adriangbot commented on PR #26973: URL: https://github.com/apache/datafusion/pull/26973#issuecomment-6569831413
π€ Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/26973#issuecomment-6569557125) **Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) Comparing codex/dictionary-only-pages (03cf12c6e4f600700ca3077899f756deafbf0352) to 0e314cc (merge-base) [diff](https://github.com/apache/datafusion/compare/0e314cc3c904b394d9711251c39f6dad376d187f..03cf12c6e4f600700ca3077899f756deafbf0352) <details><summary>Run configuration</summary> ```yaml run benchmark clickbench_partitioned env: DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true" DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G" ``` </details> <details><summary>CPU Details (lscpu)</summary> ``` Architecture: aarch64 CPU op-mode(s): 64-bit Byte Order: Little Endian CPU(s): 16 On-line CPU(s) list: 0-15 Vendor ID: ARM Model name: Neoverse-V2 Model: 1 Thread(s) per core: 1 Core(s) per cluster: 16 Socket(s): - Cluster(s): 1 Stepping: r0p1 BogoMIPS: 2000.00 Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti L1d cache: 1 MiB (16 instances) L1i cache: 1 MiB (16 instances) L2 cache: 32 MiB (16 instances) L3 cache: 80 MiB (1 instance) NUMA node(s): 1 NUMA node0 CPU(s): 0-15 Vulnerability Gather data sampling: Not affected Vulnerability Indirect target selection: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; __user pointer sanitization Vulnerability Spectre v2: Mitigation; CSV2, BHB Vulnerability Srbds: Not affected Vulnerability Tsa: Not affected Vulnerability Tsx async abort: Not affected Vulnerability Vmscape: Not affected ``` </details> <details><summary>Details</summary> <p> ``` Comparing HEAD and codex_dictionary-only-pages -------------------- Benchmark clickbench_partitioned.json -------------------- βββββββββββββ³ββββββββββββββ³ββββββββββββββββββββββββββββββ³ββββββββββββββββ β Query β HEAD β codex_dictionary-only-pages β Change β β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ© β QQuery 0 β 1.13 ms β 1.17 ms β no change β β QQuery 1 β 10.85 ms β 11.49 ms β 1.06x slower β β QQuery 2 β 35.98 ms β 36.01 ms β no change β β QQuery 3 β 31.04 ms β 31.50 ms β no change β β QQuery 4 β 233.17 ms β 237.86 ms β no change β β QQuery 5 β 1467.23 ms β 275.90 ms β +5.32x faster β β QQuery 6 β 1.16 ms β 1.20 ms β no change β β QQuery 7 β 11.95 ms β 12.58 ms β 1.05x slower β β QQuery 8 β 335.27 ms β 322.28 ms β no change β β QQuery 9 β 466.34 ms β 468.17 ms β no change β β QQuery 10 β 197.55 ms β 199.18 ms β no change β β QQuery 11 β 208.22 ms β 209.75 ms β no change β β QQuery 12 β 1488.91 ms β 266.40 ms β +5.59x faster β β QQuery 13 β 2047.91 ms β 367.14 ms β +5.58x faster