forked from NVIDIA/cccl
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmatrix.yaml
328 lines (300 loc) · 14.8 KB
/
matrix.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
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
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
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
317
318
319
320
321
322
323
324
325
326
327
328
ctk_11_1: &ctk_11_1 '11.1'
ctk_11_8: &ctk_11_8 '11.8'
ctk_12_0: &ctk_12_0 '12.0'
ctk_curr: &ctk_curr '12.4'
# The version of the devcontainer images to use from https://hub.docker.com/r/rapidsai/devcontainers
devcontainer_version: '24.06'
# gcc compiler configurations
gcc6: &gcc6 { name: 'gcc', version: '6', exe: 'g++' }
gcc7: &gcc7 { name: 'gcc', version: '7', exe: 'g++' }
gcc8: &gcc8 { name: 'gcc', version: '8', exe: 'g++' }
gcc9: &gcc9 { name: 'gcc', version: '9', exe: 'g++' }
gcc10: &gcc10 { name: 'gcc', version: '10', exe: 'g++' }
gcc11: &gcc11 { name: 'gcc', version: '11', exe: 'g++' }
gcc12: &gcc12 { name: 'gcc', version: '12', exe: 'g++' }
gcc-oldest: &gcc-oldest { name: 'gcc', version: '6', exe: 'g++' }
gcc-newest: &gcc-newest { name: 'gcc', version: '12', exe: 'g++' }
# LLVM Compiler configurations
llvm9: &llvm9 { name: 'llvm', version: '9', exe: 'clang++' }
llvm10: &llvm10 { name: 'llvm', version: '10', exe: 'clang++' }
llvm11: &llvm11 { name: 'llvm', version: '11', exe: 'clang++' }
llvm12: &llvm12 { name: 'llvm', version: '12', exe: 'clang++' }
llvm13: &llvm13 { name: 'llvm', version: '13', exe: 'clang++' }
llvm14: &llvm14 { name: 'llvm', version: '14', exe: 'clang++' }
llvm15: &llvm15 { name: 'llvm', version: '15', exe: 'clang++' }
llvm16: &llvm16 { name: 'llvm', version: '16', exe: 'clang++' }
llvm-oldest: &llvm-oldest { name: 'llvm', version: '9', exe: 'clang++' }
llvm-newest: &llvm-newest { name: 'llvm', version: '16', exe: 'clang++' }
# MSVC configs
msvc2017: &msvc2017 { name: 'cl', version: '14.16', exe: 'cl++' }
msvc2019: &msvc2019 { name: 'cl', version: '14.29', exe: 'cl++' }
msvc2022_1436: &msvc2022_1436 { name: 'cl', version: '14.36', exe: 'cl++' }
msvc2022: &msvc2022 { name: 'cl', version: '14.39', exe: 'cl++' }
# oneAPI configs
oneapi: &oneapi { name: 'oneapi', version: '2023.2.0', exe: 'icpc' }
# GHA Workflow job matrices:
workflows:
# If any jobs appear here, they will be executed instead of `pull_request' for PRs.
# This is useful for limiting resource usage when a full matrix is not needed.
# The branch protection checks will fail when using this override workflow.
#
# Example:
# override:
# - {jobs: ['build'], project['thrust'], std: 17, ctk: *ctk_curr, cxx: [*gcc12, *llvm16]}
#
override:
pull_request:
# default_projects: nvcc
- {jobs: ['build'], std: 'all', ctk: *ctk_11_1, cxx: [*gcc6, *gcc7, *gcc8, *gcc9, *llvm9, *msvc2017]}
- {jobs: ['build'], std: 'all', ctk: *ctk_11_8, cxx: [*gcc11], sm: '60;70;80;90'}
- {jobs: ['build'], std: 'all', cxx: [*gcc7, *gcc8, *gcc9, *gcc10, *gcc11]}
- {jobs: ['build'], std: 'all', cxx: [*llvm9, *llvm10, *llvm11, *llvm12, *llvm13, *llvm14, *llvm15]}
- {jobs: ['test'], std: 'all', cxx: [*gcc12, *llvm16]}
- {jobs: ['build'], std: 'all', cxx: [*gcc12, *llvm16], cpu: 'arm64'}
- {jobs: ['build'], std: 'all', cxx: [*gcc12], sm: '90a'}
- {jobs: ['build'], std: 'all', cxx: [*oneapi]}
- {jobs: ['build'], std: 'all', cxx: [*msvc2019, *msvc2022]}
# default_projects: clang-cuda
- {jobs: ['build'], std: [17, 20], cudacxx: *llvm-newest, cxx: *llvm-newest}
# nvrtc:
- {jobs: ['nvrtc'], project: 'libcudacxx', std: 'all'}
# verify-codegen:
- {jobs: ['verify_codegen'], project: 'libcudacxx'}
# cudax has different CTK reqs:
# - {jobs: ['build'], project: 'cudax', ctk: [*ctk_12_0, *ctk_curr], std: 'all', cxx: [*gcc9, *gcc10, *gcc11]}
# - {jobs: ['build'], project: 'cudax', ctk: [*ctk_12_0, *ctk_curr], std: 'all', cxx: [*llvm9, *llvm10, *llvm11, *llvm12, *llvm13, *llvm14]}
# - {jobs: ['build'], project: 'cudax', ctk: [ *ctk_curr], std: 'all', cxx: [*llvm15]}
# - {jobs: ['build'], project: 'cudax', ctk: [*ctk_12_0, ], std: 'all', cxx: [*msvc2022_1436]}
# - {jobs: ['build'], project: 'cudax', ctk: [ *ctk_curr], std: 'all', cxx: [*msvc2022]}
# - {jobs: ['build'], project: 'cudax', ctk: [*ctk_12_0 ], std: 17, cxx: [*gcc12], sm: "90"}
# - {jobs: ['build'], project: 'cudax', ctk: [ *ctk_curr], std: 17, cxx: [*gcc12], sm: "90a"}
# - {jobs: ['build'], project: 'cudax', ctk: [ *ctk_curr], std: 'all', cxx: [*gcc12, *llvm16], cpu: 'arm64'}
# - {jobs: ['build'], project: 'cudax', ctk: [ *ctk_curr], std: 17, cxx: [*oneapi]}
# - {jobs: ['test'], project: 'cudax', ctk: [*ctk_12_0, *ctk_curr], std: 'all', cxx: [*gcc12]}
# - {jobs: ['test'], project: 'cudax', ctk: [*ctk_12_0 ], std: 'all', cxx: [*llvm14]}
# - {jobs: ['test'], project: 'cudax', ctk: [ *ctk_curr], std: 'all', cxx: [*llvm16]}
# cccl-infra:
- {jobs: ['infra'], project: 'cccl', ctk: *ctk_11_1, cxx: [*gcc-oldest, *llvm-oldest]}
- {jobs: ['infra'], project: 'cccl', ctk: *ctk_curr, cxx: [*gcc-newest, *llvm-newest]}
nightly:
# libcudacxx build fails, CUB tests fail:
- {jobs: ['build'], ctk: *ctk_11_1, gpu: 'v100', sm: 'gpu', cxx: *gcc6, std: [11], project: ['cub']}
- {jobs: ['test'], ctk: *ctk_11_1, gpu: 'v100', sm: 'gpu', cxx: *gcc6, std: [11], project: ['thrust']}
# - {jobs: ['test'], ctk: *ctk_11_1, gpu: 'v100', sm: 'gpu', cxx: *gcc6, std: [11] }
# libcudacxx build fails, CUB tests fail:
- {jobs: ['build'], ctk: *ctk_11_1, gpu: 't4', sm: 'gpu', cxx: *llvm9, std: [17], project: ['cub']}
