Skip to content
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.

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

[BugFix] Add stream synchronize for nccl migration backend #20

Merged
merged 2 commits into from
Aug 29, 2024

Conversation

KuilongCui
Copy link
Contributor

@KuilongCui KuilongCui commented Aug 28, 2024

  1. bugfix for migration stream sync problem
  2. add cupy install tool

after dst worker recv data from col.recv, it need to copy data to self.gpu_cache. Before copying, it must wait cuda stream in col.recv. As nccl migration backend use cupy cuda stream, in this pr change migration stream to cupy stream for synchronize.

@s5u13b s5u13b changed the title [CI] add test for migration backend and worker [CI] Add test for migration backend and worker Aug 28, 2024
llumnix/backends/vllm/worker.py Outdated Show resolved Hide resolved
tests/backends/vllm/test_migration_backend.py Outdated Show resolved Hide resolved
tests/backends/vllm/test_worker.py Outdated Show resolved Hide resolved
@KuilongCui KuilongCui changed the title [CI] Add test for migration backend and worker [BugFix] Add stream synchronize for nccl migration backend Aug 28, 2024
@KuilongCui KuilongCui merged commit be0d326 into main Aug 29, 2024
4 checks passed
@KuilongCui KuilongCui deleted the worker_test branch September 3, 2024 07:19
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants