Rich-T-kid commented on PR #26348: URL: https://github.com/apache/datafusion/pull/26348#issuecomment-6184221799
Corrected results for Arrow `c7f60c61 -> 55d9957a` with RLE dictionary reads enabled and the NDV cap set to 2000. Lower is better; times are sums of the per-query means. | Suite | Base | PR | Delta | Faster / slower / unchanged | | --- | ---: | ---: | ---: | ---: | | [TPCH](https://github.com/apache/datafusion/pull/26348#issuecomment-6183971619) | 765.26 ms | 755.23 ms | 1.31% faster | 1 / 0 / 21 | | [TPC-DS](https://github.com/apache/datafusion/pull/26348#issuecomment-6183985466) | 8,779.81 ms | 8,802.51 ms | 0.26% slower | 2 / 1 / 96 | | [ClickBench partitioned](https://github.com/apache/datafusion/pull/26348#issuecomment-6184024537) | 20,139.17 ms | 20,314.42 ms | 0.87% slower | 7 / 4 / 32 | | [Wide schema](https://github.com/apache/datafusion/pull/26348#issuecomment-6184208966) | 3,984.8 ms | 3,995.8 ms | 0.28% slower | all 8 cases within ~1% | All queries completed; there were no failures. The direct Arrow reader benchmark shows the intended RLE dictionary path improving by roughly 1.4x-2.3x, but that work is not a large enough share of these end-to-end queries to produce a broad suite-level win. TPCH shows a small improvement; the other suites are within noise. -- 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]
