Draft: Test non exact size batch

[zhuqi-lucas]

Which issue does this PR close?

Testing for:
#17105 (comment)

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

[zhuqi-lucas]

This is the testing PR for non-exact size, FYI @alamb.

The upstream code is here:
apache/arrow-rs#8112

[alamb]

Kicking off becnhmarks

[alamb]

🤖 ./gh_compare_branch.shBenchmark Script Running
Linux aal-dev 6.11.0-1016-gcp #16~24.04.1-Ubuntu SMP Wed May 28 02:40:52 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
Comparing test_non_exact_size_batch (f642391) to 7d52145diff using: tpch_mem
Results will be posted here when complete

[alamb]

🤖: Benchmark completed

Details

Comparing HEAD and test_non_exact_size_batch
--------------------
Benchmark tpch_mem_sf1.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ test_non_exact_size_batch ┃ Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1 │ 96.18 ms │ 93.30 ms │ no change │
│ QQuery 2 │ 20.96 ms │ 22.99 ms │ 1.10x slower │
│ QQuery 3 │ 32.14 ms │ 36.01 ms │ 1.12x slower │
│ QQuery 4 │ 18.39 ms │ 18.60 ms │ no change │
│ QQuery 5 │ 48.92 ms │ 55.22 ms │ 1.13x slower │
│ QQuery 6 │ 11.99 ms │ 11.63 ms │ no change │
│ QQuery 7 │ 81.98 ms │ 93.94 ms │ 1.15x slower │
│ QQuery 8 │ 24.78 ms │ 25.59 ms │ no change │
│ QQuery 9 │ 53.99 ms │ 54.99 ms │ no change │
│ QQuery 10 │ 40.09 ms │ 49.73 ms │ 1.24x slower │
│ QQuery 11 │ 11.19 ms │ 12.19 ms │ 1.09x slower │
│ QQuery 12 │ 30.34 ms │ 34.79 ms │ 1.15x slower │
│ QQuery 13 │ 25.57 ms │ 27.45 ms │ 1.07x slower │
│ QQuery 14 │ 9.60 ms │ 12.17 ms │ 1.27x slower │
│ QQuery 15 │ 18.88 ms │ 23.25 ms │ 1.23x slower │
│ QQuery 16 │ 17.43 ms │ 21.64 ms │ 1.24x slower │
│ QQuery 17 │ 94.13 ms │ 115.21 ms │ 1.22x slower │
│ QQuery 18 │ 189.20 ms │ 236.46 ms │ 1.25x slower │
│ QQuery 19 │ 25.37 ms │ 31.06 ms │ 1.22x slower │
│ QQuery 20 │ 31.24 ms │ 37.38 ms │ 1.20x slower │
│ QQuery 21 │ 141.50 ms │ 174.59 ms │ 1.23x slower │
│ QQuery 22 │ 13.79 ms │ 15.71 ms │ 1.14x slower │
└──────────────┴───────────┴───────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD) │ 1037.67ms │
│ Total Time (test_non_exact_size_batch) │ 1203.91ms │
│ Average Time (HEAD) │ 47.17ms │
│ Average Time (test_non_exact_size_batch) │ 54.72ms │
│ Queries Faster │ 0 │
│ Queries Slower │ 17 │
│ Queries with No Change │ 5 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴───────────┘

[alamb]

🤖 ./gh_compare_branch.shBenchmark Script Running
Linux aal-dev 6.11.0-1016-gcp #16~24.04.1-Ubuntu SMP Wed May 28 02:40:52 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
Comparing test_non_exact_size_batch (f642391) to 7d52145diff using: clickbench_partitioned
Results will be posted here when complete

[zhuqi-lucas]

🤖: Benchmark completed

Details

Comparing HEAD and test_non_exact_size_batch
--------------------
Benchmark tpch_mem_sf1.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ test_non_exact_size_batch ┃ Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1 │ 96.18 ms │ 93.30 ms │ no change │
│ QQuery 2 │ 20.96 ms │ 22.99 ms │ 1.10x slower │
│ QQuery 3 │ 32.14 ms │ 36.01 ms │ 1.12x slower │
│ QQuery 4 │ 18.39 ms │ 18.60 ms │ no change │
│ QQuery 5 │ 48.92 ms │ 55.22 ms │ 1.13x slower │
│ QQuery 6 │ 11.99 ms │ 11.63 ms │ no change │
│ QQuery 7 │ 81.98 ms │ 93.94 ms │ 1.15x slower │
│ QQuery 8 │ 24.78 ms │ 25.59 ms │ no change │
│ QQuery 9 │ 53.99 ms │ 54.99 ms │ no change │
│ QQuery 10 │ 40.09 ms │ 49.73 ms │ 1.24x slower │
│ QQuery 11 │ 11.19 ms │ 12.19 ms │ 1.09x slower │
│ QQuery 12 │ 30.34 ms │ 34.79 ms │ 1.15x slower │
│ QQuery 13 │ 25.57 ms │ 27.45 ms │ 1.07x slower │
│ QQuery 14 │ 9.60 ms │ 12.17 ms │ 1.27x slower │
│ QQuery 15 │ 18.88 ms │ 23.25 ms │ 1.23x slower │
│ QQuery 16 │ 17.43 ms │ 21.64 ms │ 1.24x slower │
│ QQuery 17 │ 94.13 ms │ 115.21 ms │ 1.22x slower │
│ QQuery 18 │ 189.20 ms │ 236.46 ms │ 1.25x slower │
│ QQuery 19 │ 25.37 ms │ 31.06 ms │ 1.22x slower │
│ QQuery 20 │ 31.24 ms │ 37.38 ms │ 1.20x slower │
│ QQuery 21 │ 141.50 ms │ 174.59 ms │ 1.23x slower │
│ QQuery 22 │ 13.79 ms │ 15.71 ms │ 1.14x slower │
└──────────────┴───────────┴───────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD) │ 1037.67ms │
│ Total Time (test_non_exact_size_batch) │ 1203.91ms │
│ Average Time (HEAD) │ 47.17ms │
│ Average Time (test_non_exact_size_batch) │ 54.72ms │
│ Queries Faster │ 0 │
│ Queries Slower │ 17 │
│ Queries with No Change │ 5 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴───────────┘

