Prefetch Parquet row groups with a bounded I/O budget by Dandandan · Pull Request #23492 · apache/datafusion

Dandandan · 2026-07-12T12:18:26Z

Which issue does this PR close?

None; this is an experimental performance enhancement.

Rationale for this change

Sequential Parquet row-group reads are expensive when object-store requests have meaningful latency. Batching projected ranges can hide much of that latency without eagerly decoding data.

What changes are included in this PR?

Add an optional datafusion.execution.parquet.prefetch_size byte budget, disabled by default.
Prefetch complete projected row groups through the push decoder while accounting for bytes already buffered.
Keep prefetching I/O-only; decoding remains lazy and required reads are never constrained by the budget.
Add tests, configuration/protobuf support, documentation, and full per-query benchmark results.

Results

With a 20 MiB budget, simulated object-store latency is 2-3x (see results brlow).

With latency disabled, performance remained effectively neutral: 1.02x for TPC-H and 1.01x for ClickBench. Raising the budget to 100 MiB provided no further aggregate benefit.

Are these changes tested?

Are there any user-facing changes?

Users can opt in by setting datafusion.execution.parquet.prefetch_size; the default remains disabled.

github-actions · 2026-07-12T12:32:09Z

Thank you for opening this pull request!

Reviewer note: cargo-semver-checks reported the current version number is not SemVer-compatible with the changes in this pull request (compared against the base branch).

Details

     Cloning apache/main
    Building datafusion-common v54.0.0 (current)
       Built [  35.865s] (current)
     Parsing datafusion-common v54.0.0 (current)
      Parsed [   0.059s] (current)
    Building datafusion-common v54.0.0 (baseline)
       Built [  34.527s] (baseline)
     Parsing datafusion-common v54.0.0 (baseline)
      Parsed [   0.061s] (baseline)
    Checking datafusion-common v54.0.0 -> v54.0.0 (no change; assume patch)
     Checked [   0.713s] 223 checks: 222 pass, 1 fail, 0 warn, 30 skip

--- failure constructible_struct_adds_field: externally-constructible struct adds field ---

Description:
A pub struct constructible with a struct literal has a new pub field. Existing struct literals must be updated to include the new field.
        ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.48.0/src/lints/constructible_struct_adds_field.ron

Failed in:
  field ParquetOptions.prefetch_size in /home/runner/work/datafusion/datafusion/datafusion/common/src/config.rs:1103

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [  73.784s] datafusion-common
    Building datafusion-datasource-parquet v54.0.0 (current)
       Built [  44.590s] (current)
     Parsing datafusion-datasource-parquet v54.0.0 (current)
      Parsed [   0.031s] (current)
    Building datafusion-datasource-parquet v54.0.0 (baseline)
       Built [  44.336s] (baseline)
     Parsing datafusion-datasource-parquet v54.0.0 (baseline)
      Parsed [   0.031s] (baseline)
    Checking datafusion-datasource-parquet v54.0.0 -> v54.0.0 (no change; assume patch)
     Checked [   0.148s] 223 checks: 223 pass, 30 skip
     Summary no semver update required
    Finished [  90.281s] datafusion-datasource-parquet
    Building datafusion-proto v54.0.0 (current)
       Built [  60.972s] (current)
     Parsing datafusion-proto v54.0.0 (current)
      Parsed [   0.018s] (current)
    Building datafusion-proto v54.0.0 (baseline)
       Built [  61.071s] (baseline)
     Parsing datafusion-proto v54.0.0 (baseline)
      Parsed [   0.019s] (baseline)
    Checking datafusion-proto v54.0.0 -> v54.0.0 (no change; assume patch)
     Checked [   0.258s] 223 checks: 223 pass, 30 skip
     Summary no semver update required
    Finished [ 124.169s] datafusion-proto
    Building datafusion-proto-common v54.0.0 (current)
       Built [  21.787s] (current)
     Parsing datafusion-proto-common v54.0.0 (current)
      Parsed [   0.048s] (current)
    Building datafusion-proto-common v54.0.0 (baseline)
       Built [  21.809s] (baseline)
     Parsing datafusion-proto-common v54.0.0 (baseline)
      Parsed [   0.049s] (baseline)
    Checking datafusion-proto-common v54.0.0 -> v54.0.0 (no change; assume patch)
     Checked [   1.140s] 223 checks: 222 pass, 1 fail, 0 warn, 30 skip

--- failure constructible_struct_adds_field: externally-constructible struct adds field ---

Description:
A pub struct constructible with a struct literal has a new pub field. Existing struct literals must be updated to include the new field.
        ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.48.0/src/lints/constructible_struct_adds_field.ron

Failed in:
  field ParquetOptions.prefetch_size_opt in /home/runner/work/datafusion/datafusion/datafusion/proto-common/src/generated/prost.rs:908
  field ParquetOptions.prefetch_size_opt in /home/runner/work/datafusion/datafusion/datafusion/proto-common/src/generated/prost.rs:908
  field ParquetOptions.prefetch_size_opt in /home/runner/work/datafusion/datafusion/datafusion/proto-common/src/generated/prost.rs:908

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [  46.342s] datafusion-proto-common
    Building datafusion-proto-models v54.0.0 (current)
       Built [  25.179s] (current)
     Parsing datafusion-proto-models v54.0.0 (current)
      Parsed [   0.130s] (current)
    Building datafusion-proto-models v54.0.0 (baseline)
       Built [  25.033s] (baseline)
     Parsing datafusion-proto-models v54.0.0 (baseline)
      Parsed [   0.139s] (baseline)
    Checking datafusion-proto-models v54.0.0 -> v54.0.0 (no change; assume patch)
     Checked [   1.774s] 223 checks: 222 pass, 1 fail, 0 warn, 30 skip

--- failure constructible_struct_adds_field: externally-constructible struct adds field ---

Description:
A pub struct constructible with a struct literal has a new pub field. Existing struct literals must be updated to include the new field.
        ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.48.0/src/lints/constructible_struct_adds_field.ron

Failed in:
  field ParquetOptions.prefetch_size_opt in /home/runner/work/datafusion/datafusion/datafusion/proto-models/src/generated/datafusion_proto_common.rs:908
  field ParquetOptions.prefetch_size_opt in /home/runner/work/datafusion/datafusion/datafusion/proto-models/src/generated/datafusion_proto_common.rs:908

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [  53.270s] datafusion-proto-models
    Building datafusion-sqllogictest v54.0.0 (current)
       Built [ 185.421s] (current)
     Parsing datafusion-sqllogictest v54.0.0 (current)
      Parsed [   0.021s] (current)
    Building datafusion-sqllogictest v54.0.0 (baseline)
       Built [ 185.314s] (baseline)
     Parsing datafusion-sqllogictest v54.0.0 (baseline)
      Parsed [   0.022s] (baseline)
    Checking datafusion-sqllogictest v54.0.0 -> v54.0.0 (no change; assume patch)
     Checked [   0.089s] 223 checks: 223 pass, 30 skip
     Summary no semver update required
    Finished [ 375.084s] datafusion-sqllogictest

Dandandan · 2026-07-12T13:53:52Z

run benchmarks

env:
    DATAFUSION_EXECUTION_PARQUET_PREFETCH_SIZE: "20971520"
    SIMULATE_LATENCY: "true"

adriangbot · 2026-07-12T13:56:34Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4951427817-1009-cr5t7 6.12.85+ #1 SMP Mon May 11 08:17:35 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 codex/parquet-pushdecoder-prefetch (4d02e1e) to 2880e10 (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-07-12T13:56:40Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4951427817-1008-n2rgl 6.12.85+ #1 SMP Mon May 11 08:17:35 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 codex/parquet-pushdecoder-prefetch (4d02e1e) to 2880e10 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-07-12T14:08:59Z