β β QQuery 14 β 1637.22 ms β 280.61 ms β +5.83x faster β β QQuery 15 β 289.73 ms β 283.04 ms β no change β β QQuery 16 β 3656.78 ms β 612.24 ms β +5.97x faster β β QQuery 17 β 3700.07 ms β 620.79 ms β +5.96x faster β β QQuery 18 β 6755.99 ms β 1274.92 ms β +5.30x faster β β QQuery 19 β 26.81 ms β 27.03 ms β no change β β QQuery 20 β 1040.49 ms β 517.74 ms β +2.01x faster β β QQuery 21 β 1363.14 ms β 508.87 ms β +2.68x faster β β QQuery 22 β 3223.31 ms β 977.45 ms β +3.30x faster β β QQuery 23 β 9118.71 ms β 6063.67 ms β +1.50x faster β β QQuery 24 β 95.74 ms β 40.87 ms β +2.34x faster β β QQuery 25 β 349.57 ms β 105.39 ms β +3.32x faster β β QQuery 26 β 96.96 ms β 42.27 ms β +2.29x faster β β QQuery 27 β 1136.95 ms β 509.59 ms β +2.23x faster β β QQuery 28 β 2823.63 ms β 2847.95 ms β no change β β QQuery 29 β 38.23 ms β 38.63 ms β no change β β QQuery 30 β 545.83 ms β 303.43 ms β +1.80x faster β β QQuery 31 β 534.22 ms β 274.53 ms β +1.95x faster β β QQuery 32 β 1049.92 ms β 927.37 ms β +1.13x faster β β QQuery 33 β 13790.22 ms β 1452.16 ms β +9.50x faster β β QQuery 34 β 13987.14 ms β 1485.38 ms β +9.42x faster β β QQuery 35 β 296.29 ms β 297.11 ms β no change β β QQuery 36 β 171.89 ms β 66.24 ms β +2.59x faster β β QQuery 37 β 95.35 ms β 35.75 ms β +2.67x faster β β QQuery 38 β 78.98 ms β 39.85 ms β +1.98x faster β β QQuery 39 β 399.61 ms β 141.50 ms β +2.82x faster β β QQuery 40 β 14.11 ms β 14.67 ms β no change β β QQuery 41 β 13.66 ms β 14.04 ms β no change β β QQuery 42 β 12.98 ms β 13.39 ms β no change β βββββββββββββ΄ββββββββββββββ΄ββββββββββββββββββββββββββββββ΄ββββββββββββββββ ββββββββββββββββββββββββββββββββββββββββββββββ³βββββββββββββ β Benchmark Summary β β β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ© β Total Time (HEAD) β 72880.22ms β β Total Time (codex_dictionary-only-pages) β 22257.13ms β β Average Time (HEAD) β 1694.89ms β β Average Time (codex_dictionary-only-pages) β 517.61ms β β Queries Faster β 24 β β Queries Slower β 2 β β Queries with No Change β 17 β β Queries with Failure β 0 β ββββββββββββββββββββββββββββββββββββββββββββββ΄βββββββββββββ Distribution per query (min / mean Β±stddev / max): Comparing HEAD and codex_dictionary-only-pages -------------------- Benchmark clickbench_partitioned.json -------------------- βββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββββ β Query β HEAD β codex_dictionary-only-pages β Change β β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ© β QQuery 0 β 1.13 / 4.08 Β±5.82 / 15.72 ms β 1.17 / 4.18 Β±5.88 / 15.93 ms β no change β β QQuery 1 β 10.85 / 11.02 Β±0.11 / 11.16 ms β 11.49 / 11.66 Β±0.18 / 12.00 ms β 1.06x slower β β QQuery 2 β 35.98 / 36.41 Β±0.32 / 36.94 ms β 