wangyum opened a new pull request, #38534: URL: https://github.com/apache/spark/pull/38534
### What changes were proposed in this pull request? This PR makes `HashAggregateExec` adaptively skip partial aggregation to avoid spilling if partial aggregation do not reduce the number of output rows too much. By setting `spark.sql.aggregate.adaptivePartialAggregationThreshold` to 0 this feature can be disabled. ### Why are the changes needed? Improve partial aggregation phase performance and we can implement these 2 features after this PR: 1. SPARK-36245: Partial deduplicate the right side of left semi/anti join 2. SPARK-38506: Push partial aggregation through join ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Unit test and TPC-H 5T benchmark test. SQL | Before this PR(Seconds) | After this PR(Seconds) -- | -- | -- q15 | 66 | 51 q17 | 86 | 80 q18 | 129 | 122 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