🤖 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 codex_parquet-pushdecoder-prefetch
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ codex_parquet-pushdecoder-prefetch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     517.84 / 577.10 ±50.58 / 643.57 ms │  211.96 / 226.97 ±9.68 / 238.87 ms │ +2.54x faster │
│ QQuery 2  │     870.15 / 905.54 ±19.99 / 925.84 ms │ 651.23 / 734.80 ±57.90 / 798.91 ms │ +1.23x faster │
│ QQuery 3  │   902.74 / 1024.11 ±85.66 / 1112.70 ms │ 509.18 / 559.06 ±36.85 / 606.43 ms │ +1.83x faster │
│ QQuery 4  │     784.11 / 826.71 ±45.88 / 915.11 ms │  397.51 / 402.62 ±8.13 / 418.79 ms │ +2.05x faster │
│ QQuery 5  │  1081.22 / 1211.12 ±87.80 / 1344.85 ms │ 707.30 / 777.56 ±42.82 / 832.47 ms │ +1.56x faster │
│ QQuery 6  │     528.82 / 572.60 ±52.23 / 668.77 ms │  196.44 / 207.11 ±8.48 / 214.11 ms │ +2.76x faster │
│ QQuery 7  │ 1022.18 / 1177.67 ±123.30 / 1339.96 ms │ 603.47 / 771.48 ±98.70 / 886.04 ms │ +1.53x faster │
│ QQuery 8  │ 1143.26 / 1284.65 ±125.69 / 1501.18 ms │ 838.64 / 877.50 ±20.78 / 899.98 ms │ +1.46x faster │
│ QQuery 9  │  1207.19 / 1267.35 ±38.18 / 1320.28 ms │ 836.22 / 857.64 ±19.65 / 886.92 ms │ +1.48x faster │
│ QQuery 10 │    846.80 / 940.00 ±78.14 / 1067.06 ms │ 519.85 / 550.40 ±26.86 / 590.87 ms │ +1.71x faster │
│ QQuery 11 │     702.82 / 834.58 ±83.73 / 902.45 ms │ 616.14 / 692.04 ±44.30 / 753.16 ms │ +1.21x faster │
│ QQuery 12 │     720.05 / 839.08 ±69.01 / 930.83 ms │  400.48 / 405.38 ±7.68 / 420.65 ms │ +2.07x faster │
│ QQuery 13 │     401.34 / 410.46 ±12.36 / 434.45 ms │ 319.66 / 364.57 ±37.40 / 414.03 ms │ +1.13x faster │
│ QQuery 14 │     640.00 / 705.61 ±65.75 / 821.83 ms │ 232.00 / 326.13 ±57.74 / 405.80 ms │ +2.16x faster │
│ QQuery 15 │ 1095.10 / 1240.39 ±105.32 / 1369.28 ms │ 440.01 / 541.29 ±82.00 / 616.86 ms │ +2.29x faster │
│ QQuery 16 │     358.91 / 410.30 ±47.75 / 479.39 ms │ 264.25 / 327.11 ±58.98 / 396.35 ms │ +1.25x faster │
│ QQuery 17 │  1083.62 / 1230.32 ±78.95 / 1319.39 ms │ 510.97 / 568.71 ±40.03 / 633.36 ms │ +2.16x faster │
│ QQuery 18 │  1431.65 / 1552.91 ±68.27 / 1626.25 ms │  799.67 / 803.02 ±2.82 / 808.09 ms │ +1.93x faster │
│ QQuery 19 │     618.32 / 670.67 ±40.47 / 717.47 ms │ 293.37 / 347.77 ±39.60 / 415.20 ms │ +1.93x faster │
│ QQuery 20 │   985.73 / 1076.16 ±64.94 / 1176.67 ms │ 529.80 / 568.83 ±30.99 / 612.27 ms │ +1.89x faster │
│ QQuery 21 │ 2043.66 / 2163.86 ±141.43 / 2391.79 ms │  948.09 / 953.00 ±3.99 / 958.58 ms │ +2.27x faster │
│ QQuery 22 │     497.18 / 556.86 ±56.29 / 652.88 ms │  496.39 / 498.38 ±2.02 / 502.18 ms │ +1.12x faster │
└───────────┴────────────────────────────────────────┴────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                 ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                 │ 21478.03ms │
│ Total Time (codex_parquet-pushdecoder-prefetch)   │ 12361.38ms │
│ Average Time (HEAD)                               │   976.27ms │
│ Average Time (codex_parquet-pushdecoder-prefetch) │   561.88ms │
│ Queries Faster                                    │         22 │
│ Queries Slower                                    │          0 │
│ Queries with No Change                            │          0 │
│ Queries with Failure                              │          0 │
└───────────────────────────────────────────────────┴────────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	110.0s
Peak memory	857.4 MiB
Avg memory	418.9 MiB
CPU user	31.2s
CPU sys	3.7s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	65.0s
Peak memory	925.4 MiB
Avg memory	452.7 MiB
CPU user	30.2s
CPU sys	3.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-07-12T14:18:38Z

🤖 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 codex_parquet-pushdecoder-prefetch
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃     codex_parquet-pushdecoder-prefetch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.25 / 4.10 ±5.57 / 15.24 ms │           1.23 / 4.05 ±5.50 / 15.06 ms │     no change │
│ QQuery 1  │ 1885.45 / 1929.06 ±41.60 / 2007.55 ms │     728.11 / 746.96 ±19.05 / 778.40 ms │ +2.58x faster │
│ QQuery 2  │ 2160.87 / 2232.44 ±54.64 / 2285.33 ms │     782.88 / 801.80 ±13.32 / 821.66 ms │ +2.78x faster │
│ QQuery 3  │ 2151.58 / 2227.08 ±47.31 / 2292.47 ms │     777.39 / 810.45 ±19.35 / 833.64 ms │ +2.75x faster │
│ QQuery 4  │ 2345.35 / 2363.44 ±22.41 / 2406.06 ms │     891.99 / 922.88 ±16.07 / 935.92 ms │ +2.56x faster │
│ QQuery 5  │ 2358.82 / 2398.87 ±29.03 / 2437.10 ms │   946.88 / 1019.02 ±40.32 / 1056.94 ms │ +2.35x faster │
│ QQuery 6  │           1.26 / 1.40 ±0.22 / 1.84 ms │            1.25 / 1.41 ±0.21 / 1.83 ms │     no change │
│ QQuery 7  │ 1908.66 / 1986.85 ±57.91 / 2079.80 ms │     731.80 / 766.37 ±30.54 / 816.76 ms │ +2.59x faster │
│ QQuery 8  │ 2404.68 / 2473.29 ±58.82 / 2557.16 ms │   998.26 / 1058.74 ±33.99 / 1087.24 ms │ +2.34x faster │
│ QQuery 9  │ 2651.86 / 2756.05 ±65.72 / 2829.02 ms │  1172.96 / 1230.49 ±34.58 / 1271.37 ms │ +2.24x faster │
│ QQuery 10 │ 2211.22 / 2270.46 ±53.06 / 2368.39 ms │     802.09 / 851.58 ±25.53 / 873.93 ms │ +2.67x faster │
│ QQuery 11 │ 2219.12 / 2297.36 ±56.50 / 2358.53 ms │     828.28 / 848.52 ±15.92 / 872.29 ms │ +2.71x faster │
│ QQuery 12 │ 2410.96 / 2453.10 ±34.18 / 2499.73 ms │   984.75 / 1010.82 ±16.27 / 1034.15 ms │ +2.43x faster │
│ QQuery 13 │ 2447.61 / 2487.79 ±33.35 / 2539.14 ms │   1131.03 / 1140.07 ±6.50 / 1148.41 ms │ +2.18x faster │
│ QQuery 14 │ 2409.09 / 2438.69 ±27.81 / 2473.81 ms │  1013.24 / 1030.28 ±10.32 / 1045.66 ms │ +2.37x faster │
│ QQuery 15 │ 2367.54 / 2426.65 ±52.78 / 2514.29 ms │    941.85 / 976.09 ±26.80 / 1023.85 ms │ +2.49x faster │
│ QQuery 16 │ 2614.23 / 2704.85 ±69.70 / 2829.38 ms │  1293.08 / 1309.61 ±16.06 / 1335.14 ms │ +2.07x faster │
│ QQuery 17 │ 2649.61 / 2704.22 ±46.41 / 2788.36 ms │   1330.93 / 1339.38 ±8.27 / 1353.98 ms │ +2.02x faster │
│ QQuery 18 │ 3162.40 / 3208.70 ±32.99 / 3260.77 ms │  1846.66 / 1907.69 ±32.31 / 1933.62 ms │ +1.68x faster │
│ QQuery 19 │ 1757.14 / 1833.47 ±58.17 / 1932.49 ms │     735.76 / 762.86 ±16.76 / 781.71 ms │ +2.40x faster │
│ QQuery 20 │ 2646.42 / 2711.44 ±58.78 / 2797.96 ms │  1612.82 / 1657.04 ±47.64 / 1742.93 ms │ +1.64x faster │
│ QQuery 21 │ 2683.30 / 2732.72 ±32.84 / 2784.88 ms │  1794.34 / 1886.24 ±53.65 / 1960.83 ms │ +1.45x faster │
│ QQuery 22 │ 3125.74 / 3151.86 ±18.61 / 3170.79 ms │  2812.73 / 2902.76 ±59.74 / 2983.72 ms │ +1.09x faster │
│ QQuery 23 │ 4946.09 / 5056.73 ±72.63 / 5168.18 ms │ 4921.29 / 5017.94 ±118.21 / 5235.22 ms │     no change │
│ QQuery 24 │    487.49 / 508.92 ±23.92 / 541.81 ms │     300.63 / 350.32 ±32.65 / 401.14 ms │ +1.45x faster │
│ QQuery 25 │ 2256.13 / 2349.35 ±77.43 / 2474.04 ms │     841.84 / 884.28 ±32.17 / 926.18 ms │ +2.66x faster │
│ QQuery 26 │    468.86 / 522.48 ±60.67 / 635.82 ms │     319.28 / 354.91 ±20.74 / 380.38 ms │ +1.47x faster │
│ QQuery 27 │ 2736.42 / 2841.89 ±63.70 / 2926.75 ms │  1759.48 / 1805.63 ±53.21 / 1908.44 ms │ +1.57x faster │
│ QQuery 28 │ 5061.24 / 5158.44 ±62.10 / 5217.45 ms │  4083.56 / 4168.71 ±74.39 / 4271.46 ms │ +1.24x faster │
│ QQuery 29 │ 2110.07 / 2222.69 ±59.98 / 2287.86 ms │     771.02 / 827.21 ±29.57 / 854.70 ms │ +2.69x faster │
│ QQuery 30 │ 2444.41 / 2480.86 ±35.41 / 2547.41 ms │  1047.40 / 1093.91 ±26.86 / 1115.31 ms │ +2.27x faster │
│ QQuery 31 │ 2336.98 / 2417.41 ±48.46 / 2489.87 ms │   992.90 / 1039.70 ±27.15 / 1070.41 ms │ +2.33x faster │
│ QQuery 32 │ 2555.46 / 2624.33 ±50.05 / 2694.92 ms │  1147.51 / 1240.17 ±59.53 / 1314.85 ms │ +2.12x faster │
│ QQuery 33 │  3432.63 / 3447.52 ±9.56 / 3458.87 ms │  2452.68 / 2497.96 ±34.08 / 2540.05 ms │ +1.38x faster │
│ QQuery 34 │ 3435.15 / 3482.94 ±31.26 / 3521.14 ms │  2480.25 / 2509.02 ±19.83 / 2535.72 ms │ +1.39x faster │
│ QQuery 35 │ 2395.62 / 2430.96 ±35.98 / 2499.42 ms │   948.92 / 1034.53 ±46.85 / 1086.25 ms │ +2.35x faster │
│ QQuery 36 │    139.60 / 295.06 ±90.18 / 398.56 ms │     129.13 / 217.17 ±60.47 / 288.94 ms │ +1.36x faster │
│ QQuery 37 │    134.98 / 252.58 ±81.79 / 369.73 ms │     121.08 / 164.34 ±41.78 / 234.09 ms │ +1.54x faster │
│ QQuery 38 │    130.10 / 273.66 ±90.29 / 392.73 ms │     143.51 / 187.91 ±42.97 / 247.31 ms │ +1.46x faster │
│ QQuery 39 │    192.64 / 341.74 ±91.71 / 440.39 ms │     270.19 / 343.21 ±56.97 / 426.38 ms │     no change │
│ QQuery 40 │    117.85 / 248.10 ±84.36 / 350.64 ms │     114.04 / 147.16 ±29.34 / 195.90 ms │ +1.69x faster │
│ QQuery 41 │    116.66 / 242.80 ±81.00 / 344.59 ms │      85.70 / 155.74 ±49.67 / 215.60 ms │ +1.56x faster │
│ QQuery 42 │    113.88 / 240.39 ±81.54 / 344.86 ms │     113.67 / 153.26 ±37.74 / 220.17 ms │ +1.57x faster │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                 ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                 │ 89232.75ms │
│ Total Time (codex_parquet-pushdecoder-prefetch)   │ 49178.19ms │
│ Average Time (HEAD)                               │  2075.18ms │
│ Average Time (codex_parquet-pushdecoder-prefetch) │  1143.68ms │
│ Queries Faster                                    │         39 │
│ Queries Slower                                    │          0 │
│ Queries with No Change                            │          4 │
│ Queries with Failure                              │          0 │
└───────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	450.1s
Peak memory	9.3 GiB
Avg memory	1.8 GiB
CPU user	1011.3s
CPU sys	114.3s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	250.1s
Peak memory	10.3 GiB
Avg memory	2.3 GiB
CPU user	1001.7s
CPU sys	96.3s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-07-12T14:18:48Z

