CMakeLists.txt 18.9 KB
Newer Older
1
cmake_minimum_required(VERSION 3.14)
2
3
4
5
if(POLICY CMP0140)
  # policies CMP0140 not known to CMake until 3.25
  cmake_policy(SET CMP0140 NEW)
endif()
JD's avatar
JD committed
6

7
8
9
10
11
12
13
14
15
16
17
18
# 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()
JD's avatar
JD committed
19

20
set(version 1.1.0)
JD's avatar
JD committed
21
# Check support for CUDA/HIP in Cmake
22
project(composable_kernel VERSION ${version})
Chao Liu's avatar
Chao Liu committed
23

24
25
list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")

26
if (DTYPES)
27
28
29
30
31
32
33
34
35
    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")
    endif()
36
37
38
    if (DTYPES MATCHES "bf8")
        add_definitions(-DCK_ENABLE_BF8)
        set(CK_ENABLE_BF8 "ON")
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
    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}")
57
else()
58
    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP8 -DCK_ENABLE_BF8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16)
59
    set(CK_ENABLE_ALL_DTYPES "ON")
60
61
endif()

62
63
64
#for f8/bf8_t type
add_compile_options(-Wno-bit-int-extension)

65
66
if(DL_KERNELS)
    add_definitions(-DDL_KERNELS)
67
    set(CK_ENABLE_DL_KERNELS "ON")
68
69
70
71
endif()

if(INSTANCES_ONLY)
    add_definitions(-DINSTANCES_ONLY)
72
    set(CK_ENABLE_INSTANCES_ONLY "ON")
73
74
endif()

75
76
77
78
79
80
81
82
# 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
83
84
enable_testing()

85
set(ROCM_SYMLINK_LIBS OFF)
Anthony Chang's avatar
Anthony Chang committed
86
find_package(ROCM REQUIRED PATHS /opt/rocm)
JD's avatar
JD committed
87
88
89
90
91
92

include(ROCMInstallTargets)
include(ROCMPackageConfigHelpers)
include(ROCMSetupVersion)
include(ROCMInstallSymlinks)
include(ROCMCreatePackage)
Chao Liu's avatar
Chao Liu committed
93
include(CheckCXXCompilerFlag)
94
include(ROCMCheckTargetIds)
JD's avatar
JD committed
95
include(TargetFlags)
96
97
98

rocm_setup_version(VERSION ${version})

JD's avatar
JD committed
99
100
list(APPEND CMAKE_PREFIX_PATH ${CMAKE_INSTALL_PREFIX} ${CMAKE_INSTALL_PREFIX}/llvm ${CMAKE_INSTALL_PREFIX}/hip /opt/rocm /opt/rocm/llvm /opt/rocm/hip)

101
102
103
104
105
106
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
107
108
if(NOT PROFILER_ONLY)
    rocm_check_target_ids(DEFAULT_GPU_TARGETS
109
        TARGETS "gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx1100;gfx1101;gfx1102")
110
111
else()
    add_definitions(-DPROFILER_ONLY)
112
    set(GPU_TARGETS "" CACHE STRING "" FORCE)
113
    if(GPU_TARGETS)
114
        message(FATAL_ERROR "For PROFILE_ONLY build, please do not set GPU_TARGETS, use GPU_ARCH = gfx90, gfx94, gfx10, or gfx11")
115
    endif()
116
    if(GPU_ARCH MATCHES "gfx90")
117
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx908;gfx90a")
118
119
    elseif(GPU_ARCH MATCHES "gfx94")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx940;gfx941;gfx942")
120
121
122
123
124
    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()
125
        message(FATAL_ERROR "For PROFILE_ONLY build, please specify GPU_ARCH as gfx90, gfx94, gfx10, or gfx11")
126
    endif()
127
    set(GPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
128
129
endif()

130
message("Supported GPU_TARGETS= ${DEFAULT_GPU_TARGETS}")
131

132
set(AMDGPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
133
134
135
136
137
138

if(GPU_TARGETS)
    message("Building CK for the following targets: ${GPU_TARGETS}")
else()
    message("Building CK for the following targets: ${AMDGPU_TARGETS}")
endif()
139

