A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 4, 15:00 UTC
[Docker] Install numactl CLI in CUDA runtime image (#41032)
Signed-off-by: Zhewen Li <zhewenli@inferact.ai> Co-authored-by: Zhewen Li <zhewenli@inferact.ai> Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Zhewen Li committed
Apr 27, 2026 at 10:58 UTC
6232fb4b66b42c8e5f4ef1cc4c5163442cc99208
1 file changed
+3
-1
docker/Dockerfile
+3
-1
@@ -540,7 +540,9 @@ RUN CUDA_VERSION_DASH=$(echo $CUDA_VERSION | cut -d. -f1,2 | tr '.' '-') && \
540
libcurand-dev-${CUDA_VERSION_DASH} \
541
libcublas-dev-${CUDA_VERSION_DASH} \
542
# Required by fastsafetensors (fixes #20384)
543
- libnuma-dev && \
543
+ libnuma-dev \
544
+ # numactl CLI for NUMA binding at runtime
545
+ numactl && \
546
# Fixes nccl_allocator requiring nccl.h at runtime
547
# https://github.com/vllm-project/vllm/blob/1336a1ea244fa8bfd7e72751cabbdb5b68a0c11a/vllm/distributed/device_communicators/pynccl_allocator.py#L22
548
# NCCL packages don't use the cuda-MAJOR-MINOR naming convention,