-
Notifications
You must be signed in to change notification settings - Fork 230
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
Group-by aggregation based optimization for UNBOUNDED collect_set
window function
#10248
Merged
revans2
merged 59 commits into
NVIDIA:branch-24.04
from
mythrocks:revans-agg-for-window
Feb 12, 2024
Merged
Group-by aggregation based optimization for UNBOUNDED collect_set
window function
#10248
revans2
merged 59 commits into
NVIDIA:branch-24.04
from
mythrocks:revans-agg-for-window
Feb 12, 2024
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
Signed-off-by: Robert (Bobby) Evans <bobby@apache.org>
Signed-off-by: Robert (Bobby) Evans <bobby@apache.org>
Signed-off-by: Robert (Bobby) Evans <bobby@apache.org>
Signed-off-by: Robert (Bobby) Evans <bobby@apache.org>
Signed-off-by: Robert (Bobby) Evans <bobby@apache.org>
Signed-off-by: MithunR <mythrocks@gmail.com>
No need to partition any more than the last group. All the other groups shake out during the merge.
Signed-off-by: MithunR <mythrocks@gmail.com>
Also, clarified group-by column ordering.
mythrocks
added
performance
A performance related task/issue
reliability
Features to improve reliability or bugs that severly impact the reliability of the plugin
labels
Jan 23, 2024
Sorry, I thought I'd upmerged. Yes, I should resolve these before we merge. |
Still getting errors on compile. Looks like some style issues with the imports. |
I did some quick scale tests and it looks really good. I'll try to look through all of the code on Monday. |
Build |
Build |
Build |
revans2
reviewed
Feb 6, 2024
...gin/src/main/scala/com/nvidia/spark/rapids/window/GpuUnboundedToUnboundedAggWindowExec.scala
Outdated
Show resolved
Hide resolved
...gin/src/main/scala/com/nvidia/spark/rapids/window/GpuUnboundedToUnboundedAggWindowExec.scala
Show resolved
Hide resolved
Build |
revans2
previously approved these changes
Feb 7, 2024
Build |
revans2
approved these changes
Feb 12, 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
reliability
Features to improve reliability or bugs that severly impact the reliability of the plugin
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.
Fixes #10114.
This change introduces an optimization for the
collect_set()
window function, when run as an fullyUNBOUNDED
window function, as follows:It might be possible to solve most UNBOUNDED window aggregations via group-wise aggregations. This change addresses
collect_set
first.This change currently builds on the changes @revans2 made over at #10158, and includes those changes here.
A new window-exec is introduced specifically to address fully UNBOUNDED window functions. Each input batch is processed via group-wise aggregations. For groups that are known to be complete, the output is returned. The results for incomplete groups are cached, and finalized only when the end of the group is encountered.