🤖 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 codex_parquet-pushdecoder-prefetch
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃     codex_parquet-pushdecoder-prefetch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     106.58 / 173.19 ±37.74 / 207.13 ms │     106.62 / 173.05 ±37.83 / 207.06 ms │     no change │
│ QQuery 2  │     520.99 / 648.79 ±69.02 / 707.70 ms │     522.66 / 656.64 ±71.45 / 707.45 ms │     no change │
│ QQuery 3  │     381.19 / 426.92 ±31.95 / 473.61 ms │     382.84 / 429.45 ±30.32 / 473.13 ms │     no change │
│ QQuery 4  │  1201.76 / 1268.72 ±55.15 / 1345.91 ms │  1184.88 / 1280.51 ±69.32 / 1355.09 ms │     no change │
│ QQuery 5  │     516.56 / 583.71 ±53.80 / 677.18 ms │     554.15 / 595.17 ±49.79 / 689.32 ms │     no change │
│ QQuery 6  │    545.98 / 696.44 ±102.40 / 822.34 ms │    541.11 / 700.27 ±105.90 / 817.30 ms │     no change │
│ QQuery 7  │     566.53 / 692.09 ±70.16 / 759.43 ms │     547.07 / 663.07 ±72.36 / 772.69 ms │     no change │
│ QQuery 8  │     572.51 / 608.18 ±31.15 / 652.73 ms │     483.31 / 569.71 ±57.82 / 642.67 ms │ +1.07x faster │
│ QQuery 9  │  1372.25 / 1491.12 ±96.71 / 1592.41 ms │  1378.59 / 1491.36 ±93.22 / 1589.76 ms │     no change │
│ QQuery 10 │     728.49 / 756.17 ±21.51 / 785.52 ms │     751.64 / 783.04 ±31.77 / 841.55 ms │     no change │
│ QQuery 11 │   928.13 / 1030.56 ±74.92 / 1136.87 ms │   930.96 / 1007.77 ±44.29 / 1051.23 ms │     no change │
│ QQuery 12 │     346.62 / 374.99 ±26.30 / 418.19 ms │     343.62 / 372.35 ±27.19 / 418.87 ms │     no change │
│ QQuery 13 │     661.82 / 742.25 ±47.83 / 796.12 ms │     678.82 / 749.29 ±42.15 / 789.21 ms │     no change │
│ QQuery 14 │  4051.16 / 4130.38 ±40.94 / 4166.10 ms │ 3750.80 / 3998.85 ±191.80 / 4285.01 ms │     no change │
│ QQuery 15 │     407.43 / 498.97 ±49.31 / 547.76 ms │     405.36 / 498.39 ±49.92 / 548.52 ms │     no change │
│ QQuery 16 │     108.07 / 173.96 ±37.55 / 208.04 ms │     108.08 / 174.86 ±38.29 / 211.34 ms │     no change │
│ QQuery 17 │     618.80 / 708.98 ±48.71 / 758.61 ms │     636.25 / 709.18 ±41.46 / 756.50 ms │     no change │
│ QQuery 18 │     716.99 / 800.69 ±60.78 / 903.69 ms │     718.64 / 804.29 ±61.47 / 908.25 ms │     no change │
│ QQuery 19 │     533.49 / 599.91 ±56.47 / 676.26 ms │     485.21 / 581.13 ±60.99 / 648.99 ms │     no change │
│ QQuery 20 │     326.95 / 400.03 ±41.53 / 452.35 ms │     325.16 / 398.38 ±40.75 / 447.53 ms │     no change │
│ QQuery 21 │     507.32 / 572.91 ±49.77 / 634.03 ms │     508.42 / 571.64 ±47.90 / 628.08 ms │     no change │
│ QQuery 22 │     473.16 / 526.56 ±32.61 / 573.88 ms │     477.99 / 527.70 ±30.10 / 570.34 ms │     no change │
│ QQuery 23 │  2546.90 / 2680.19 ±87.28 / 2759.78 ms │ 2467.53 / 2689.09 ±129.37 / 2827.59 ms │     no change │
│ QQuery 24 │ 2178.14 / 2353.72 ±107.81 / 2470.57 ms │ 2192.59 / 2357.34 ±104.69 / 2477.86 ms │     no change │
│ QQuery 25 │     683.41 / 731.96 ±35.37 / 770.57 ms │     687.55 / 732.63 ±37.10 / 777.97 ms │     no change │
│ QQuery 26 │     554.11 / 612.15 ±31.67 / 638.26 ms │     559.93 / 621.93 ±48.89 / 701.66 ms │     no change │
│ QQuery 27 │     145.32 / 254.81 ±82.74 / 359.16 ms │     141.28 / 254.27 ±84.30 / 359.61 ms │     no change │
│ QQuery 28 │     340.14 / 404.43 ±35.21 / 440.08 ms │     344.10 / 401.00 ±31.88 / 433.98 ms │     no change │
│ QQuery 29 │     721.09 / 785.39 ±42.21 / 823.69 ms │     720.83 / 788.90 ±49.25 / 845.60 ms │     no change │
│ QQuery 30 │     498.15 / 633.53 ±83.59 / 745.07 ms │     550.99 / 632.52 ±74.36 / 743.39 ms │     no change │
│ QQuery 31 │  1217.23 / 1316.93 ±63.89 / 1411.04 ms │  1209.55 / 1303.29 ±48.47 / 1350.40 ms │     no change │
│ QQuery 32 │     414.37 / 443.80 ±25.90 / 484.75 ms │     413.47 / 443.27 ±26.32 / 485.71 ms │     no change │
│ QQuery 33 │     520.67 / 594.34 ±43.53 / 644.76 ms │     522.71 / 594.73 ±41.87 / 640.84 ms │     no change │
│ QQuery 34 │     194.44 / 306.30 ±59.75 / 358.52 ms │     193.27 / 305.72 ±60.22 / 359.48 ms │     no change │
│ QQuery 35 │     713.37 / 751.72 ±32.52 / 801.63 ms │     652.37 / 757.92 ±71.76 / 874.26 ms │     no change │
│ QQuery 36 │     107.47 / 173.96 ±38.15 / 210.44 ms │     107.29 / 174.01 ±38.03 / 207.89 ms │     no change │
│ QQuery 37 │     221.50 / 276.50 ±30.10 / 310.05 ms │     221.14 / 276.54 ±30.37 / 310.38 ms │     no change │
│ QQuery 38 │     522.74 / 653.52 ±69.70 / 725.65 ms │     661.80 / 714.27 ±52.79 / 813.39 ms │  1.09x slower │
│ QQuery 39 │ 1851.53 / 2010.97 ±112.63 / 2154.59 ms │ 1853.46 / 2011.00 ±111.72 / 2147.70 ms │     no change │
│ QQuery 40 │     466.61 / 493.06 ±20.59 / 510.48 ms │     466.43 / 493.10 ±21.19 / 510.99 ms │     no change │
│ QQuery 41 │     176.03 / 278.26 ±73.02 / 380.60 ms │     176.22 / 279.11 ±73.17 / 381.42 ms │     no change │
│ QQuery 42 │     314.71 / 406.43 ±65.22 / 474.04 ms │     313.84 / 406.07 ±65.70 / 468.16 ms │     no change │
│ QQuery 43 │     106.56 / 153.47 ±40.37 / 207.44 ms │     106.76 / 152.76 ±40.64 / 206.50 ms │     no change │
│ QQuery 44 │     197.33 / 262.38 ±47.60 / 302.46 ms │     197.54 / 262.85 ±47.52 / 303.47 ms │     no change │
│ QQuery 45 │     487.64 / 567.00 ±51.27 / 644.98 ms │     452.22 / 541.16 ±67.39 / 647.78 ms │     no change │
│ QQuery 46 │     309.53 / 351.40 ±34.83 / 399.07 ms │     290.02 / 345.95 ±41.70 / 397.85 ms │     no change │
│ QQuery 47 │  1233.32 / 1273.36 ±41.79 / 1331.29 ms │  1243.07 / 1277.95 ±34.25 / 1325.17 ms │     no change │
│ QQuery 48 │     589.33 / 703.40 ±63.71 / 767.41 ms │     596.08 / 705.31 ±65.07 / 774.86 ms │     no change │
│ QQuery 49 │     581.69 / 634.95 ±48.32 / 702.23 ms │     585.40 / 647.06 ±45.19 / 706.56 ms │     no change │
│ QQuery 50 │     541.38 / 633.13 ±49.73 / 678.52 ms │     536.47 / 638.78 ±54.24 / 691.36 ms │     no change │
│ QQuery 51 │     512.78 / 539.91 ±18.29 / 557.91 ms │     503.40 / 539.53 ±28.83 / 565.73 ms │     no change │
│ QQuery 52 │     313.13 / 406.88 ±67.05 / 475.32 ms │     308.24 / 402.52 ±65.90 / 469.19 ms │     no change │
│ QQuery 53 │     412.65 / 453.24 ±47.30 / 515.52 ms │     413.16 / 453.60 ±48.61 / 513.85 ms │     no change │
