Skip to content

Conversation

@mrava87
Copy link
Contributor

@mrava87 mrava87 commented Jul 1, 2025

This is a WIP PR to add tests for various nccl utility routines in pylops_mpi.utils._nccl for primitive communication calls.

The need for this tests stems from the realization that if we pass send_buf of different size to nccl_allgather we get a deadlock... so we should make sure to check and raise an error before this happens.

Note that the last test is currently commented out because mpiexec will hang if running this test - I believe this is due to the fact that nccl_comm.allGather does not cope with send_buf of different size between ranks

@mrava87 mrava87 requested a review from tharittk July 1, 2025 08:52
@mrava87
Copy link
Contributor Author

mrava87 commented Jul 3, 2025

Superseded by #148

@mrava87 mrava87 closed this Jul 3, 2025
@mrava87 mrava87 deleted the test-ncclutils branch July 3, 2025 16:53
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.

1 participant