Io dynamic by adriangb · Pull Request #21571 · apache/datafusion

adriangb · 2026-04-12T13:25:00Z

Which issue does this PR close?

Closes #.

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

This reverts commit c479ed68e02a4337eb9cbafa8ec1d70b4e5c2345.

Applies apache#20363 on top of the io-dynamic branch, resolving conflicts in opener.rs and source.rs by adapting the new predicate_conjuncts/selectivity_tracker fields to the direct ParquetMorselizer construction pattern used on this branch. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

adriangb · 2026-04-12T13:30:52Z

run benchmark clickbench_partitioned

env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: true
  DATAFUSION_EXECUTION_PARQUET_REORDER_FILTERS: true
baseline:
  ref: 9124f2b31994c2de24935cd207bd8dd50ffdef40
changed:
  ref: 2d4f79bbb959f0461fc19bc5acf13e699872167c

adriangbot · 2026-04-12T13:33:29Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

Cloning into '/workspace/datafusion-branch'...
io-dynamic
From https://github.com/apache/datafusion
 * [new ref]         refs/pull/21571/head -> io-dynamic
 * branch            main                 -> FETCH_HEAD
Switched to branch 'io-dynamic'
4389f14e70daa859c8ec41f9f09437c0b8e2bb55
HEAD is now at 2d4f79b Apply PR #20363: Adaptive filter selectivity evaluation
Cloning into '/workspace/datafusion-base'...
fatal: reference is not a tree: 9124f2b31994c2de24935cd207bd8dd50ffdef40

File an issue against this benchmark runner

adriangb · 2026-04-12T13:55:20Z

run benchmark clickbench_partitioned

env:
  DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: true
  DATAFUSION_EXECUTION_PARQUET_REORDER_FILTERS: true
baseline:
  ref: 9124f2b31994c2de24935cd207bd8dd50ffdef40
changed:
  ref: 2d4f79bbb959f0461fc19bc5acf13e699872167c

adriangbot · 2026-04-12T13:58:00Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4231643483-1117-z5rsk 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing 2d4f79b (2d4f79b) to 9124f2b diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-12T14:16:19Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