Adam Osewski's avatar
Adam Osewski committed
140
option(USE_BITINT_EXTENSION_INT4, "Whether to enable clang's BitInt extension to provide int4 data type." OFF)
141
option(USE_OPT_NAVI3X, "Whether to enable LDS cumode and Wavefront32 mode for NAVI3X silicons." OFF)
Adam Osewski's avatar
Adam Osewski committed
142
143
144
145
146
147
148

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()

149
150
151
152
153
154
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()

155
156
157
158
159
## Threads
set(THREADS_PREFER_PTHREAD_FLAG ON)
find_package(Threads REQUIRED)
link_libraries(Threads::Threads)

160
## C++
Chao Liu's avatar
Chao Liu committed
161
enable_language(CXX)
Chao Liu's avatar
Chao Liu committed
162
set(CMAKE_CXX_STANDARD 17)
Chao Liu's avatar
Chao Liu committed
163
set(CMAKE_CXX_STANDARD_REQUIRED ON)
Chao Liu's avatar
Chao Liu committed
164
set(CMAKE_CXX_EXTENSIONS OFF)
Chao Liu's avatar
Chao Liu committed
165
166
message("CMAKE_CXX_COMPILER_ID: ${CMAKE_CXX_COMPILER_ID}")

Alan Turner's avatar
Alan Turner committed
167
option(CK_BUILD_JIT_LIB, "Only build the CK JIT Helper Library" OFF)
Alan Turner's avatar
Alan Turner committed
168
if (NOT CK_BUILD_JIT_LIB)
Umang Yadav's avatar
Umang Yadav committed
169
170
171
172
173
174
175
176
177
	find_package(hip)
	# 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}")
	if(${hip_VERSION_FLAT} GREATER 500723302)
	   message("Adding the fno-offload-uniform-block compiler flag")
	   add_compile_options(-fno-offload-uniform-block)
	endif()
178

Alan Turner's avatar
Alan Turner committed
179
    option(USE_BITINT_EXTENSION_INT4, "Whether to enable clang's BitInt extension to provide int4 data type." OFF)
Alan Turner's avatar
Alan Turner committed
180
    option(USE_OPT_NAVI3X, "Whether to enable LDS cumode and Wavefront32 mode for NAVI3X silicons." OFF)
Chao Liu's avatar
Chao Liu committed
181

Alan Turner's avatar
Alan Turner committed
182
183
184
185
186
    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()
Chao Liu's avatar
Chao Liu committed
187

Alan Turner's avatar
Alan Turner committed
188
189
190
191
192
    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()
Chao Liu's avatar
Chao Liu committed
193

Alan Turner's avatar
Alan Turner committed
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
    ## Threads
    set(THREADS_PREFER_PTHREAD_FLAG ON)
    find_package(Threads REQUIRED)
    link_libraries(Threads::Threads)
    
    ## OpenMP
    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
211

Alan Turner's avatar
Alan Turner committed
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
    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}")

    link_libraries(${OpenMP_gomp_LIBRARY})
    link_libraries(${OpenMP_pthread_LIBRARY})

    ## HIP
    find_package(HIP REQUIRED)
    # 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}")
    link_libraries(hip::device)
    add_compile_definitions(__HIP_PLATFORM_HCC__=1)
JD's avatar
JD committed
243
endif()
244

Chao Liu's avatar
Chao Liu committed
245
246
## tidy
include(EnableCompilerWarnings)
JD's avatar
JD committed
247
set(CK_TIDY_ERRORS ERRORS * -readability-inconsistent-declaration-parameter-name)
Chao Liu's avatar
Chao Liu committed
248
if(CMAKE_CXX_COMPILER MATCHES ".*hcc" OR CMAKE_CXX_COMPILER MATCHES ".*clang\\+\\+")
JD's avatar
JD committed
249
    set(CK_TIDY_CHECKS -modernize-use-override -readability-non-const-parameter)
Chao Liu's avatar
Chao Liu committed
250
# Enable tidy on hip
JD's avatar
JD committed
251
252
elseif(CK_BACKEND STREQUAL "HIP" OR CK_BACKEND STREQUAL "HIPNOGPU")
    set(CK_TIDY_ERRORS ALL)
Chao Liu's avatar
Chao Liu committed
253
254
endif()

