CMakeLists.txt 20.1 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
get_property(_GENERATOR_IS_MULTI_CONFIG GLOBAL PROPERTY GENERATOR_IS_MULTI_CONFIG)

9
10
# 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
11
12
13
14
15
16
if(_GENERATOR_IS_MULTI_CONFIG)
    set(CMAKE_CONFIGURATION_TYPES "Debug;Release;RelWithDebInfo;MinSizeRel" CACHE STRING
            "Available build types (configurations) on multi-config generators")
else()
    set(CMAKE_BUILD_TYPE Release CACHE STRING
            "Choose the type of build, options are: None Debug Release RelWithDebInfo MinSizeRel.")
17
18
19
endif()

# Default installation path
20
if(NOT WIN32)
21
22
23
    set(CMAKE_INSTALL_PREFIX "/opt/rocm" CACHE PATH "")
endif()

24
set(version 1.1.0)
JD's avatar
JD committed
25
# Check support for CUDA/HIP in Cmake
26
project(composable_kernel VERSION ${version} LANGUAGES CXX HIP)
27
include(CTest)
Chao Liu's avatar
Chao Liu committed
28

29
find_package(Python3 3.6 COMPONENTS Interpreter REQUIRED)
carlushuang's avatar
carlushuang committed
30

31
32
list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")

33
if (DTYPES)
34
35
36
37
38
39
40
41
    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")
42
43
44
45
    endif()
    if (DTYPES MATCHES "bf8")
        add_definitions(-DCK_ENABLE_BF8)
        set(CK_ENABLE_BF8 "ON")
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
    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}")
64
else()
65
    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP8 -DCK_ENABLE_BF8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16)
66
    set(CK_ENABLE_ALL_DTYPES "ON")
67
68
endif()

69
70
#for f8/bf8_t type
add_compile_options(-Wno-bit-int-extension)
71
add_compile_options(-Wno-pass-failed)
72
add_compile_options(-Wno-switch-default)
73

74
75
if(DL_KERNELS)
    add_definitions(-DDL_KERNELS)
76
    set(CK_ENABLE_DL_KERNELS "ON")
77
78
79
80
endif()

if(INSTANCES_ONLY)
    add_definitions(-DINSTANCES_ONLY)
81
    set(CK_ENABLE_INSTANCES_ONLY "ON")
82
83
endif()

84
85
include(getopt)

86
87
88
# 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)
89
configure_file(include/ck/version.h.in ${CMAKE_CURRENT_BINARY_DIR}/include/ck/version.h)
JD's avatar
JD committed
90

91
set(ROCM_SYMLINK_LIBS OFF)
Anthony Chang's avatar
Anthony Chang committed
92
find_package(ROCM REQUIRED PATHS /opt/rocm)
JD's avatar
JD committed
93
94
95
96
97
98

include(ROCMInstallTargets)
include(ROCMPackageConfigHelpers)
include(ROCMSetupVersion)
include(ROCMInstallSymlinks)
include(ROCMCreatePackage)
Chao Liu's avatar
Chao Liu committed
99
include(CheckCXXCompilerFlag)
100
include(ROCMCheckTargetIds)
JD's avatar
JD committed
101
include(TargetFlags)
102
103
104

rocm_setup_version(VERSION ${version})

105
list(APPEND CMAKE_PREFIX_PATH ${CMAKE_INSTALL_PREFIX} ${CMAKE_INSTALL_PREFIX}/llvm ${CMAKE_INSTALL_PREFIX}/hip /opt/rocm /opt/rocm/llvm /opt/rocm/hip "$ENV{ROCM_PATH}" "$ENV{HIP_PATH}")
JD's avatar
JD committed
106

107
108
109
110
111
112
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
113
if(NOT PROFILER_ONLY)
114
115
  if(NOT ENABLE_ASAN_PACKAGING)
    #build CK for all supported targets
116
    rocm_check_target_ids(DEFAULT_GPU_TARGETS
117
        TARGETS "gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx1100;gfx1101;gfx1102;gfx1200;gfx1201")
118
119
120
121
122
123
  else()
    #build CK only for xnack-supported targets
    rocm_check_target_ids(DEFAULT_GPU_TARGETS
        TARGETS "gfx908:xnack+;gfx90a:xnack+;gfx940:xnack+;gfx941:xnack+;gfx942:xnack+")
    set(GPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
  endif()
124
125
else()
    add_definitions(-DPROFILER_ONLY)
126
    set(GPU_TARGETS "" CACHE STRING "" FORCE)
127
    if(GPU_TARGETS)
128
        message(FATAL_ERROR "For PROFILE_ONLY build, please do not set GPU_TARGETS, use GPU_ARCH = gfx90, gfx94, gfx10, gfx11 or gfx12")
129
    endif()
130
    if(GPU_ARCH MATCHES "gfx90")
131
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx908;gfx90a")
132
133
    elseif(GPU_ARCH MATCHES "gfx94")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx940;gfx941;gfx942")
134
135
136
137
    elseif(GPU_ARCH MATCHES "gfx10")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1030")
    elseif(GPU_ARCH MATCHES "gfx11")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1100;gfx1101;gfx1102")
138
139
    elseif(GPU_ARCH MATCHES "gfx12")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1200;gfx1201")
140
    else()
141
        message(FATAL_ERROR "For PROFILE_ONLY build, please specify GPU_ARCH as gfx90, gfx94, gfx10, gfx11 or gfx12")
142
    endif()
143
    set(GPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
144
145
endif()

