[c10d] Fix extra CUDA context created by barrier #152834
Merged
+23
−21
Conversation
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
Fixes #149119. In ProcessGroup.hpp, we create a dummy tensor for dispatching. This requires a correct device index. This PR uses `device_id` given by user when calling `init_process_group`. This PR also uses `torch._C._get_accelerator()` to determine the device type. ghstack-source-id: 96c32b9 Pull Request resolved: #149144
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/152834
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 1 New Failure, 4 Cancelled Jobs, 1 Unrelated FailureAs of commit 99138ee with merge base 924a247 ( NEW FAILURE - The following job has failed:
CANCELLED JOBS - The following jobs were cancelled. Please retry:
FLAKY - The following job failed but was likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
oncall: distributed
ngimel
reviewed
May 5, 2025
| ) | ||
| # Detect the accelerator on the machine. If no accelerator is available, it | ||
| # returns CPU. | ||
| device = torch._C._get_accelerator() |
There was a problem hiding this comment.
_get_accelerator poisons the context on the current device, to just get the accelerator on the machine it's better to use _accelerator_getAccelerator.
There was a problem hiding this comment.
Per @albanD torch.accelerator.current_accelerator() is also non-poisoning
albanD
reviewed
May 5, 2025
| # may use default device 0, causing issues like hang or all processes | ||
| # creating context on device 0. | ||
| opts.device = device | ||
| warnings.warn( # warn only once |
There was a problem hiding this comment.
Does it actually warn only once by default?
There was a problem hiding this comment.
atalman
approved these changes
May 27, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Fixes #149119.
In ProcessGroup.hpp, we create a dummy tensor for dispatching. This requires a correct device index. This PR uses
device_idgiven by user when callinginit_process_group.This PR also uses
torch._C._get_accelerator()to determine the device type.ghstack-source-id: 96c32b9565794d995c26bd1794856d1ef7961652
Pull Request resolved: #149144
cc @H-Huang @awgu @wanchaol @fegin @fduwjj @wz337 @wconstab @d4l3k