│ QQuery 54 │     903.69 / 938.74 ±33.52 / 996.88 ms │    919.95 / 978.05 ±40.19 / 1036.27 ms │     no change │
│ QQuery 55 │     377.89 / 424.99 ±30.57 / 466.72 ms │     376.16 / 424.66 ±31.53 / 469.85 ms │     no change │
│ QQuery 56 │     539.08 / 614.93 ±52.77 / 691.87 ms │     523.17 / 584.63 ±35.25 / 623.56 ms │     no change │
│ QQuery 57 │  1111.73 / 1193.23 ±60.25 / 1276.10 ms │  1083.06 / 1190.79 ±80.41 / 1308.30 ms │     no change │
│ QQuery 58 │  1155.87 / 1326.89 ±96.27 / 1423.05 ms │  1154.35 / 1327.87 ±97.50 / 1425.56 ms │     no change │
│ QQuery 59 │    907.17 / 928.63 ±41.79 / 1012.20 ms │    906.65 / 932.97 ±48.49 / 1029.84 ms │     no change │
│ QQuery 60 │     580.21 / 669.50 ±60.32 / 742.09 ms │     575.99 / 657.50 ±54.62 / 740.92 ms │     no change │
│ QQuery 61 │      393.54 / 394.93 ±0.73 / 395.63 ms │      394.41 / 395.06 ±0.48 / 395.58 ms │     no change │
│ QQuery 62 │     580.07 / 600.53 ±24.64 / 631.75 ms │     579.20 / 600.48 ±25.13 / 631.92 ms │     no change │
│ QQuery 63 │     416.63 / 440.79 ±28.49 / 475.76 ms │     417.81 / 441.41 ±27.97 / 478.06 ms │     no change │
│ QQuery 64 │  3163.42 / 3300.85 ±92.90 / 3408.81 ms │  3185.37 / 3291.29 ±74.53 / 3385.49 ms │     no change │
│ QQuery 65 │     639.73 / 741.84 ±60.79 / 813.56 ms │     626.46 / 739.25 ±66.74 / 828.17 ms │     no change │
│ QQuery 66 │     562.08 / 592.08 ±46.81 / 685.41 ms │     562.65 / 595.68 ±41.36 / 675.49 ms │     no change │
│ QQuery 67 │     638.53 / 703.83 ±44.33 / 751.49 ms │     618.40 / 686.97 ±45.85 / 737.67 ms │     no change │
│ QQuery 68 │     283.56 / 317.83 ±19.44 / 338.42 ms │     284.24 / 313.24 ±15.64 / 326.57 ms │     no change │
│ QQuery 69 │    624.68 / 774.30 ±109.53 / 931.83 ms │     697.82 / 779.06 ±53.48 / 862.06 ms │     no change │
│ QQuery 70 │     684.00 / 759.38 ±61.29 / 856.30 ms │     694.83 / 763.45 ±56.21 / 855.44 ms │     no change │
│ QQuery 71 │     486.71 / 564.89 ±62.46 / 668.23 ms │     514.52 / 591.98 ±70.70 / 676.42 ms │     no change │
│ QQuery 72 │ 2916.19 / 3106.91 ±104.87 / 3207.29 ms │  2944.67 / 3067.50 ±77.33 / 3173.44 ms │     no change │
│ QQuery 73 │     194.76 / 308.62 ±58.89 / 359.19 ms │     194.33 / 307.21 ±60.82 / 357.65 ms │     no change │
│ QQuery 74 │     681.88 / 800.19 ±85.43 / 936.92 ms │     741.89 / 810.40 ±59.17 / 918.42 ms │     no change │
│ QQuery 75 │    941.40 / 982.65 ±31.65 / 1039.29 ms │    939.80 / 997.15 ±40.34 / 1045.65 ms │     no change │
│ QQuery 76 │     515.11 / 598.93 ±56.05 / 657.67 ms │     516.97 / 598.22 ±56.89 / 655.35 ms │     no change │
│ QQuery 77 │     643.49 / 718.76 ±59.96 / 813.47 ms │     685.82 / 747.77 ±44.19 / 797.45 ms │     no change │
│ QQuery 78 │     709.42 / 859.81 ±82.97 / 961.14 ms │     706.01 / 855.88 ±81.92 / 954.60 ms │     no change │
│ QQuery 79 │     451.54 / 500.08 ±45.23 / 582.67 ms │     452.44 / 505.92 ±50.57 / 599.34 ms │     no change │
│ QQuery 80 │     638.23 / 710.80 ±64.69 / 794.09 ms │     640.47 / 716.23 ±50.13 / 794.67 ms │     no change │
│ QQuery 81 │     488.87 / 604.10 ±70.91 / 685.21 ms │     477.80 / 638.71 ±81.07 / 690.62 ms │  1.06x slower │
│ QQuery 82 │     557.08 / 643.60 ±47.53 / 682.13 ms │     557.41 / 644.79 ±47.92 / 682.65 ms │     no change │
│ QQuery 83 │   982.04 / 1015.36 ±45.48 / 1098.42 ms │   860.55 / 1013.13 ±87.71 / 1097.65 ms │     no change │
│ QQuery 84 │     440.52 / 503.36 ±57.85 / 595.64 ms │     432.77 / 501.41 ±59.59 / 595.80 ms │     no change │
│ QQuery 85 │     822.30 / 891.55 ±50.67 / 951.27 ms │     814.27 / 886.32 ±51.04 / 951.85 ms │     no change │
│ QQuery 86 │     252.76 / 315.20 ±52.68 / 390.24 ms │     246.79 / 314.23 ±55.37 / 391.06 ms │     no change │
│ QQuery 87 │     659.90 / 682.30 ±30.63 / 740.93 ms │     568.16 / 679.97 ±84.74 / 780.02 ms │     no change │
│ QQuery 88 │      737.47 / 739.69 ±2.32 / 743.70 ms │      738.41 / 739.84 ±1.36 / 741.94 ms │     no change │
│ QQuery 89 │     425.29 / 460.88 ±28.19 / 485.29 ms │     426.70 / 459.86 ±27.04 / 482.95 ms │     no change │
│ QQuery 90 │     426.72 / 466.89 ±26.55 / 504.33 ms │     425.27 / 466.40 ±25.76 / 499.66 ms │     no change │
│ QQuery 91 │    523.47 / 663.29 ±112.45 / 807.77 ms │    526.00 / 664.16 ±112.23 / 806.66 ms │     no change │
│ QQuery 92 │     372.92 / 425.33 ±44.86 / 487.90 ms │     371.37 / 418.19 ±55.20 / 486.88 ms │     no change │
│ QQuery 93 │     465.94 / 481.16 ±15.90 / 509.12 ms │     470.08 / 482.52 ±13.05 / 504.31 ms │     no change │
│ QQuery 94 │     481.27 / 528.77 ±41.59 / 585.26 ms │     479.38 / 529.30 ±43.87 / 587.84 ms │     no change │
│ QQuery 95 │     691.53 / 776.13 ±46.19 / 825.17 ms │     694.67 / 778.67 ±45.34 / 823.84 ms │     no change │
│ QQuery 96 │     407.64 / 446.64 ±47.11 / 506.77 ms │     407.93 / 447.25 ±47.57 / 507.98 ms │     no change │
│ QQuery 97 │     459.05 / 519.48 ±49.68 / 610.74 ms │     454.41 / 518.76 ±50.53 / 610.06 ms │     no change │
│ QQuery 98 │     395.69 / 440.96 ±32.33 / 486.50 ms │     395.72 / 441.72 ±29.93 / 485.33 ms │     no change │
│ QQuery 99 │     508.23 / 611.98 ±56.66 / 658.90 ms │     508.99 / 620.09 ±63.68 / 695.76 ms │     no change │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                 ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                 │ 75138.17ms │
│ Total Time (codex_parquet-pushdecoder-prefetch)   │ 75086.23ms │
│ Average Time (HEAD)                               │   758.97ms │
│ Average Time (codex_parquet-pushdecoder-prefetch) │   758.45ms │
│ Queries Faster                                    │          1 │
│ Queries Slower                                    │          2 │
│ Queries with No Change                            │         96 │
│ Queries with Failure                              │          0 │
└───────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	380.1s
Peak memory	1.0 GiB
Avg memory	513.9 MiB
CPU user	239.0s
CPU sys	12.0s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	380.1s
Peak memory	1.1 GiB
Avg memory	578.0 MiB
CPU user	236.4s
CPU sys	11.0s
Peak spill	0 B

