Make CUDACachingAllocator::recordStream() a no-op on null ptrs #20658
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
pytorchbot
added
module: cuda
mrshenli
added
the
triaged
Contributor
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@mrshenli has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
Contributor
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@mrshenli has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
colesbury
approved these changes
May 17, 2019
Contributor
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #20651
Communication collectives in
torch.distributedcallCUDACachingAllocator::recordStream()on input and output tensors to prevent their memory blocks being freed too early.CUDACachingAllocatoruses tensor's data pointer to track memory blocks, which does not accept null pointers. However, empty tensor'sstorage().data()might be null. In this case, as there is no associated memory block for the empty tensor, it should be fine to makerecordStream()a no-op.Tests only cover
broadcastempty tensors for GLOO backend, because GLOO does not support empty inputs (pytorch/gloo/issues/179). It can be addressed in eitherProcessGroupGlooor GLOO itself. Will add more tests when that gap is filled.