ARROW-9873: [C++][Compute] Optimize mode kernel for integers in small value range #8091
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
kou
changed the title
pitrou
reviewed
Sep 1, 2020
Member
There was a problem hiding this comment.
I don't think polymorphism is a good idea for performance. Instead, you should probably use templated code.
Also, when null_count == length, the implementation can be trivial.
Contributor
Author
There was a problem hiding this comment.
Changed to template. Did get better performance. Thanks.
Contributor
Author
|
Latest benchmark result after re-implementation. |
Contributor
Author
|
"RTools 35" CI failure is not related |
Member
|
I get similar speedups on AMD Ryzen. |
… value range For int16/32/64 arrays with reasonable length, scan the array to find min/max values first. If (max-min) is within some threshold, instead of general hashmap, using a value indexed array can improve performance significantly. To be compatible with chunked array, value count array is transferred to hashmap before merging with others. This is an overhead for short array. Finding min/max may also introduce performance penalty in some cases. Please note it's hard to benefit all use cases. By applying this patch: - about 2x performance uplift for integers in small value range - no obvious performance drop for normal cases - non-trivial performance drop in some cases * 40% drop for short int8 array (8k length) * 10% drop for sparse array (few distinct values, big value gap)
pitrou
approved these changes
Sep 2, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
For int16/32/64 arrays with reasonable length, scan the array to find
min/max values first. If (max-min) is within some threshold, instead
of general hashmap, using a value indexed array can improve performance
significantly.
To be compatible with chunked array, value count array is transferred to
hashmap before merging with others. This is an overhead for short array.
Finding min/max may also introduce performance penalty in some cases.
Please note it's hard to benefit all use cases. By applying this patch: