[CI/Build] Cross python wheel (#6394)

Author: robertgshaw2-redhat

.buildkite/release-pipeline.yaml

@@ -1,51 +1,17 @@
 steps:
-  - label: "Build wheel default - Python {{matrix.python_version}}, CUDA {{matrix.cuda_version}}" 
+  - label: "Build wheel - CUDA {{matrix.cuda_version}}"
     agents:
       queue: cpu_queue
     commands:
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg CUDA_VERSION={{matrix.cuda_version}} --build-arg PYTHON_VERSION={{matrix.python_version}} --tag vllm-ci:build-image --target build --progress plain ."
+      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg CUDA_VERSION={{matrix.cuda_version}} --tag vllm-ci:build-image --target build --progress plain ."
       - "mkdir artifacts"
       - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
       # rename the files to change linux -> manylinux1
       - "for f in artifacts/dist/*.whl; do mv -- \"$$f\" \"$${f/linux/manylinux1}\"; done"
       - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/$BUILDKITE_COMMIT/"
       - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/nightly/"
-    matrix:
-      setup:
-        cuda_version:
-          - "12.1.0"
-        python_version:
-          - "3.10"
-          - "3.11"
-
-  - block: "Build wheels full"
-
-  - label: "Build wheel - Python {{matrix.python_version}}, CUDA {{matrix.cuda_version}}"
-    agents:
-      queue: cpu_queue
-    commands:
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg CUDA_VERSION={{matrix.cuda_version}} --build-arg PYTHON_VERSION={{matrix.python_version}} --tag vllm-ci:build-image --target build --progress plain ."
-      - "mkdir artifacts"
-      - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      - "for f in artifacts/dist/*.whl; do mv -- \"$$f\" \"$${f/linux/manylinux1}\"; done"
-      - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/$BUILDKITE_COMMIT/"
-      - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/nightly/"
     matrix:
       setup:
         cuda_version:
           - "11.8.0"
           - "12.1.0"
-        python_version:
-          - "3.8"
-          - "3.9"
-          - "3.10"
-          - "3.11"
-      adjustments:
-      - with:
-          cuda_version: "12.1.0"
-          python_version: "3.10"
-        skip: true
-      - with:
-          cuda_version: "12.1.0"
-          python_version: "3.11"
-        skip: true

Dockerfile

@@ -104,15 +104,15 @@ RUN --mount=type=cache,target=/root/.cache/pip \
         && export SCCACHE_REGION=us-west-2 \
         && export CMAKE_BUILD_TYPE=Release \
         && sccache --show-stats \
-        && python3 setup.py bdist_wheel --dist-dir=dist \
+        && python3 setup.py bdist_wheel --dist-dir=dist --py-limited-api=cp38 \
         && sccache --show-stats; \
     fi
 
 ENV CCACHE_DIR=/root/.cache/ccache
 RUN --mount=type=cache,target=/root/.cache/ccache \
     --mount=type=cache,target=/root/.cache/pip \
     if [ "$USE_SCCACHE" != "1" ]; then \
-        python3 setup.py bdist_wheel --dist-dir=dist; \
+        python3 setup.py bdist_wheel --dist-dir=dist --py-limited-api=cp38; \
     fi
 
 # check the size of the wheel, we cannot upload wheels larger than 100MB