Repository navigation
Conversation
…s, nvtx tags - copy_gradients_to_partial_buffer: replace the per-parameter .copy_() loop with torch._foreach_copy_() over batched destination/source lists. - DBuffer.get_local_tensor: cache narrow/view results keyed by tensor index; invalidate on storage resize (release/reallocate). Views are stable within a buffer lifetime since storage is preserved on resize. - FsdpModule: add nvtx ranges for the all-gather (unshard) and reduce-grad phases so M-FSDP v2 comm is visible in nsys traces. Signed-off-by: Jianbin Chang <jianbinc@nvidia.com>
shjwudp
force-pushed
the
split/pr-e-cpu-overhead
branch
from
August 12, 2026 18:11
f6a3dd4 to
6afab9d
Compare
shjwudp
changed the title
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Reduce M-FSDP v2 CPU overhead in the backward gradient path and make M-FSDP v2 communication visible in nsys traces.
Changes
copy_gradients_to_partial_buffer: replace the per-parameter.copy_()loop withtorch._foreach_copy_()over batched destination/source lists — one kernel launch instead of O(params) launches.DBuffer.get_local_tensor: cache narrow/view results keyed by tensor index; invalidate on storage resize (release/reallocate). Views are stable within a buffer lifetime since storage is preserved on resize.FsdpModule: add nvtx ranges for the all-gather (unshard) and reduce-grad phases so M-FSDP v2 comm is visible in nsys traces.Notes
Split out of PR #6197; standalone (based on main, no dependency on the overlap PR).
Test plan