biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[ROCm][Feature] Enable Pipeline Parallelism with Ray Compiled Graph on ROCm (#24275)

Browse Source 
Signed-off-by: charlifu <charlifu@amd.com>
This commit is contained in:
Charlie Fu
2025-09-09 18:27:35 -05:00
committed by GitHub
parent fb1a8f932a
commit 73e688cb79
3 changed files with 16 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
docker/Dockerfile.rocm
View File

@@ -104,6 +104,7 @@ COPY --from=export_vllm /examples ${COMMON_WORKDIR}/vllm/examples
COPY --from=export_vllm /docker ${COMMON_WORKDIR}/vllm/docker
ENV RAY_EXPERIMENTAL_NOSET_ROCR_VISIBLE_DEVICES=1
ENV RAY_EXPERIMENTAL_NOSET_HIP_VISIBLE_DEVICES=1
ENV TOKENIZERS_PARALLELISM=false
# ENV that can improve safe tensor loading, and end-to-end time