- {jobs: ['test'], ctk: *ctk_11_1, gpu: 't4', sm: 'gpu', cxx: *llvm9, std: [17], project: ['thrust']}
# - {jobs: ['test'], ctk: *ctk_11_1, gpu: 't4', sm: 'gpu', cxx: *llvm9, std: [17] }
# CUB + libcudacxx tests fails:
- {jobs: ['build'], ctk: *ctk_11_8, gpu: 'rtx2080', sm: 'gpu', cxx: *gcc11, std: [17], project: ['libcudacxx', 'cub']}
- {jobs: ['test'], ctk: *ctk_11_8, gpu: 'rtx2080', sm: 'gpu', cxx: *gcc11, std: [17], project: ['thrust']}
# - {jobs: ['test'], ctk: *ctk_11_8, gpu: 'rtx2080', sm: 'gpu', cxx: *gcc11, std: [17] }
# libcudacxx tests fail:
- {jobs: ['build'], ctk: *ctk_curr, gpu: 'rtxa6000', sm: 'gpu', cxx: *gcc7, std: [14], project: ['libcudacxx']}
- {jobs: ['build'], ctk: *ctk_curr, gpu: 'l4', sm: 'gpu', cxx: *gcc12, std: 'all', project: ['libcudacxx']}
- {jobs: ['build'], ctk: *ctk_curr, gpu: 'rtx4090', sm: 'gpu', cxx: *llvm9, std: [11], project: ['libcudacxx']}
- {jobs: ['build'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *gcc12, std: [11, 20], project: ['libcudacxx']}
- {jobs: ['build'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *llvm16, std: [17], project: ['libcudacxx']}
- {jobs: ['test'], ctk: *ctk_curr, gpu: 'rtxa6000', sm: 'gpu', cxx: *gcc7, std: [14], project: ['cub', 'thrust']}
- {jobs: ['test'], ctk: *ctk_curr, gpu: 'l4', sm: 'gpu', cxx: *gcc12, std: 'all', project: ['cub', 'thrust']}
- {jobs: ['test'], ctk: *ctk_curr, gpu: 'rtx4090', sm: 'gpu', cxx: *llvm9, std: [11], project: ['cub', 'thrust']}
- {jobs: ['test'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *gcc12, std: [11, 20], project: ['cub', 'thrust']}
- {jobs: ['test'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *llvm16, std: [17], project: ['cub', 'thrust']}
# - {jobs: ['test'], ctk: *ctk_curr, gpu: 'rtxa6000', sm: 'gpu', cxx: *gcc7, std: [14] }
# - {jobs: ['test'], ctk: *ctk_curr, gpu: 'l4', sm: 'gpu', cxx: *gcc12, std: 'all' }
# - {jobs: ['test'], ctk: *ctk_curr, gpu: 'rtx4090', sm: 'gpu', cxx: *llvm9, std: [11] }
# - {jobs: ['test'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *gcc12, std: [11, 20] }
# - {jobs: ['test'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *llvm16, std: [17] }
# nvrtc:
- {jobs: ['nvrtc'], ctk: *ctk_curr, gpu: 't4', sm: 'gpu', cxx: *gcc12, std: [20], project: ['libcudacxx']}
- {jobs: ['nvrtc'], ctk: *ctk_curr, gpu: 'rtxa6000', sm: 'gpu', cxx: *gcc12, std: [20], project: ['libcudacxx']}
- {jobs: ['nvrtc'], ctk: *ctk_curr, gpu: 'l4', sm: 'gpu', cxx: *gcc12, std: 'all', project: ['libcudacxx']}
# Fails on h100:
# - {jobs: ['nvrtc'], ctk: *ctk_curr, gpu: 'h100', sm: 'gpu', cxx: *gcc12, std: [11, 20], project: ['libcudacxx']}
#
# Resources for compute_matrix.py. These can be modified to add new jobs, etc.
#
# Jobs are executed by running scripts:
# - Linux: 'ci/<job>_<project>.sh`
# - Windows: `ci/windows/<job>_<project>.bat`
# A matrix entry must have the following tag.
required_tags:
- 'jobs' # A list of job types to run (e.g. 'build', 'test', 'nvrtc', 'infra', 'verify_codegen', ...) for
# the specified configuration(s).
# If a matrix entry omits these tags, a default value (defined later in `default_<tag>`) is used.
defaulted_tags:
- 'ctk' # CUDA ToolKit version. Will be exploded if a list.
