Fix AsyncTP enabling#3869
Open
jenniew wants to merge 1 commit into
Open
Conversation
meta-cla
Bot
added
the
CLA Signed
|
|
Contributor
|
@jenniew pytorch/pytorch#172163, calling |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
After we set enable_async_tensor_parallel in torchtitan, it would run micro_pipeline_tp_pass to fuse_all_gather_matmul and fuse_matmul_reduce_scatter. fuse_all_gather_matmul and fuse_matmul_reduce_scatter would check is_symm_mem_enabled_for_group is True or False. If is_symm_mem_enabled_for_group is False, fuse_all_gather_matmul and fuse_matmul_reduce_scatter would not replace matched patterns with torch.ops.symm_mem.fused_all_gather_matmul and torch.ops.symm_mem.fused_scaled_matmul_reduce_scatter.
In torchtitan, we haven't enabled symm_mem_enabled_for_group during initialization, so the check would be fail, and symmetric fused APIs would not be used.
This PR is to enable symm_mem_enabled_for_group during the torchtitan trainner initialization.