[ARROW-9605] [C++] Optimize performance for aggregate min/max compute kernels

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.0.0
• Component/s: C++
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/25667
• Assignee: Frank Du
• Reporter: Frank Du
• Created: 31/Jul/20 03:06
• Updated: 11/Jan/23 08:07
• Resolved: 02/Sep/20 13:35

Description

1. Use BitBlockCounter to speedup the performance for typical 0.01% null-able data.
2. Enable AVX compiler auto vectorize for no-nulls on int types. Float/Double use fmin/fmax to handle NaN which can't be vectorize by compiler.

Issue Links

• links to: GitHub Pull Request #7871