[DPER] Introduce barrier operation to force synchronization of threads in async execution#49322
Closed
kennyhorror wants to merge 1 commit intopytorch:masterfrom
Closed
[DPER] Introduce barrier operation to force synchronization of threads in async execution#49322kennyhorror wants to merge 1 commit intopytorch:masterfrom
kennyhorror wants to merge 1 commit intopytorch:masterfrom
Conversation
kennyhorror
force-pushed
the
export-D24933471
branch
from
07f31e1 to
e44e222
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D24933471 |
kennyhorror
force-pushed
the
export-D24933471
branch
from
e44e222 to
943298d
Compare
…s in async execution (pytorch#49322) Summary: Pull Request resolved: pytorch#49322 In some cases async execution might loose dependencies (Alias like ops) or produce suboptimal scheduling when there is an option which parts to schedule first. Example of the later behavior can happen in ModelParallel training where copy can get lower priority compared to the rest of the execution on the given GPU, which will caused other GPUs to starve. This operator allows to address these issues by introducing extra explicit dependencies between ops. Test Plan: Unit-test/ E2E testing in the future diffs. Reviewed By: xianjiec Differential Revision: D24933471 fbshipit-source-id: 18e29c0899a97183115339528dc5c3c8b090205a
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D24933471 |
kennyhorror
force-pushed
the
export-D24933471
branch
from
943298d to
c6bb365
Compare
Codecov Report
@@ Coverage Diff @@
## master #49322 +/- ##
=======================================
Coverage 80.56% 80.56%
=======================================
Files 1875 1875
Lines 202701 202701
=======================================
+ Hits 163307 163309 +2
+ Misses 39394 39392 -2 |
Contributor
|
This pull request has been merged in 46debe7. |
hwangdeyu
pushed a commit
to hwangdeyu/pytorch
that referenced
this pull request
Jan 6, 2021
…s in async execution (pytorch#49322) Summary: Pull Request resolved: pytorch#49322 In some cases async execution might loose dependencies (Alias like ops) or produce suboptimal scheduling when there is an option which parts to schedule first. Example of the later behavior can happen in ModelParallel training where copy can get lower priority compared to the rest of the execution on the given GPU, which will caused other GPUs to starve. This operator allows to address these issues by introducing extra explicit dependencies between ops. Test Plan: Unit-test/ E2E testing in the future diffs. Reviewed By: xianjiec Differential Revision: D24933471 fbshipit-source-id: 1668994c7856d73926cde022378a99e1e8db3567
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.
Summary:
In some cases async execution might loose dependencies (Alias like ops) or produce suboptimal scheduling when there is an option which parts to schedule first. Example of the later behavior can happen in ModelParallel training where copy can get lower priority compared to the rest of the execution on the given GPU, which will caused other GPUs to starve.
This operator allows to address these issues by introducing extra explicit dependencies between ops.
Test Plan:
Unit-test/
E2E testing in the future diffs.
Reviewed By: xianjiec
Differential Revision: D24933471