File an issue against this benchmark runner

Dandandan · 2026-07-12T15:40:41Z

run benchmarks

env:
    DATAFUSION_EXECUTION_PARQUET_PREFETCH_SIZE: "20971520"
    SIMULATE_LATENCY: "true"
    PUSHDOWN_FILTERS: "true"
    REORDER_FILTERS: "true"

adriangbot · 2026-07-12T15:42:47Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4951762927-1013-kpnxx 6.12.85+ #1 SMP Mon May 11 08:17:35 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 codex/parquet-pushdecoder-prefetch (4d02e1e) to 2880e10 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-07-12T15:43:33Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4951762927-1014-vdhd9 6.12.85+ #1 SMP Mon May 11 08:17:35 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 codex/parquet-pushdecoder-prefetch (4d02e1e) to 2880e10 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-07-12T15:43:38Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4951762927-1015-9dkvq 6.12.85+ #1 SMP Mon May 11 08:17:35 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 codex/parquet-pushdecoder-prefetch (4d02e1e) to 2880e10 (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-07-12T15:55:51Z

🤖 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 codex_parquet-pushdecoder-prefetch
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ codex_parquet-pushdecoder-prefetch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     573.50 / 616.27 ±35.03 / 662.98 ms │ 212.90 / 225.87 ±10.62 / 238.27 ms │ +2.73x faster │
│ QQuery 2  │     787.96 / 875.87 ±49.01 / 935.71 ms │ 692.85 / 747.96 ±43.03 / 799.29 ms │ +1.17x faster │
│ QQuery 3  │   947.45 / 1011.14 ±46.33 / 1074.99 ms │ 511.42 / 558.97 ±36.95 / 609.07 ms │ +1.81x faster │
│ QQuery 4  │     734.55 / 804.89 ±66.27 / 902.88 ms │  397.37 / 401.78 ±7.89 / 417.55 ms │ +2.00x faster │
│ QQuery 5  │ 1074.50 / 1210.11 ±109.24 / 1373.13 ms │ 703.02 / 775.31 ±41.85 / 824.53 ms │ +1.56x faster │
│ QQuery 6  │     531.42 / 575.99 ±39.13 / 631.95 ms │ 194.29 / 208.36 ±10.83 / 221.72 ms │ +2.76x faster │
│ QQuery 7  │  981.75 / 1136.33 ±105.76 / 1247.37 ms │ 604.58 / 772.00 ±97.78 / 886.68 ms │ +1.47x faster │
│ QQuery 8  │ 1121.66 / 1280.33 ±136.01 / 1458.29 ms │ 833.47 / 874.11 ±21.27 / 895.77 ms │ +1.46x faster │
│ QQuery 9  │  1126.28 / 1209.32 ±51.09 / 1278.37 ms │  839.29 / 841.14 ±1.16 / 842.49 ms │ +1.44x faster │
│ QQuery 10 │   913.75 / 1014.22 ±54.61 / 1060.92 ms │ 526.77 / 551.86 ±24.97 / 589.36 ms │ +1.84x faster │
│ QQuery 11 │    669.93 / 818.52 ±100.21 / 968.57 ms │ 615.77 / 691.70 ±44.58 / 754.40 ms │ +1.18x faster │
│ QQuery 12 │     746.30 / 840.99 ±60.69 / 909.27 ms │  403.69 / 408.94 ±6.21 / 420.56 ms │ +2.06x faster │
│ QQuery 13 │     399.96 / 427.32 ±30.87 / 465.21 ms │ 320.65 / 361.38 ±36.00 / 402.71 ms │ +1.18x faster │
│ QQuery 14 │     662.16 / 704.04 ±45.71 / 787.09 ms │ 236.25 / 327.04 ±57.34 / 409.68 ms │ +2.15x faster │
│ QQuery 15 │  1165.64 / 1228.56 ±58.47 / 1304.60 ms │ 443.52 / 540.92 ±78.98 / 613.88 ms │ +2.27x faster │
│ QQuery 16 │     377.21 / 413.33 ±37.05 / 483.09 ms │ 264.62 / 326.51 ±57.74 / 394.73 ms │ +1.27x faster │
│ QQuery 17 │ 1098.42 / 1309.29 ±147.52 / 1519.69 ms │ 515.28 / 563.14 ±35.57 / 623.78 ms │ +2.32x faster │
│ QQuery 18 │ 1379.50 / 1546.22 ±120.75 / 1718.06 ms │  801.71 / 809.93 ±5.35 / 818.15 ms │ +1.91x faster │
│ QQuery 19 │     572.62 / 680.62 ±84.83 / 822.46 ms │ 278.85 / 345.19 ±42.94 / 414.57 ms │ +1.97x faster │
│ QQuery 20 │   883.88 / 1027.48 ±76.20 / 1103.98 ms │ 539.15 / 576.08 ±34.98 / 636.89 ms │ +1.78x faster │
│ QQuery 21 │  2049.11 / 2185.75 ±81.40 / 2279.74 ms │  945.61 / 952.67 ±4.70 / 959.98 ms │ +2.29x faster │
│ QQuery 22 │     496.08 / 536.31 ±31.22 / 584.09 ms │  494.79 / 497.12 ±1.55 / 499.04 ms │ +1.08x faster │
└───────────┴────────────────────────────────────────┴────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                 ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                 │ 21452.91ms │
│ Total Time (codex_parquet-pushdecoder-prefetch)   │ 12357.99ms │
│ Average Time (HEAD)                               │   975.13ms │
│ Average Time (codex_parquet-pushdecoder-prefetch) │   561.73ms │
│ Queries Faster                                    │         22 │
│ Queries Slower                                    │          0 │
│ Queries with No Change                            │          0 │
│ Queries with Failure                              │          0 │
└───────────────────────────────────────────────────┴────────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	110.0s
Peak memory	801.3 MiB
Avg memory	402.1 MiB
CPU user	29.8s
CPU sys	3.8s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	65.0s
Peak memory	892.1 MiB
Avg memory	450.9 MiB
CPU user	29.4s
CPU sys	3.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-07-12T16:05:24Z

