Rich-T-kid commented on PR #12126: URL: https://github.com/apache/arrow-rs/pull/12126#issuecomment-6725919142
ClickBench results for the Arrow reader change in apache/arrow-rs#12126. [Raw JSON for the balanced baseline/candidate/candidate/baseline runs](https://gist.github.com/Rich-T-kid/01a4a5d59e08c28a436cc43704a1ed9a) - DataFusion: `32dd1d2c` - Arrow baseline: `62a3c43f` - Arrow candidate: `ea86d977` (the runtime code in the current PR head is unchanged) - RLE-to-dictionary enabled in both builds - 100-file, 14.7 GB partitioned dataset; 12 partitions; 32 GiB limit - 3 iterations in each run, giving 6 samples per query for each binary All 43 queries succeeded. The sum of per-query medians was 31.96 s for the baseline and 23.96 s for the candidate, a 25.0% reduction. | Query | Baseline | Candidate | Change | |---|---:|---:|---:| | Q10 | 137.86 ms | 45.78 ms | 3.01x faster | | Q11 | 142.91 ms | 55.01 ms | 2.60x faster | | Q22 | 3,049.41 ms | 2,061.16 ms | 1.48x faster | | Q23 | 8,029.85 ms | 4,389.27 ms | 1.83x faster | | Q25 | 337.21 ms | 221.83 ms | 1.52x faster | | Q33 | 2,517.46 ms | 2,104.94 ms | 1.20x faster | | Q34 | 2,702.52 ms | 2,274.52 ms | 1.19x faster | | Q37 | 77.10 ms | 45.93 ms | 1.68x faster | -- 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]
