[ARROW-5705] [Java] Optimize BaseValueVector#computeCombinedBufferSize logic

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 0.14.0
• Component/s: Java
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/22135
• Assignee: Ji Liu
• Reporter: Ji Liu
• Created: 24/Jun/19 08:47
• Updated: 11/Jan/23 07:42
• Resolved: 27/Jun/19 07:21

Description

Now in BaseValueVector#computeCombinedBufferSize, it computes validity buffer size as follow:

roundUp8(getValidityBufferSizeFromCount(valueCount))

which can be be expanded to 

(((valueCount + 7) >> 3 + 7) / 8) * 8

Seems there's no need to compute bufferSize first and expression above could be replaced with:

(valueCount + 63) / 64 * 8

In this way, performance of computeCombinedBufferSize would be improved. Performance test:

Before:
BaseValueVectorBenchmarks.testC_omputeCombinedBufferSize_ avgt 5 4083.180 ± 180.363 ns/op

After:

BaseValueVectorBenchmarks.testC_omputeCombinedBufferSize_ avgt 5 3808.635 ± 162.347 ns/op

 

Issue Links

• links to: GitHub Pull Request #4671