146
message("Supported GPU_TARGETS= ${DEFAULT_GPU_TARGETS}")
147
148
149
150

if(GPU_TARGETS)
    message("Building CK for the following targets: ${GPU_TARGETS}")
else()
151
    message("Building CK for the default targets: ${DEFAULT_GPU_TARGETS}")
152
endif()
153
154
155
156
157
158

if (GPU_TARGETS)
    if (GPU_TARGETS MATCHES "gfx9")
        add_definitions(-DCK_USE_XDL)
        set(CK_USE_XDL "ON")
    endif()
159
    if (GPU_TARGETS MATCHES "gfx11" OR GPU_TARGETS MATCHES "gfx12")
160
161
162
163
164
165
166
167
168
        add_definitions(-DCK_USE_WMMA)
        set(CK_USE_WMMA "ON")
    endif()
else()
    add_definitions(-DCK_USE_WMMA -DCK_USE_XDL)
    set(CK_USE_XDL "ON")
    set(CK_USE_WMMA "ON")
endif()

169
170
171
# CK config file to record supported datatypes, etc.
configure_file(include/ck/config.h.in ${CMAKE_CURRENT_BINARY_DIR}/include/ck/config.h)

172
find_package(hip)
173
174
175
176
# 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}")
177
if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 500723302)
178
179
180
   message("Adding the fno-offload-uniform-block compiler flag")
   add_compile_options(-fno-offload-uniform-block)
endif()
181
182
183
184
if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 600140090)
   message("Adding the enable-post-misched=0 compiler flag")
   add_compile_options(-mllvm -enable-post-misched=0)
endif()
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
#
# Seperate linking jobs from compiling
# Too many concurrent linking jobs can break the build
# Copied from LLVM
set(CK_PARALLEL_LINK_JOBS "" CACHE STRING
  "Define the maximum number of concurrent link jobs (Ninja only).")
if(CMAKE_GENERATOR MATCHES "Ninja")
  if(CK_PARALLEL_LINK_JOBS)
    set_property(GLOBAL APPEND PROPERTY JOB_POOLS link_job_pool=${CK_PARALLEL_LINK_JOBS})
    set(CMAKE_JOB_POOL_LINK link_job_pool)
  endif()
elseif(CK_PARALLEL_LINK_JOBS)
  message(WARNING "Job pooling is only available with Ninja generators.")
endif()
# Similar for compiling
set(CK_PARALLEL_COMPILE_JOBS "" CACHE STRING
  "Define the maximum number of concurrent compile jobs (Ninja only).")
if(CMAKE_GENERATOR MATCHES "Ninja")
  if(CK_PARALLEL_COMPILE_JOBS)
    set_property(GLOBAL APPEND PROPERTY JOB_POOLS compile_job_pool=${CK_PARALLEL_COMPILE_JOBS})
    set(CMAKE_JOB_POOL_COMPILE compile_job_pool)
  endif()
elseif(CK_PARALLEL_COMPILE_JOBS)
  message(WARNING "Job pooling is only available with Ninja generators.")
endif()


212
option(USE_BITINT_EXTENSION_INT4 "Whether to enable clang's BitInt extension to provide int4 data type." OFF)
Illia Silin's avatar
Illia Silin committed
213
option(USE_OPT_GFX11 "Whether to enable LDS cumode and Wavefront32 mode for GFX11 silicons." OFF)
Adam Osewski's avatar
Adam Osewski committed
214
215
216
217
218
219
220

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()

Illia Silin's avatar
Illia Silin committed
221
if(USE_OPT_GFX11)
222
223
    add_compile_options(-mcumode)
    add_compile_options(-mno-wavefrontsize64)
Illia Silin's avatar
Illia Silin committed
224
    message("CK compiled with USE_OPT_GFX11 set to ${USE_OPT_GFX11}")
225
226
endif()

227
228
229
230
231
## Threads
set(THREADS_PREFER_PTHREAD_FLAG ON)
find_package(Threads REQUIRED)
link_libraries(Threads::Threads)

