Skip to content

[DO NOT MERGE] Benchmark arrow-rs #12126 - #27249

Closed
Rich-T-kid wants to merge 2 commits into
apache:mainfrom
Rich-T-kid:rich-T-kid/arrow-rs-12126-benchmark
Closed

[DO NOT MERGE] Benchmark arrow-rs #12126#27249
Rich-T-kid wants to merge 2 commits into
apache:mainfrom
Rich-T-kid:rich-T-kid/arrow-rs-12126-benchmark

Conversation

@Rich-T-kid

@Rich-T-kid Rich-T-kid commented Nov 8, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Which issue does this PR close?

Rationale for this change

Benchmark the DataFusion impact of preserving the requested dictionary value type in Arrow.

What changes are included in this PR?

The first commit pins Arrow to the PR parent (62a3c43f); the tip pins Arrow to the PR head (83ae72c6). There are no DataFusion runtime changes.

What is the testing strategy for this PR?

Exact-head hosted suites with RLE dictionary reads enabled on both sides:

  • TPC-H: 796.01 → 799.79 ms (0.5% higher).
  • TPC-DS: 9.02 → 8.88 s (1.6% lower).
  • ClickBench: 38.26 → 38.16 s (0.3% lower).

Are there any user-facing changes?

No.

@Rich-T-kid

Rich-T-kid commented Nov 8, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark tpch tpcds clickbench_partitioned

env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"

@Rich-T-kid

Rich-T-kid commented Nov 8, 2026

Copy link
Copy Markdown
Contributor Author

Reopening this exact dependency-only branch as a regular PR so the hosted benchmark service can process the trigger.

@Rich-T-kid Rich-T-kid closed this Nov 8, 2026
@Rich-T-kid Rich-T-kid reopened this Nov 8, 2026
@Rich-T-kid Rich-T-kid closed this Nov 8, 2026
@Rich-T-kid Rich-T-kid reopened this Nov 8, 2026
@adriangbot

adriangbot commented Nov 8, 2026

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6726201402-4635-rhwss 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 1293f9c (1293f9c) to 0b062c8 diff

Run configuration 
run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"

Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot

adriangbot commented Nov 8, 2026

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6726201402-4634-xzqxc 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 1293f9c (1293f9c) to 0b062c8 diff

Run configuration 
run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"

Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot

adriangbot commented Nov 8, 2026

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6726201402-4636-dwf9l 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 1293f9c (1293f9c) to 0b062c8 diff

Run configuration 
run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"

Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot

adriangbot commented Nov 8, 2026

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing 1293f9c (1293f9c) to 0b062c8 diff

