software-engineering/spark-vllm-docker
2
0
Fork 0
Code Issues Actions Packages Projects Releases Wiki Activity

moved deps buster

Browse Source
This commit is contained in:
Eugene Rakhmatulin
2026-01-30 15:25:54 -08:00
parent 008af21383
commit 518dc0108b
1 changed files with 5 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
Dockerfile
View File

@@ -58,9 +58,7 @@ WORKDIR $VLLM_BASE_DIR
ENV TORCH_CUDA_ARCH_LIST=12.1a ENV TORCH_CUDA_ARCH_LIST=12.1a
ENV TRITON_PTXAS_PATH=/usr/local/cuda/bin/ptxas ENV TRITON_PTXAS_PATH=/usr/local/cuda/bin/ptxas
# --- CACHE BUSTER ---
# Change this argument to force a re-download of FlashInfer
ARG CACHEBUST_DEPS=1
# 3. Install Python Dependencies with Cache Mounts # 3. Install Python Dependencies with Cache Mounts
# Using --mount=type=cache ensures that even if this layer invalidates, # Using --mount=type=cache ensures that even if this layer invalidates,
@@ -109,6 +107,10 @@ ENV FLASHINFER_CUDA_ARCH_LIST="12.1f"
WORKDIR $VLLM_BASE_DIR WORKDIR $VLLM_BASE_DIR
ARG FLASHINFER_REF=main ARG FLASHINFER_REF=main
# --- CACHE BUSTER ---
# Change this argument to force a re-download of FlashInfer
ARG CACHEBUST_DEPS=1
RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \ RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
uv pip install nvidia-nvshmem-cu13 "apache-tvm-ffi<0.2" uv pip install nvidia-nvshmem-cu13 "apache-tvm-ffi<0.2"