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

Fix comm ready check in ncclIbIsend and ncclIbIrecv #1489

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

WWeiOne
Copy link

@WWeiOne WWeiOne commented Oct 19, 2024

ncclIbIsend & ncclIbIrecv contains redundant checks for comm->base.ready. These checks were made redundant due to a change in commit 5d3ab08, which removed a call to ncclSendCheck that might alter the readiness state of the communicator. The second check for comm->base.ready now will never executes.

Commit 5d3ab08 Feb 27 2023

  if (comm->ready == 0) NCCLCHECK(ncclSendCheck(comm));
  if (comm->ready == 0) { *request = NULL; return ncclSuccess; }

Also, the *request = NULL; should be kept, as none request is successfully added into the fifo slot, it's safer to assign NULL and consistent with the check if (sub->requests[buffSlot] != NULL) in sendProxyProgress.

@kiskra-nvidia
Copy link
Member

Thank you for your submission! We'll consider it for a future release (it missed the 2.24 window).

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.

2 participants