JD's avatar
JD committed
255

Chao Liu's avatar
Chao Liu committed
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
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
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
347
348
        ${CK_TIDY_CHECKS}
        ${CK_TIDY_ERRORS}
Chao Liu's avatar
Chao Liu committed
349
350
351
    HEADER_FILTER
        "\.hpp$"
    EXTRA_ARGS
JD's avatar
JD committed
352
        -DCK_USE_CLANG_TIDY
Chao Liu's avatar
Chao Liu committed
353
354
)

Alan Turner's avatar
Alan Turner committed
355

Chao Liu's avatar
Chao Liu committed
356
357
358
359
360
361
362
363
364
365
366
367
368
include(CppCheck)
enable_cppcheck(
    CHECKS
        warning
        style
        performance
        portability
    SUPPRESS
        ConfigurationNotChecked
        constStatement
        duplicateCondition
        noExplicitConstructor
        passedByValue
Chao Liu's avatar
Chao Liu committed
369
        preprocessorErrorDirective
Chao Liu's avatar
Chao Liu committed
370
371
372
373
374
375
376
        shadowVariable
        unusedFunction
        unusedPrivateFunction
        unusedStructMember
        unmatchedSuppression
    FORCE
    SOURCES
Chao Liu's avatar
Chao Liu committed
377
        library/src
Chao Liu's avatar
Chao Liu committed
378
379
380
    INCLUDE
        ${CMAKE_CURRENT_SOURCE_DIR}/include
        ${CMAKE_CURRENT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
381
        ${CMAKE_CURRENT_SOURCE_DIR}/library/include
Chao Liu's avatar
Chao Liu committed
382
383
384
385
    DEFINE
        CPPCHECK=1
        __linux__=1
)
Chao Liu's avatar
Chao Liu committed
386

JD's avatar
JD committed
387
388
389
390
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)

391
# set CK project include directories
Chao Liu's avatar
Chao Liu committed
392
include_directories(BEFORE
393
    ${PROJECT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
394
395
    ${PROJECT_SOURCE_DIR}/include
    ${PROJECT_SOURCE_DIR}/library/include
396
    ${HIP_INCLUDE_DIRS}
JD's avatar
JD committed
397
)
Chao Liu's avatar
Chao Liu committed
398

Alan Turner's avatar
Alan Turner committed
399

Anthony Chang's avatar
Anthony Chang committed
400
add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND} --output-on-failure -C ${CMAKE_CFG_INTDIR})
Alan Turner's avatar
Alan Turner committed
401
if (NOT CK_BUILD_JIT_LIB)
Alan Turner's avatar
Alan Turner committed
402
403
    SET(BUILD_DEV ON CACHE BOOL "BUILD_DEV")
    if(BUILD_DEV)
404
        add_compile_options(-Werror -Weverything)
Alan Turner's avatar
Alan Turner committed
405
    endif()
406
407
    #add flags to reduce the size of binaries
    add_compile_options(-Oz -flto=thin)
Alan Turner's avatar
Alan Turner committed
408
409
    message("CMAKE_CXX_FLAGS: ${CMAKE_CXX_FLAGS}")

