| 1 | group: Model Executor Intel |
| 2 | depends_on: |
| 3 | - image-build-xpu |
| 4 | steps: |
| 5 | - label: Model Executor (Intel) |
| 6 | key: model-executor-intel |
| 7 | timeout_in_minutes: 45 |
| 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/engine/arg_utils.py |
| 21 | - vllm/config/model.py |
| 22 | - vllm/model_executor |
| 23 | - tests/model_executor |
| 24 | commands: |
| 25 | - >- |
| 26 | bash .buildkite/scripts/hardware_ci/run-intel-test.sh |
| 27 | 'apt-get update && apt-get install -y curl libsodium23 && |
| 28 | export VLLM_WORKER_MULTIPROC_METHOD=spawn && |
| 29 | export PYTHONFAULTHANDLER=1 && |
| 30 | cd tests && |
| 31 | pytest -v -s model_executor -m "not slow_test" --ignore="model_executor/layers/test_rocm_unquantized_gemm.py" --deselect="tests/model_executor/model_loader/test_reload.py::test_kv_scale_reload"' |