CMakeLists.txt 17.3 KB
Newer Older
1
cmake_minimum_required(VERSION 3.14)
JD's avatar
JD committed
2

3
4
5
6
7
8
9
10
11
12
13
14
15
# This has to be initialized before the project() command appears
# Set the default of CMAKE_BUILD_TYPE to be release, unless user specifies with -D.  MSVC_IDE does not use CMAKE_BUILD_TYPE
if( NOT MSVC_IDE AND NOT CMAKE_BUILD_TYPE )
    set( CMAKE_BUILD_TYPE Release CACHE STRING "Choose the type of build, options are: None Debug Release RelWithDebInfo MinSizeRel." )
endif()

# Default installation path
if(WIN32)
    set(CMAKE_INSTALL_PREFIX "/opt/rocm/x86_64-w64-mingw32" CACHE PATH "")
else()
    set(CMAKE_INSTALL_PREFIX "/opt/rocm" CACHE PATH "")
endif()

16
set(version 1.1.0)
JD's avatar
JD committed
17
# Check support for CUDA/HIP in Cmake
18
project(composable_kernel VERSION ${version})
Chao Liu's avatar
Chao Liu committed
19

20
21
list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")

22
if (DTYPES)
23
24
25
26
27
28
29
30
    add_definitions(-DDTYPES)
    if (DTYPES MATCHES "int8")
        add_definitions(-DCK_ENABLE_INT8)
        set(CK_ENABLE_INT8 "ON")
    endif()
    if (DTYPES MATCHES "fp8")
        add_definitions(-DCK_ENABLE_FP8)
        set(CK_ENABLE_FP8 "ON")
31
32
33
34
35
36
        add_compile_options(-Wno-bit-int-extension)
    endif()
    if (DTYPES MATCHES "bf8")
        add_definitions(-DCK_ENABLE_BF8)
        set(CK_ENABLE_BF8 "ON")
        add_compile_options(-Wno-bit-int-extension)
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
    endif()
    if (DTYPES MATCHES "fp16")
        add_definitions(-DCK_ENABLE_FP16)
        set(CK_ENABLE_FP16 "ON")
    endif()
    if (DTYPES MATCHES "fp32")
        add_definitions(-DCK_ENABLE_FP32)
        set(CK_ENABLE_FP32 "ON")
    endif()
    if (DTYPES MATCHES "fp64")
        add_definitions(-DCK_ENABLE_FP64)
        set(CK_ENABLE_FP64 "ON")
    endif()
    if (DTYPES MATCHES "bf16")
        add_definitions(-DCK_ENABLE_BF16)
        set(CK_ENABLE_BF16 "ON")
    endif()
    message("DTYPES macro set to ${DTYPES}")
55
else()
56
    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP8 -DCK_ENABLE_BF8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16)
57
    set(CK_ENABLE_ALL_DTYPES "ON")
58
    add_compile_options(-Wno-bit-int-extension) # enable fp8 and bf8
59
60
endif()

61
62
if(DL_KERNELS)
    add_definitions(-DDL_KERNELS)
63
    set(CK_ENABLE_DL_KERNELS "ON")
64
65
66
67
endif()

if(INSTANCES_ONLY)
    add_definitions(-DINSTANCES_ONLY)
68
    set(CK_ENABLE_INSTANCES_ONLY "ON")
69
70
endif()

71
72
73
74
75
76
77
78
# CK config file to record supported datatypes, etc.
configure_file("${PROJECT_SOURCE_DIR}/include/ck/config.h.in" "${PROJECT_BINARY_DIR}/include/ck/config.h")

# CK version file to record release version as well as git commit hash
find_package(Git REQUIRED)
execute_process(COMMAND "${GIT_EXECUTABLE}" rev-parse HEAD OUTPUT_VARIABLE COMMIT_ID OUTPUT_STRIP_TRAILING_WHITESPACE)
configure_file("${PROJECT_SOURCE_DIR}/include/ck/version.h.in" "${PROJECT_BINARY_DIR}/include/ck/version.h")

