[WIP][xpu][fix] improve XPU host time in tuning by creating less events #1230
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
meta-cla
bot
added
the
CLA Signed
jianyizh
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
I found XPU host is slow on host when creating too many events. It may stuck ~60s after some iteration. I optimize it by using less events. i.e. when I bench rms_norm, the first shape is [2048, 1024], it will bench ~150 kernels with repeat=1000. It will create 300k events. We can reuse events instead of maintaining so many.
Before change:
maintain #kernels * #repeat * 2 events, use 1 synchronize
Example of host time of each iteration when benchmark 200 kernels with repeat = 100:
After change:
maintain #kernels * 2 events, use #repeat synchronizes
cc: @EikanWang