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

fix: add temporary patch for CUDA graphs estimation

Browse Source
This commit is contained in:
Eugene Rakhmatulin
2026-03-08 22:43:41 -07:00
parent 9724619dbd
commit 63b2a8dbed
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
Dockerfile
View File

@@ -194,6 +194,8 @@ RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
# TEMPORARY PATCH for broken vLLM build (unguarded Hopper code) - reverting PR #34758 and #34302
RUN curl -L https://patch-diff.githubusercontent.com/raw/vllm-project/vllm/pull/34758.diff | patch -p1 -R || echo "Cannot revert PR #34758, skipping"
RUN curl -L https://patch-diff.githubusercontent.com/raw/vllm-project/vllm/pull/34302.diff | patch -p1 -R || echo "Cannot revert PR #34302, skipping"
# TEMPORARY PATCH for CUDA graphs estimation
RUN curl -L https://patch-diff.githubusercontent.com/raw/vllm-project/vllm/pull/36416.diff | patch -p1 || echo "Cannot apply PR #36416, skipping"
# Final Compilation
RUN --mount=type=cache,id=ccache,target=/root/.ccache \