Dandandan commented on issue #5276: URL: https://github.com/apache/arrow-datafusion/issues/5276#issuecomment-1434991533
@dudzicp The repartitioning shouldn't be on grouping key, as hash repartitioning itself is CPU intensive. So for optimal concurrency, you want that to be parallelized as well. The round-robin partitioning is very cheap to do in DataFusion. In certain cases it might be beneficial to do hash repartition before the (partial)`AggregateExec` and only perform a single aggregation, now we're basically performing the work twice. -- 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]
