CMakeLists.txt 21.9 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
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()
aska-0096's avatar
aska-0096 committed
65
66
67
68
69
70
71
72
73
74
75
    add_definitions(-DCK_ENABLE_INT8 -DCK_ENABLE_FP16 -DCK_ENABLE_FP32 -DCK_ENABLE_FP64 -DCK_ENABLE_BF16)
    set(CK_ENABLE_INT8 "ON")
    set(CK_ENABLE_FP16 "ON")
    set(CK_ENABLE_FP32 "ON")
    set(CK_ENABLE_FP64 "ON")
    set(CK_ENABLE_BF16 "ON")
    if (GPU_TARGETS MATCHES "gfx94")
        add_definitions(-DCK_ENABLE_FP8 -DCK_ENABLE_BF8)
        set(CK_ENABLE_FP8 "ON")
        set(CK_ENABLE_BF8 "ON")
    endif()
76
77
endif()

78
79
#for f8/bf8_t type
add_compile_options(-Wno-bit-int-extension)
80
add_compile_options(-Wno-pass-failed)
81
add_compile_options(-Wno-switch-default)
82

83
84
if(DL_KERNELS)
    add_definitions(-DDL_KERNELS)
85
    set(CK_ENABLE_DL_KERNELS "ON")
86
87
88
89
endif()

if(INSTANCES_ONLY)
    add_definitions(-DINSTANCES_ONLY)
90
    set(CK_ENABLE_INSTANCES_ONLY "ON")
91
92
endif()

93
94
include(getopt)

95
96
97
# 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)
98
configure_file(include/ck/version.h.in ${CMAKE_CURRENT_BINARY_DIR}/include/ck/version.h)
JD's avatar
JD committed
99

100
set(ROCM_SYMLINK_LIBS OFF)
Anthony Chang's avatar
Anthony Chang committed
101
find_package(ROCM REQUIRED PATHS /opt/rocm)
JD's avatar
JD committed
102
103
104
105
106
107

include(ROCMInstallTargets)
include(ROCMPackageConfigHelpers)
include(ROCMSetupVersion)
include(ROCMInstallSymlinks)
include(ROCMCreatePackage)
Chao Liu's avatar
Chao Liu committed
108
include(CheckCXXCompilerFlag)
109
include(ROCMCheckTargetIds)
JD's avatar
JD committed
110
include(TargetFlags)
111
112
113

rocm_setup_version(VERSION ${version})

114
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
115

116
117
message("GPU_TARGETS= ${GPU_TARGETS}")

118
119
120
121
122
123
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}")

124
125
126
127
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
128
if(NOT PROFILER_ONLY)
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
    if(NOT ENABLE_ASAN_PACKAGING)
        #build CK for all supported targets
        if(NOT WIN32 AND ${hip_VERSION_FLAT} LESS 600300000)
            # WORKAROUND: compiler does not yet fully support gfx12 targets, need to fix version above
            rocm_check_target_ids(DEFAULT_GPU_TARGETS
                TARGETS "gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx1100;gfx1101;gfx1102")
        else()
            rocm_check_target_ids(DEFAULT_GPU_TARGETS
                TARGETS "gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx1100;gfx1101;gfx1102;gfx1200;gfx1201")
        endif()
    else()
        #build CK only for xnack-supported targets
        rocm_check_target_ids(DEFAULT_GPU_TARGETS
            TARGETS "gfx908:xnack+;gfx90a:xnack+;gfx940:xnack+;gfx941:xnack+;gfx942:xnack+")
        set(GPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
    endif()
145
146
else()
    add_definitions(-DPROFILER_ONLY)
147
    set(GPU_TARGETS "" CACHE STRING "" FORCE)
148
    if(GPU_TARGETS)
149
        message(FATAL_ERROR "For PROFILE_ONLY build, please do not set GPU_TARGETS, use GPU_ARCH = gfx90, gfx94, gfx10, gfx11 or gfx12")
150
    endif()
151
    if(GPU_ARCH MATCHES "gfx90")
152
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx908;gfx90a")
153
154
    elseif(GPU_ARCH MATCHES "gfx94")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx940;gfx941;gfx942")
155
156
157
158
    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")
159
160
    elseif(GPU_ARCH MATCHES "gfx12")
        rocm_check_target_ids(DEFAULT_GPU_TARGETS TARGETS "gfx1200;gfx1201")
161
    else()
162
        message(FATAL_ERROR "For PROFILE_ONLY build, please specify GPU_ARCH as gfx90, gfx94, gfx10, gfx11 or gfx12")
163
    endif()
164
    set(GPU_TARGETS "${DEFAULT_GPU_TARGETS}" CACHE STRING " " FORCE)
165
166
endif()