232
## C++
Chao Liu's avatar
Chao Liu committed
233
set(CMAKE_CXX_STANDARD 17)
Chao Liu's avatar
Chao Liu committed
234
set(CMAKE_CXX_STANDARD_REQUIRED ON)
Chao Liu's avatar
Chao Liu committed
235
set(CMAKE_CXX_EXTENSIONS OFF)
236
237
238
239
240
241
242
message("CMAKE_CXX_COMPILER: ${CMAKE_CXX_COMPILER}")

## HIP
set(CMAKE_HIP_PLATFORM amd)
set(CMAKE_HIP_COMPILER ${CMAKE_CXX_COMPILER})
set(CMAKE_HIP_EXTENSIONS ON)
message("CMAKE_HIP_COMPILER: ${CMAKE_HIP_COMPILER}")
Chao Liu's avatar
Chao Liu committed
243

244
## OpenMP
Chao Liu's avatar
Chao Liu committed
245
246
247
248
249
250
251
252
253
254
255
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
256

Chao Liu's avatar
Chao Liu committed
257
258
259
260
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
261

Chao Liu's avatar
Chao Liu committed
262
263
link_libraries(${OpenMP_gomp_LIBRARY})
link_libraries(${OpenMP_pthread_LIBRARY})
Chao Liu's avatar
Chao Liu committed
264

265
266
## HIP
find_package(HIP REQUIRED)
JD's avatar
JD committed
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
# 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}")
286
link_libraries(hip::device)
287
288
289
290
291
if(CK_hip_VERSION VERSION_GREATER_EQUAL 6.0.23494)
    add_compile_definitions(__HIP_PLATFORM_AMD__=1)
else()
    add_compile_definitions(__HIP_PLATFORM_HCC__=1)
endif()
292

Chao Liu's avatar
Chao Liu committed
293
294
## tidy
include(EnableCompilerWarnings)
JD's avatar
JD committed
295
set(CK_TIDY_ERRORS ERRORS * -readability-inconsistent-declaration-parameter-name)
Chao Liu's avatar
Chao Liu committed
296
if(CMAKE_CXX_COMPILER MATCHES ".*hcc" OR CMAKE_CXX_COMPILER MATCHES ".*clang\\+\\+")
JD's avatar
JD committed
297
    set(CK_TIDY_CHECKS -modernize-use-override -readability-non-const-parameter)
Chao Liu's avatar
Chao Liu committed
298
# Enable tidy on hip
JD's avatar
JD committed
299
300
elseif(CK_BACKEND STREQUAL "HIP" OR CK_BACKEND STREQUAL "HIPNOGPU")
    set(CK_TIDY_ERRORS ALL)
Chao Liu's avatar
Chao Liu committed
301
302
endif()

JD's avatar
JD committed
303

Chao Liu's avatar
Chao Liu committed
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
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
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
395
396
        ${CK_TIDY_CHECKS}
        ${CK_TIDY_ERRORS}
Chao Liu's avatar
Chao Liu committed
397
398
399
    HEADER_FILTER
        "\.hpp$"
    EXTRA_ARGS
JD's avatar
JD committed
400
        -DCK_USE_CLANG_TIDY
Chao Liu's avatar
Chao Liu committed
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
)

include(CppCheck)
enable_cppcheck(
    CHECKS
        warning
        style
        performance
        portability
    SUPPRESS
        ConfigurationNotChecked
        constStatement
        duplicateCondition
        noExplicitConstructor
        passedByValue
Chao Liu's avatar
Chao Liu committed
416
        preprocessorErrorDirective
Chao Liu's avatar
Chao Liu committed
417
418
419
420
421
422
423
        shadowVariable
        unusedFunction
        unusedPrivateFunction
        unusedStructMember
        unmatchedSuppression
    FORCE
    SOURCES
Chao Liu's avatar
Chao Liu committed
424
        library/src
Chao Liu's avatar
Chao Liu committed
425
426
427
    INCLUDE
        ${CMAKE_CURRENT_SOURCE_DIR}/include
        ${CMAKE_CURRENT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
428
        ${CMAKE_CURRENT_SOURCE_DIR}/library/include
Chao Liu's avatar
Chao Liu committed
429
430
431
432
    DEFINE
        CPPCHECK=1
        __linux__=1
)
Chao Liu's avatar
Chao Liu committed
433

JD's avatar
JD committed
434
435
436
437
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)

438
# set CK project include directories
Chao Liu's avatar
Chao Liu committed
439
include_directories(BEFORE
440
    ${PROJECT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
441
442
    ${PROJECT_SOURCE_DIR}/include
    ${PROJECT_SOURCE_DIR}/library/include
443
    ${HIP_INCLUDE_DIRS}
JD's avatar
JD committed
444
)
Chao Liu's avatar
Chao Liu committed
445

