ci/config.yaml at master · InfiniTensor/ci · GitHub

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
repo:
  url: https://github.com/InfiniTensor/InfiniOps.git
  branch: master

github:
  status_context_prefix: "ci/infiniops"

# Uncomment and replace the URLs below with actual host IPs to dispatch jobs to remote
# machines via `agent.py run`. Required on the trigger machine when each platform's
# agent runs on a separate host. See the README for multi-machine deployment details.
# agents:
#   nvidia:
#     url: http://nvidia-host:8080
#   iluvatar:
#     url: http://iluvatar-host:8080
#   metax:
#     url: http://metax-host:8080
#   moore:
#     url: http://moore-host:8080
#   cambricon:
#     url: http://cambricon-host:8080

platforms:
  nvidia:
    image:
      dockerfile: .ci/images/nvidia/
      build_args:
        BASE_IMAGE: nvcr.io/nvidia/pytorch:25.12-py3
    setup: pip install .[dev] --no-build-isolation
    jobs:
      gpu:
        type: unittest
        resources:
          ngpus: 1                         # Scheduler auto-picks this many free GPUs
          memory: 32GB
          shm_size: 16g                    # Prevent PyTorch default 64MB shared memory limit
          timeout: 3600
        # env:                             # Uncomment to inject extra env vars into the container.
        #   MY_VAR: value
        stages:
          - name: test
            run: pytest tests/ -n 8 -v --tb=short --junitxml=/workspace/results/test-results.xml
      gpu_smoketest:
        type: smoketest
        docker_args:
          - "--gpus all"
          - "-u root"
          - "--network host"
          - "--privileged"
          - "--cap-add=ALL"
          - "--pid=host"
          - "--ipc=host"
          - "--entrypoint=''"
          - "--workdir=/workspace"
          - "-e CUDA_DEVICE_ORDER=PCI_BUS_ID"
        volumes:
          - /dev:/dev
          - /data/shared/limingge/weight:/home/weight
          - /home/zkjh/workspace:/workspace
        resources:
          shm_size: 80g
        env:
          MODEL_LIST: Qwen3-32B-FP8
          IMAGE: v0.10.0
          ENGINE: vLLM
      gpu_performancetest:
        type: performancetest
        docker_args:
          - "--gpus all"
          - "-u root"
          - "--network host"
          - "--privileged"
          - "--cap-add=ALL"
          - "--pid=host"
          - "--ipc=host"
          - "--entrypoint=''"
          - "--workdir=/workspace"
          - "-e CUDA_DEVICE_ORDER=PCI_BUS_ID"
        volumes:
          - /dev:/dev
          - /data/shared/limingge/weight:/home/weight
          - /home/zkjh/workspace:/workspace
        resources:
          shm_size: 80g
        env:
          MODEL_LIST: Qwen3-32B-FP8
          IMAGE: v0.10.0
          ENGINE: vLLM

  iluvatar:
    image:
      dockerfile: .ci/images/iluvatar/
      build_args:
        BASE_IMAGE: corex:qs_pj20250825
        APT_MIRROR: http://archive.ubuntu.com/ubuntu
        PIP_INDEX_URL: https://pypi.org/simple
    docker_args:
      - "--privileged"
      - "--cap-add=ALL"
      - "--pid=host"
      - "--ipc=host"
    volumes:
      - /dev:/dev
      - /lib/firmware:/lib/firmware
      - /usr/src:/usr/src
      - /lib/modules:/lib/modules
    setup: pip install .[dev] --no-build-isolation
    jobs:
      gpu:
        type: unittest
        resources:
          ngpus: 1
          gpu_style: none
          memory: 32GB
          shm_size: 16g
          timeout: 3600
        stages:
          - name: test
            run: pytest tests/ -n 8 -v --tb=short --junitxml=/workspace/results/test-results.xml

  metax:
    image:
      dockerfile: .ci/images/metax/
      build_args:
        BASE_IMAGE: cr.metax-tech.com/public-library/maca-pytorch:3.2.1.4-torch2.4-py310-ubuntu22.04-amd64
        APT_MIRROR: http://archive.ubuntu.com/ubuntu
        PIP_INDEX_URL: https://pypi.org/simple
    docker_args:
      - "--privileged"
      - "--ulimit=memlock=-1"
      - "--ulimit=stack=67108864"
    setup: pip install .[dev] --no-build-isolation
    jobs:
      gpu:
        type: unittest
        resources:
          ngpus: 1
          gpu_style: none                  # MetaX: passthrough via --privileged, no CUDA_VISIBLE_DEVICES
          memory: 32GB
          shm_size: 16g
          timeout: 3600
        stages:
          - name: test
            run: pytest tests/ -n 4 -v --tb=short --junitxml=/workspace/results/test-results.xml

  moore:
    image:
      dockerfile: .ci/images/moore/
      build_args:
        BASE_IMAGE: sh-harbor.mthreads.com/mcctest/vllm_musa:20251112_hygon
        APT_MIRROR: http://archive.ubuntu.com/ubuntu
        PIP_INDEX_URL: https://pypi.org/simple
    docker_args:
      - "--privileged"
    setup: pip install .[dev] --no-build-isolation
    jobs:
      gpu:
        type: unittest
        resources:
          ngpus: 1
          gpu_style: none                  # Moore: passthrough via --privileged, MTHREADS_VISIBLE_DEVICES set by base image
          memory: 32GB
          shm_size: 16g
          timeout: 3600
        stages:
          - name: test
            run: pytest tests/test_add.py tests/test_gemm.py tests/test_swiglu.py -n 4 -v --tb=short --junitxml=/workspace/results/test-results.xml