36.01 / 36.68 Β±0.77 / 38.19 ms β no change β β QQuery 3 β 31.04 / 31.88 Β±0.84 / 33.22 ms β 31.50 / 31.76 Β±0.25 / 32.22 ms β no change β β QQuery 4 β 233.17 / 239.89 Β±3.48 / 242.59 ms β 237.86 / 240.88 Β±2.47 / 245.10 ms β no change β β QQuery 5 β 1467.23 / 1496.97 Β±25.30 / 1543.56 ms β 275.90 / 281.14 Β±4.15 / 286.53 ms β +5.32x faster β β QQuery 6 β 1.16 / 1.33 Β±0.25 / 1.82 ms β 1.20 / 1.37 Β±0.25 / 1.86 ms β no change β β QQuery 7 β 11.95 / 15.85 Β±7.40 / 30.65 ms β 12.58 / 12.77 Β±0.15 / 12.98 ms β +1.24x faster β β QQuery 8 β 335.27 / 337.01 Β±2.07 / 341.07 ms β 322.28 / 333.74 Β±7.53 / 342.98 ms β no change β β QQuery 9 β 466.34 / 483.78 Β±12.82 / 504.82 ms β 468.17 / 489.51 Β±11.05 / 499.75 ms β no change β β QQuery 10 β 197.55 / 198.09 Β±0.67 / 199.38 ms β 199.18 / 200.08 Β±0.90 / 201.80 ms β no change β β QQuery 11 β 208.22 / 209.34 Β±0.70 / 210.28 ms β 209.75 / 210.51 Β±0.70 / 211.44 ms β no change β β QQuery 12 β 1488.91 / 1540.43 Β±46.60 / 1599.22 ms β 266.40 / 271.00 Β±5.27 / 281.13 ms β +5.68x faster β β QQuery 13 β 2047.91 / 2438.75 Β±344.54 / 2880.24 ms β 367.14 / 374.14 Β±4.99 / 380.63 ms β +6.52x faster β β QQuery 14 β 1637.22 / 1658.73 Β±23.55 / 1703.12 ms β 280.61 / 287.10 Β±4.53 / 293.10 ms β +5.78x faster β β QQuery 15 β 289.73 / 302.42 Β±8.71 / 315.69 ms β 283.04 / 287.79 Β±5.31 / 295.95 ms β no change β β QQuery 16 β 3656.78 / 3754.49 Β±99.68 / 3916.41 ms β 612.24 / 627.48 Β±12.52 / 650.50 ms β +5.98x faster β β QQuery 17 β 3700.07 / 3823.48 Β±103.15 / 3998.55 ms β 620.79 / 631.51 Β±8.94 / 645.84 ms β +6.05x faster β β QQuery 18 β 6755.99 / 7198.87 Β±373.85 / 7834.20 ms β 1274.92 / 1283.86 Β±8.73 / 1294.81 ms β +5.61x faster β β QQuery 19 β 26.81 / 50.11 Β±39.68 / 128.73 ms β 27.03 / 27.48 Β±0.28 / 27.74 ms β +1.82x faster β β QQuery 20 β 1040.49 / 1052.85 Β±8.63 / 1067.37 ms β 517.74 / 533.28 Β±20.77 / 574.32 ms β +1.97x faster β β QQuery 21 β 1363.14 / 1374.98 Β±8.78 / 1387.29 ms β 508.87 / 516.17 Β±3.80 / 519.20 ms β +2.66x faster β β QQuery 22 β 3223.31 / 3265.72 Β±37.21 / 3325.26 ms β 977.45 / 984.87 Β±6.98 / 993.43 ms β +3.32x faster β β QQuery 23 β 9118.71 / 9397.05 Β±188.19 / 9648.57 ms β 6063.67 / 6155.10 Β±48.76 / 6204.22 ms β +1.53x faster β β QQuery 24 β 95.74 / 160.50 Β±112.56 / 384.49 ms β 40.87 / 45.83 Β±5.88 / 57.29 ms β +3.50x faster β β QQuery 25 β 349.57 / 354.04 Β±3.34 / 359.81 ms β 105.39 / 107.40 Β±2.09 / 111.32 ms β +3.30x faster β β QQuery 26 β 96.96 / 99.88 Β±2.23 / 102.81 ms β 42.27 / 47.86 Β±9.53 / 66.88 ms β +2.09x faster β β QQuery 27 β 1136.95 / 1160.42 Β±26.71 / 1210.68 ms β 509.59 / 513.20 Β±5.14 / 523.35 ms β +2.26x faster β β QQuery 28 β 2823.63 / 2849.08 Β±35.43 / 2916.33 ms β 2847.95 / 2897.26 Β±36.81 / 2958.47 ms β no change β β QQuery 29 β 38.23 / 38.91 Β±0.62 / 39.73 ms β 38.63 / 48.56 Β±19.69 / 87.93 ms β 1.25x slower β β QQuery 30 β 545.83 / 555.81 Β±6.29 / 565.59 ms β 303.43 / 309.93 Β±5.05 / 319.01 ms β +1.79x faster β β QQuery 31 β 534.22 / 551.01 Β±15.18 / 570.93 ms β 274.53 / 300.94 Β±17.15 / 327.13 ms β +1.83x faster β β QQuery 32 β 1049.92 / 1087.56 Β±44.21 / 1169.58 ms β 927.37 / 979.85 Β±38.15 / 1043.19 ms β +1.11x faster β β QQuery 33 β 13790.22 / 14304.81 Β±479.37 / 15059.29 ms β 1452.16 / 1480.67 Β±20.27 / 1506.85 ms β +9.66x faster β β QQuery 34 β 13987.14 / 14716.86 Β±685.63 / 15563.68 ms β 1485.38 / 1528.77 Β±39.60 / 1602.34 ms β +9.63x faster β β QQuery 35 β 296.29 / 383.18 Β±99.99 / 531.01 ms β 297.11 / 335.57 Β±48.49 / 420.54 ms β +1.14x faster β β QQuery 36 β 171.89 / 190.10 Β±12.41 / 208.42 ms β 66.24 / 75.55 Β±7.11 / 86.62 ms β +2.52x faster β β QQuery 37 β 95.35 / 101.63 Β±7.18 / 114.89 ms β 35.75 / 43.45 Β±12.82 / 68.96 ms β +2.34x faster β β QQuery 38 β 78.98 / 84.06 Β±4.58 / 92.54 ms β 39.85 / 41.84 Β±1.17 / 43.42 ms β +2.01x faster β β QQuery 39 β 399.61 / 415.52 Β±8.85 / 424.19 ms β 141.50 / 156.40 Β±8.30 / 166.23 ms β +2.66x faster β β QQuery 40 β 14.11 / 14.89 Β±0.72 / 15.78 ms β 14.67 / 15.10 Β±0.36 / 15.75 ms β no change β β QQuery 41 β 13.66 / 14.03 Β±0.71 / 15.44 ms β 14.04 / 14.47 Β±0.30 / 14.80 ms β no change β β QQuery 42 β 12.98 / 13.25 Β±0.17 / 13.45 ms β 13.39 / 13.45 Β±0.07 / 13.56 ms β no change β βββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββββ ββββββββββββββββββββββββββββββββββββββββββββββ³βββββββββββββ β Benchmark Summary β β β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ© β Total Time (HEAD) β 76019.11ms β β Total Time (codex_dictionary-only-pages) β 22790.16ms β β Average Time (HEAD) β 1767.89ms β β Average Time (codex_dictionary-only-pages) β 530.00ms β β Queries Faster β 27 β β Queries Slower β 2 β β Queries with No Change β 14 β β Queries with Failure β 0 β ββββββββββββββββββββββββββββββββββββββββββββββ΄βββββββββββββ ``` </p> </details> <details><summary>Memory Pool Peaks</summary> Peak `MemoryPool` reservation per query β what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the `dfbench` suites, and the suites `bench.sh` runs through the Criterion SQL harness since apache/datafusion#25644), and only under a memory limit: `DATAFUSION_RUNTIME_MEMORY_LIMIT`, or one the suite sets itself. Base: `0e314cc (merge-base)` | Changed: `codex/dictionary-only-pages` **`clickbench_partitioned`** | Query | Base | Changed | Change | | --- | --- | --- | --- | | Query 0 | 0 B | 0 B | 0.0% | | Query 1 | 104 B | 104 B | +0.0% | | Query 2 | 936 B | 936 B | +0.0% | | Query 