410

Alan Turner's avatar
Alan Turner committed
411
412
413
414
    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 ${dir_list})
415
416
417
418
419
    set(target_dir)
    IF(IS_DIRECTORY "${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/${subdir_path}/CMakeLists.txt" cmake_instance)
        set(add_inst 0)
420
        if(("${cmake_instance}" MATCHES "fp8" OR "${cmake_instance}" MATCHES "_f8") AND DTYPES MATCHES "fp8")
421
422
            set(add_inst 1)
        endif()
423
        if(("${cmake_instance}" MATCHES "bf8" OR "${cmake_instance}" MATCHES "_b8") AND DTYPES MATCHES "bf8")
424
425
            set(add_inst 1)
        endif()
426
        if(("${cmake_instance}" MATCHES "fp16" OR "${cmake_instance}" MATCHES "_f16") AND DTYPES MATCHES "fp16")
427
428
            set(add_inst 1)
        endif()
429
        if(("${cmake_instance}" MATCHES "fp32" OR "${cmake_instance}" MATCHES "_f32") AND DTYPES MATCHES "fp32")
430
431
            set(add_inst 1)
        endif()
432
        if(("${cmake_instance}" MATCHES "fp64" OR "${cmake_instance}" MATCHES "_f64") AND DTYPES MATCHES "fp64")
433
434
            set(add_inst 1)
        endif()
435
436
437
438
        if(("${cmake_instance}" MATCHES "bf16" OR "${cmake_instance}" MATCHES "_b16") AND DTYPES MATCHES "bf16")
            set(add_inst 1)
        endif()
        if(("${cmake_instance}" MATCHES "int8" OR "${cmake_instance}" MATCHES "_i8") AND DTYPES MATCHES "int8")
439
440
441
442
443
444
445
446
447
            set(add_inst 1)
        endif()
        if(NOT "${cmake_instance}" MATCHES "DTYPES")
            set(add_inst 1)
        endif()
        if(add_inst EQUAL 1 OR NOT DEFINED DTYPES)
            list(APPEND CK_DEVICE_INSTANCES device_${subdir_path}_instance)
        endif()
    ENDIF()
Alan Turner's avatar
Alan Turner committed
448
    ENDFOREACH()
449

450
451
        add_custom_target(instances DEPENDS utility;${CK_DEVICE_INSTANCES}  SOURCES ${INSTANCE_FILES})
        add_subdirectory(library)
Anthony Chang's avatar
Anthony Chang committed
452

453
454
455
456
457
458
        if(NOT DEFINED INSTANCES_ONLY)
        if(NOT DEFINED PROFILER_ONLY)
        rocm_package_setup_component(tests
                LIBRARY_NAME composablekernel
                PACKAGE_NAME tests # Prevent -static suffix on package name
        )
459

460
461
462
463
464
465
        rocm_package_setup_component(examples
                LIBRARY_NAME composablekernel
                PACKAGE_NAME examples
        )
        add_subdirectory(example)
        add_subdirectory(test)
466

Alan Turner's avatar
Alan Turner committed
467
468
    rocm_package_setup_component(profiler
            LIBRARY_NAME composablekernel
469
            PACKAGE_NAME ckprofiler
Alan Turner's avatar
Alan Turner committed
470
471
    )
    add_subdirectory(profiler)
472
473
474
475
    else()
        #When building PROFILER_ONLY, label the package with GPU_ARCH
        rocm_package_setup_component(profiler
        LIBRARY_NAME composablekernel
476
        PACKAGE_NAME ckprofiler_${GPU_ARCH}
477
478
479
480
        )
        add_subdirectory(profiler)
    endif()
    endif()
481
else()
Alan Turner's avatar
Alan Turner committed
482
    rocm_package_setup_component(jit_library
483
484
    LIBRARY_NAME composablekernel
    PACKAGE_NAME jit_library
Alan Turner's avatar
Alan Turner committed
485
    )
486
487
    add_subdirectory(library)
    add_subdirectory(test)
Alan Turner's avatar
Alan Turner committed
488
endif()
JD's avatar
JD committed
489
490
491
492
493
494
495
496
497
498

#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
499
configure_package_config_file(${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in
500
501
502
    "${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
503
504
)

505
rocm_install(FILES
JD's avatar
JD committed
506
507
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
Anthony Chang's avatar
Anthony Chang committed
508
    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
JD's avatar
JD committed
509
)
510

511
# Install CK version and configuration files
512
rocm_install(FILES
513
514
515
516
517
    ${PROJECT_BINARY_DIR}/include/ck/version.h
    ${PROJECT_BINARY_DIR}/include/ck/config.h
    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/ck/
)

518
519
520
521
522
523
524
525
526
527
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
)