59db6a4285
Signed-off-by: tzhouam <tzhouam@connect.ust.hk> Signed-off-by: Gao Han <hgaoaf@connect.ust.hk> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-authored-by: Gao Han <hgaoaf@connect.ust.hk>
54 lines
2.0 KiB
Docker
54 lines
2.0 KiB
Docker
# Target vLLM version: v0.30.0. Keep VLLM_VERSION aligned with the target vLLM
|
|
# release on every rebase.
|
|
#
|
|
# This file layers vLLM-Omni on top of the published upstream XPU image, which
|
|
# already ships a self-consistent XPU runtime (torch-xpu, oneCCL/UCX+NIXL, the
|
|
# pinned UMD/compute-runtime stack, and vLLM itself) built from
|
|
# https://github.com/vllm-project/vllm/blob/main/docker/Dockerfile.xpu. Docker
|
|
# pulls that base automatically, so there is nothing to build here beyond the
|
|
# Omni layers. Mirrors how docker/Dockerfile.ci consumes vllm/vllm-openai.
|
|
#
|
|
# If the tag does not exist yet (targeting an unreleased vLLM commit), build the
|
|
# base from upstream's own Dockerfile and point VLLM_BASE at the result:
|
|
#
|
|
# git clone --depth 1 --branch $VLLM_VERSION https://github.com/vllm-project/vllm /tmp/vllm
|
|
# docker build -t vllm-openai-xpu:local --target vllm-openai \
|
|
# -f /tmp/vllm/docker/Dockerfile.xpu /tmp/vllm
|
|
# docker build -f docker/Dockerfile.xpu --build-arg VLLM_BASE=vllm-openai-xpu:local .
|
|
#
|
|
# .buildkite/intel/scripts/run-xpu-test.sh automates that pull-then-build fallback.
|
|
ARG VLLM_BASE_IMAGE=vllm/vllm-openai-xpu
|
|
ARG VLLM_VERSION=v0.30.0
|
|
ARG VLLM_BASE=${VLLM_BASE_IMAGE}:${VLLM_VERSION}
|
|
|
|
FROM ${VLLM_BASE} AS vllm-omni
|
|
|
|
# The base image ships ENTRYPOINT ["vllm", "serve"]; clear it so this stage stays
|
|
# usable as a plain shell/test container regardless of which VLLM_BASE was used.
|
|
ENTRYPOINT []
|
|
CMD ["/bin/bash"]
|
|
|
|
RUN apt-get update && \
|
|
apt-get install -y espeak-ng ffmpeg git jq && \
|
|
apt-get clean && \
|
|
rm -rf /var/lib/apt/lists/*
|
|
|
|
WORKDIR /workspace/vllm-omni
|
|
COPY . .
|
|
|
|
ENV VLLM_OMNI_TARGET_DEVICE=xpu
|
|
# Keep the upstream oneCCL packages paired with the XPU torch runtime.
|
|
RUN --mount=type=cache,target=/root/.cache/uv \
|
|
uv pip install --no-build-isolation \
|
|
".[dev]" && \
|
|
uv pip uninstall triton && \
|
|
uv pip install triton-xpu==3.7.2 --reinstall
|
|
|
|
FROM vllm-omni AS vllm-omni-openai
|
|
|
|
RUN ln -sf /usr/bin/python3 /usr/bin/python
|
|
|
|
ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
|
|
|
|
ENTRYPOINT ["vllm", "serve", "--omni"]
|