@tduncan / Nvidia-Optimization-Stack-VLLM
Code Commits Issues Pulls Cloud Agent CI/CD
main
main
yaml 152 lines 4.6 KB
Raw
1 group: Models - Multimodal
2 depends_on:
3 - image-build-xpu
4 steps:
5 - label: "Multi-Modal Models (Standard) 1: qwen2"
6 key: multi-modal-models-standard-1-qwen2
7 timeout_in_minutes: 70
8 device: intel_gpu
9 agent_tags:
10 label: production
11 gpu: 1+
12 mem: 24+
13 no_plugin: true
14 working_dir: "."
15 env:
16 REGISTRY: "public.ecr.aws/q9t5s3a7"
17 REPO: "vllm-ci-test-repo"
18 VLLM_TEST_DEVICE: "xpu"
19 source_file_dependencies:
20 - vllm/
21 - tests/models/multimodal
22 commands:
23 - >-
24 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
25 'cd tests &&
26 pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "qwen2" &&
27 pytest -v -s models/multimodal/generation/test_ultravox.py -m core_model'
28
29 - label: "Multi-Modal Models (Standard) 2: qwen3 + gemma"
30 key: multi-modal-models-standard-2-qwen3-gemma
31 timeout_in_minutes: 90
32 device: intel_gpu
33 agent_tags:
34 label: production
35 gpu: 1+
36 mem: 24+
37 no_plugin: true
38 working_dir: "."
39 env:
40 REGISTRY: "public.ecr.aws/q9t5s3a7"
41 REPO: "vllm-ci-test-repo"
42 VLLM_TEST_DEVICE: "xpu"
43 source_file_dependencies:
44 - vllm/
45 - tests/models/multimodal
46 commands:
47 - >-
48 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
49 'cd tests &&
50 pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "qwen3 or gemma" &&
51 pytest -v -s models/multimodal/generation/test_mm_prefix_lm.py -m core_model &&
52 pytest -v -s models/multimodal/generation/test_qwen2_5_vl.py -m core_model'
53
54 - label: "Multi-Modal Models (Standard) 3: llava + qwen2_vl"
55 key: multi-modal-models-standard-3-llava-qwen2-vl
56 timeout_in_minutes: 65
57 device: intel_gpu
58 agent_tags:
59 label: production
60 gpu: 1+
61 mem: 24+
62 no_plugin: true
63 working_dir: "."
64 env:
65 REGISTRY: "public.ecr.aws/q9t5s3a7"
66 REPO: "vllm-ci-test-repo"
67 VLLM_TEST_DEVICE: "xpu"
68 source_file_dependencies:
69 - vllm/
70 - tests/models/multimodal
71 commands:
72 - >-
73 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
74 'cd tests &&
75 pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "not qwen2 and not qwen3 and not gemma" &&
76 pytest -v -s models/multimodal/generation/test_qwen2_vl.py -m core_model'
77
78 - label: "Multi-Modal Models (Standard) 4: other + whisper"
79 key: multi-modal-models-standard-4-other-whisper
80 timeout_in_minutes: 45
81 device: intel_gpu
82 agent_tags:
83 label: production
84 gpu: 1+
85 mem: 24+
86 no_plugin: true
87 working_dir: "."
88 env:
89 REGISTRY: "public.ecr.aws/q9t5s3a7"
90 REPO: "vllm-ci-test-repo"
91 VLLM_TEST_DEVICE: "xpu"
92 source_file_dependencies:
93 - vllm/
94 - tests/models/multimodal
95 commands:
96 - >-
97 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
98 'cd tests &&
99 pytest -v -s models/multimodal -m core_model --ignore models/multimodal/generation/test_common.py --ignore models/multimodal/generation/test_ultravox.py --ignore models/multimodal/generation/test_qwen2_5_vl.py --ignore models/multimodal/generation/test_qwen2_vl.py --ignore models/multimodal/generation/test_whisper.py --ignore models/multimodal/generation/test_memory_leak.py --ignore models/multimodal/processing'
100
101 - label: Multi-Modal Processor # 44min
102 key: multi-modal-processor
103 timeout_in_minutes: 60
104 device: intel_gpu
105 agent_tags:
106 label: production
107 gpu: 1+
108 mem: 24+
109 no_plugin: true
110 working_dir: "."
111 env:
112 REGISTRY: "public.ecr.aws/q9t5s3a7"
113 REPO: "vllm-ci-test-repo"
114 VLLM_TEST_DEVICE: "xpu"
115 source_file_dependencies:
116 - vllm/
117 - tests/models/multimodal
118 - tests/models/registry.py
119 commands:
120 - >-
121 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
122 'cd tests &&
123 pytest -v -s models/multimodal/processing/test_tensor_schema.py
124 --deselect="tests/models/multimodal/processing/test_tensor_schema.py::test_model_tensor_schema[deepseek-ai/DeepSeek-V4.1-Flash]"
125 --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB'
126 parallelism: 4
127
128 - label: Multi-Modal Models (Extended Generation 1)
129 key: multi-modal-models-extended-generation-1
130 timeout_in_minutes: 60
131 device: intel_gpu
132 agent_tags:
133 label: production
134 gpu: 1+
135 mem: 24+
136 no_plugin: true
137 working_dir: "."
138 env:
139 REGISTRY: "public.ecr.aws/q9t5s3a7"
140 REPO: "vllm-ci-test-repo"
141 VLLM_TEST_DEVICE: "xpu"
142 optional: true
143 source_file_dependencies:
144 - vllm/
145 - "!vllm/distributed/kv_transfer/"
146 - tests/models/multimodal/generation
147 - tests/models/multimodal/test_mapping.py
148 commands:
149 - >-
150 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
151 'cd tests &&
152 pytest -v -s models/multimodal/test_mapping.py'