JD's avatar
JD committed
446
447
SET(BUILD_DEV ON CACHE BOOL "BUILD_DEV")
if(BUILD_DEV)
448
449
    add_compile_options(-Werror)
    add_compile_options(-Weverything)
JD's avatar
JD committed
450
451
452
endif()
message("CMAKE_CXX_FLAGS: ${CMAKE_CXX_FLAGS}")

453
454
455
456
457
458
459
if("${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang")
    add_compile_options(-fcolor-diagnostics)
endif()
if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU" AND CMAKE_CXX_COMPILER_VERSION VERSION_GREATER 4.9)
    add_compile_options(-fdiagnostics-color=always)
endif()

Anthony Chang's avatar
Anthony Chang committed
460
461
add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND} --output-on-failure -C ${CMAKE_CFG_INTDIR})

462
463
464
465
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 ${dir_list})
466
467
468
469
470
set(target_dir)
IF(IS_DIRECTORY "${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/${subdir_path}/CMakeLists.txt" cmake_instance)
    set(add_inst 0)
471
    if(("${cmake_instance}" MATCHES "fp8" OR "${cmake_instance}" MATCHES "_f8") AND DTYPES MATCHES "fp8")
472
        set(add_inst 1)
473
    endif()
474
    if(("${cmake_instance}" MATCHES "bf8" OR "${cmake_instance}" MATCHES "_b8") AND DTYPES MATCHES "bf8")
475
476
        set(add_inst 1)
    endif()
477
    if(("${cmake_instance}" MATCHES "fp16" OR "${cmake_instance}" MATCHES "_f16") AND DTYPES MATCHES "fp16")
478
        set(add_inst 1)
479
    endif()
480
    if(("${cmake_instance}" MATCHES "fp32" OR "${cmake_instance}" MATCHES "_f32") AND DTYPES MATCHES "fp32")
481
        set(add_inst 1)
482
    endif()
483
    if(("${cmake_instance}" MATCHES "fp64" OR "${cmake_instance}" MATCHES "_f64") AND DTYPES MATCHES "fp64")
484
        set(add_inst 1)
485
    endif()
486
    if(("${cmake_instance}" MATCHES "bf16" OR "${cmake_instance}" MATCHES "_b16") AND DTYPES MATCHES "bf16")
487
        set(add_inst 1)
488
    endif()
489
    if(("${cmake_instance}" MATCHES "int8" OR "${cmake_instance}" MATCHES "_i8") AND DTYPES MATCHES "int8")
490
        set(add_inst 1)
491
492
    endif()
    if(NOT "${cmake_instance}" MATCHES "DTYPES")
493
        set(add_inst 1)
494
495
    endif()
    if(add_inst EQUAL 1 OR NOT DEFINED DTYPES)
496
        list(APPEND CK_DEVICE_INSTANCES device_${subdir_path}_instance)
497
498
    endif()
ENDIF()
499
ENDFOREACH()
500

501
add_custom_target(instances DEPENDS utility;${CK_DEVICE_INSTANCES}  SOURCES ${INSTANCE_FILES})
502
add_subdirectory(library)
503

504
if(NOT DEFINED INSTANCES_ONLY)
505
 if(NOT DEFINED PROFILER_ONLY)
506
   rocm_package_setup_component(tests
507
508
        LIBRARY_NAME composablekernel
        PACKAGE_NAME tests # Prevent -static suffix on package name
509
   )
510

511
   rocm_package_setup_component(examples
512
513
        LIBRARY_NAME composablekernel
        PACKAGE_NAME examples
514
   )
515
   add_subdirectory(example)
516
517
518
   if(BUILD_TESTING)
      add_subdirectory(test)
   endif()
519

520
   rocm_package_setup_component(profiler
521
        LIBRARY_NAME composablekernel
522
        PACKAGE_NAME ckprofiler
523
524
   )
   add_subdirectory(profiler)
525
526
527
528
  else()
    #When building PROFILER_ONLY, label the package with GPU_ARCH
    rocm_package_setup_component(profiler
       LIBRARY_NAME composablekernel
529
       PACKAGE_NAME ckprofiler_${GPU_ARCH}
530
531
532
    )
    add_subdirectory(profiler)
  endif()
533
endif()
JD's avatar
JD committed
534
535
536
537
538
539
540
541
542
543

#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
544
configure_package_config_file(${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in
545
546
547
    "${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
548
549
)

550
rocm_install(FILES
JD's avatar
JD committed
551
552
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
Anthony Chang's avatar
Anthony Chang committed
553
    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
JD's avatar
JD committed
554
)
555

556
# Install CK version and configuration files
557
rocm_install(FILES
558
559
560
561
562
    ${PROJECT_BINARY_DIR}/include/ck/version.h
    ${PROJECT_BINARY_DIR}/include/ck/config.h
    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/ck/
)

563
564
565
566
567
568
569
570
571
572
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
)