Apply forwarding in pjit linearization rule to avoid intermediate copies #27735
Conversation
This file contains 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
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Outdated suggestions cannot be applied.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.
Many linearization rules require forwarding inputs as residuals and we should avoid instantiating copies of those arguments. We have discussed adding forwarding information to the signature of the linearize rules but, in the meantime, it's probably useful to explicitly apply our input forwarding logic when linearizing pjit.
This PR also adds a DCE pass to linearize_jaxpr to prune some other unused residuals.
These two changes fix some OOM crashes with real workloads when using direct linearize for AD.