JD's avatar
JD committed
79
80
enable_testing()

81
set(ROCM_SYMLINK_LIBS OFF)
Anthony Chang's avatar
Anthony Chang committed
82
find_package(ROCM REQUIRED PATHS /opt/rocm)
JD's avatar
JD committed
83
84
85
86
87
88

include(ROCMInstallTargets)
include(ROCMPackageConfigHelpers)
include(ROCMSetupVersion)
include(ROCMInstallSymlinks)
include(ROCMCreatePackage)
Chao Liu's avatar
Chao Liu committed
89
include(CheckCXXCompilerFlag)
90
include(ROCMCheckTargetIds)
JD's avatar
JD committed
91
include(TargetFlags)
92
93
94

rocm_setup_version(VERSION ${version})

JD's avatar
JD committed
95
96
list(APPEND CMAKE_PREFIX_PATH ${CMAKE_INSTALL_PREFIX} ${CMAKE_INSTALL_PREFIX}/llvm ${CMAKE_INSTALL_PREFIX}/hip /opt/rocm /opt/rocm/llvm /opt/rocm/hip)

97
98
99
100
101
102
message("GPU_TARGETS= ${GPU_TARGETS}")

message("checking which targets are supported")
#This is the list of targets to be used in case GPU_TARGETS is not set on command line
#These targets will be filtered and only supported ones will be used
#Setting GPU_TARGETS on command line will override this list
103
104
105
106
107
if(NOT PROFILER_ONLY)
    rocm_check_target_ids(DEFAULT_GPU_TARGETS
        TARGETS "gfx900;gfx906;gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx1100;gfx1101;gfx1102")
else()
    add_definitions(-DPROFILER_ONLY)
108
    set(GPU_TARGETS "" CACHE STRING "" FORCE)
109
110
111
112
113
114
115
116
117
118
119
120
    if(GPU_TARGETS)
        message(FATAL_ERROR "For PROFILE_ONLY build, please do not set GPU_TARGETS, use GPU_ARCH = gfx9, gfx10, or gfx11")
    endif()
    if(GPU_ARCH MATCHES "gfx9")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx900;gfx906;gfx908;gfx90a;gfx940;gfx941;gfx942")
    elseif(GPU_ARCH MATCHES "gfx10")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1030")
    elseif(GPU_ARCH MATCHES "gfx11")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1100;gfx1101;gfx1102")
    else()
        message(FATAL_ERROR "For PROFILE_ONLY build, please specify GPU_ARCH as gfx9, gfx10, or gfx11")
    endif()
121
    set(GPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
122
123
endif()

124
message("Supported GPU_TARGETS= ${DEFAULT_GPU_TARGETS}")
125

126
set(AMDGPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
127
128
129
130
131
132

if(GPU_TARGETS)
    message("Building CK for the following targets: ${GPU_TARGETS}")
else()
    message("Building CK for the following targets: ${AMDGPU_TARGETS}")
endif()
133
find_package(hip)
134
135
136
137
# No assumption that HIP kernels are launched with uniform block size for backward compatibility
# SWDEV-413293 and https://reviews.llvm.org/D155213
math(EXPR hip_VERSION_FLAT "(${hip_VERSION_MAJOR} * 1000 + ${hip_VERSION_MINOR}) * 100000 + ${hip_VERSION_PATCH}")
message("hip_version_flat=${hip_VERSION_FLAT}")
138
if(${hip_VERSION_FLAT} GREATER 500723302)
139
140
141
   message("Adding the fno-offload-uniform-block compiler flag")
   add_compile_options(-fno-offload-uniform-block)
endif()
142