3 | 312 B | 312 B | +0.0% | | Query 4 | 846.0 MiB | 811.2 MiB | -4.1% | | Query 5 | 1.4 GiB | 1.2 GiB | -15.8% | | Query 6 | 0 B | 0 B | 0.0% | | Query 7 | 60.0 MiB | 60.0 MiB | +0.0% | | Query 8 | 967.0 MiB | 969.7 MiB | +0.3% | | Query 9 | 594.0 MiB | 636.0 MiB | +7.1% | | Query 10 | 74.9 MiB | 72.3 MiB | -3.5% | | Query 11 | 77.7 MiB | 76.9 MiB | -1.1% | | Query 12 | 1.6 GiB | 1.3 GiB | -16.7% | | Query 13 | 1.9 GiB | 1.6 GiB | -15.5% | | Query 14 | 1.9 GiB | 1.3 GiB | -32.1% | | Query 15 | 1.2 GiB | 1.3 GiB | +4.6% | | Query 16 | 3.3 GiB | 3.4 GiB | +2.9% | | Query 17 | 3.4 GiB | 3.4 GiB | -0.1% | | Query 18 | 7.2 GiB | 8.1 GiB | +11.6% | | Query 19 | 0 B | 0 B | 0.0% | | Query 20 | 104 B | 104 B | +0.0% | | Query 21 | 1.3 MiB | 3.7 MiB | +177.0% | | Query 22 | 2.2 MiB | 2.9 MiB | +31.0% | | Query 23 | 12.4 MiB | 11.7 MiB | -5.1% | | Query 24 | 6.9 MiB | 60.6 MiB | +780.1% | | Query 25 | 7.8 MiB | 170.1 MiB | +2075.2% | | Query 26 | 8.0 MiB | 64.4 MiB | +700.8% | | Query 27 | 2.5 MiB | 2.2 MiB | -8.9% | | Query 28 | 1.9 GiB | 1.7 GiB | -10.8% | | Query 29 | 624 B | 624 B | +0.0% | | Query 30 | 648.6 MiB | 697.5 MiB | +7.5% | | Query 31 | 1.4 GiB | 1.5 GiB | +4.9% | | Query 32 | 7.1 GiB | 7.1 GiB | -0.3% | | Query 33 | 12.7 GiB | 8.0 GiB | -37.5% | | Query 34 | 12.4 GiB | 8.2 GiB | -33.5% | | Query 35 | 668.8 MiB | 664.8 MiB | -0.6% | | Query 36 | 474.1 MiB | 121.0 MiB | -74.5% | | Query 37 | 8.1 MiB | 7.3 MiB | -10.3% | | Query 38 | 5.3 MiB | 5.8 MiB | +10.2% | | Query 39 | 450.8 MiB | 296.7 MiB | -34.2% | | Query 40 | 2.6 MiB | 2.3 MiB | -11.5% | | Query 41 | 3.3 MiB | 3.3 MiB | +0.0% | | Query 42 | 1.7 MiB | 2.0 MiB | +17.3% | **Pool accounting vs. process RSS** Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it. | Benchmark | Side | Max pool peak | Peak RSS | Gap | RSS / pool | | --- | --- | --- | --- | --- | --- | | clickbench_partitioned | base (`0e314cc (merge-base)`) | 12.7 GiB | 12.6 GiB | β115.7 MiB | 1.0Γ | | clickbench_partitioned | changed (`codex/dictionary-only-pages`) | 8.2 GiB | 12.7 GiB | 4.5 GiB | 1.5Γ | </details> <details><summary>Resource Usage</summary> **clickbench_partitioned β base (merge-base)** | Metric | Value | |--------|-------| | Wall time | 385.1s | | Peak memory | 12.6 GiB | | Avg memory | 5.9 GiB | | CPU user | 3307.2s | | CPU sys | 206.3s | | Peak spill | 0 B | **clickbench_partitioned β branch** | Metric | Value | |--------|-------| | Wall time | 115.0s | | Peak memory | 12.7 GiB | | Avg memory | 4.6 GiB | | CPU user | 1183.5s | | CPU sys | 73.7s | | Peak spill | 0 B | </details> --- [File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner -- 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]