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

Comparing HEAD and io-dynamic
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                            io-dynamic ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.39 / 4.77 ±6.58 / 17.93 ms │          1.36 / 4.74 ±6.56 / 17.86 ms │     no change │
│ QQuery 1  │        13.25 / 13.44 ±0.19 / 13.70 ms │        13.47 / 14.32 ±0.76 / 15.71 ms │  1.07x slower │
│ QQuery 2  │        37.43 / 37.95 ±0.29 / 38.26 ms │        39.80 / 40.17 ±0.30 / 40.51 ms │  1.06x slower │
│ QQuery 3  │        35.81 / 36.91 ±1.11 / 38.90 ms │        35.56 / 36.75 ±1.35 / 38.48 ms │     no change │
│ QQuery 4  │     292.58 / 298.34 ±4.02 / 305.06 ms │     305.63 / 311.69 ±5.85 / 319.52 ms │     no change │
│ QQuery 5  │     336.87 / 340.49 ±3.93 / 346.10 ms │     334.45 / 340.53 ±3.70 / 344.48 ms │     no change │
│ QQuery 6  │           6.37 / 6.97 ±0.54 / 7.78 ms │           6.91 / 7.64 ±0.73 / 9.01 ms │  1.10x slower │
│ QQuery 7  │        17.23 / 17.67 ±0.26 / 18.00 ms │        15.68 / 16.17 ±0.50 / 17.12 ms │ +1.09x faster │
│ QQuery 8  │     416.25 / 420.33 ±2.91 / 423.57 ms │     418.75 / 426.91 ±8.31 / 442.48 ms │     no change │
│ QQuery 9  │    567.82 / 582.55 ±15.25 / 606.29 ms │     585.66 / 596.48 ±7.38 / 604.80 ms │     no change │
│ QQuery 10 │     105.35 / 107.31 ±1.96 / 110.32 ms │     106.17 / 108.51 ±2.59 / 113.32 ms │     no change │
│ QQuery 11 │     116.43 / 117.04 ±0.69 / 118.39 ms │     116.78 / 118.21 ±1.28 / 119.73 ms │     no change │
│ QQuery 12 │     358.33 / 362.82 ±4.03 / 368.91 ms │     322.94 / 331.43 ±6.95 / 342.22 ms │ +1.09x faster │
│ QQuery 13 │    511.86 / 529.79 ±14.67 / 548.71 ms │     513.09 / 527.97 ±8.03 / 537.25 ms │     no change │
│ QQuery 14 │     373.72 / 379.97 ±6.43 / 391.18 ms │     377.84 / 380.48 ±2.56 / 385.00 ms │     no change │
│ QQuery 15 │     371.15 / 380.19 ±7.37 / 392.00 ms │    371.58 / 394.32 ±24.79 / 436.26 ms │     no change │
│ QQuery 16 │     748.10 / 755.21 ±6.55 / 766.23 ms │    759.59 / 770.74 ±15.46 / 801.04 ms │     no change │
│ QQuery 17 │     746.22 / 750.93 ±3.75 / 755.51 ms │     752.32 / 759.06 ±5.06 / 766.20 ms │     no change │
│ QQuery 18 │ 1483.01 / 1599.53 ±64.21 / 1651.80 ms │ 1526.99 / 1627.09 ±67.43 / 1733.07 ms │     no change │
│ QQuery 19 │        33.19 / 34.87 ±1.68 / 38.06 ms │        31.48 / 35.92 ±7.75 / 51.41 ms │     no change │
│ QQuery 20 │    527.55 / 543.64 ±15.48 / 568.47 ms │     532.77 / 542.90 ±8.41 / 554.41 ms │     no change │
│ QQuery 21 │     593.09 / 599.21 ±7.13 / 612.61 ms │     595.38 / 603.90 ±6.00 / 614.05 ms │     no change │
│ QQuery 22 │  980.20 / 1001.87 ±15.29 / 1024.90 ms │     829.01 / 837.37 ±6.23 / 844.80 ms │ +1.20x faster │
