[CPU] Experimentally enable Triton and MRV2 (#43225)

Signed-off-by: jiang1.li <jiang1.li@intel.com>
This commit is contained in:
Li, Jiang
2026-05-22 01:48:17 -07:00
committed by GitHub
parent 2380bfc210
commit 65b7a812a2
11 changed files with 169 additions and 13 deletions
+5 -6
View File
@@ -27,11 +27,14 @@ WORKDIR /workspace
ARG PYTHON_VERSION=3.12
ARG PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu"
ARG max_jobs=32
ENV MAX_JOBS=${max_jobs}
# Install minimal dependencies and uv
RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
--mount=type=cache,target=/var/lib/apt,sharing=locked \
apt-get update -y \
&& apt-get install -y --no-install-recommends sudo ccache git curl wget ca-certificates \
&& apt-get install -y --no-install-recommends sudo ccache git curl wget ca-certificates zlib1g-dev \
gcc-12 g++-12 libtcmalloc-minimal4 libnuma-dev ffmpeg libsm6 libxext6 libgl1 jq lsof make xz-utils \
&& update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-12 10 --slave /usr/bin/g++ g++ /usr/bin/g++-12 \
&& curl -LsSf https://astral.sh/uv/install.sh | sh
@@ -123,9 +126,6 @@ RUN --mount=type=cache,target=/root/.cargo/registry \
######################### BUILD IMAGE #########################
FROM base AS vllm-build
ARG max_jobs=32
ENV MAX_JOBS=${max_jobs}
ARG GIT_REPO_CHECK=0
# Support for cross-compilation with x86 ISA including AVX2 and AVX512: docker build --build-arg VLLM_CPU_X86="true" ...
ARG VLLM_CPU_X86=0
@@ -257,8 +257,7 @@ WORKDIR /vllm-workspace
RUN --mount=type=cache,target=/root/.cache/uv \
--mount=type=cache,target=/root/.cache/ccache \
--mount=type=bind,from=vllm-build,src=/vllm-workspace/dist,target=dist \
uv pip install dist/*.whl && \
uv pip install "vllm[audio]"
uv pip install "$(realpath dist/*.whl)[audio,triton-cpu]"
# Add labels to document build configuration
LABEL org.opencontainers.image.title="vLLM CPU"