🤖 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 codex_parquet-pushdecoder-prefetch
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃     codex_parquet-pushdecoder-prefetch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.26 / 4.09 ±5.54 / 15.18 ms │           1.21 / 3.95 ±5.38 / 14.72 ms │     no change │
│ QQuery 1  │ 1899.70 / 1928.55 ±32.19 / 1989.64 ms │     727.32 / 746.70 ±19.51 / 779.84 ms │ +2.58x faster │
│ QQuery 2  │ 2204.53 / 2273.36 ±38.79 / 2319.14 ms │     783.30 / 802.52 ±13.48 / 824.53 ms │ +2.83x faster │
│ QQuery 3  │ 2158.89 / 2188.58 ±44.10 / 2275.07 ms │     778.60 / 810.43 ±18.59 / 833.55 ms │ +2.70x faster │
│ QQuery 4  │ 2281.15 / 2342.49 ±57.58 / 2427.21 ms │     878.76 / 923.04 ±24.05 / 941.99 ms │ +2.54x faster │
│ QQuery 5  │ 2355.92 / 2422.78 ±51.29 / 2514.29 ms │   943.11 / 1015.14 ±39.52 / 1053.13 ms │ +2.39x faster │
│ QQuery 6  │           1.27 / 1.41 ±0.22 / 1.84 ms │            1.26 / 1.42 ±0.22 / 1.86 ms │     no change │
│ QQuery 7  │ 1944.37 / 2010.23 ±51.28 / 2102.32 ms │     729.56 / 764.30 ±29.57 / 812.02 ms │ +2.63x faster │
│ QQuery 8  │ 2450.40 / 2477.87 ±16.59 / 2497.54 ms │  1023.30 / 1048.95 ±16.22 / 1071.79 ms │ +2.36x faster │
│ QQuery 9  │ 2640.44 / 2665.85 ±25.37 / 2710.84 ms │  1171.90 / 1226.77 ±32.59 / 1263.43 ms │ +2.17x faster │
│ QQuery 10 │ 2239.57 / 2309.75 ±44.74 / 2374.34 ms │     805.67 / 850.76 ±23.52 / 871.67 ms │ +2.71x faster │
│ QQuery 11 │ 2220.86 / 2275.82 ±60.15 / 2384.89 ms │     826.03 / 846.34 ±14.87 / 868.25 ms │ +2.69x faster │
│ QQuery 12 │ 2377.35 / 2426.38 ±41.60 / 2496.82 ms │   964.30 / 1006.74 ±23.88 / 1035.05 ms │ +2.41x faster │
│ QQuery 13 │ 2464.66 / 2479.72 ±14.66 / 2507.36 ms │  1090.73 / 1119.45 ±21.85 / 1151.59 ms │ +2.22x faster │
│ QQuery 14 │ 2396.46 / 2428.45 ±17.05 / 2447.56 ms │   995.99 / 1030.51 ±20.34 / 1054.54 ms │ +2.36x faster │
│ QQuery 15 │ 2384.87 / 2399.28 ±15.16 / 2425.58 ms │    947.36 / 977.71 ±18.76 / 1002.11 ms │ +2.45x faster │
│ QQuery 16 │ 2651.99 / 2689.71 ±42.67 / 2759.98 ms │  1291.91 / 1315.76 ±17.13 / 1337.12 ms │ +2.04x faster │
│ QQuery 17 │ 2627.83 / 2685.44 ±34.42 / 2719.89 ms │  1305.72 / 1327.67 ±15.19 / 1341.59 ms │ +2.02x faster │
│ QQuery 18 │ 3111.48 / 3158.40 ±37.01 / 3212.31 ms │  1862.76 / 1892.20 ±30.64 / 1938.50 ms │ +1.67x faster │
│ QQuery 19 │ 1745.03 / 1843.89 ±72.05 / 1947.99 ms │     744.10 / 760.88 ±11.09 / 778.79 ms │ +2.42x faster │
│ QQuery 20 │ 2654.25 / 2718.17 ±66.88 / 2828.85 ms │  1578.60 / 1644.03 ±35.38 / 1680.86 ms │ +1.65x faster │
│ QQuery 21 │ 2630.85 / 2707.82 ±52.83 / 2768.85 ms │  1799.08 / 1867.70 ±63.52 / 1951.95 ms │ +1.45x faster │
│ QQuery 22 │ 3087.79 / 3126.72 ±26.75 / 3169.67 ms │  2821.08 / 2885.40 ±63.93 / 2989.55 ms │ +1.08x faster │
│ QQuery 23 │ 4847.33 / 4973.43 ±89.52 / 5090.72 ms │  4978.92 / 5059.30 ±59.61 / 5147.01 ms │     no change │
│ QQuery 24 │    451.32 / 501.07 ±35.77 / 544.63 ms │     331.96 / 366.13 ±24.47 / 407.11 ms │ +1.37x faster │
│ QQuery 25 │ 2349.38 / 2390.68 ±27.54 / 2428.31 ms │     877.12 / 904.10 ±17.12 / 924.30 ms │ +2.64x faster │
│ QQuery 26 │    490.78 / 524.32 ±27.69 / 557.09 ms │     322.02 / 363.55 ±27.92 / 407.38 ms │ +1.44x faster │
│ QQuery 27 │ 2823.29 / 2902.44 ±57.27 / 2970.89 ms │  1746.98 / 1857.53 ±93.08 / 2000.17 ms │ +1.56x faster │
│ QQuery 28 │ 5097.80 / 5171.14 ±41.48 / 5217.47 ms │  4052.74 / 4203.06 ±91.21 / 4296.42 ms │ +1.23x faster │
│ QQuery 29 │ 2243.67 / 2286.67 ±46.56 / 2363.37 ms │     784.78 / 802.25 ±12.88 / 816.47 ms │ +2.85x faster │
│ QQuery 30 │ 2485.53 / 2581.20 ±62.38 / 2649.36 ms │  1004.64 / 1059.36 ±36.28 / 1098.36 ms │ +2.44x faster │
│ QQuery 31 │ 2401.05 / 2460.50 ±52.83 / 2557.62 ms │  1024.67 / 1054.40 ±34.10 / 1114.16 ms │ +2.33x faster │
│ QQuery 32 │ 2629.65 / 2677.64 ±42.58 / 2741.76 ms │ 1188.31 / 1286.67 ±110.57 / 1494.72 ms │ +2.08x faster │
│ QQuery 33 │ 3459.14 / 3509.71 ±44.86 / 3583.61 ms │  2442.04 / 2460.03 ±14.33 / 2485.70 ms │ +1.43x faster │
│ QQuery 34 │ 3424.32 / 3516.37 ±87.80 / 3659.75 ms │  2432.51 / 2500.24 ±50.42 / 2569.33 ms │ +1.41x faster │
│ QQuery 35 │ 2392.22 / 2471.95 ±54.06 / 2548.18 ms │     948.08 / 977.51 ±18.58 / 997.31 ms │ +2.53x faster │
│ QQuery 36 │    143.63 / 290.07 ±89.56 / 403.94 ms │     180.29 / 223.38 ±29.19 / 270.53 ms │ +1.30x faster │
│ QQuery 37 │    135.20 / 251.61 ±80.12 / 357.99 ms │      86.98 / 139.51 ±52.40 / 239.23 ms │ +1.80x faster │
│ QQuery 38 │    130.91 / 269.91 ±85.68 / 376.13 ms │     144.69 / 190.72 ±38.78 / 234.58 ms │ +1.42x faster │
│ QQuery 39 │    195.96 / 327.52 ±95.29 / 453.54 ms │     256.46 / 336.27 ±53.83 / 414.90 ms │     no change │
│ QQuery 40 │    115.94 / 239.26 ±84.77 / 347.74 ms │     119.71 / 152.82 ±30.90 / 204.32 ms │ +1.57x faster │
│ QQuery 41 │    115.06 / 234.56 ±83.74 / 345.18 ms │     116.09 / 171.36 ±41.77 / 237.72 ms │ +1.37x faster │
│ QQuery 42 │    113.43 / 240.32 ±81.75 / 344.73 ms │      82.09 / 124.47 ±46.54 / 214.38 ms │ +1.93x faster │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                 ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                 │ 89385.15ms │
│ Total Time (codex_parquet-pushdecoder-prefetch)   │ 49101.06ms │
│ Average Time (HEAD)                               │  2078.72ms │
│ Average Time (codex_parquet-pushdecoder-prefetch) │  1141.89ms │
│ Queries Faster                                    │         39 │
│ Queries Slower                                    │          0 │
│ Queries with No Change                            │          4 │
│ Queries with Failure                              │          0 │
└───────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	450.1s
Peak memory	9.2 GiB
Avg memory	1.8 GiB
CPU user	999.2s
CPU sys	109.3s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	250.1s
Peak memory	10.4 GiB
Avg memory	2.3 GiB
CPU user	995.2s
CPU sys	92.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-07-12T16:05:42Z