Adam Osewski's avatar
Adam Osewski committed
143
option(USE_BITINT_EXTENSION_INT4, "Whether to enable clang's BitInt extension to provide int4 data type." OFF)
144
option(USE_OPT_NAVI3X, "Whether to enable LDS cumode and Wavefront32 mode for NAVI3X silicons." OFF)
Adam Osewski's avatar
Adam Osewski committed
145
146
147
148
149
150
151

if(USE_BITINT_EXTENSION_INT4)
    add_compile_definitions(CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4)
    add_compile_options(-Wno-bit-int-extension)
    message("CK compiled with USE_BITINT_EXTENSION_INT4 set to ${USE_BITINT_EXTENSION_INT4}")
endif()

152
153
154
155
156
157
if(USE_OPT_NAVI3X)
    add_compile_options(-mcumode)
    add_compile_options(-mno-wavefrontsize64)
    message("CK compiled with USE_OPT_NAVI3X set to ${USE_OPT_NAVI3X}")
endif()

158
159
160
161
162
## Threads
set(THREADS_PREFER_PTHREAD_FLAG ON)
find_package(Threads REQUIRED)
link_libraries(Threads::Threads)

163
## C++
Chao Liu's avatar
Chao Liu committed
164
enable_language(CXX)
Chao Liu's avatar
Chao Liu committed
165
set(CMAKE_CXX_STANDARD 17)
Chao Liu's avatar
Chao Liu committed
166
set(CMAKE_CXX_STANDARD_REQUIRED ON)
Chao Liu's avatar
Chao Liu committed
167
set(CMAKE_CXX_EXTENSIONS OFF)
Chao Liu's avatar
Chao Liu committed
168
169
message("CMAKE_CXX_COMPILER_ID: ${CMAKE_CXX_COMPILER_ID}")

170
## OpenMP
Chao Liu's avatar
Chao Liu committed
171
172
173
174
175
176
177
178
179
180
181
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()
Chao Liu's avatar
Chao Liu committed
182

Chao Liu's avatar
Chao Liu committed
183
184
185
186
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}")
Chao Liu's avatar
Chao Liu committed
187

Chao Liu's avatar
Chao Liu committed
188
189
link_libraries(${OpenMP_gomp_LIBRARY})
link_libraries(${OpenMP_pthread_LIBRARY})
Chao Liu's avatar
Chao Liu committed
190

191
192
## HIP
find_package(HIP REQUIRED)
JD's avatar
JD committed
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
# Override HIP version in config.h, if necessary.
# The variables set by find_package() can't be overwritten,
# therefore let's use intermediate variables.
set(CK_HIP_VERSION_MAJOR "${HIP_VERSION_MAJOR}")
set(CK_HIP_VERSION_MINOR "${HIP_VERSION_MINOR}")
set(CK_HIP_VERSION_PATCH "${HIP_VERSION_PATCH}")
if( DEFINED CK_OVERRIDE_HIP_VERSION_MAJOR )
    set(CK_HIP_VERSION_MAJOR "${CK_OVERRIDE_HIP_VERSION_MAJOR}")
    message(STATUS "CK_HIP_VERSION_MAJOR overriden with ${CK_OVERRIDE_HIP_VERSION_MAJOR}")
endif()
if( DEFINED CK_OVERRIDE_HIP_VERSION_MINOR )
    set(CK_HIP_VERSION_MINOR "${CK_OVERRIDE_HIP_VERSION_MINOR}")
    message(STATUS "CK_HIP_VERSION_MINOR overriden with ${CK_OVERRIDE_HIP_VERSION_MINOR}")
endif()
if( DEFINED CK_OVERRIDE_HIP_VERSION_PATCH )
    set(CK_HIP_VERSION_PATCH "${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}")
212
213
link_libraries(hip::device)
add_compile_definitions(__HIP_PLATFORM_HCC__=1)
214

