diff --git a/.buildkite/scripts/hardware_ci/run-gh200-test.sh b/.buildkite/scripts/hardware_ci/run-gh200-test.sh index 06e0f7af87..c0d1081f05 100644 --- a/.buildkite/scripts/hardware_ci/run-gh200-test.sh +++ b/.buildkite/scripts/hardware_ci/run-gh200-test.sh @@ -15,7 +15,6 @@ DOCKER_BUILDKIT=1 docker build . \ -t gh200-test \ --build-arg max_jobs=66 \ --build-arg nvcc_threads=2 \ - --build-arg RUN_WHEEL_CHECK=false \ --build-arg torch_cuda_arch_list="9.0+PTX" # Setup cleanup diff --git a/docker/Dockerfile b/docker/Dockerfile index c5d8f006c9..693b500b96 100644 --- a/docker/Dockerfile +++ b/docker/Dockerfile @@ -585,13 +585,14 @@ COPY --from=extensions-build /tmp/ep_kernels_workspace/dist /tmp/ep_kernels_work RUN sha256sum /tmp/ep_kernels_workspace/dist/*.whl \ > /tmp/ep_kernels_workspace/dist/wheels.sha256 -# Check the size of the wheel if RUN_WHEEL_CHECK is true +# Check the size of the CUDA 13 wheel uploaded to PyPI COPY .buildkite/check-wheel-size.py check-wheel-size.py # sync the default value with .buildkite/check-wheel-size.py ARG VLLM_MAX_SIZE_MB=500 ENV VLLM_MAX_SIZE_MB=$VLLM_MAX_SIZE_MB ARG RUN_WHEEL_CHECK=true -RUN if [ "$RUN_WHEEL_CHECK" = "true" ]; then \ +RUN if [ "$RUN_WHEEL_CHECK" = "true" ] && \ + [ "${CUDA_VERSION%%.*}" = "13" ]; then \ python3 check-wheel-size.py dist; \ else \ echo "Skipping wheel size check."; \ diff --git a/docs/getting_started/installation/gpu.cuda.inc.md b/docs/getting_started/installation/gpu.cuda.inc.md index 0e86c0e604..7755ad6278 100644 --- a/docs/getting_started/installation/gpu.cuda.inc.md +++ b/docs/getting_started/installation/gpu.cuda.inc.md @@ -384,8 +384,7 @@ A docker container can be built for aarch64 systems such as the Nvidia Grace-Hop -t vllm/vllm-gh200-openai:latest \ --build-arg max_jobs=66 \ --build-arg nvcc_threads=2 \ - --build-arg torch_cuda_arch_list="9.0 10.0+PTX" \ - --build-arg RUN_WHEEL_CHECK=false + --build-arg torch_cuda_arch_list="9.0 10.0+PTX" ``` For (G)B300, we recommend using CUDA 13, as shown in the following command. @@ -398,7 +397,6 @@ For (G)B300, we recommend using CUDA 13, as shown in the following command. --build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.2-devel-ubuntu22.04 \ --build-arg max_jobs=256 \ --build-arg nvcc_threads=2 \ - --build-arg RUN_WHEEL_CHECK=false \ --build-arg torch_cuda_arch_list='9.0 10.0+PTX' \ --platform "linux/arm64" \ --tag vllm/vllm-gb300-openai:latest \