🤖 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 codex_parquet-pushdecoder-prefetch
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃     codex_parquet-pushdecoder-prefetch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │     106.46 / 173.02 ±37.70 / 207.17 ms │     106.62 / 172.84 ±37.73 / 207.03 ms │     no change │
│ QQuery 2  │     512.17 / 649.37 ±72.17 / 706.23 ms │     520.41 / 650.36 ±70.15 / 706.73 ms │     no change │
│ QQuery 3  │     382.08 / 427.22 ±31.86 / 474.42 ms │     382.53 / 429.64 ±32.22 / 475.25 ms │     no change │
│ QQuery 4  │  1127.00 / 1271.27 ±73.97 / 1324.92 ms │  1199.36 / 1257.81 ±42.96 / 1308.72 ms │     no change │
│ QQuery 5  │     499.24 / 572.13 ±64.22 / 678.20 ms │     501.02 / 582.53 ±61.11 / 687.24 ms │     no change │
│ QQuery 6  │     610.02 / 715.94 ±86.66 / 820.80 ms │    535.95 / 701.22 ±108.14 / 820.67 ms │     no change │
│ QQuery 7  │     595.60 / 692.52 ±57.35 / 754.06 ms │     595.18 / 696.20 ±57.75 / 759.28 ms │     no change │
│ QQuery 8  │     571.17 / 620.00 ±28.46 / 651.69 ms │     536.56 / 605.87 ±44.45 / 655.30 ms │     no change │
│ QQuery 9  │  1378.71 / 1488.86 ±91.19 / 1590.00 ms │  1377.87 / 1492.37 ±94.70 / 1592.12 ms │     no change │
│ QQuery 10 │     696.01 / 792.23 ±64.55 / 879.34 ms │     687.82 / 764.24 ±53.86 / 843.21 ms │     no change │
│ QQuery 11 │   822.97 / 996.86 ±113.92 / 1138.17 ms │   919.53 / 1015.79 ±67.15 / 1092.81 ms │     no change │
│ QQuery 12 │     348.35 / 373.29 ±25.24 / 417.51 ms │     350.50 / 373.86 ±24.76 / 418.28 ms │     no change │
│ QQuery 13 │     661.66 / 740.18 ±45.96 / 790.37 ms │     687.86 / 746.73 ±39.16 / 797.30 ms │     no change │
│ QQuery 14 │ 3685.00 / 4002.01 ±174.29 / 4209.75 ms │ 3821.54 / 4135.10 ±219.09 / 4419.67 ms │     no change │
│ QQuery 15 │     408.92 / 499.40 ±48.91 / 549.28 ms │     411.73 / 507.26 ±62.08 / 582.23 ms │     no change │
│ QQuery 16 │     107.40 / 174.72 ±38.51 / 210.86 ms │     108.12 / 174.73 ±38.35 / 210.63 ms │     no change │
│ QQuery 17 │     626.97 / 704.89 ±44.77 / 756.82 ms │     627.71 / 709.19 ±45.39 / 758.83 ms │     no change │
│ QQuery 18 │     718.02 / 803.99 ±61.11 / 907.82 ms │     715.09 / 805.10 ±61.43 / 905.24 ms │     no change │
│ QQuery 19 │     466.23 / 577.04 ±66.27 / 643.79 ms │     531.85 / 596.96 ±56.47 / 683.26 ms │     no change │
│ QQuery 20 │     326.71 / 397.93 ±39.82 / 446.05 ms │     326.76 / 399.39 ±39.99 / 445.76 ms │     no change │
│ QQuery 21 │     508.24 / 570.39 ±48.32 / 627.75 ms │     505.64 / 570.52 ±49.82 / 633.34 ms │     no change │
│ QQuery 22 │     477.60 / 516.17 ±21.02 / 541.48 ms │     474.87 / 528.96 ±32.65 / 570.98 ms │     no change │
│ QQuery 23 │ 2495.22 / 2723.94 ±147.71 / 2891.86 ms │ 2432.43 / 2711.02 ±152.23 / 2860.42 ms │     no change │
│ QQuery 24 │ 2192.55 / 2352.81 ±102.38 / 2468.57 ms │ 2185.65 / 2357.55 ±110.15 / 2483.78 ms │     no change │
│ QQuery 25 │     686.18 / 728.57 ±35.45 / 777.66 ms │     684.78 / 732.06 ±33.89 / 777.28 ms │     no change │
│ QQuery 26 │     610.10 / 649.61 ±33.45 / 705.15 ms │     603.33 / 633.33 ±21.57 / 669.76 ms │     no change │
│ QQuery 27 │     140.19 / 253.79 ±84.30 / 358.70 ms │     140.30 / 253.30 ±84.35 / 358.82 ms │     no change │
│ QQuery 28 │     339.09 / 399.49 ±33.51 / 432.17 ms │     342.58 / 398.10 ±32.36 / 433.31 ms │     no change │
│ QQuery 29 │     724.44 / 780.89 ±39.29 / 823.42 ms │     729.67 / 785.85 ±42.76 / 834.89 ms │     no change │
│ QQuery 30 │     487.18 / 617.17 ±91.76 / 740.43 ms │     514.54 / 625.58 ±83.70 / 743.63 ms │     no change │
│ QQuery 31 │  1192.95 / 1266.50 ±58.80 / 1342.18 ms │ 1134.15 / 1262.18 ±107.89 / 1405.83 ms │     no change │
│ QQuery 32 │     413.79 / 443.57 ±26.85 / 487.62 ms │     372.95 / 433.79 ±50.77 / 487.67 ms │     no change │
│ QQuery 33 │     540.55 / 617.52 ±55.74 / 703.32 ms │     522.54 / 585.54 ±39.85 / 643.33 ms │ +1.05x faster │
│ QQuery 34 │     193.47 / 304.98 ±59.67 / 359.86 ms │     194.23 / 306.29 ±59.81 / 358.35 ms │     no change │
│ QQuery 35 │     684.68 / 739.15 ±42.04 / 803.13 ms │     723.34 / 751.07 ±29.29 / 801.66 ms │     no change │
│ QQuery 36 │     108.08 / 173.33 ±37.50 / 207.36 ms │     107.56 / 173.41 ±37.61 / 207.83 ms │     no change │
│ QQuery 37 │     221.67 / 276.34 ±30.19 / 310.39 ms │     222.05 / 276.33 ±29.98 / 310.24 ms │     no change │
│ QQuery 38 │     602.18 / 665.90 ±33.89 / 700.49 ms │     598.40 / 667.90 ±77.51 / 816.02 ms │     no change │
│ QQuery 39 │ 1849.20 / 2008.00 ±112.90 / 2142.61 ms │ 1841.51 / 2006.02 ±111.28 / 2141.24 ms │     no change │
│ QQuery 40 │     464.51 / 492.49 ±22.73 / 512.50 ms │     467.79 / 493.57 ±20.86 / 511.99 ms │     no change │
│ QQuery 41 │     176.30 / 279.08 ±73.21 / 381.09 ms │     175.16 / 278.59 ±73.59 / 381.23 ms │     no change │
│ QQuery 42 │     313.82 / 404.78 ±64.46 / 464.30 ms │     313.95 / 404.34 ±65.01 / 472.85 ms │     no change │
│ QQuery 43 │     106.71 / 153.28 ±40.84 / 206.41 ms │     106.75 / 152.93 ±40.18 / 206.42 ms │     no change │
│ QQuery 44 │     198.42 / 262.14 ±47.05 / 302.33 ms │     197.42 / 261.93 ±47.23 / 302.38 ms │     no change │
│ QQuery 45 │     416.40 / 541.07 ±82.20 / 646.27 ms │     451.73 / 526.45 ±48.13 / 578.98 ms │     no change │
│ QQuery 46 │     289.65 / 345.37 ±41.26 / 398.87 ms │     309.42 / 351.92 ±34.87 / 399.62 ms │     no change │
│ QQuery 47 │  1227.52 / 1257.84 ±34.23 / 1323.00 ms │  1230.91 / 1274.29 ±35.29 / 1321.20 ms │     no change │
│ QQuery 48 │     585.05 / 704.80 ±65.83 / 771.69 ms │     606.65 / 707.91 ±62.24 / 776.07 ms │     no change │
│ QQuery 49 │     579.54 / 629.86 ±50.21 / 706.35 ms │     522.95 / 603.49 ±61.66 / 712.86 ms │     no change │
│ QQuery 50 │     526.90 / 635.40 ±57.04 / 689.66 ms │     541.26 / 635.29 ±50.21 / 683.07 ms │     no change │
│ QQuery 51 │     504.35 / 540.06 ±27.21 / 563.52 ms │     522.92 / 542.69 ±15.34 / 559.04 ms │     no change │
│ QQuery 52 │     313.96 / 403.67 ±63.80 / 463.39 ms │     314.78 / 405.64 ±64.87 / 471.49 ms │     no change │
│ QQuery 53 │     407.93 / 453.22 ±50.83 / 515.99 ms │     409.80 / 453.29 ±47.74 / 514.03 ms │     no change │
│ QQuery 54 │    905.92 / 946.53 ±46.20 / 1033.73 ms │     902.46 / 937.26 ±34.75 / 997.29 ms │     no change │
│ QQuery 55 │     377.06 / 423.62 ±31.73 / 467.88 ms │     377.66 / 422.84 ±32.42 / 471.79 ms │     no change │
│ QQuery 56 │     539.40 / 604.00 ±40.03 / 657.80 ms │     533.92 / 590.93 ±40.14 / 643.35 ms │     no change │
│ QQuery 57 │  1081.57 / 1195.64 ±77.51 / 1306.15 ms │  1082.64 / 1193.48 ±76.15 / 1303.68 ms │     no change │