│ QQuery 23 │     127.01 / 136.76 ±8.30 / 150.75 ms │    178.70 / 233.44 ±49.22 / 299.12 ms │  1.71x slower │
│ QQuery 24 │        44.14 / 46.13 ±1.59 / 48.85 ms │        41.55 / 47.91 ±5.79 / 58.79 ms │     no change │
│ QQuery 25 │     151.80 / 153.17 ±1.05 / 154.57 ms │     122.21 / 124.21 ±1.45 / 126.66 ms │ +1.23x faster │
│ QQuery 26 │        64.19 / 65.90 ±1.36 / 67.34 ms │        59.67 / 62.90 ±2.59 / 66.84 ms │     no change │
│ QQuery 27 │     744.73 / 751.84 ±6.49 / 762.05 ms │     662.01 / 667.77 ±6.34 / 680.03 ms │ +1.13x faster │
│ QQuery 28 │ 3202.90 / 3265.29 ±31.69 / 3287.85 ms │ 3124.71 / 3156.36 ±17.47 / 3176.21 ms │     no change │
│ QQuery 29 │        45.76 / 50.35 ±4.28 / 57.99 ms │        46.90 / 52.11 ±5.76 / 59.65 ms │     no change │
│ QQuery 30 │     368.57 / 371.64 ±4.15 / 379.69 ms │     367.09 / 373.29 ±6.33 / 384.59 ms │     no change │
│ QQuery 31 │     362.77 / 366.04 ±3.06 / 371.36 ms │     357.76 / 361.10 ±4.11 / 368.71 ms │     no change │
│ QQuery 32 │ 1278.08 / 1391.83 ±69.42 / 1487.40 ms │ 1151.03 / 1163.58 ±15.97 / 1193.32 ms │ +1.20x faster │
│ QQuery 33 │ 1654.35 / 1680.21 ±21.98 / 1709.05 ms │ 1634.98 / 1650.90 ±10.31 / 1667.40 ms │     no change │
│ QQuery 34 │ 1639.96 / 1673.43 ±21.57 / 1703.51 ms │ 1652.16 / 1736.06 ±50.15 / 1810.26 ms │     no change │
│ QQuery 35 │     379.69 / 386.83 ±5.29 / 396.07 ms │     377.82 / 382.07 ±3.68 / 387.45 ms │     no change │
│ QQuery 36 │        75.55 / 78.56 ±3.46 / 83.59 ms │        67.67 / 69.74 ±1.77 / 72.21 ms │ +1.13x faster │
│ QQuery 37 │        40.97 / 45.75 ±3.51 / 51.15 ms │        37.79 / 39.01 ±0.93 / 40.63 ms │ +1.17x faster │
│ QQuery 38 │        38.38 / 41.60 ±2.52 / 44.54 ms │        43.27 / 44.59 ±1.44 / 47.18 ms │  1.07x slower │
│ QQuery 39 │     135.79 / 139.39 ±3.67 / 146.23 ms │     127.44 / 137.88 ±6.17 / 146.71 ms │     no change │
│ QQuery 40 │        23.33 / 26.00 ±1.58 / 27.74 ms │        21.31 / 23.86 ±2.37 / 28.27 ms │ +1.09x faster │
│ QQuery 41 │        19.11 / 20.90 ±1.44 / 23.17 ms │        18.63 / 20.98 ±2.09 / 24.69 ms │     no change │
│ QQuery 42 │        16.15 / 16.85 ±0.47 / 17.32 ms │        16.79 / 17.17 ±0.25 / 17.54 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)         │ 19630.26ms │
│ Total Time (io-dynamic)   │ 19198.24ms │
│ Average Time (HEAD)       │   456.52ms │
│ Average Time (io-dynamic) │   446.47ms │
│ Queries Faster            │          9 │
│ Queries Slower            │          5 │
│ Queries with No Change    │         29 │
│ Queries with Failure      │          0 │
└───────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	99.4s
Peak memory	41.9 GiB
Avg memory	29.3 GiB
CPU user	1003.4s
CPU sys	101.0s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	97.2s
Peak memory	39.8 GiB
Avg memory	30.4 GiB
CPU user	991.7s
CPU sys	87.3s
Peak spill	0 B

