Dandandan opened a new pull request #9826: URL: https://github.com/apache/arrow/pull/9826
This is better for the common case where (almost) every value in the build side will be unique, and thus will be store once per this list, avoiding heap allocations and extra indirections (not saving the index inline but in a different `Vec`). this is a > 10% improvement for TCP-H query 5 (SF=1, 16 partitions/threads, loaded in memory). master: `Query 5 avg time: 169.63 ms` PR: `Query 5 avg time: 151.07 ms` -- 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. For queries about this service, please contact Infrastructure at: [email protected]
