTYPES)
|
|
set(test 1)
|
|
endif()
|
|
if(test EQUAL 1)
|
|
message(DEBUG "removing example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
endforeach()
|
|
endif()
|
|
|
|
set(EX_TARGETS ${SUPPORTED_GPU_TARGETS})
|
|
|
|
set(source_name_list "")
|
|
foreach(source IN LISTS FILE_NAME)
|
|
get_filename_component(source_name ${source} NAME)
|
|
#Do not build any DL examples if DL_KERNELS not set
|
|
if(NOT DEFINED DL_KERNELS AND source_name MATCHES "_dl")
|
|
message(DEBUG "removing dl example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any XDL examples if gfx9 targets are not on the list
|
|
if(NOT EX_TARGETS MATCHES "gfx9" AND source_name MATCHES "_xdl")
|
|
message(DEBUG "removing xdl example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
#Do not build any WMMA examples if gfx11 targets are not on the list
|
|
if(NOT EX_TARGETS MATCHES "gfx11" AND NOT EX_TARGETS MATCHES "gfx12" AND source_name MATCHES "_wmma")
|
|
message(DEBUG "removing wmma example ${source} ")
|
|
list(REMOVE_ITEM FILE_NAME "${source}")
|
|
endif()
|
|
list(APPEND source_name_list ${source_name})
|
|
endforeach()
|
|
#only continue if there are some source files left on the list
|
|
if(FILE_NAME)
|
|
if(source_name_list MATCHES "_xdl")
|
|
list(REMOVE_ITEM EX_TARGETS gfx900 gfx906 gfx906:xnack- gfx1030 gfx1100 gfx1101 gfx1102 gfx1103 gfx1150 gfx1151 gfx1152 gfx1200 gfx1201 gfx10-3-generic gfx11-generic gfx12-generic)
|
|
elseif(source_name_list MATCHES "_wmma")
|
|
list(REMOVE_ITEM EX_TARGETS gfx900 gfx906 gfx906:xnack- gfx908:xnack+ gfx908:xnack- gfx90a:xnack+ gfx90a:xnack- gfx908 gfx90a gfx942 gfx1030 gfx950)
|
|
endif()
|
|
set_source_files_properties(${FILE_NAME} PROPERTIES LANGUAGE HIP)
|
|
add_executable(${EXAMPLE_NAME} ${FILE_NAME})
|
|
target_link_libraries(${EXAMPLE_NAME} PRIVATE utility)
|
|
add_dependencies(examples ${EXAMPLE_NAME})
|
|
set_property(TARGET ${EXAMPLE_NAME} PROPERTY HIP_ARCHITECTURES ${EX_TARGETS})
|
|
rocm_install(TARGETS ${EXAMPLE_NAME} COMPONENT examples)
|
|
set(result 0)
|
|
endif()
|
|
|
|
message(DEBUG "add_example returns ${result}")
|
|
set(result ${result} PARENT_SCOPE)
|
|
|
|
endfunction(add_example_executable_no_testing EXAMPLE_NAME)
|
|
|
|
function(example_compile_options EXAMPLE_NAME)
|
|
if(TARGET ${EXAMPLE_NAME})
|
|
target_compile_options(${EXAMPLE_NAME} ${ARGN})
|
|
endif()
|
|
endfunction(example_compile_options)
|
|
|
|
# add all example subdir
|
|
file(GLOB dir_list LIST_DIRECTORIES true *)
|
|
FOREACH(subdir ${dir_list})
|
|
if(IS_DIRECTORY "${subdir}" AND EXISTS "${subdir}/CMakeLists.txt")
|
|
add_subdirectory(${subdir})
|
|
ENDIF()
|
|
ENDFOREACH()
|