Skip to content

fix: Pin ARM version for VLLM FlashInfer #2358

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
zaristei wants to merge 4 commits into from
Closed

fix: Pin ARM version for VLLM FlashInfer #2358

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
99c662b
hotfix for flashinfer gb200 version
zaristei Aug 7, 2025
149c9da
Merge branch 'main' into zaristei/flashinfer-arm-hotfix
zaristei Aug 7, 2025
023bdae
adjust tag to accomodate flashinfer versioning typo
zaristei Aug 7, 2025
fe9c2d5
Merge branch 'main' into zaristei/flashinfer-arm-hotfix
dmitry-tokarev-nv Aug 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion container/Dockerfile.vllm
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@ ARG TORCH_BACKEND="cu128"
# Pinned to commit before https://github.com/deepseek-ai/DeepGEMM/pull/112 for DeepGEMM which seems to break on H100:
# "RuntimeError: Failed: CUDA runtime error csrc/jit/kernel_runtime.hpp:108 '98'"
ARG DEEPGEMM_REF="03d0be3"
ARG FLASHINF_REF="v0.2.8rc1"
ARG FLASHINF_REF="v0.2.8.rc1"

# Define general architecture ARGs for supporting both x86 and aarch64 builds.
# ARCH: Used for package suffixes (e.g., amd64, arm64)
Expand Down
14 changes: 5 additions & 9 deletions container/deps/vllm/install_vllm.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,14 +167,10 @@ python setup.py install


# Install Flash Infer
if [ "$ARCH" = "arm64" ]; then
uv pip install flashinfer-python
else
cd $INSTALLATION_DIR
git clone https://github.com/flashinfer-ai/flashinfer.git --recursive
cd flashinfer
git checkout $FLASHINF_REF
python -m pip install -v .
fi
cd $INSTALLATION_DIR
git clone https://github.com/flashinfer-ai/flashinfer.git --recursive
cd flashinfer
git checkout $FLASHINF_REF
uv pip install -v .

echo "vllm installation completed successfully"
Loading