  cambricon:
    image:
      dockerfile: .ci/images/cambricon/
      build_args:
        BASE_IMAGE: cambricon/pytorch:v1.25.3-torch2.1-anolisos8.8-py310
        PIP_INDEX_URL: https://pypi.org/simple
    docker_args:
      - "--privileged"
    setup: pip install .[dev] --no-build-isolation
    jobs:
      gpu:
        type: unittest
        resources:
          ngpus: 1
          gpu_style: mlu                   # Cambricon: passthrough via --privileged, MLU_VISIBLE_DEVICES for device control
          memory: 32GB
          shm_size: 16g
          timeout: 3600
        stages:
          - name: test
            run: pytest tests/test_gemm.py -n 4 -v --tb=short --junitxml=/workspace/results/test-results.xml

  ascend:
    image:
      dockerfile: .ci/images/ascend/
      build_args:
        BASE_IMAGE: swr.cn-south-1.myhuaweicloud.com/ascendhub/mindie:2.1.RC1-800I-A2-py311-openeuler24.03-lts
    jobs:
      npu:
        type: unittest
        resources:
          ngpus: 1                         # Scheduler auto-picks this many free GPUs
          gpu_style: none
          memory: 32GB
          shm_size: 16g                    # Prevent PyTorch default 64MB shared memory limit
          timeout: 3600
        # env:                             # Uncomment to inject extra env vars into the container.
        #   MY_VAR: value
        stages:
          - name: test
            run: pytest tests/ -n 8 -v --tb=short --junitxml=/workspace/results/test-results.xml
      npu_smoketest:
        type: smoketest
        docker_args:
          - "--privileged"
          - "--cap-add=ALL"
          - "--pid=host"
          - "--ipc=host"
          - "--net=host"
          - "--device=/dev/davinci0"
          - "--device=/dev/davinci1"
          - "--device=/dev/davinci2"
          - "--device=/dev/davinci3"
          - "--device=/dev/davinci4"
          - "--device=/dev/davinci5"
          - "--device=/dev/davinci6"
          - "--device=/dev/davinci7"
          - "--device=/dev/davinci_manager"
          - "--device=/dev/hisi_hdc"
          - "--device=/dev/devmm_svm"
        volumes:
          - /usr/local/Ascend/driver/lib64:/usr/local/Ascend/driver/lib64
          - /usr/local/Ascend/driver/include:/usr/local/Ascend/driver/include
          - /usr/local/Ascend/driver/tools:/usr/local/Ascend/driver/tools
          - /usr/local/Ascend/driver:/usr/local/Ascend/driver
          - /usr/local/Ascend/firmware:/usr/local/Ascend/firmware
          - /usr/local/sbin/npu-smi:/usr/local/sbin/npu-smi
          - /usr/local/sbin:/usr/local/sbin
          - /etc/hccn.conf:/etc/hccn.conf
          - /home/zkjh/weight:/home/weight
          - /home/zkjh:/home/zkjh
        resources:
          shm_size: 500g
        env:
          MODEL_LIST: DeepSeek-R1-Distill-Qwen-1.5B
          IMAGE: 2.1.RC1-800I-A2-py311-openeuler24.03-lts
          ENGINE: MindIE
      npu_performancetest:
        type: performancetest
        docker_args:
          - "--privileged"
          - "--cap-add=ALL"
          - "--pid=host"
          - "--ipc=host"
          - "--net=host"
          - "--device=/dev/davinci0"
          - "--device=/dev/davinci1"
          - "--device=/dev/davinci2"
          - "--device=/dev/davinci3"
          - "--device=/dev/davinci4"
          - "--device=/dev/davinci5"
          - "--device=/dev/davinci6"
          - "--device=/dev/davinci7"
          - "--device=/dev/davinci_manager"
          - "--device=/dev/hisi_hdc"
          - "--device=/dev/devmm_svm"
        volumes:
          - /usr/local/Ascend/driver/lib64:/usr/local/Ascend/driver/lib64
          - /usr/local/Ascend/driver/include:/usr/local/Ascend/driver/include
          - /usr/local/Ascend/driver/tools:/usr/local/Ascend/driver/tools
          - /usr/local/Ascend/driver:/usr/local/Ascend/driver
          - /usr/local/Ascend/firmware:/usr/local/Ascend/firmware
          - /usr/local/sbin/npu-smi:/usr/local/sbin/npu-smi
          - /usr/local/sbin:/usr/local/sbin
          - /etc/hccn.conf:/etc/hccn.conf
          - /home/zkjh/weight:/home/weight
          - /home/zkjh:/home/zkjh
        resources:
          shm_size: 500g
        env:
          MODEL_LIST: DeepSeek-R1-Distill-Qwen-1.5B
          IMAGE: 2.1.RC1-800I-A2-py311-openeuler24.03-lts
          ENGINE: MindIE