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.

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

Jump to bottom

Question about dynamic SMs allocation #2

Open
initzhang opened this issue Oct 25, 2024 · 0 comments
Open

Question about dynamic SMs allocation #2

initzhang opened this issue Oct 25, 2024 · 0 comments

Comments

@initzhang
Copy link

Hello Team,

Thank you for your outstanding work!

I have a question regarding the SM allocation in MuxServe. In your paper, you mentioned:

... parallel runtime dynamically assigns SMs to each job at runtime rather than statically allocating ...

Does this imply that the SM configuration for each process is adjusted on demand during the inference of incoming requests? If that’s the case, could you please explain how this is accomplished? Is it done by setting the CUDA set_active_thread_percentage <PID> <percentage> from time to time for each process?
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@initzhang