mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-04-19 14:29:05 +00:00
004784ef98
* chore(copyright) update library wide CMakeLists.txt files copyright header template * Fix build --------- Co-authored-by: Sami Remes <samremes@amd.com>
57 lines
2.6 KiB
CMake
57 lines
2.6 KiB
CMake
# Copyright (c) Advanced Micro Devices, Inc., or its affiliates.
|
|
# SPDX-License-Identifier: MIT
|
|
|
|
function(create_tile_layernorm2d_fwd SUFFIX)
|
|
set(TEST_CK_TILE_LAYERNORM2D_FWD "test_ck_tile_layernorm2d_fwd_${SUFFIX}")
|
|
|
|
message(DEBUG "adding example ${TEST_CK_TILE_LAYERNORM2D_FWD}")
|
|
add_test_executable(${TEST_CK_TILE_LAYERNORM2D_FWD} layernorm2d_fwd_${SUFFIX}.cpp)
|
|
target_include_directories(${TEST_CK_TILE_LAYERNORM2D_FWD} PRIVATE ${CMAKE_CURRENT_LIST_DIR})
|
|
target_sources(${TEST_CK_TILE_LAYERNORM2D_FWD} PRIVATE ${LAYERNORM2D_FWD_GEN_BLOBS})
|
|
|
|
set(TEST_CK_TILE_LAYERNORM2D_FWD_COMPILE_OPTIONS)
|
|
|
|
# NOTE: we turn off undefined-func-template to let source compile without explicit declare function specializations
|
|
list(APPEND TEST_CK_TILE_LAYERNORM2D_FWD_COMPILE_OPTIONS -Wno-undefined-func-template -Wno-float-equal --offload-compress)
|
|
|
|
target_compile_options(${TEST_CK_TILE_LAYERNORM2D_FWD} PRIVATE ${TEST_CK_TILE_LAYERNORM2D_FWD_COMPILE_OPTIONS})
|
|
endfunction()
|
|
|
|
if(GPU_TARGETS MATCHES "gfx9|gfx11|gfx12")
|
|
set(LAYERNORM2D_FWD_KNOWN_APIS "fwd;bwd")
|
|
set(LAYERNORM2D_FWD_ENABLE_APIS "fwd" CACHE STRING
|
|
"semicolon-separated list of APIs to generate (${LAYERNORM2D_FWD_KNOWN_APIS}) & link, or \"all\".")
|
|
if(LAYERNORM2D_FWD_ENABLE_APIS STREQUAL "all")
|
|
set(LAYERNORM2D_FWD_ENABLE_APIS ${LAYERNORM2D_FWD_KNOWN_APIS})
|
|
endif()
|
|
|
|
# generate a list of kernels, but not actually emit files at config sta
|
|
execute_process(
|
|
COMMAND ${Python3_EXECUTABLE} ${CMAKE_CURRENT_LIST_DIR}/generate.py
|
|
--api ${LAYERNORM2D_FWD_ENABLE_APIS} --working_path ${CMAKE_CURRENT_BINARY_DIR} --list_blobs
|
|
RESULT_VARIABLE ret
|
|
)
|
|
if(ret AND NOT ret EQUAL 0)
|
|
message( FATAL_ERROR "Fail to generate kernels via Python. ${ret}")
|
|
endif()
|
|
|
|
file(STRINGS ${CMAKE_CURRENT_BINARY_DIR}/layernorm2d_fwd_blobs.txt LAYERNORM2D_FWD_GEN_BLOBS)
|
|
|
|
add_custom_command(
|
|
OUTPUT ${LAYERNORM2D_FWD_GEN_BLOBS}
|
|
COMMAND ${Python3_EXECUTABLE} ${CMAKE_CURRENT_LIST_DIR}/generate.py
|
|
--api ${LAYERNORM2D_FWD_ENABLE_APIS} --working_path ${CMAKE_CURRENT_BINARY_DIR} --gen_blobs
|
|
)
|
|
|
|
create_tile_layernorm2d_fwd("fp16")
|
|
create_tile_layernorm2d_fwd("bf16")
|
|
|
|
# TODO: we have to turn off this global prop, otherwise the progress bar generated
|
|
# by cmake will print too many files, execvp: /bin/sh: Argument list too long
|
|
# however, this property may affect global
|
|
# TODO: consider codegen a makefile by us
|
|
set_property(GLOBAL PROPERTY RULE_MESSAGES OFF)
|
|
else()
|
|
message(DEBUG "Skipping ck tile add_rmsnorm2d_rdquant_fwd tests for current target")
|
|
endif()
|