Chao Liu's avatar
Chao Liu committed
215
216
## tidy
include(EnableCompilerWarnings)
JD's avatar
JD committed
217
set(CK_TIDY_ERRORS ERRORS * -readability-inconsistent-declaration-parameter-name)
Chao Liu's avatar
Chao Liu committed
218
if(CMAKE_CXX_COMPILER MATCHES ".*hcc" OR CMAKE_CXX_COMPILER MATCHES ".*clang\\+\\+")
JD's avatar
JD committed
219
    set(CK_TIDY_CHECKS -modernize-use-override -readability-non-const-parameter)
Chao Liu's avatar
Chao Liu committed
220
# Enable tidy on hip
JD's avatar
JD committed
221
222
elseif(CK_BACKEND STREQUAL "HIP" OR CK_BACKEND STREQUAL "HIPNOGPU")
    set(CK_TIDY_ERRORS ALL)
Chao Liu's avatar
Chao Liu committed
223
224
endif()

JD's avatar
JD committed
225

Chao Liu's avatar
Chao Liu committed
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
include(ClangTidy)
enable_clang_tidy(
    CHECKS
        *
        -abseil-*
        -android-cloexec-fopen
        # Yea we shouldn't be using rand()
        -cert-msc30-c
        -bugprone-exception-escape
        -bugprone-macro-parentheses
        -cert-env33-c
        -cert-msc32-c
        -cert-msc50-cpp
        -cert-msc51-cpp
        -cert-dcl37-c
        -cert-dcl51-cpp
        -clang-analyzer-alpha.core.CastToStruct
        -clang-analyzer-optin.performance.Padding
        -clang-diagnostic-deprecated-declarations
        -clang-diagnostic-extern-c-compat
        -clang-diagnostic-unused-command-line-argument
        -cppcoreguidelines-avoid-c-arrays
        -cppcoreguidelines-avoid-magic-numbers
        -cppcoreguidelines-explicit-virtual-functions
        -cppcoreguidelines-init-variables
        -cppcoreguidelines-macro-usage
        -cppcoreguidelines-non-private-member-variables-in-classes
        -cppcoreguidelines-pro-bounds-array-to-pointer-decay
        -cppcoreguidelines-pro-bounds-constant-array-index
        -cppcoreguidelines-pro-bounds-pointer-arithmetic
        -cppcoreguidelines-pro-type-member-init
        -cppcoreguidelines-pro-type-reinterpret-cast
        -cppcoreguidelines-pro-type-union-access
        -cppcoreguidelines-pro-type-vararg
        -cppcoreguidelines-special-member-functions
        -fuchsia-*
        -google-explicit-constructor
        -google-readability-braces-around-statements
        -google-readability-todo
        -google-runtime-int
        -google-runtime-references
        -hicpp-vararg
        -hicpp-braces-around-statements
        -hicpp-explicit-conversions
        -hicpp-named-parameter
        -hicpp-no-array-decay
        # We really shouldn't use bitwise operators with signed integers, but
        # opencl leaves us no choice
        -hicpp-avoid-c-arrays
        -hicpp-signed-bitwise
        -hicpp-special-member-functions
        -hicpp-uppercase-literal-suffix
        -hicpp-use-auto
        -hicpp-use-equals-default
        -hicpp-use-override
        -llvm-header-guard
        -llvm-include-order
        #-llvmlibc-*
        -llvmlibc-restrict-system-libc-headers
        -llvmlibc-callee-namespace
        -llvmlibc-implementation-in-namespace
        -llvm-else-after-return
        -llvm-qualified-auto
        -misc-misplaced-const
        -misc-non-private-member-variables-in-classes
        -misc-no-recursion
        -modernize-avoid-bind
        -modernize-avoid-c-arrays
        -modernize-pass-by-value
        -modernize-use-auto
        -modernize-use-default-member-init
        -modernize-use-equals-default
        -modernize-use-trailing-return-type
        -modernize-use-transparent-functors
        -performance-unnecessary-value-param
        -readability-braces-around-statements
        -readability-else-after-return
        # we are not ready to use it, but very useful
        -readability-function-cognitive-complexity
        -readability-isolate-declaration
        -readability-magic-numbers
        -readability-named-parameter
        -readability-uppercase-literal-suffix
        -readability-convert-member-functions-to-static
        -readability-qualified-auto
        -readability-redundant-string-init
        # too many narrowing conversions in our code
        -bugprone-narrowing-conversions
        -cppcoreguidelines-narrowing-conversions
        -altera-struct-pack-align
        -cppcoreguidelines-prefer-member-initializer
JD's avatar
JD committed
317
318
        ${CK_TIDY_CHECKS}
        ${CK_TIDY_ERRORS}
Chao Liu's avatar
Chao Liu committed
319
320
321
    HEADER_FILTER
        "\.hpp$"
    EXTRA_ARGS
JD's avatar
JD committed
322
        -DCK_USE_CLANG_TIDY
Chao Liu's avatar
Chao Liu committed
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
)

