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

Possible RedistributeGPU optimizations #4179

Open
atmyers opened this issue Oct 2, 2024 · 0 comments
Open

Possible RedistributeGPU optimizations #4179

atmyers opened this issue Oct 2, 2024 · 0 comments
Assignees
@atmyers
Labels
GPU performance

Comments

@atmyers
Copy link
Member

atmyers commented Oct 2, 2024

@AlexanderSinn has suggested a few optimizations to try for RedistributeGPU:

  1. When doing the handshake step, we could use new single-sided communication features in MPI-3 to avoid doing MPI_Reduce_Scatter.
  2. When doing the above, we communicate the number of bytes of data as well as the connectivity.
  3. We could reduce the number of MPI_Send calls by using MPI_Probe to get the number of bytes.
  4. We could use MPI_isend instead of MPI_Send.
@atmyers atmyers self-assigned this Oct 2, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@atmyers atmyers

Labels
GPU performance
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@ax3l @atmyers