Rich-T-kid commented on PR #27252: URL: https://github.com/apache/datafusion/pull/27252#issuecomment-6729225753
## Released Arrow result; next: live Arrow with views disabled With `schema_force_view_types=false`, released Arrow compares plain `Utf8` to `Dictionary<_, Utf8>`: - TPCH: `766.18 -> 779.51 ms` (`+1.74%`) - TPCDS: `9681.08 -> 8869.12 ms` (`-8.39%`) - ClickBench partitioned: `22618.67 -> 41792.30 ms` (`+84.77%`); wall time `115 -> 210s` This is distinct from the earlier default-view run. Disabling views makes TPCH much closer, makes dictionaries clearly faster on TPCDS, and reduces—but does not remove—the ClickBench regression. Next round: live Arrow main at `fa57639f4` on both sides, `schema_force_view_types=false` on both sides, with only `enable_rle_to_dictionary=false` versus `true`. This directly measures plain `Utf8` versus `Dictionary<_, Utf8>` on live Arrow. -- 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]