Thank you @alamb for benchmark, it's worse than the exact batch, it may due to the extra allocation because we reserve for fixed batch_size.

I need to investigate it more.

[alamb]

🤖: Benchmark completed

Details

Comparing HEAD and test_non_exact_size_batch
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ test_non_exact_size_batch ┃ Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0 │ 2.16 ms │ 2.41 ms │ 1.11x slower │
│ QQuery 1 │ 27.70 ms │ 27.56 ms │ no change │
│ QQuery 2 │ 72.12 ms │ 74.69 ms │ no change │
│ QQuery 3 │ 86.23 ms │ 86.55 ms │ no change │
│ QQuery 4 │ 563.40 ms │ 574.67 ms │ no change │
│ QQuery 5 │ 829.14 ms │ 848.11 ms │ no change │
│ QQuery 6 │ 2.13 ms │ 2.32 ms │ 1.09x slower │
│ QQuery 7 │ 31.36 ms │ 32.30 ms │ no change │
│ QQuery 8 │ 820.00 ms │ 843.20 ms │ no change │
│ QQuery 9 │ 1125.91 ms │ 1135.55 ms │ no change │
│ QQuery 10 │ 224.83 ms │ 220.14 ms │ no change │
│ QQuery 11 │ 251.48 ms │ 250.18 ms │ no change │
│ QQuery 12 │ 810.17 ms │ 827.53 ms │ no change │
│ QQuery 13 │ 1173.81 ms │ 1146.57 ms │ no change │
│ QQuery 14 │ 755.95 ms │ 776.39 ms │ no change │
│ QQuery 15 │ 757.28 ms │ 776.32 ms │ no change │
│ QQuery 16 │ 1577.92 ms │ 1584.93 ms │ no change │
│ QQuery 17 │ 1551.37 ms │ 1589.69 ms │ no change │
│ QQuery 18 │ 2979.57 ms │ 2860.07 ms │ no change │
│ QQuery 19 │ 79.11 ms │ 80.25 ms │ no change │
│ QQuery 20 │ 1117.55 ms │ 1120.36 ms │ no change │
│ QQuery 21 │ 1268.95 ms │ 1265.01 ms │ no change │
│ QQuery 22 │ 2079.90 ms │ 2092.64 ms │ no change │
│ QQuery 23 │ 7333.93 ms │ 7329.52 ms │ no change │
│ QQuery 24 │ 389.95 ms │ 390.15 ms │ no change │
│ QQuery 25 │ 277.49 ms │ 260.52 ms │ +1.07x faster │
│ QQuery 26 │ 390.78 ms │ 380.57 ms │ no change │
│ QQuery 27 │ 1514.84 ms │ 1507.57 ms │ no change │
│ QQuery 28 │ 12714.12 ms │ 12888.81 ms │ no change │
│ QQuery 29 │ 520.11 ms │ 520.26 ms │ no change │
│ QQuery 30 │ 748.09 ms │ 766.14 ms │ no change │
│ QQuery 31 │ 760.21 ms │ 779.52 ms │ no change │
│ QQuery 32 │ 2351.95 ms │ 2446.23 ms │ no change │
│ QQuery 33 │ 3151.50 ms │ 3208.65 ms │ no change │
│ QQuery 34 │ 3243.75 ms │ 3272.06 ms │ no change │
│ QQuery 35 │ 1235.44 ms │ 1253.88 ms │ no change │
│ QQuery 36 │ 119.02 ms │ 117.32 ms │ no change │
│ QQuery 37 │ 51.38 ms │ 50.01 ms │ no change │
│ QQuery 38 │ 117.30 ms │ 114.31 ms │ no change │
│ QQuery 39 │ 187.59 ms │ 188.05 ms │ no change │
│ QQuery 40 │ 40.27 ms │ 40.98 ms │ no change │
│ QQuery 41 │ 42.51 ms │ 38.16 ms │ +1.11x faster │
│ QQuery 42 │ 31.63 ms │ 31.84 ms │ no change │
└──────────────┴─────────────┴───────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 53409.93ms │
│ Total Time (test_non_exact_size_batch) │ 53801.99ms │
│ Average Time (HEAD) │ 1242.09ms │
│ Average Time (test_non_exact_size_batch) │ 1251.21ms │
│ Queries Faster │ 2 │
│ Queries Slower │ 2 │
│ Queries with No Change │ 39 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴────────────┘

[github-actions]

Thank you for your contribution. Unfortunately, this pull request is stale because it has been open 60 days with no activity. Please remove the stale label or comment or this will be closed in 7 days.