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

Dont redirect to another allreduce algorithm. #12552

Conversation

bosilca
Copy link
Member

@bosilca bosilca commented May 16, 2024

The decision to switch to a different algorithm based on count does not belong to the base algorithms, it should be done in the upper level decision functions.

We need to validate the performance for small allreduce at mid-range scale to make sure this does not negatively affect the allreduce.

Fixes #12550.

@wenduwan
Copy link
Contributor

@bosilca Could you please resolve the conflict?

The decision to switch to a different algorithm based on count
does not belong to the base algorithms, it should be done in the
upper level decision functions.

Signed-off-by: George Bosilca <[email protected]>
@bosilca bosilca force-pushed the fix/no_linear_allreduce_in_base_allreduce_scatter_allgather branch from ee35f84 to 7a1a4e6 Compare May 23, 2024 16:24
@bosilca bosilca merged commit 75f5045 into open-mpi:main May 28, 2024
14 checks passed
@bosilca bosilca deleted the fix/no_linear_allreduce_in_base_allreduce_scatter_allgather branch May 28, 2024 15:13
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Base Allreduce Algorithm Selection/Performance Issue
2 participants