Complete Implement fast Vectorized grouping for high cardinality #6889
Description
Is your feature request related to a problem or challenge?
This is a summary issue tracking the work to complete the design described in #4973 (comment)
Describe the solution you'd like
- Create a POC demonstrating sufficient improvement: RFC: Demonstrate new
GroupHashAggregatestream approach (runs more than 2x faster!) #6800 - Create a PR to use the new GroupsAggregators: Vectorized hash grouping #6904
- Regression: bit_and aggregate function is incorrect #6955
- File ticket for Optimize inner loop / null calculation performance Improve grouping performance via better vectorization in accumulate functions #7066
- Add tests for min/max aggregates on Strings: Minor: Add String/Binary aggregate tests #6962
- Reduce duplication between
BoundedAggregateStreamandGroupedHashAggregateStream#6798 - Remove RowAggregators and the now unused row format Remove RowAccumulators and datafusion-row #6968
- Write a blog post fast Vectorized grouping for high cardinality #6988
- File a ticket for avoiding reallocations in accumulators (suggested y @yjshen on RFC: Demonstrate new
GroupHashAggregatestream approach (runs more than 2x faster!) #6800 (comment)): Improve aggregate performance with adaptive sizing in accumulators / avoiding reallocations in accumulators #7065 - File a ticket for fast min/max accumulator for strings : Implement fast min/max accumulator for binary / strings (now it uses the slower path) #6906
- File a ticket about special casing single column group by queries and avoiding the row conversion (suggested by @Dandandan on RFC: Demonstrate new
GroupHashAggregatestream approach (runs more than 2x faster!) #6800 (comment)): Improve aggregate performance by special casing single group keys #6969
Describe alternatives you've considered
No response
Additional context
No response