[WideEP] Update NCCL to 2.30.7 to enable DeepEPv2 in the vllm/vllm-openai image (#45321)

Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Signed-off-by: Tyler Michael Smith <tyler@vllm.ai>
Signed-off-by: Tyler Michael Smith <tyler@tylermsmith.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Ilya Markov <ilmarkov@users.noreply.github.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Codex <noreply@openai.com>
This commit is contained in:
Tyler Michael Smith
2026-07-24 13:00:02 -07:00
committed by GitHub
co-authored by Claude Ilya Markov OpenAI Codex Codex
parent e222c33f2f
commit 5d8e90a966
13 changed files with 338 additions and 78 deletions
+25 -8
View File
@@ -25,6 +25,10 @@
ARG CUDA_VERSION=13.0.3
ARG PYTHON_VERSION=3.12
ARG UBUNTU_VERSION=22.04
# DeepEPv2 requires NCCL >= 2.30.4 (GIN backend).
# This version is only used for CUDA 13+ builds; CUDA 12 falls back to
# the default NCCL version shipped with the base image.
ARG NCCL_VERSION=2.30.7
# By parameterizing the base images, we allow third-party to use their own
# base images. One use case is hermetic builds with base images stored in
@@ -477,10 +481,17 @@ WORKDIR /workspace
# Build DeepEP wheels
COPY tools/ep_kernels/install_python_libraries.sh /tmp/install_python_libraries.sh
# Defaults moved here from tools/ep_kernels/install_python_libraries.sh for centralized version management
ARG DEEPEP_COMMIT_HASH=73b6ea4
ARG DEEPEP_COMMIT_HASH=d4f41e4e93
ARG NVSHMEM_VER
ARG NCCL_VERSION
RUN --mount=type=cache,target=/opt/uv/cache \
mkdir -p /tmp/ep_kernels_workspace/dist && \
CUDA_MAJOR=$(echo $CUDA_VERSION | cut -d. -f1) && \
if [ "$CUDA_MAJOR" -ge 13 ] && [ -n "$NCCL_VERSION" ]; then \
echo "nvidia-nccl-cu${CUDA_MAJOR}==${NCCL_VERSION}" \
> /tmp/nccl-override.txt && \
export UV_OVERRIDE=/tmp/nccl-override.txt; \
fi && \
export TORCH_CUDA_ARCH_LIST='9.0a 10.0a' && \
/tmp/install_python_libraries.sh \
--workspace /tmp/ep_kernels_workspace \
@@ -644,6 +655,7 @@ FROM ${FINAL_BASE_IMAGE} AS vllm-base
ARG CUDA_VERSION
ARG PYTHON_VERSION
ARG NCCL_VERSION
ARG DEADSNAKES_MIRROR_URL
ARG DEADSNAKES_GPGKEY_URL
ARG GET_PIP_URL
@@ -696,7 +708,6 @@ RUN apt-get update -y \
# Install CUDA development tools for runtime JIT compilation
# (FlashInfer, DeepGEMM, EP kernels all require compilation at runtime)
RUN CUDA_VERSION_DASH=$(echo $CUDA_VERSION | cut -d. -f1,2 | tr '.' '-') && \
CUDA_VERSION_SHORT=$(echo $CUDA_VERSION | cut -d. -f1,2) && \
apt-get update -y && \
apt-get install -y --no-install-recommends --allow-change-held-packages \
cuda-nvcc-${CUDA_VERSION_DASH} \
@@ -709,12 +720,6 @@ RUN CUDA_VERSION_DASH=$(echo $CUDA_VERSION | cut -d. -f1,2 | tr '.' '-') && \
libnuma-dev \
# numactl CLI for NUMA binding at runtime
numactl && \
# Fixes nccl_allocator requiring nccl.h at runtime
# https://github.com/vllm-project/vllm/blob/1336a1ea244fa8bfd7e72751cabbdb5b68a0c11a/vllm/distributed/device_communicators/pynccl_allocator.py#L22
# NCCL packages don't use the cuda-MAJOR-MINOR naming convention,
# so we pin the version to match our CUDA version
NCCL_VER=$(apt-cache madison libnccl-dev | grep "+cuda${CUDA_VERSION_SHORT}" | head -1 | awk -F'|' '{gsub(/^ +| +$/, "", $2); print $2}') && \
apt-get install -y --no-install-recommends --allow-change-held-packages libnccl-dev=${NCCL_VER} libnccl2=${NCCL_VER} && \
rm -rf /var/lib/apt/lists/*
# Install uv for faster pip installs
@@ -734,6 +739,18 @@ RUN mkdir -p "${UV_PYTHON_INSTALL_DIR}" "${UV_CACHE_DIR}" \
&& chgrp -R 0 /opt/uv \
&& chmod -R g+rwX,a+rX /opt/uv
# DeepEPv2 GIN requires NCCL >= 2.30.4 at both compile and runtime. torch pins
# an older version as a transitive dep; this override forces uv to use our
# pinned version whenever nvidia-nccl-cu* is resolved. Empty on CUDA 12 (no-op).
RUN CUDA_MAJOR=$(echo $CUDA_VERSION | cut -d. -f1) && \
if [ "$CUDA_MAJOR" -ge 13 ]; then \
echo "nvidia-nccl-cu${CUDA_MAJOR}==${NCCL_VERSION}" \
> /etc/uv-overrides.txt; \
else \
touch /etc/uv-overrides.txt; \
fi
ENV UV_OVERRIDE=/etc/uv-overrides.txt
# ----------------------------------------------------------------------
# Non-root support (opt-in)
# ----------------------------------------------------------------------