Skip to content

Add optimization barrier as no op#16439

Closed
shraiysh wants to merge 1 commit intoopenxla:mainfrom
shraiysh:fix_pgle_latency_scheduler
Closed

Add optimization barrier as no op#16439
shraiysh wants to merge 1 commit intoopenxla:mainfrom
shraiysh:fix_pgle_latency_scheduler

Conversation

@shraiysh
Copy link
Contributor

For PGLE latency hiding scheduler, optimization barrier should be considered a no-op because it is never really executed.

@shraiysh shraiysh added the kokoro:force-run Forces CI to rerun label Aug 24, 2024
@kokoro-team kokoro-team removed the kokoro:force-run Forces CI to rerun label Aug 24, 2024
For PGLE latency hiding scheduler, optimization barrier should be
considered a no-op because it is never really executed.
@NaiyerRizz NaiyerRizz self-assigned this Aug 26, 2024
copybara-service bot pushed a commit to tensorflow/tensorflow that referenced this pull request Aug 26, 2024
Imported from GitHub PR openxla/xla#16439

For PGLE latency hiding scheduler, optimization barrier should be considered a no-op because it is never really executed.
Copybara import of the project:

--
44bab12c6fb0b0c4d60ac62113eae7c959c05536 by Shraiysh Vaishay <svaishay@nvidia.com>:

Add optimization barrier as no op

For PGLE latency hiding scheduler, optimization barrier should be
considered a no-op because it is never really executed.

Merging this change closes #16439

FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#16439 from shraiysh:fix_pgle_latency_scheduler 44bab12c6fb0b0c4d60ac62113eae7c959c05536
PiperOrigin-RevId: 667505975
copybara-service bot pushed a commit to tensorflow/tensorflow that referenced this pull request Aug 26, 2024
Imported from openxla/triton#8

FUTURE_COPYBARA_INTEGRATE_REVIEW=openxla/xla#16439 from shraiysh:fix_pgle_latency_scheduler 44bab12c6fb0b0c4d60ac62113eae7c959c05536
PiperOrigin-RevId: 665336874
copybara-service bot pushed a commit to tensorflow/tensorflow that referenced this pull request Aug 26, 2024
Imported from GitHub PR openxla/xla#16439

For PGLE latency hiding scheduler, optimization barrier should be considered a no-op because it is never really executed.
Copybara import of the project:

--
44bab12c6fb0b0c4d60ac62113eae7c959c05536 by Shraiysh Vaishay <svaishay@nvidia.com>:

Add optimization barrier as no op

For PGLE latency hiding scheduler, optimization barrier should be
considered a no-op because it is never really executed.

Merging this change closes #16439

PiperOrigin-RevId: 667515007
@shraiysh shraiysh deleted the fix_pgle_latency_scheduler branch September 26, 2024 08:20
@shraiysh shraiysh restored the fix_pgle_latency_scheduler branch September 26, 2024 08:20
@shraiysh shraiysh deleted the fix_pgle_latency_scheduler branch November 18, 2024 22:14
@shraiysh shraiysh restored the fix_pgle_latency_scheduler branch November 18, 2024 22:14
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants

Comments