File an issue against this benchmark runner

adriangb · 2026-04-12T14:19:17Z

run benchmark clickbench_partitioned

baseline:
  ref: 9124f2b31994c2de24935cd207bd8dd50ffdef40
  env:
    DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: false
    DATAFUSION_EXECUTION_PARQUET_REORDER_FILTERS: true
changed:
  ref: 2d4f79bbb959f0461fc19bc5acf13e699872167c
  env:
    DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: true
    DATAFUSION_EXECUTION_PARQUET_REORDER_FILTERS: true

adriangbot · 2026-04-12T14:21:56Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4231682622-1118-pq5fj 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing 2d4f79b (2d4f79b) to 9124f2b diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-12T14:36:07Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

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

Comparing HEAD and io-dynamic
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                            io-dynamic ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.26 / 4.60 ±6.51 / 17.61 ms │          1.22 / 4.53 ±6.45 / 17.43 ms │      no change │
│ QQuery 1  │        12.53 / 13.13 ±0.31 / 13.42 ms │        12.39 / 13.01 ±0.31 / 13.23 ms │      no change │
│ QQuery 2  │        38.23 / 38.72 ±0.36 / 39.20 ms │        36.48 / 37.00 ±0.30 / 37.31 ms │      no change │
│ QQuery 3  │        32.39 / 35.39 ±2.08 / 37.55 ms │        33.50 / 34.71 ±0.76 / 35.46 ms │      no change │
│ QQuery 4  │     269.67 / 279.53 ±5.60 / 285.70 ms │     267.37 / 269.89 ±1.98 / 273.04 ms │      no change │
│ QQuery 5  │     313.83 / 317.92 ±3.11 / 322.83 ms │     308.87 / 313.61 ±3.14 / 317.33 ms │      no change │
│ QQuery 6  │           6.15 / 6.83 ±0.50 / 7.61 ms │           5.44 / 6.49 ±0.94 / 8.03 ms │      no change │
│ QQuery 7  │        14.58 / 14.74 ±0.12 / 14.89 ms │        14.42 / 14.77 ±0.21 / 15.05 ms │      no change │
│ QQuery 8  │     378.35 / 385.08 ±4.08 / 390.17 ms │     373.64 / 382.81 ±5.15 / 389.14 ms │      no change │
│ QQuery 9  │    525.90 / 543.86 ±16.03 / 564.13 ms │     533.26 / 536.94 ±4.06 / 544.61 ms │      no change │
│ QQuery 10 │        76.53 / 78.08 ±1.79 / 81.57 ms │     101.09 / 102.45 ±1.71 / 105.69 ms │   1.31x slower │
│ QQuery 11 │        87.33 / 88.51 ±0.63 / 89.15 ms │     110.53 / 111.22 ±0.82 / 112.78 ms │   1.26x slower │
│ QQuery 12 │     302.56 / 305.39 ±1.68 / 307.76 ms │     300.25 / 305.66 ±5.77 / 312.93 ms │      no change │
│ QQuery 13 │    430.22 / 445.50 ±11.43 / 464.67 ms │    472.68 / 504.02 ±31.86 / 563.48 ms │   1.13x slower │
│ QQuery 14 │     318.91 / 322.13 ±2.69 / 325.18 ms │     347.64 / 349.49 ±1.61 / 352.45 ms │   1.08x slower │
│ QQuery 15 │     338.02 / 353.60 ±9.60 / 364.37 ms │    330.72 / 343.88 ±12.70 / 367.36 ms │      no change │
│ QQuery 16 │    706.86 / 719.54 ±17.58 / 753.82 ms │     691.90 / 697.01 ±5.44 / 706.96 ms │      no change │
│ QQuery 17 │     694.96 / 701.44 ±5.09 / 708.54 ms │     687.80 / 693.17 ±6.16 / 702.85 ms │      no change │
│ QQuery 18 │ 1393.11 / 1496.93 ±76.61 / 1617.77 ms │ 1402.93 / 1503.29 ±79.65 / 1632.49 ms │      no change │
│ QQuery 19 │        30.28 / 31.02 ±0.87 / 32.68 ms │        30.15 / 31.39 ±1.68 / 34.70 ms │      no change │
