Skip to content

chore: fix build nvlink allocator with nvcc #669

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
wants to merge 1 commit into from
Closed

chore: fix build nvlink allocator with nvcc #669

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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 mooncake-transfer-engine/nvlink-allocator/build.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@ CPP_FILE=$(dirname $(readlink -f $0))/nvlink_allocator.cpp # get cpp file path,

# Use nvcc or g++ based on flag
if [ "$USE_NVCC" = true ]; then
nvcc "$CPP_FILE" -o "$OUTPUT_DIR/nvlink_allocator.so" -shared -Xcompiler -fPIC -lcuda -I/usr/local/cuda/include
nvcc "$CPP_FILE" -o "$OUTPUT_DIR/nvlink_allocator.so" -shared -Xcompiler -fPIC -I/usr/local/cuda/include
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since we have CUDA in the env now, should we just use g++? The early compiled result of nvcc is not functioning well, not really sure about this version.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we didn't have full cuda support in ci. and we tried to remove cuda refs, which broke the release (the pr i reverted). so change this to make ci happy.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

But the release workload in https://github.com/kvcache-ai/Mooncake/blob/main/.github/workflows/release.yaml#L71C1-L75C90 will use nvcc and produce a nvlink_allocator.so which is not functioning as expected.

else
g++ "$CPP_FILE" -o "$OUTPUT_DIR/nvlink_allocator.so" --shared -fPIC -lcuda -I/usr/local/cuda/include
fi
Expand Down
Loading