-
Notifications
You must be signed in to change notification settings - Fork 5.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[air] Randomize block order by default to avoid hotspots #25870
Conversation
if config.use_stream_api: | ||
dataset = dataset.random_shuffle_each_window() | ||
else: | ||
dataset = dataset.random_shuffle() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Bug fix!
if conf.randomize_block_order: | ||
dataset = dataset.randomize_block_order() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should we add randomize_block_order_each_window
for the pipelined case?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, I was debating that. I guess we may as well in case it helps with randomization, even though it doesn't help with hotspots.
Putting a hold on this, since we identified that this is breaking stage fusion from |
Closing this for now. |
This should be good to go now that fusion is fixed. |
Why are these changes needed?
Enable block order randomization by default to avoid ingest hotspots when running concurrent trials.