│ QQuery 20 │     516.34 / 525.29 ±8.74 / 538.27 ms │     515.60 / 522.79 ±6.66 / 531.47 ms │      no change │
│ QQuery 21 │     596.65 / 598.55 ±1.63 / 601.52 ms │     576.02 / 584.46 ±5.58 / 593.11 ms │      no change │
│ QQuery 22 │  1061.48 / 1068.86 ±6.73 / 1080.88 ms │     815.35 / 823.96 ±8.90 / 841.00 ms │  +1.30x faster │
│ QQuery 23 │ 3340.87 / 3373.93 ±26.02 / 3419.41 ms │    175.49 / 256.78 ±55.69 / 304.02 ms │ +13.14x faster │
│ QQuery 24 │        44.67 / 46.24 ±1.50 / 48.63 ms │        39.23 / 42.79 ±3.27 / 48.47 ms │  +1.08x faster │
│ QQuery 25 │     114.48 / 115.68 ±1.42 / 118.15 ms │     118.66 / 120.71 ±1.40 / 122.53 ms │      no change │
│ QQuery 26 │        42.77 / 43.82 ±0.89 / 45.46 ms │        58.82 / 61.17 ±1.91 / 63.82 ms │   1.40x slower │
│ QQuery 27 │     664.71 / 673.93 ±7.64 / 687.22 ms │     653.68 / 657.77 ±2.35 / 660.72 ms │      no change │
│ QQuery 28 │ 3068.29 / 3094.08 ±16.93 / 3110.33 ms │  3064.52 / 3073.89 ±7.36 / 3082.55 ms │      no change │
│ QQuery 29 │        45.67 / 50.42 ±3.95 / 55.83 ms │        44.04 / 47.78 ±2.89 / 52.33 ms │  +1.06x faster │
│ QQuery 30 │     342.28 / 344.92 ±1.64 / 346.84 ms │     340.90 / 344.82 ±3.69 / 351.44 ms │      no change │
│ QQuery 31 │     347.41 / 350.39 ±2.53 / 354.86 ms │     337.75 / 341.70 ±3.12 / 345.81 ms │      no change │
│ QQuery 32 │ 1267.73 / 1316.10 ±47.37 / 1406.70 ms │ 1050.67 / 1065.47 ±10.60 / 1083.63 ms │  +1.24x faster │
│ QQuery 33 │ 1607.10 / 1688.19 ±68.92 / 1815.57 ms │  1504.95 / 1522.13 ±9.01 / 1529.49 ms │  +1.11x faster │
│ QQuery 34 │ 1533.96 / 1592.64 ±47.43 / 1645.93 ms │ 1531.92 / 1573.93 ±28.78 / 1610.10 ms │      no change │
│ QQuery 35 │     325.63 / 332.27 ±4.59 / 339.21 ms │     333.22 / 337.86 ±5.83 / 348.65 ms │      no change │
│ QQuery 36 │        64.54 / 68.38 ±2.45 / 71.97 ms │        62.24 / 64.50 ±1.59 / 67.01 ms │  +1.06x faster │
│ QQuery 37 │        37.10 / 40.27 ±2.12 / 43.58 ms │        36.68 / 38.34 ±0.91 / 39.23 ms │      no change │
│ QQuery 38 │        41.98 / 43.33 ±1.02 / 45.03 ms │        38.12 / 39.56 ±0.96 / 40.53 ms │  +1.10x faster │
│ QQuery 39 │     134.04 / 136.01 ±1.27 / 137.60 ms │     126.83 / 128.66 ±2.20 / 132.84 ms │  +1.06x faster │
│ QQuery 40 │        14.92 / 19.57 ±2.90 / 23.59 ms │        22.02 / 23.28 ±1.12 / 24.96 ms │   1.19x slower │
│ QQuery 41 │        14.38 / 15.53 ±0.75 / 16.32 ms │        17.46 / 18.08 ±0.44 / 18.59 ms │   1.16x slower │
│ QQuery 42 │        14.26 / 14.58 ±0.23 / 14.91 ms │        15.65 / 16.11 ±0.24 / 16.28 ms │   1.10x slower │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)         │ 21734.92ms │
│ Total Time (io-dynamic)   │ 17961.88ms │
│ Average Time (HEAD)       │   505.46ms │
│ Average Time (io-dynamic) │   417.72ms │
│ Queries Faster            │          9 │
│ Queries Slower            │          8 │
│ Queries with No Change    │         26 │
│ Queries with Failure      │          0 │
└───────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	109.8s
Peak memory	38.2 GiB
Avg memory	28.2 GiB
CPU user	1115.4s
CPU sys	99.8s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	91.0s
Peak memory	40.5 GiB
Avg memory	31.4 GiB
CPU user	928.7s
CPU sys	79.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangb · 2026-04-12T14:40:19Z