167
message("Supported GPU_TARGETS= ${DEFAULT_GPU_TARGETS}")
168
169
170
171

if(GPU_TARGETS)
    message("Building CK for the following targets: ${GPU_TARGETS}")
else()
172
    message("Building CK for the default targets: ${DEFAULT_GPU_TARGETS}")
173
endif()
174
175
176
177
178
179

if (GPU_TARGETS)
    if (GPU_TARGETS MATCHES "gfx9")
        add_definitions(-DCK_USE_XDL)
        set(CK_USE_XDL "ON")
    endif()
180
    if (GPU_TARGETS MATCHES "gfx11" OR GPU_TARGETS MATCHES "gfx12")
181
182
183
184
185
186
187
188
189
        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()

190
191
192
# CK config file to record supported datatypes, etc.
configure_file(include/ck/config.h.in ${CMAKE_CURRENT_BINARY_DIR}/include/ck/config.h)

193
if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 500723302)
194
195
196
   message("Adding the fno-offload-uniform-block compiler flag")
   add_compile_options(-fno-offload-uniform-block)
endif()
197
198
if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 600140090)
   message("Adding the enable-post-misched=0 compiler flag")
199
200
   add_compile_options("SHELL: -mllvm -enable-post-misched=0")
endif()
201
202
203
set(check-coerce)
check_cxx_compiler_flag(" -mllvm -amdgpu-coerce-illegal-types=1" check-coerce)
if(NOT WIN32 AND check-coerce AND ${hip_VERSION_FLAT} GREATER 600241132 AND ${hip_VERSION_FLAT} LESS 600300000)
204
205
206
207
   message("Adding the amdgpu-coerce-illegal-types=1")
   add_compile_options("SHELL: -mllvm -amdgpu-coerce-illegal-types=1")
endif()
if(NOT WIN32 AND ${hip_VERSION_FLAT} GREATER 600241132)
Haocong WANG's avatar
Haocong WANG committed
208
209
   message("Adding -amdgpu-early-inline-all=true, -amdgpu-function-calls=false and -greedy-reverse-local-assignment=1")
   add_compile_options("SHELL: -mllvm -greedy-reverse-local-assignment=1")
210
211
   add_compile_options("SHELL: -mllvm -amdgpu-early-inline-all=true")
   add_compile_options("SHELL: -mllvm -amdgpu-function-calls=false")
212
endif()
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
#
# Seperate linking jobs from compiling
# Too many concurrent linking jobs can break the build
# Copied from LLVM
set(CK_PARALLEL_LINK_JOBS "" CACHE STRING
  "Define the maximum number of concurrent link jobs (Ninja only).")
if(CMAKE_GENERATOR MATCHES "Ninja")
  if(CK_PARALLEL_LINK_JOBS)
    set_property(GLOBAL APPEND PROPERTY JOB_POOLS link_job_pool=${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()


240
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
241
option(USE_OPT_GFX11 "Whether to enable LDS cumode and Wavefront32 mode for GFX11 silicons." OFF)
Adam Osewski's avatar
Adam Osewski committed
242
243
244
245
246
247
248

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
249
if(USE_OPT_GFX11)
250
251
    add_compile_options(-mcumode)
    add_compile_options(-mno-wavefrontsize64)
Illia Silin's avatar
Illia Silin committed
252
    message("CK compiled with USE_OPT_GFX11 set to ${USE_OPT_GFX11}")
253
254
endif()

255
256
257
258
259
## Threads
set(THREADS_PREFER_PTHREAD_FLAG ON)
find_package(Threads REQUIRED)
link_libraries(Threads::Threads)

260
## C++
Chao Liu's avatar
Chao Liu committed
261
set(CMAKE_CXX_STANDARD 17)
Chao Liu's avatar
Chao Liu committed
262
set(CMAKE_CXX_STANDARD_REQUIRED ON)
Chao Liu's avatar
Chao Liu committed
263
set(CMAKE_CXX_EXTENSIONS OFF)
264
265
message("CMAKE_CXX_COMPILER: ${CMAKE_CXX_COMPILER}")