include(CppCheck)
enable_cppcheck(
    CHECKS
        warning
        style
        performance
        portability
    SUPPRESS
        ConfigurationNotChecked
        constStatement
        duplicateCondition
        noExplicitConstructor
        passedByValue
Chao Liu's avatar
Chao Liu committed
338
        preprocessorErrorDirective
Chao Liu's avatar
Chao Liu committed
339
340
341
342
343
344
345
        shadowVariable
        unusedFunction
        unusedPrivateFunction
        unusedStructMember
        unmatchedSuppression
    FORCE
    SOURCES
Chao Liu's avatar
Chao Liu committed
346
        library/src
Chao Liu's avatar
Chao Liu committed
347
348
349
    INCLUDE
        ${CMAKE_CURRENT_SOURCE_DIR}/include
        ${CMAKE_CURRENT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
350
        ${CMAKE_CURRENT_SOURCE_DIR}/library/include
Chao Liu's avatar
Chao Liu committed
351
352
353
354
    DEFINE
        CPPCHECK=1
        __linux__=1
)
Chao Liu's avatar
Chao Liu committed
355

JD's avatar
JD committed
356
357
358
359
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/lib)
set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/lib)
set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/bin)

360
# set CK project include directories
Chao Liu's avatar
Chao Liu committed
361
include_directories(BEFORE
362
    ${PROJECT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
363
364
    ${PROJECT_SOURCE_DIR}/include
    ${PROJECT_SOURCE_DIR}/library/include
365
    ${HIP_INCLUDE_DIRS}
JD's avatar
JD committed
366
)
Chao Liu's avatar
Chao Liu committed
367

JD's avatar
JD committed
368
369
SET(BUILD_DEV ON CACHE BOOL "BUILD_DEV")
if(BUILD_DEV)
Chao Liu's avatar
Chao Liu committed
370
371
    add_compile_options(-Werror)
    add_compile_options(-Weverything)
JD's avatar
JD committed
372
373
374
endif()
message("CMAKE_CXX_FLAGS: ${CMAKE_CXX_FLAGS}")

Anthony Chang's avatar
Anthony Chang committed
375
376
add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND} --output-on-failure -C ${CMAKE_CFG_INTDIR})