run benchmark clickbench_partitioned

baseline:
  ref: 9124f2b31994c2de24935cd207bd8dd50ffdef40
  env:
    DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: false
    DATAFUSION_EXECUTION_PARQUET_REORDER_FILTERS: true
changed:
  ref: 9124f2b31994c2de24935cd207bd8dd50ffdef40
  env:
    DATAFUSION_EXECUTION_PARQUET_PUSHDOWN_FILTERS: true
    DATAFUSION_EXECUTION_PARQUET_REORDER_FILTERS: true

adriangbot · 2026-04-12T14:43:08Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4231718947-1119-m5ptf 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing 9124f2b (9124f2b) to 9124f2b diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-12T14:57:23Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

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

Comparing HEAD and io-dynamic
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                             io-dynamic ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.20 / 4.40 ±6.30 / 17.00 ms │           1.20 / 4.41 ±6.32 / 17.05 ms │      no change │
│ QQuery 1  │        11.96 / 12.35 ±0.21 / 12.57 ms │         12.19 / 12.80 ±0.61 / 13.98 ms │      no change │
│ QQuery 2  │        36.07 / 36.41 ±0.26 / 36.80 ms │         36.01 / 36.27 ±0.18 / 36.44 ms │      no change │
│ QQuery 3  │        31.58 / 34.34 ±2.03 / 36.90 ms │         32.82 / 34.81 ±1.56 / 37.17 ms │      no change │
│ QQuery 4  │     253.52 / 258.71 ±2.92 / 261.73 ms │      259.33 / 260.90 ±1.67 / 264.04 ms │      no change │
│ QQuery 5  │     295.84 / 300.06 ±3.01 / 303.79 ms │      298.62 / 304.41 ±7.49 / 319.07 ms │      no change │
│ QQuery 6  │         7.75 / 10.67 ±4.48 / 19.50 ms │           6.55 / 8.21 ±1.85 / 11.21 ms │  +1.30x faster │
│ QQuery 7  │        13.34 / 13.65 ±0.19 / 13.92 ms │         16.46 / 17.14 ±0.72 / 18.50 ms │   1.26x slower │
│ QQuery 8  │     362.11 / 368.59 ±5.90 / 377.82 ms │      390.10 / 398.75 ±5.63 / 406.71 ms │   1.08x slower │
│ QQuery 9  │    512.03 / 529.99 ±11.04 / 542.64 ms │     531.94 / 552.29 ±13.03 / 565.99 ms │      no change │
│ QQuery 10 │        76.66 / 78.56 ±2.13 / 82.73 ms │       97.78 / 100.41 ±3.46 / 107.19 ms │   1.28x slower │
│ QQuery 11 │        86.55 / 87.93 ±0.97 / 89.16 ms │      108.29 / 109.43 ±0.76 / 110.48 ms │   1.24x slower │
│ QQuery 12 │    292.87 / 302.44 ±10.49 / 320.38 ms │      324.58 / 327.07 ±2.63 / 331.13 ms │   1.08x slower │
│ QQuery 13 │     447.61 / 453.87 ±6.30 / 464.37 ms │     454.76 / 467.83 ±14.82 / 493.78 ms │      no change │
│ QQuery 14 │     319.06 / 324.06 ±3.67 / 329.32 ms │      334.53 / 335.22 ±0.62 / 336.21 ms │      no change │
│ QQuery 15 │    328.19 / 338.01 ±12.50 / 361.17 ms │     319.86 / 340.83 ±19.13 / 375.54 ms │      no change │
│ QQuery 16 │    670.84 / 695.84 ±17.30 / 717.84 ms │      670.76 / 674.68 ±2.81 / 679.36 ms │      no change │
│ QQuery 17 │    669.85 / 697.19 ±21.36 / 719.87 ms │      668.85 / 673.16 ±2.84 / 676.81 ms │      no change │
│ QQuery 18 │ 1327.20 / 1373.56 ±43.16 / 1430.10 ms │  1429.94 / 1510.64 ±62.61 / 1593.94 ms │   1.10x slower │
│ QQuery 19 │        28.45 / 29.82 ±1.58 / 32.70 ms │        30.09 / 42.59 ±17.48 / 76.93 ms │   1.43x slower │
│ QQuery 20 │     512.35 / 525.16 ±8.05 / 535.05 ms │      512.47 / 521.26 ±7.95 / 534.57 ms │      no change │
│ QQuery 21 │     591.78 / 595.77 ±3.40 / 601.49 ms │     577.45 / 590.90 ±11.72 / 607.61 ms │      no change │
│ QQuery 22 │  1050.93 / 1062.40 ±8.90 / 1078.12 ms │      973.38 / 985.36 ±7.94 / 995.70 ms │  +1.08x faster │
│ QQuery 23 │ 3283.30 / 3359.79 ±40.46 / 3401.99 ms │      120.84 / 130.77 ±8.99 / 146.66 ms │ +25.69x faster │
│ QQuery 24 │        42.28 / 45.76 ±2.71 / 49.94 ms │         43.27 / 44.46 ±0.84 / 45.51 ms │      no change │
