Fix intermittent CUDA activity flush issue #95
Closed
Conversation
This file contains 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
Summary: We recently observed that no Cuda activities appears in libkineto traces for certain setups. The issue is that cuptiActivityFlushAll sometimes produces no buffers. Setting the CUPTI_ACTIVITY_FLAG_FLUSH_FORCED appears to give us consistent results. We're still figuring out which setups are affected. Differential Revision: D26717258 fbshipit-source-id: ac3f39841de1f64d7800393f2bad5d105897a708
|
This pull request was exported from Phabricator. Differential Revision: D26717258 |
|
This pull request has been merged in fa0f2d5. |
Kingwl
pushed a commit
to Kingwl/kineto
that referenced
this pull request
Mar 4, 2021
Summary: Pull Request resolved: pytorch#95 We recently observed that no Cuda activities appears in libkineto traces for certain setups. The issue is that cuptiActivityFlushAll sometimes produces no buffers. Setting the CUPTI_ACTIVITY_FLAG_FLUSH_FORCED appears to give us consistent results. We're still figuring out why it suddenly started happening and why it affects some setups and not others. Reviewed By: chauhang Differential Revision: D26717258 fbshipit-source-id: d96df0eccaaa7ef865d0314eca6dc43c4ad1d648
liuziyue
pushed a commit
to liuziyue/kineto
that referenced
this pull request
Mar 10, 2021
Summary: Pull Request resolved: pytorch#95 We recently observed that no Cuda activities appears in libkineto traces for certain setups. The issue is that cuptiActivityFlushAll sometimes produces no buffers. Setting the CUPTI_ACTIVITY_FLAG_FLUSH_FORCED appears to give us consistent results. We're still figuring out why it suddenly started happening and why it affects some setups and not others. Reviewed By: chauhang Differential Revision: D26717258 fbshipit-source-id: d96df0eccaaa7ef865d0314eca6dc43c4ad1d648
ilia-cher
pushed a commit
that referenced
this pull request
Mar 16, 2021
Summary: Pull Request resolved: #95 We recently observed that no Cuda activities appears in libkineto traces for certain setups. The issue is that cuptiActivityFlushAll sometimes produces no buffers. Setting the CUPTI_ACTIVITY_FLAG_FLUSH_FORCED appears to give us consistent results. We're still figuring out why it suddenly started happening and why it affects some setups and not others. Reviewed By: chauhang Differential Revision: D26717258 fbshipit-source-id: d96df0eccaaa7ef865d0314eca6dc43c4ad1d648
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:
We recently observed that no Cuda activities appears in libkineto traces for certain setups.
The issue is that cuptiActivityFlushAll sometimes produces no buffers.
Setting the CUPTI_ACTIVITY_FLAG_FLUSH_FORCED appears to give us consistent results.
We're still figuring out which setups are affected.
Differential Revision: D26717258