ctsk commented on PR #15981: URL: https://github.com/apache/datafusion/pull/15981#issuecomment-2859359952
I've ran clickbench_partitioned and tpch_mem10 - on a machine with 16 cores. The clickbench results are pretty much the same, tpch_mem10 ran significantly faster. <details> <summary> data </summary> ```bash Comparing baseline and experiment -------------------- Benchmark clickbench_partitioned.json -------------------- ┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Query ┃ baseline ┃ experiment ┃ Change ┃ ┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ QQuery 0 │ 1.43ms │ 1.53ms │ 1.08x slower │ │ QQuery 1 │ 12.53ms │ 12.35ms │ no change │ │ QQuery 2 │ 30.19ms │ 27.21ms │ +1.11x faster │ │ QQuery 3 │ 50.88ms │ 54.24ms │ 1.07x slower │ │ QQuery 4 │ 358.15ms │ 351.98ms │ no change │ │ QQuery 5 │ 352.41ms │ 355.79ms │ no change │ │ QQuery 6 │ 1.61ms │ 1.44ms │ +1.12x faster │ │ QQuery 7 │ 14.28ms │ 14.10ms │ no change │ │ QQuery 8 │ 442.55ms │ 442.55ms │ no change │ │ QQuery 9 │ 459.15ms │ 461.75ms │ no change │ │ QQuery 10 │ 107.97ms │ 106.25ms │ no change │ │ QQuery 11 │ 114.65ms │ 115.48ms │ no change │ │ QQuery 12 │ 419.74ms │ 428.35ms │ no change │ │ QQuery 13 │ 710.78ms │ 706.78ms │ no change │ │ QQuery 14 │ 442.09ms │ 446.72ms │ no change │ │ QQuery 15 │ 434.88ms │ 428.77ms │ no change │ │ QQuery 16 │ 938.56ms │ 937.20ms │ no change │ │ QQuery 17 │ 910.94ms │ 909.49ms │ no change │ │ QQuery 18 │ 1794.56ms │ 1771.61ms │ no change │ │ QQuery 19 │ 43.31ms │ 44.30ms │ no change │ │ QQuery 20 │ 473.76ms │ 497.42ms │ no change │ │ QQuery 21 │ 622.46ms │ 615.66ms │ no change │ │ QQuery 22 │ 1337.53ms │ 1342.03ms │ no change │ │ QQuery 23 │ 6692.44ms │ 6711.11ms │ no change │ │ QQuery 24 │ 234.47ms │ 235.70ms │ no change │ │ QQuery 25 │ 170.02ms │ 173.35ms │ no change │ │ QQuery 26 │ 283.54ms │ 281.97ms │ no change │ │ QQuery 27 │ 803.14ms │ 819.98ms │ no change │ │ QQuery 28 │ 3756.01ms │ 3729.24ms │ no change │ │ QQuery 29 │ 163.74ms │ 167.47ms │ no change │ │ QQuery 30 │ 394.33ms │ 392.14ms │ no change │ │ QQuery 31 │ 549.66ms │ 545.99ms │ no change │ │ QQuery 32 │ 1975.25ms │ 1898.85ms │ no change │ │ QQuery 33 │ 1989.53ms │ 2008.24ms │ no change │ │ QQuery 34 │ 2074.40ms │ 2030.66ms │ no change │ │ QQuery 35 │ 600.63ms │ 583.67ms │ no change │ │ QQuery 36 │ 63.53ms │ 61.90ms │ no change │ │ QQuery 37 │ 28.54ms │ 29.79ms │ no change │ │ QQuery 38 │ 62.20ms │ 62.75ms │ no change │ │ QQuery 39 │ 99.07ms │ 100.04ms │ no change │ │ QQuery 40 │ 22.92ms │ 24.33ms │ 1.06x slower │ │ QQuery 41 │ 21.77ms │ 23.69ms │ 1.09x slower │ │ QQuery 42 │ 18.75ms │ 19.79ms │ 1.06x slower │ └──────────────┴───────────┴────────────┴───────────────┘ ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓ ┃ Benchmark Summary ┃ ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩ │ Total Time (baseline) │ 30078.35ms │ │ Total Time (experiment) │ 29973.68ms │ │ Average Time (baseline) │ 699.50ms │ │ Average Time (experiment) │ 697.06ms │ │ Queries Faster │ 2 │ │ Queries Slower │ 5 │ │ Queries with No Change │ 36 │ └───────────────────────────┴────────────┘ -------------------- Benchmark tpch_mem_sf10.json -------------------- ┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Query ┃ baseline ┃ experiment ┃ Change ┃ ┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ QQuery 1 │ 614.84ms │ 631.16ms │ no change │ │ QQuery 2 │ 123.17ms │ 108.61ms │ +1.13x faster │ │ QQuery 3 │ 265.83ms │ 246.60ms │ +1.08x faster │ │ QQuery 4 │ 78.03ms │ 70.74ms │ +1.10x faster │ │ QQuery 5 │ 454.91ms │ 416.73ms │ +1.09x faster │ │ QQuery 6 │ 41.99ms │ 41.35ms │ no change │ │ QQuery 7 │ 951.96ms │ 928.08ms │ no change │ │ QQuery 8 │ 351.56ms │ 301.93ms │ +1.16x faster │ │ QQuery 9 │ 808.58ms │ 717.11ms │ +1.13x faster │ │ QQuery 10 │ 381.84ms │ 368.03ms │ no change │ │ QQuery 11 │ 110.79ms │ 95.48ms │ +1.16x faster │ │ QQuery 12 │ 177.93ms │ 180.48ms │ no change │ │ QQuery 13 │ 164.45ms │ 149.70ms │ +1.10x faster │ │ QQuery 14 │ 44.06ms │ 43.21ms │ no change │ │ QQuery 15 │ 93.61ms │ 90.69ms │ no change │ │ QQuery 16 │ 89.70ms │ 86.08ms │ no change │ │ QQuery 17 │ 670.70ms │ 593.42ms │ +1.13x faster │ │ QQuery 18 │ 1805.03ms │ 1783.89ms │ no change │ │ QQuery 19 │ 102.52ms │ 102.40ms │ no change │ │ QQuery 20 │ 198.15ms │ 192.78ms │ no change │ │ QQuery 21 │ 1120.13ms │ 1064.13ms │ no change │ │ QQuery 22 │ 68.61ms │ 64.01ms │ +1.07x faster │ └──────────────┴───────────┴────────────┴───────────────┘ ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓ ┃ Benchmark Summary ┃ ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩ │ Total Time (baseline) │ 8718.38ms │ │ Total Time (experiment) │ 8276.63ms │ │ Average Time (baseline) │ 396.29ms │ │ Average Time (experiment) │ 376.21ms │ │ Queries Faster │ 10 │ │ Queries Slower │ 0 │ │ Queries with No Change │ 12 │ └───────────────────────────┴───────────┘ ``` </details> I think it will be useful to also check for performance changes for lower and higher target partition values. -- 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: github-unsubscr...@datafusion.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: github-unsubscr...@datafusion.apache.org For additional commands, e-mail: github-h...@datafusion.apache.org
