[megatron] Make FileSystemReader patch idempotent - #10034
Merged
tastelikefeet merged 1 commit intoSep 4, 2026
Merged
Conversation
tastelikefeet
approved these changes
Sep 4, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What this PR does
_patch_torch_FileSystemReader()when the Swift reader wrapper is already installed.init_megatron_env()calls from nesting reader thread pools.Why
Importing
swift.megatronalready callsinit_megatron_env(), while the initializer is public and is also called explicitly in the NPU setup documentation. Before this change, every call captured the previously patchedFileSystemReader.read_dataand wrapped it again. With the default 16 workers, a second initialization could therefore fan out into 16 × 16 reader tasks.Validation
swift/megatron/init.py.MCORE_READER_MAX_WORKERS=2verifies that two patch attempts retain the same wrapper and invoke the underlying reader twice rather than four times.