[Autograd] expand_as instead of clone to get AccumulateGrad#96356
Closed
awgu wants to merge 1 commit intogh/awgu/345/basefrom
Closed
[Autograd] expand_as instead of clone to get AccumulateGrad#96356awgu wants to merge 1 commit intogh/awgu/345/basefrom
expand_as instead of clone to get AccumulateGrad#96356awgu wants to merge 1 commit intogh/awgu/345/basefrom
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/96356
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f71c335: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
awgu
added
topic: not user facing
Contributor
|
I don't think there's any reason to clone, good catch! |
soulitzer
approved these changes
Mar 9, 2023
Collaborator
Author
|
@pytorchbot merge |
Collaborator
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
This was referenced Mar 10, 2023
cyyever
pushed a commit
to cyyever/pytorch_private
that referenced
this pull request
Mar 12, 2023
…6356) This PR makes a minor change to the multi-grad hook implementation. This should decrease peak memory since we avoid one `clone()` per tensor passed into the multi-grad hook. Let me know if there are technical reasons why we need to clone. If so, is there a way for some use cases to not clone? Before with `clone()`:  After with `expand_as()` -- no more "Memcpy DtoD" kernels:  Pull Request resolved: pytorch/pytorch#96356 Approved by: https://github.com/soulitzer
ydwu4
added a commit
to ydwu4/pytorch
that referenced
this pull request
Mar 13, 2023
…torch#96356) This PR makes a minor change to the multi-grad hook implementation. This should decrease peak memory since we avoid one `clone()` per tensor passed into the multi-grad hook. Let me know if there are technical reasons why we need to clone. If so, is there a way for some use cases to not clone? Before with `clone()`:  After with `expand_as()` -- no more "Memcpy DtoD" kernels:  Pull Request resolved: pytorch#96356 Approved by: https://github.com/soulitzer
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
Stack from ghstack (oldest at bottom):
expand_asinstead ofcloneto getAccumulateGrad#96356This PR makes a minor change to the multi-grad hook implementation. This should decrease peak memory since we avoid one
clone()per tensor passed into the multi-grad hook. Let me know if there are technical reasons why we need to clone. If so, is there a way for some use cases to not clone?Before with
clone():After with
expand_as()-- no more "Memcpy DtoD" kernels:cc @ezyang @albanD @zou3519 @gqchen @pearu @nikitaved @soulitzer @lezcano @Varal7