266
267
268
269
270
271
272
273
274
275
# https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_macros.html
# _GLIBCXX_ASSERTIONS
# Undefined by default. When defined, enables extra error checking in the form of
# precondition assertions, such as bounds checking in strings and null pointer
# checks when dereferencing smart pointers
option(USE_GLIBCXX_ASSERTIONS "Turn on additional c++ library checks." OFF)
if(USE_GLIBCXX_ASSERTIONS)
  add_compile_options(-Wp,-D_GLIBCXX_ASSERTIONS)
endif()

276
277
278
279
280
## 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
281

282
## OpenMP
Chao Liu's avatar
Chao Liu committed
283
284
285
286
287
288
289
290
291
292
293
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
294

Chao Liu's avatar
Chao Liu committed
295
296
297
298
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
299

Chao Liu's avatar
Chao Liu committed
300
301
link_libraries(${OpenMP_gomp_LIBRARY})
link_libraries(${OpenMP_pthread_LIBRARY})
Chao Liu's avatar
Chao Liu committed
302

303
304
## HIP
find_package(HIP REQUIRED)
JD's avatar
JD committed
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
# 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}")
324
link_libraries(hip::device)
325
326
327
328
329
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()
330

Chao Liu's avatar
Chao Liu committed
331
332
## tidy
include(EnableCompilerWarnings)
JD's avatar
JD committed
333
set(CK_TIDY_ERRORS ERRORS * -readability-inconsistent-declaration-parameter-name)
Chao Liu's avatar
Chao Liu committed
334
if(CMAKE_CXX_COMPILER MATCHES ".*hcc" OR CMAKE_CXX_COMPILER MATCHES ".*clang\\+\\+")
JD's avatar
JD committed
335
    set(CK_TIDY_CHECKS -modernize-use-override -readability-non-const-parameter)
Chao Liu's avatar
Chao Liu committed
336
# Enable tidy on hip
JD's avatar
JD committed
337
338
elseif(CK_BACKEND STREQUAL "HIP" OR CK_BACKEND STREQUAL "HIPNOGPU")
    set(CK_TIDY_ERRORS ALL)
Chao Liu's avatar
Chao Liu committed
339
340
endif()

JD's avatar
JD committed
341

Chao Liu's avatar
Chao Liu committed
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
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
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
433
434
        ${CK_TIDY_CHECKS}
        ${CK_TIDY_ERRORS}
Chao Liu's avatar
Chao Liu committed
435
436
437
    HEADER_FILTER
        "\.hpp$"
    EXTRA_ARGS
JD's avatar
JD committed
438
        -DCK_USE_CLANG_TIDY
Chao Liu's avatar
Chao Liu committed
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
)

include(CppCheck)
enable_cppcheck(
    CHECKS
        warning
        style
        performance
        portability
    SUPPRESS
        ConfigurationNotChecked
        constStatement
        duplicateCondition
        noExplicitConstructor
        passedByValue
Chao Liu's avatar
Chao Liu committed
454
        preprocessorErrorDirective
Chao Liu's avatar
Chao Liu committed
455
456
457
458
459
460
461
        shadowVariable
        unusedFunction
        unusedPrivateFunction
        unusedStructMember
        unmatchedSuppression
    FORCE
    SOURCES
Chao Liu's avatar
Chao Liu committed
462
        library/src
Chao Liu's avatar
Chao Liu committed
463
464
465
    INCLUDE
        ${CMAKE_CURRENT_SOURCE_DIR}/include
        ${CMAKE_CURRENT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
466
        ${CMAKE_CURRENT_SOURCE_DIR}/library/include
Chao Liu's avatar
Chao Liu committed
467
468
469
470
    DEFINE
        CPPCHECK=1
        __linux__=1
)
Chao Liu's avatar
Chao Liu committed
471

JD's avatar
JD committed
472
473
474
475
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)

476
# set CK project include directories
Chao Liu's avatar
Chao Liu committed
477
include_directories(BEFORE
478
    ${PROJECT_BINARY_DIR}/include
Chao Liu's avatar
Chao Liu committed
479
480
    ${PROJECT_SOURCE_DIR}/include
    ${PROJECT_SOURCE_DIR}/library/include
481
    ${HIP_INCLUDE_DIRS}
JD's avatar
JD committed
482
)
Chao Liu's avatar
Chao Liu committed
483

JD's avatar
JD committed
484
485
SET(BUILD_DEV ON CACHE BOOL "BUILD_DEV")
if(BUILD_DEV)
486
487
    add_compile_options(-Werror)
    add_compile_options(-Weverything)
JD's avatar
JD committed
488
489
490
endif()
message("CMAKE_CXX_FLAGS: ${CMAKE_CXX_FLAGS}")