- 'cpu' # CPU architecture. Will be exploded if a list.
- 'gpu' # GPU model. Will be exploded if a list.
- 'cxx' # Host compiler {name, version, exe}. Will be exploded if a list.
- 'cudacxx' # Device compiler as {name, version, exe} or 'nvcc' to use nvcc from the specified `ctk`.
# Will be exploded if a list.
- 'project' # Project name (e.g. libcudacxx, cub, thrust, cccl). Will be exploded if a list.
- 'os' # Operating system. Will be exploded if a list.
# These tags will only exist if needed:
optional_tags:
- 'std' # C++ standard. Passed to script with `-std <std>`. Will be exploded if a list.
# If set to 'all', all stds supported by the host/device compiler are used.
- 'sm' # `CMAKE_CUDA_ARCHITECTURES` Passed to script with `-arch <sm>`.
# Defaults to use the settings in the CMakePresets.json file.
# Set to 'gpu' to only target the GPU in the `gpu` tag.
# Can pass multiple architectures via "60;70-real;80-virtual"
# Will be exploded if a list (e.g. `sm: ['60;70;80;90', '90a']` creates two jobs)
- 'cmake_options' # Additional CMake options to pass to the build. Passed to script with `-cmake_options "<cmake_options>"`.
# Will be exploded if a list.
# `default_<tag>`: Used when the tag is omitted.
default_ctk: *ctk_curr
default_cudacxx: 'nvcc'
default_cxx: *gcc12
default_cpu: 'amd64'
default_gpu: 'v100'
default_project:
- 'libcudacxx'
- 'cub'
- 'thrust'
# Special handling: lookup os from ctk/cxx info
# See `matrix.yml` at https://github.com/rapidsai/devcontainers
default_os_lookup:
'ctk11.1-gcc6': 'ubuntu18.04'
'ctk11.1-gcc7': 'ubuntu18.04'
'ctk11.1-gcc8': 'ubuntu18.04'
'ctk11.1-gcc9': 'ubuntu18.04'
'ctk11.1-llvm9': 'ubuntu18.04'
'ctk11.1-cl14.16': 'windows2022'
'ctk11.8-gcc11': 'ubuntu22.04'
'ctk12.0-gcc7': 'ubuntu20.04'
'ctk12.0-gcc8': 'ubuntu20.04'
'ctk12.0-gcc9': 'ubuntu20.04'
'ctk12.0-gcc10': 'ubuntu20.04'
'ctk12.0-gcc11': 'ubuntu22.04'
'ctk12.0-gcc12': 'ubuntu22.04'
'ctk12.0-llvm9': 'ubuntu20.04'
'ctk12.0-llvm10': 'ubuntu20.04'
'ctk12.0-llvm11': 'ubuntu20.04'
'ctk12.0-llvm12': 'ubuntu20.04'
'ctk12.0-llvm13': 'ubuntu20.04'
'ctk12.0-llvm14': 'ubuntu20.04'
'ctk12.0-llvm15': 'ubuntu22.04'
'ctk12.0-llvm16': 'ubuntu22.04'
'ctk12.0-cl14.29': 'windows2022'
'ctk12.0-cl14.36': 'windows2022'
'ctk12.0-cl14.39': 'windows2022'
'ctk12.0-oneapi2023.2.0': 'ubuntu22.04'
'ctk12.4-gcc7': 'ubuntu20.04'
'ctk12.4-gcc8': 'ubuntu20.04'
'ctk12.4-gcc9': 'ubuntu20.04'
'ctk12.4-gcc10': 'ubuntu20.04'
'ctk12.4-gcc11': 'ubuntu22.04'
'ctk12.4-gcc12': 'ubuntu22.04'
'ctk12.4-llvm9': 'ubuntu20.04'
'ctk12.4-llvm10': 'ubuntu20.04'
'ctk12.4-llvm11': 'ubuntu20.04'
'ctk12.4-llvm12': 'ubuntu20.04'
'ctk12.4-llvm13': 'ubuntu20.04'
'ctk12.4-llvm14': 'ubuntu20.04'
'ctk12.4-llvm15': 'ubuntu22.04'
'ctk12.4-llvm16': 'ubuntu22.04'
'ctk12.4-cl14.29': 'windows2022'
'ctk12.4-cl14.36': 'windows2022'
'ctk12.4-cl14.39': 'windows2022'
'ctk12.4-oneapi2023.2.0': 'ubuntu22.04'
