Better support multidevice placement with stream.async.barrier
#19651
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
Barriers / transfers should have semantics that attempt to parallelize partitioning. If a value has a barrier placed it should divide partitions to avoid spaning behavior with cross device dependencies. Intermediate and ending transfers we want to place on the producing partition so that any produced operator ends by producing the value at the needed desetination For incoming transfers we place in the destination partition as these will not add a dependency on the incoming data.
benvanik
approved these changes
Jan 9, 2025
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Outdated suggestions cannot be applied.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.
Barriers / transfers should have semantics that attempt to parallelize partitioning. If a value has a barrier placed it should divide partitions to avoid spaning behavior with cross device dependencies.
Intermediate and ending transfers we want to place on the producing partition so that any produced operator ends by producing the value at the needed desetination
For incoming transfers we place in the destination partition as these will not add a dependency on the incoming data.