377
378
379
380
file(GLOB_RECURSE INSTANCE_FILES "${PROJECT_SOURCE_DIR}/*/device_*_instance.cpp")
file(GLOB dir_list RELATIVE ${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu ${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/*)
set(CK_DEVICE_INSTANCES)
FOREACH(subdir_path ${dir_list})
381
382
383
384
385
386
set(target_dir)
IF(IS_DIRECTORY "${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/${subdir_path}")
    set(cmake_instance)
    file(READ "${PROJECT_SOURCE_DIR}/library/src/tensor_operation_instance/gpu/${subdir_path}/CMakeLists.txt" cmake_instance)
    set(add_inst 0)
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp8\" " AND DTYPES MATCHES "fp8")
387
388
        #message("fp8 instance found!")
        set(add_inst 1)
389
    endif()
390
391
392
393
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"bf8\" " AND DTYPES MATCHES "bf8")
        #message("bf8 instance found!")
        set(add_inst 1)
    endif()
394
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp16\"" AND DTYPES MATCHES "fp16")
395
396
        #message("fp16 instance found!")
        set(add_inst 1)
397
398
    endif()
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp32\"" AND DTYPES MATCHES "fp32")
399
400
        #message("fp32 instance found!")
        set(add_inst 1)
401
402
    endif()
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"fp64\"" AND DTYPES MATCHES "fp64")
403
404
        #message("fp64 instance found!")
        set(add_inst 1)
405
406
    endif()
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"bf16\"" AND DTYPES MATCHES "bf16")
407
408
        #message("bf16 instance found!")
        set(add_inst 1)
409
410
    endif()
    if("${cmake_instance}" MATCHES "DTYPES MATCHES \"int8\"" AND DTYPES MATCHES "int8")
411
412
        #message("int8 instance found!")
        set(add_inst 1)
413
414
    endif()
    if(NOT "${cmake_instance}" MATCHES "DTYPES")
415
416
        #message("instance should be built for all types!")
        set(add_inst 1)
417
418
    endif()
    if(add_inst EQUAL 1 OR NOT DEFINED DTYPES)
419
        list(APPEND CK_DEVICE_INSTANCES device_${subdir_path}_instance)
420
421
    endif()
ENDIF()
422
ENDFOREACH()
423

424
add_custom_target(instances DEPENDS utility;${CK_DEVICE_INSTANCES}  SOURCES ${INSTANCE_FILES})
425
add_subdirectory(library)
426

427
if(NOT DEFINED INSTANCES_ONLY)
428
 if(NOT DEFINED PROFILER_ONLY)
429
   rocm_package_setup_component(tests
430
431
        LIBRARY_NAME composablekernel
        PACKAGE_NAME tests # Prevent -static suffix on package name
432
   )
433

434
   rocm_package_setup_component(examples
435
436
        LIBRARY_NAME composablekernel
        PACKAGE_NAME examples
437
   )
438
439
   add_subdirectory(example)
   add_subdirectory(test)
440

441
   rocm_package_setup_component(profiler
442
443
        LIBRARY_NAME composablekernel
        PACKAGE_NAME ckProfiler
444
445
   )
   add_subdirectory(profiler)
446
447
448
449
450
451
452
453
  else()
    #When building PROFILER_ONLY, label the package with GPU_ARCH
    rocm_package_setup_component(profiler
       LIBRARY_NAME composablekernel
       PACKAGE_NAME ckProfiler_${GPU_ARCH}
    )
    add_subdirectory(profiler)
  endif()
454
endif()
JD's avatar
JD committed
455
456
457
458
459
460
461
462
463
464

#Create an interface target for the include only files and call it "composablekernels"
include(CMakePackageConfigHelpers)

write_basic_package_version_file(
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
    VERSION "${version}"
    COMPATIBILITY AnyNewerVersion
)

Anthony Chang's avatar
Anthony Chang committed
465
configure_package_config_file(${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in
466
467
468
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
    INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
    NO_CHECK_REQUIRED_COMPONENTS_MACRO
JD's avatar
JD committed
469
470
)

471
rocm_install(FILES
JD's avatar
JD committed
472
473
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
Anthony Chang's avatar
Anthony Chang committed
474
    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
JD's avatar
JD committed
475
)
476

477
# Install CK version and configuration files
478
rocm_install(FILES
479
480
481
482
483
    ${PROJECT_BINARY_DIR}/include/ck/version.h
    ${PROJECT_BINARY_DIR}/include/ck/config.h
    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/ck/
)

484
485
486
487
488
489
490
491
492
493
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
)