# Lookup supported C++ standards for a given compiler when `std: 'all'`.
all_stds: [11, 14, 17, 20]
lookup_cxx_supported_stds:
'gcc6': [11, 14 ]
'gcc7': [11, 14, 17 ]
'gcc8': [11, 14, 17 ]
'gcc9': [11, 14, 17 ]
'gcc10': [11, 14, 17, 20]
'gcc11': [11, 14, 17, 20]
'gcc12': [11, 14, 17, 20]
'llvm9': [11, 14, 17 ]
'llvm10': [11, 14, 17 ]
'llvm11': [11, 14, 17, 20]
'llvm12': [11, 14, 17, 20]
'llvm13': [11, 14, 17, 20]
'llvm14': [11, 14, 17, 20]
'llvm15': [11, 14, 17, 20]
'llvm16': [11, 14, 17, 20]
'cl14.16': [ 14 ]
'cl14.29': [ 14, 17 ]
'cl14.36': [ 14, 17, 20]
'cl14.39': [ 14, 17, 20]
'oneapi2023.2.0': [11, 14, 17 ]
lookup_cudacxx_supported_stds:
'nvcc11.1': [11, 14, 17 ]
'nvcc11.8': [11, 14, 17 ]
'nvcc12.0': [11, 14, 17, 20]
'nvcc12.4': [11, 14, 17, 20]
'llvm16': [11, 14, 17, 20]
lookup_project_supported_stds:
'cccl': [11, 14, 17, 20]
'libcudacxx': [11, 14, 17, 20]
'cub': [11, 14, 17, 20]
'thrust': [11, 14, 17, 20]
'cudax': [ 17, 20]
# Tags that aren't exploded:
non_exploded_tags:
- 'jobs' # Keeping jobs as a list allows for dependency handling of build->test steps.
# Jobs that have an implied prerequisite 'build' job:
build_required_jobs:
- 'test'
# Jobs that require a GPU
gpu_required_jobs:
- 'test'
- 'nvrtc'
- 'infra' # cccl infra's example project test launches a kernel
# When --skip-tests is given to compute-matrix.py, these jobs are ignored.
skip_test_jobs:
- 'test'
- 'nvrtc'
- 'infra'
# Human readable name for jobs. Default behavior is to capitalize the first letter.
formatted_jobs:
'nvrtc': 'NVRTC'
'verify_codegen': 'VerifyCodegen'
# Human readable name for projects. Default behavior uses the project name as-is.
formatted_project_names:
'libcudacxx': 'libcu++'
'cub': 'CUB'
'thrust': 'Thrust'
'cccl': 'CCCL'
# Human readable name for compilers. Default behavior uses the "compiler.name" tag as-is.
formatted_cxx_names:
'llvm': 'clang'
'oneapi': 'Intel'
'cl': 'MSVC'
# All known GPUs
gpus:
- 'v100' # 40 runners
- 't4' # 8 runners
- 'rtx2080' # 8 runners
- 'rtxa6000' # 12 runners
- 'l4' # 48 runners
- 'rtx4090' # 10 runners
- 'h100' # 16 runners
# SM versions of GPUs
gpu_sm:
'v100': '70'
't4': '75'
'rtx2080': '75'
'rtxa6000': '86'
'l4': '89'
'rtx4090': '89'
'h100': '90'
# Memory size of GPUs
gpu_mem_gb:
'v100': '32'
't4': '16'
'rtx2080': '8'
'rtxa6000': '48'
'l4': '24'
'rtx4090': '24'
'h100': '80'
# GPUs that require `-testing` at the end of the runner pool name.
testing_pool_gpus:
- 't4'
- 'rtx2080'
- 'rtxa6000'
- 'l4'
- 'rtx4090'