Categorize DDP comm-hook bucket all-reduce and TorchRec comm ops as NET#19
Closed
Chenguang-Zhu wants to merge 1 commit into
Closed
Categorize DDP comm-hook bucket all-reduce and TorchRec comm ops as NET#19Chenguang-Zhu wants to merge 1 commit into
Chenguang-Zhu wants to merge 1 commit into
Conversation
Summary: Add four new NET-domain rules to the categorization rule table: 1. `c10d::Reducer::all_reduce_bucket` — DDP bucket all-reduce. 2. `c10d::PythonCommHook::runHook` — Python comm-hook dispatch. 3. Filename contains `torch/distributed/algorithms/ddp_comm_hooks/` — any comm-hook implementation. 4. Filename contains `torchrec/distributed/comm_ops.py` — TorchRec distributed comm primitives (alltoall_pooled, etc.). Order-of-rules detail: these rules are inserted **after** the autograd-engine BACKWARD rules — so an autograd-bound bucket allocation still goes to BACKWARD — but **before** the broad `"backward" in frame.name` substring fallback, so a comm-hook frame in an otherwise generic backward stack still wins. Differential Revision: D103992608
|
@Chenguang-Zhu has exported this pull request. If you are a Meta employee, you can view the originating Diff in D103992608. |
|
This pull request has been merged in a931358. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary:
Add four new NET-domain rules to the categorization rule table:
c10d::Reducer::all_reduce_bucket— DDP bucket all-reduce.c10d::PythonCommHook::runHook— Python comm-hook dispatch.torch/distributed/algorithms/ddp_comm_hooks/— any comm-hook implementation.
torchrec/distributed/comm_ops.py— TorchRecdistributed comm primitives (alltoall_pooled, etc.).
Order-of-rules detail: these rules are inserted after the
autograd-engine BACKWARD rules — so an autograd-bound
bucket allocation still goes to BACKWARD — but before the broad
"backward" in frame.namesubstring fallback, so a comm-hookframe in an otherwise generic backward stack still wins.
Differential Revision: D103992608