Repository navigation
[core][rdt] Add Axon tensor transport backends - #67186
Draft
Florentino-Ding wants to merge 2 commits into
Draft
Florentino-Ding wants to merge 2 commits into
Florentino-Ding wants to merge 2 commits into
Conversation
Register Axon backends for NumPy arrays and PyTorch CPU/CUDA tensors. Use receiver-initiated transfers without scheduling Ray sender tasks. Integrate transport cleanup with worker shutdown and normal process exit. Release completed-transfer actor references held by the failure monitor so actors and their resources can be reclaimed automatically. Add transfer and lifecycle tests, documentation, and optional Axon coverage in the single-node RDT microbenchmark. Signed-off-by: FlorentinoDing <dingluran@outlook.com>
This branch has not been deployed
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Register Axon backends for NumPy arrays and PyTorch CPU/CUDA tensors. Use receiver-initiated transfers without scheduling Ray sender tasks.
Integrate transport cleanup with worker shutdown and normal process exit. Release completed-transfer actor references held by the failure monitor so actors and their resources can be reclaimed automatically.
Add transfer and lifecycle tests, documentation, and optional Axon coverage in the single-node RDT microbenchmark.
Related issues
Related to #66890