Optimize shard_dim_alltoall to use alltoall_single #148868
Closed
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
as titled, previously the shard_dim_alltoall uses `all_to_all`, which essentially could incur lots of copies if the tensor become non-contiguous during splits, and alltoall itself also incur copies This PR uses alltoall_single instead, so that we could minimize tensor copies. tested on all the shard dim change tests and it works properly
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/148868
Note: Links to docs will display an error until the docs builds have been completed. ❌ 4 New Failures, 1 Pending, 5 Unrelated FailuresAs of commit c63b4a2 with merge base d789c22 ( NEW FAILURES - The following jobs have failed:
FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
UNSTABLE - The following jobs are marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
CI failures are not related to the PR |
tianyu-l
reviewed
Mar 10, 2025
Comment on lines
+555
to
+557
| std::vector<int64_t> out_split_sizes; | ||
| std::vector<int64_t> in_split_sizes; | ||
| c10d::AllToAllOptions opts; |
There was a problem hiding this comment.
these are optional, so don't have to be fed in alltoall_base?
There was a problem hiding this comment.
They are required by the alltoall_base APIs :(
|
@pytorchbot merge -i "failures are not related to the PR" |
|
❌ 🤖 pytorchbot command failed: Try |
|
@pytorchbot merge -i |
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Outdated suggestions cannot be applied.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.
as titled, previously the shard_dim_alltoall uses
all_to_all, which essentially could incur lots of copies if the tensor become non-contiguous during splits, and alltoall itself also incur copiesThis PR uses alltoall_single instead, so that we could minimize tensor copies.
tested on all the shard dim change tests and it works properly:
Fixes #ISSUE_NUMBER
cc @H-Huang @awgu @kwen2501 @fegin @fduwjj @wz337 @wconstab @d4l3k @c-p-i-o