491
492
493
494
495
496
497
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
498
499
add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND} --output-on-failure -C ${CMAKE_CFG_INTDIR})

500
501
502
503
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 ${dir_list})
504
505
506
507
508
set(target_dir)
IF(IS_DIRECTORY "${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/${subdir_path}/CMakeLists.txt" cmake_instance)
    set(add_inst 0)
509
    if(("${cmake_instance}" MATCHES "fp8" OR "${cmake_instance}" MATCHES "_f8") AND DTYPES MATCHES "fp8")
510
        set(add_inst 1)
511
    endif()
512
    if(("${cmake_instance}" MATCHES "bf8" OR "${cmake_instance}" MATCHES "_b8") AND DTYPES MATCHES "bf8")
513
514
        set(add_inst 1)
    endif()
515
    if(("${cmake_instance}" MATCHES "fp16" OR "${cmake_instance}" MATCHES "_f16") AND DTYPES MATCHES "fp16")
516
        set(add_inst 1)
517
    endif()
518
    if(("${cmake_instance}" MATCHES "fp32" OR "${cmake_instance}" MATCHES "_f32") AND DTYPES MATCHES "fp32")
519
        set(add_inst 1)
520
    endif()
521
    if(("${cmake_instance}" MATCHES "fp64" OR "${cmake_instance}" MATCHES "_f64") AND DTYPES MATCHES "fp64")
522
        set(add_inst 1)
523
    endif()
524
    if(("${cmake_instance}" MATCHES "bf16" OR "${cmake_instance}" MATCHES "_b16") AND DTYPES MATCHES "bf16")
525
        set(add_inst 1)
526
    endif()
527
    if(("${cmake_instance}" MATCHES "int8" OR "${cmake_instance}" MATCHES "_i8") AND DTYPES MATCHES "int8")
528
        set(add_inst 1)
529
530
    endif()
    if(NOT "${cmake_instance}" MATCHES "DTYPES")
531
        set(add_inst 1)
532
533
    endif()
    if(add_inst EQUAL 1 OR NOT DEFINED DTYPES)
534
        list(APPEND CK_DEVICE_INSTANCES device_${subdir_path}_instance)
535
536
    endif()
ENDIF()
537
ENDFOREACH()
538

539
add_custom_target(instances DEPENDS utility;${CK_DEVICE_INSTANCES}  SOURCES ${INSTANCE_FILES})
540
add_subdirectory(library)
541

542
if(NOT DEFINED INSTANCES_ONLY)
543
 if(NOT DEFINED PROFILER_ONLY)
544
   rocm_package_setup_component(tests
545
546
        LIBRARY_NAME composablekernel
        PACKAGE_NAME tests # Prevent -static suffix on package name
547
   )
548

549
   rocm_package_setup_component(examples
550
551
        LIBRARY_NAME composablekernel
        PACKAGE_NAME examples
552
   )
553
   add_subdirectory(example)
554
555
556
   if(BUILD_TESTING)
      add_subdirectory(test)
   endif()
557

558
   rocm_package_setup_component(profiler
559
        LIBRARY_NAME composablekernel
560
        PACKAGE_NAME ckprofiler
561
562
   )
   add_subdirectory(profiler)
563
564
565
566
  else()
    #When building PROFILER_ONLY, label the package with GPU_ARCH
    rocm_package_setup_component(profiler
       LIBRARY_NAME composablekernel
567
       PACKAGE_NAME ckprofiler_${GPU_ARCH}
568
569
570
    )
    add_subdirectory(profiler)
  endif()
571
endif()
JD's avatar
JD committed
572
573
574
575
576
577
578
579
580
581

#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
582
configure_package_config_file(${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in
583
584
585
    "${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
586
587
)

588
rocm_install(FILES
JD's avatar
JD committed
589
590
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfig.cmake"
    "${CMAKE_CURRENT_BINARY_DIR}/composable_kernelConfigVersion.cmake"
Anthony Chang's avatar
Anthony Chang committed
591
    DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/composable_kernel
JD's avatar
JD committed
592
)
593

594
# Install CK version and configuration files
595
rocm_install(FILES
596
597
598
599
600
    ${PROJECT_BINARY_DIR}/include/ck/version.h
    ${PROJECT_BINARY_DIR}/include/ck/config.h
    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/ck/
)

601
602
603
604
605
606
607
608
609
610
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
)