Skip to content
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

[FEA] See if we can optimize sort for a single batch #10570

Closed
revans2 opened this issue Mar 11, 2024 · 0 comments · Fixed by #10572
Closed

[FEA] See if we can optimize sort for a single batch #10570

revans2 opened this issue Mar 11, 2024 · 0 comments · Fixed by #10572
Assignees
Labels
performance A performance related task/issue

Comments

@revans2
Copy link
Collaborator

revans2 commented Mar 11, 2024

Is your feature request related to a problem? Please describe.
In many cases a task is going to have a single batch to process. When we sort right now, we will sort the input batch, slice it into smaller batches, and then start to merge those batches together. But most of that ends up being a noop if there is only a single batch to sort.

I would like to propose that we modify GpuOutOfCoreSortIterator so that it takes the first batch and just makes it spillable. It then asks for a second batch. Only if it gets a second batch does it go through the complicated sort/merge process. If there is only one batch we sort it and go on. It is a little more complicated than that because of retry and splits with the retry, but I think it is totally doable.

@revans2 revans2 added feature request New feature or request ? - Needs Triage Need team to review and classify performance A performance related task/issue labels Mar 11, 2024
@revans2 revans2 self-assigned this Mar 12, 2024
@sameerz sameerz removed the feature request New feature or request label Mar 12, 2024
@mattahrens mattahrens removed the ? - Needs Triage Need team to review and classify label Mar 14, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
performance A performance related task/issue
Projects
None yet
Development

Successfully merging a pull request may close this issue.

3 participants