│ QQuery 58 │  1162.77 / 1327.99 ±94.05 / 1424.96 ms │  1158.10 / 1328.50 ±97.18 / 1421.19 ms │     no change │
│ QQuery 59 │    893.20 / 926.89 ±51.48 / 1029.18 ms │    908.78 / 935.70 ±48.38 / 1032.42 ms │     no change │
│ QQuery 60 │     582.62 / 659.08 ±68.20 / 745.11 ms │     581.73 / 673.34 ±64.62 / 749.09 ms │     no change │
│ QQuery 61 │      394.89 / 395.19 ±0.22 / 395.48 ms │      394.53 / 395.41 ±0.77 / 396.27 ms │     no change │
│ QQuery 62 │     579.72 / 600.45 ±24.67 / 631.03 ms │     579.81 / 600.88 ±25.18 / 632.36 ms │     no change │
│ QQuery 63 │     416.71 / 441.63 ±28.37 / 478.44 ms │     417.70 / 441.35 ±27.63 / 475.64 ms │     no change │
│ QQuery 64 │ 3126.25 / 3277.24 ±103.91 / 3402.85 ms │  3161.09 / 3287.78 ±84.68 / 3387.49 ms │     no change │
│ QQuery 65 │     626.16 / 739.79 ±67.30 / 826.18 ms │     631.23 / 744.87 ±64.98 / 826.51 ms │     no change │
│ QQuery 66 │     564.71 / 598.14 ±41.98 / 677.42 ms │     560.96 / 589.62 ±43.97 / 677.29 ms │     no change │
│ QQuery 67 │     614.30 / 684.54 ±45.68 / 737.55 ms │     659.23 / 690.60 ±27.67 / 723.51 ms │     no change │
│ QQuery 68 │     290.02 / 320.51 ±20.76 / 354.71 ms │     289.13 / 315.30 ±15.80 / 336.05 ms │     no change │
│ QQuery 69 │     633.74 / 806.91 ±94.48 / 907.86 ms │     702.76 / 782.41 ±84.04 / 904.29 ms │     no change │
│ QQuery 70 │     674.75 / 762.16 ±66.28 / 866.95 ms │     683.18 / 758.04 ±58.78 / 853.09 ms │     no change │
│ QQuery 71 │     524.69 / 583.20 ±45.71 / 664.16 ms │     519.66 / 573.90 ±54.06 / 670.26 ms │     no change │
│ QQuery 72 │  2997.80 / 3077.43 ±67.68 / 3167.34 ms │  3117.86 / 3221.68 ±70.74 / 3322.51 ms │     no change │
│ QQuery 73 │     194.15 / 308.65 ±60.24 / 362.54 ms │     193.62 / 308.64 ±62.37 / 360.69 ms │     no change │
│ QQuery 74 │     723.81 / 819.10 ±56.80 / 892.66 ms │     640.07 / 739.76 ±75.21 / 837.79 ms │ +1.11x faster │
│ QQuery 75 │   922.46 / 1002.60 ±50.56 / 1066.75 ms │    921.85 / 981.19 ±52.56 / 1042.41 ms │     no change │
│ QQuery 76 │     477.66 / 567.71 ±68.38 / 652.90 ms │     524.93 / 609.96 ±63.72 / 686.89 ms │  1.07x slower │
│ QQuery 77 │     653.75 / 701.70 ±52.72 / 790.56 ms │     604.85 / 696.04 ±73.05 / 819.72 ms │     no change │
│ QQuery 78 │     707.72 / 856.09 ±81.14 / 951.21 ms │     715.09 / 851.98 ±77.99 / 954.32 ms │     no change │
│ QQuery 79 │     448.87 / 510.46 ±53.50 / 603.32 ms │     456.37 / 507.24 ±44.15 / 583.64 ms │     no change │
│ QQuery 80 │     651.00 / 715.34 ±45.80 / 790.76 ms │     658.42 / 722.90 ±55.39 / 794.55 ms │     no change │
│ QQuery 81 │     476.65 / 618.13 ±77.18 / 681.15 ms │     477.79 / 613.40 ±76.80 / 689.09 ms │     no change │
│ QQuery 82 │     557.52 / 644.04 ±47.23 / 682.51 ms │     557.14 / 643.67 ±47.58 / 684.34 ms │     no change │
│ QQuery 83 │    862.98 / 952.57 ±74.88 / 1028.36 ms │   860.69 / 980.51 ±105.48 / 1098.77 ms │     no change │
│ QQuery 84 │     442.23 / 502.69 ±55.57 / 591.75 ms │     439.42 / 502.05 ±56.78 / 592.05 ms │     no change │
│ QQuery 85 │     820.09 / 888.19 ±50.25 / 947.21 ms │     816.18 / 887.07 ±51.14 / 949.14 ms │     no change │
│ QQuery 86 │     247.98 / 315.19 ±55.39 / 392.54 ms │     249.19 / 315.39 ±55.99 / 397.28 ms │     no change │
│ QQuery 87 │     612.93 / 682.64 ±67.75 / 778.16 ms │     609.20 / 673.65 ±58.73 / 779.54 ms │     no change │
│ QQuery 88 │      738.12 / 739.39 ±1.14 / 741.06 ms │      737.47 / 738.34 ±0.95 / 740.13 ms │     no change │
│ QQuery 89 │     427.17 / 459.36 ±26.25 / 481.13 ms │     426.22 / 461.05 ±28.39 / 484.68 ms │     no change │
│ QQuery 90 │     427.64 / 467.66 ±25.69 / 501.91 ms │     425.25 / 466.28 ±26.42 / 502.14 ms │     no change │
│ QQuery 91 │    525.00 / 664.53 ±112.72 / 806.61 ms │    523.68 / 663.32 ±113.50 / 806.00 ms │     no change │
│ QQuery 92 │     373.67 / 446.73 ±41.22 / 487.66 ms │     371.49 / 439.06 ±37.89 / 488.83 ms │     no change │
│ QQuery 93 │     467.68 / 484.21 ±18.86 / 519.56 ms │     466.79 / 481.71 ±15.30 / 508.57 ms │     no change │
│ QQuery 94 │     479.73 / 530.13 ±43.05 / 591.50 ms │     480.42 / 530.73 ±43.38 / 588.15 ms │     no change │
│ QQuery 95 │     692.97 / 777.00 ±46.24 / 824.03 ms │     694.33 / 778.60 ±45.51 / 825.62 ms │     no change │
│ QQuery 96 │     403.44 / 446.46 ±48.92 / 507.20 ms │     403.23 / 445.88 ±48.25 / 506.93 ms │     no change │
│ QQuery 97 │     453.60 / 518.38 ±51.76 / 612.43 ms │     508.30 / 535.67 ±40.95 / 616.41 ms │     no change │
│ QQuery 98 │     394.58 / 440.31 ±32.06 / 487.87 ms │     401.63 / 442.58 ±28.26 / 483.87 ms │     no change │
│ QQuery 99 │     508.26 / 620.00 ±64.57 / 697.87 ms │     507.78 / 611.66 ±61.68 / 694.50 ms │     no change │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                 ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                 │ 74914.28ms │
│ Total Time (codex_parquet-pushdecoder-prefetch)   │ 75080.57ms │
│ Average Time (HEAD)                               │   756.71ms │
│ Average Time (codex_parquet-pushdecoder-prefetch) │   758.39ms │
│ Queries Faster                                    │          2 │
│ Queries Slower                                    │          1 │
│ Queries with No Change                            │         96 │
│ Queries with Failure                              │          0 │
└───────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	380.1s
Peak memory	1.1 GiB
Avg memory	618.5 MiB
CPU user	229.2s
CPU sys	11.2s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	380.1s
Peak memory	965.1 MiB
Avg memory	540.9 MiB
CPU user	241.1s
CPU sys	11.5s
Peak spill	0 B

File an issue against this benchmark runner

Prefetch Parquet row groups with a bounded budget

4d02e1e

github-actions Bot added documentation Improvements or additions to documentation sqllogictest SQL Logic Tests (.slt) common Related to common crate proto Related to proto crate datasource Changes to the datasource crate labels Jul 12, 2026

github-actions Bot added the auto detected api change Auto detected API change label Jul 12, 2026

Uh oh!

Conversation

Dandandan commented Jul 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Results

Are these changes tested?

Are there any user-facing changes?

Uh oh!

github-actions Bot commented Jul 12, 2026

Uh oh!

Dandandan commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

Dandandan commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

adriangbot commented Jul 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Dandandan commented Jul 12, 2026 •

edited

Loading