│ QQuery 25 │     112.20 / 113.78 ±1.18 / 115.68 ms │      144.79 / 146.86 ±2.27 / 151.15 ms │   1.29x slower │
│ QQuery 26 │        42.38 / 43.86 ±2.00 / 47.61 ms │         62.00 / 63.84 ±1.46 / 65.56 ms │   1.46x slower │
│ QQuery 27 │     657.07 / 663.06 ±5.09 / 669.70 ms │      716.98 / 725.90 ±8.30 / 740.48 ms │   1.09x slower │
│ QQuery 28 │ 2972.32 / 3038.21 ±53.87 / 3105.47 ms │  3074.12 / 3165.04 ±55.14 / 3212.18 ms │      no change │
│ QQuery 29 │        42.76 / 46.40 ±3.05 / 50.43 ms │         45.52 / 49.16 ±4.03 / 56.37 ms │   1.06x slower │
│ QQuery 30 │     323.46 / 329.97 ±7.12 / 342.52 ms │      344.86 / 351.63 ±6.82 / 364.76 ms │   1.07x slower │
│ QQuery 31 │     332.11 / 337.39 ±5.94 / 348.34 ms │      341.71 / 349.36 ±6.94 / 361.84 ms │      no change │
│ QQuery 32 │ 1038.33 / 1073.57 ±20.30 / 1092.40 ms │ 1022.12 / 1160.17 ±144.15 / 1402.54 ms │   1.08x slower │
│ QQuery 33 │ 1427.49 / 1451.83 ±40.31 / 1532.17 ms │  1423.07 / 1445.81 ±17.63 / 1466.64 ms │      no change │
│ QQuery 34 │ 1412.63 / 1479.02 ±55.23 / 1575.96 ms │  1449.73 / 1515.16 ±68.18 / 1602.07 ms │      no change │
│ QQuery 35 │     346.92 / 352.09 ±6.06 / 363.97 ms │     310.26 / 333.56 ±24.12 / 372.10 ms │  +1.06x faster │
│ QQuery 36 │        65.10 / 69.00 ±3.57 / 75.11 ms │         65.73 / 69.30 ±4.06 / 77.24 ms │      no change │
│ QQuery 37 │        37.25 / 39.56 ±1.79 / 41.93 ms │         40.58 / 43.45 ±2.36 / 47.10 ms │   1.10x slower │
│ QQuery 38 │        42.32 / 43.74 ±1.15 / 45.36 ms │         36.95 / 41.17 ±2.29 / 43.72 ms │  +1.06x faster │
│ QQuery 39 │     131.12 / 136.78 ±3.59 / 141.39 ms │      127.42 / 131.94 ±3.06 / 136.67 ms │      no change │
│ QQuery 40 │        15.20 / 17.11 ±1.41 / 18.78 ms │         18.50 / 23.13 ±3.23 / 26.47 ms │   1.35x slower │
│ QQuery 41 │        14.64 / 15.99 ±1.24 / 18.32 ms │         18.46 / 19.39 ±0.77 / 20.32 ms │   1.21x slower │
│ QQuery 42 │        14.48 / 14.91 ±0.50 / 15.78 ms │         15.28 / 15.91 ±0.74 / 17.32 ms │   1.07x slower │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)         │ 20805.60ms │
│ Total Time (io-dynamic)   │ 18135.39ms │
│ Average Time (HEAD)       │   483.85ms │
│ Average Time (io-dynamic) │   421.75ms │
│ Queries Faster            │          5 │
│ Queries Slower            │         17 │
│ Queries with No Change    │         21 │
│ Queries with Failure      │          0 │
└───────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	105.1s
Peak memory	39.6 GiB
Avg memory	33.0 GiB
CPU user	1094.7s
CPU sys	78.5s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	91.8s
Peak memory	39.8 GiB
Avg memory	29.1 GiB
CPU user	928.1s
CPU sys	90.3s
Peak spill	0 B

File an issue against this benchmark runner

alamb and others added 8 commits April 9, 2026 12:59

Rewrite FileStream in terms of Morselize API

8cd86f8

Address feedback from @adriangb

393c03f

Merge remote-tracking branch 'apache/main' into alamb/file_stream_split

9923ed9

Add Dynamic work scheduling in FileStream

97bc9c9

Revert "simple"

416570e

This reverts commit c479ed68e02a4337eb9cbafa8ec1d70b4e5c2345.

Refactor shared state creation

74032d6

spelling

9124f2b

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Io dynamic#21571

Io dynamic#21571
adriangb wants to merge 8 commits intoapache:mainfrom
adriangb:io-dynamic

adriangb commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

adriangb commented Apr 12, 2026

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangb commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

adriangbot commented Apr 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants