@tduncan / Nvidia-Optimization-Stack-VLLM
Code Commits Issues Pulls Cloud Agent CI/CD
main
main
yaml 215 lines 9.88 KB
Raw
1 group: Intel
2 steps:
3 - label: ":docker: Build XPU image"
4 soft_fail: true
5 depends_on: []
6 key: image-build-xpu
7 commands:
8 - bash -lc '.buildkite/image_build/image_build_xpu.sh "public.ecr.aws/q9t5s3a7" "vllm-ci-test-repo" "$BUILDKITE_COMMIT"'
9 env:
10 DOCKER_BUILDKIT: "1"
11 retry:
12 automatic:
13 - exit_status: -1 # Agent was lost
14 limit: 2
15 - exit_status: -10 # Agent was lost
16 limit: 2
17 - label: "XPU example Test"
18 key: xpu-example-test
19 depends_on:
20 - image-build-xpu
21 timeout_in_minutes: 70
22 device: intel_gpu
23 agent_tags:
24 label: production
25 gpu: 2+
26 mem: 24+
27 no_plugin: true
28 env:
29 REGISTRY: "public.ecr.aws/q9t5s3a7"
30 REPO: "vllm-ci-test-repo"
31 source_file_dependencies:
32 - vllm/
33 - .buildkite/intel_jobs/test-intel.yaml
34 commands:
35 - >-
36 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
37 'python3 examples/basic/offline_inference/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager &&
38 python3 examples/basic/offline_inference/generate.py --model facebook/opt-125m --block-size 64 -O3 -cc.cudagraph_mode=NONE &&
39 python3 examples/basic/offline_inference/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager -tp 2 --distributed-executor-backend mp &&
40 python3 examples/basic/offline_inference/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager --attention-backend=TRITON_ATTN &&
41 python3 examples/basic/offline_inference/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager --quantization fp8_per_tensor &&
42 python3 examples/basic/offline_inference/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager --kv-cache-dtype fp8 &&
43 python3 examples/basic/offline_inference/generate.py --model nvidia/Llama-3.1-8B-Instruct-FP8 --block-size 64 --enforce-eager --quantization modelopt --kv-cache-dtype fp8 --attention-backend TRITON_ATTN --max-model-len 4096 &&
44 python3 examples/basic/offline_inference/generate.py --model superjob/Qwen3-4B-Instruct-2507-GPTQ-Int4 --block-size 64 --enforce-eager --max-model-len 8192 &&
45 python3 examples/basic/offline_inference/generate.py --model TheBloke/TinyLlama-1.1B-Chat-v0.3-AWQ --block-size 64 --enforce-eager &&
46 python3 examples/basic/offline_inference/generate.py --model ibm-research/PowerMoE-3b --block-size 64 --enforce-eager -tp 2 &&
47 python3 examples/basic/offline_inference/generate.py --model ibm-research/PowerMoE-3b --block-size 64 --enforce-eager -tp 2 --enable-expert-parallel &&
48 python3 examples/basic/offline_inference/generate.py --model superjob/Qwen3-4B-Instruct-2507-GPTQ-Int4 --max-model-len 8192 &&
49 VLLM_XPU_FUSED_MOE_USE_REF=1 python3 examples/basic/offline_inference/generate.py --model Qwen/Qwen3-30B-A3B-Instruct-2507-FP8 --enforce-eager -tp 2 --max-model-len 8192 &&
50 python3 examples/basic/offline_inference/generate.py --model INCModel/Qwen3-30B-A3B-Instruct-2507-MXFP4-LLMC -tp 2 --max-model-len 8192
51 '
52 - label: "XPU W8A8 FP8 Linear Examples"
53 key: xpu-w8a8-fp8-linear-examples
54 depends_on:
55 - image-build-xpu
56 timeout_in_minutes: 60
57 device: intel_gpu
58 agent_tags:
59 label: production
60 gpu: 1+
61 mem: 24+
62 no_plugin: true
63 env:
64 REGISTRY: "public.ecr.aws/q9t5s3a7"
65 REPO: "vllm-ci-test-repo"
66 VLLM_TEST_DEVICE: "xpu"
67 source_file_dependencies:
68 - vllm/
69 - .buildkite/intel_jobs/test-intel.yaml
70 commands:
71 - >-
72 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
73 'python3 examples/basic/offline_inference/generate.py --linear-backend xpu --model RedHatAI/Meta-Llama-3.1-8B-Instruct-FP8 --enforce-eager --max-model-len 4096 &&
74 python3 examples/basic/offline_inference/generate.py --linear-backend torch --model RedHatAI/Meta-Llama-3.1-8B-Instruct-FP8 --enforce-eager --max-model-len 4096 &&
75 python3 examples/basic/offline_inference/generate.py --linear-backend xpu --model neuralmagic/Llama-3.2-1B-Instruct-FP8-dynamic --enforce-eager --max-model-len 4096 &&
76 python3 examples/basic/offline_inference/generate.py --linear-backend torch --model Qwen/Qwen3-4B-Instruct-2507-FP8 --enforce-eager --max-model-len 4096 &&
77 python3 examples/basic/offline_inference/generate.py --linear-backend xpu --model meta-llama/Llama-3.2-1B-Instruct --quantization fp8_per_tensor --enforce-eager --max-model-len 4096 &&
78 python3 examples/basic/offline_inference/generate.py --linear-backend torch --model meta-llama/Llama-3.2-1B-Instruct --quantization fp8_per_tensor --enforce-eager --max-model-len 4096
79 '
80 - label: "XPU V1 test"
81 key: xpu-v1-test
82 depends_on:
83 - image-build-xpu
84 timeout_in_minutes: 70
85 device: intel_gpu
86 agent_tags:
87 label: production
88 gpu: 1+
89 mem: 24+
90 no_plugin: true
91 env:
92 REGISTRY: "public.ecr.aws/q9t5s3a7"
93 REPO: "vllm-ci-test-repo"
94 source_file_dependencies:
95 - vllm/
96 - .buildkite/intel_jobs/test-intel.yaml
97 commands:
98 - >-
99 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
100 'cd tests && pip install dpctl &&
101 pytest -v -s v1/core --ignore=v1/core/test_reset_prefix_cache_e2e.py --ignore=v1/core/test_scheduler_e2e.py &&
102 pytest -v -s v1/engine --ignore=v1/engine/test_output_processor.py --ignore=v1/engine/test_core_engine_actor_manager.py &&
103 pytest -v -s v1/sample --ignore=v1/sample/test_logprobs.py --ignore=v1/sample/test_logprobs_e2e.py -k "not test_topk_only and not test_topp_only and not test_topk_and_topp" &&
104 pytest -v -s v1/worker --ignore=v1/worker/test_gpu_model_runner.py --ignore=v1/worker/test_worker_memory_snapshot.py &&
105 pytest -v -s v1/structured_output &&
106 pytest -v -s v1/test_serial_utils.py &&
107 pytest -v -s v1/e2e/general/test_correctness_sliding_window.py &&
108 pytest -v -s v1/spec_decode --ignore=v1/spec_decode/test_max_len.py --ignore=v1/spec_decode/test_speculators_eagle3.py --ignore=v1/spec_decode/test_acceptance_length.py --ignore=v1/spec_decode/test_speculators_correctness.py --deselect=tests/v1/spec_decode/test_mtp.py::test_glm_mtp_defers_lm_head &&
109 pytest -v -s v1/kv_connector/unit --ignore=v1/kv_connector/unit/test_lmcache_integration.py --ignore=v1/kv_connector/unit/test_hf3fs_client.py --ignore=v1/kv_connector/unit/test_hf3fs_connector.py --ignore=v1/kv_connector/unit/test_hf3fs_metadata_server.py --ignore=v1/kv_connector/unit/test_offloading_connector.py --ignore=v1/kv_connector/unit/test_nixl_connector_hma.py --deselect=tests/v1/kv_connector/unit/test_nixl_connector.py::test_abort_timeout_on_prefiller[ray]'
110 parallelism: 2
111 - label: "XPU server test"
112 key: xpu-server-test
113 depends_on:
114 - image-build-xpu
115 timeout_in_minutes: 45
116 device: intel_gpu
117 agent_tags:
118 label: production
119 gpu: 1+
120 mem: 16+
121 no_plugin: true
122 env:
123 REGISTRY: "public.ecr.aws/q9t5s3a7"
124 REPO: "vllm-ci-test-repo"
125 source_file_dependencies:
126 - vllm/
127 - .buildkite/intel_jobs/test-intel.yaml
128 commands:
129 - >-
130 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
131 'cd tests &&
132 pytest -v -s entrypoints/multimodal/openai/chat_completion/test_audio_in_video.py &&
133 pytest -v -s benchmarks/test_serve_cli.py'
134 - label: "XPU quantization test"
135 key: xpu-quantization-test
136 depends_on:
137 - image-build-xpu
138 timeout_in_minutes: 30
139 device: intel_gpu
140 agent_tags:
141 label: production
142 gpu: 1+
143 mem: 16+
144 no_plugin: true
145 env:
146 REGISTRY: "public.ecr.aws/q9t5s3a7"
147 REPO: "vllm-ci-test-repo"
148 source_file_dependencies:
149 - vllm/
150 - .buildkite/intel_jobs/test-intel.yaml
151 commands:
152 - >-
153 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
154 'cd tests &&
155 pytest -v -s quantization/test_auto_round.py &&
156 pytest -v -s quantization/test_online.py --deselect=tests/quantization/test_online.py::test_online_quantization_loads_real_weights'
157 - label: "XPU compressed tensors FP8 test"
158 key: xpu-compressed-tensors-fp8-test
159 depends_on:
160 - image-build-xpu
161 timeout_in_minutes: 60
162 device: intel_gpu
163 agent_tags:
164 label: production
165 gpu: 1+
166 mem: 16+
167 no_plugin: true
168 env:
169 REGISTRY: "public.ecr.aws/q9t5s3a7"
170 REPO: "vllm-ci-test-repo"
171 VLLM_TEST_DEVICE: "xpu"
172 source_file_dependencies:
173 - vllm/
174 - tests/quantization/test_compressed_tensors.py
175 - .buildkite/intel_jobs/test-intel.yaml
176 commands:
177 - >-
178 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
179 'cd tests &&
180 pytest -v -s quantization/test_compressed_tensors.py::test_compressed_tensors_fp8 &&
181 pytest -v -s quantization/test_compressed_tensors.py::test_compressed_tensors_fp8_block_enabled'
182 - label: "XPU Graph"
183 key: xpu-graph
184 depends_on:
185 - image-build-xpu
186 timeout_in_minutes: 60
187 device: intel_gpu
188 agent_tags:
189 label: production
190 gpu: 1+
191 mem: 32+
192 no_plugin: true
193 env:
194 REGISTRY: "public.ecr.aws/q9t5s3a7"
195 REPO: "vllm-ci-test-repo"
196 VLLM_TEST_DEVICE: "xpu"
197 source_file_dependencies:
198 - vllm/compilation/
199 - vllm/config/compilation.py
200 - vllm/platforms/xpu.py
201 - vllm/v1/cudagraph_dispatcher.py
202 - vllm/v1/worker/encoder_cudagraph.py
203 - vllm/v1/worker/encoder_cudagraph_defs.py
204 - vllm/v1/worker/xpu_model_runner.py
205 - vllm/v1/worker/xpu_worker.py
206 - tests/v1/cudagraph
207 - .buildkite/intel_jobs/test-intel.yaml
208 commands:
209 - >-
210 bash .buildkite/scripts/hardware_ci/run-intel-test.sh
211 'cd tests &&
212 pytest -v -s v1/cudagraph/test_cudagraph_dispatch.py &&
213 pytest -v -s v1/cudagraph/test_cudagraph_mode.py &&
214 pytest -v -s v1/cudagraph/test_breakable_cudagraph.py &&
215 pytest -v -s v1/cudagraph/test_encoder_cudagraph.py'