delete unimplemented broadcast transport path [need discussion]#24
Open
TianyeGGBond wants to merge 1 commit into
Open
delete unimplemented broadcast transport path [need discussion]#24TianyeGGBond wants to merge 1 commit into
TianyeGGBond wants to merge 1 commit into
Conversation
sync_selected_workers advertised a two-path transport via a public broadcast_local_ranks kwarg, but any non-empty value immediately raised NotImplementedError because the cache-owner sender-side NCCL path was never wired. In practice only the cpu_serialize path runs. Remove the broadcast kwarg, validation, and NotImplementedError so the service exposes only the path it actually implements. The wire plan is unchanged: broadcast_local_ranks is still emitted (empty), comm_ranks are all 0, and world_size is 1 — exactly the values the cpu_serialize-only path already produced, so the MILES run_sync_session contract is untouched. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Context
MilesModelUpdateService.sync_selected_workerswas named/documented as atwo-path weight transport (cpu_serialize + NCCL broadcast) and exposed a
public
broadcast_local_rankskwarg. But any non-empty value immediatelyraised
NotImplementedError: the cache-owner sender-side NCCL path(
init_process_group+ per-bucketdist.broadcast) was never wired — onlythe receiver-side fan-out existed.
So the only path that ever runs is cpu_serialize. The broadcast kwarg,
subset validation, NCCL rank assignment, and
world_sizemath are deadcode that makes the surface look like it offers something it doesn't.
Change
pipeline/miles_model_update_service.py:broadcast_local_rankskwarg fromsync_selected_workers,along with the subset validation and the
NotImplementedErrorbranch.Every target now goes over cpu_serialize.
broadcast_setparameter threaded through_run_atomic_unitand
_build_plan, and the broadcast field from the done-log line._build_plan, replace the broadcast NCCL rank loop /world_sizecalculation with single-path constants.
Wire compatibility
The wire plan shape is intentionally unchanged.
broadcast_local_ranksisstill present (empty),
comm_ranksare all 0, andworld_sizeis 1 — theexact values the cpu_serialize-only path already produced. The MILES
run_sync_sessionplan-key validator sees no difference.Net: +20 / -60 lines, no behavior change on the live path.