vibe coded bmm sharder - #3399
Closed
suyoggupta wants to merge 1 commit into
Closed
Conversation
Signed-off-by: Suyog Gupta <suyogg@nvidia.com>
lucaslie
reviewed
Apr 9, 2025
Collaborator
Author
We'd need to add all_reduce if we're sharding a dimension that is reduced. The current sharding strategy is to shard the 0th dimension, which happens to be of size num_experts. So the strategy of bmm1->all_gather->bmm2->all_gather works. The first all_gather can be optimized away if all ops between bmm1 and bmm2 are elementwise ops. Does that make sense? |
Collaborator
Author
|
open this one instead. nv-auto-deploy#1. |
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
No description provided.