Run configuration 
run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"
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 rich-T-kid_arrow-rs-12126-benchmark
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ rich-T-kid_arrow-rs-12126-benchmark ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 40.34 ms │                            40.57 ms │    no change │
│ QQuery 2  │ 19.70 ms │                            19.94 ms │    no change │
│ QQuery 3  │ 29.69 ms │                            30.09 ms │    no change │
│ QQuery 4  │ 18.23 ms │                            18.08 ms │    no change │
│ QQuery 5  │ 38.61 ms │                            38.43 ms │    no change │
│ QQuery 6  │ 14.53 ms │                            14.46 ms │    no change │
│ QQuery 7  │ 45.79 ms │                            43.99 ms │    no change │
│ QQuery 8  │ 41.91 ms │                            42.57 ms │    no change │
│ QQuery 9  │ 53.48 ms │                            52.86 ms │    no change │
│ QQuery 10 │ 55.89 ms │                            58.26 ms │    no change │
│ QQuery 11 │ 14.67 ms │                            14.63 ms │    no change │
│ QQuery 12 │ 23.21 ms │                            23.04 ms │    no change │
│ QQuery 13 │ 50.81 ms │                            53.42 ms │ 1.05x slower │
│ QQuery 14 │ 25.59 ms │                            25.53 ms │    no change │
│ QQuery 15 │ 33.20 ms │                            33.46 ms │    no change │
│ QQuery 16 │ 14.61 ms │                            14.82 ms │    no change │
│ QQuery 17 │ 79.12 ms │                            78.79 ms │    no change │
│ QQuery 18 │ 54.87 ms │                            55.06 ms │    no change │
│ QQuery 19 │ 30.38 ms │                            30.21 ms │    no change │
│ QQuery 20 │ 36.10 ms │                            36.49 ms │    no change │
│ QQuery 21 │ 58.16 ms │                            58.37 ms │    no change │
│ QQuery 22 │ 17.13 ms │                            16.73 ms │    no change │
└───────────┴──────────┴─────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                  ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                  │ 796.01ms │
│ Total Time (rich-T-kid_arrow-rs-12126-benchmark)   │ 799.79ms │
│ Average Time (HEAD)                                │  36.18ms │
│ Average Time (rich-T-kid_arrow-rs-12126-benchmark) │  36.35ms │
│ Queries Faster                                     │        0 │
│ Queries Slower                                     │        1 │
│ Queries with No Change                             │       21 │
│ Queries with Failure                               │        0 │
└────────────────────────────────────────────────────┴──────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and rich-T-kid_arrow-rs-12126-benchmark
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ rich-T-kid_arrow-rs-12126-benchmark ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │ 40.34 / 41.26 ±1.02 / 43.24 ms │      40.57 / 41.90 ±1.47 / 44.17 ms │     no change │
│ QQuery 2  │ 19.70 / 19.86 ±0.16 / 20.17 ms │      19.94 / 20.09 ±0.15 / 20.29 ms │     no change │
│ QQuery 3  │ 29.69 / 30.27 ±0.40 / 30.76 ms │      30.09 / 30.40 ±0.23 / 30.78 ms │     no change │
│ QQuery 4  │ 18.23 / 18.73 ±0.57 / 19.78 ms │      18.08 / 18.54 ±0.68 / 19.90 ms │     no change │
│ QQuery 5  │ 38.61 / 39.20 ±0.34 / 39.54 ms │      38.43 / 39.06 ±0.56 / 40.08 ms │     no change │
│ QQuery 6  │ 14.53 / 14.66 ±0.11 / 14.82 ms │      14.46 / 14.67 ±0.33 / 15.33 ms │     no change │
│ QQuery 7  │ 45.79 / 47.32 ±1.03 / 48.25 ms │      43.99 / 45.81 ±1.19 / 47.50 ms │     no change │
│ QQuery 8  │ 41.91 / 43.11 ±1.17 / 45.28 ms │      42.57 / 43.00 ±0.29 / 43.41 ms │     no change │
│ QQuery 9  │ 53.48 / 54.27 ±0.69 / 55.55 ms │      52.86 / 54.08 ±0.97 / 55.41 ms │     no change │
│ QQuery 10 │ 55.89 / 57.37 ±1.24 / 59.56 ms │      58.26 / 58.86 ±0.44 / 59.41 ms │     no change │
│ QQuery 11 │ 14.67 / 15.18 ±0.60 / 16.31 ms │      14.63 / 14.85 ±0.21 / 15.22 ms │     no change │
│ QQuery 12 │ 23.21 / 24.60 ±1.97 / 28.44 ms │      23.04 / 24.54 ±2.01 / 28.52 ms │     no change │
│ QQuery 13 │ 50.81 / 51.99 ±1.06 / 53.92 ms │      53.42 / 54.27 ±1.05 / 56.32 ms │     no change │
│ QQuery 14 │ 25.59 / 25.83 ±0.23 / 26.11 ms │      25.53 / 25.71 ±0.11 / 25.84 ms │     no change │
│ QQuery 15 │ 33.20 / 33.99 ±0.71 / 35.01 ms │      33.46 / 34.42 ±0.83 / 35.75 ms │     no change │
│ QQuery 16 │ 14.61 / 15.08 ±0.58 / 16.20 ms │      14.82 / 15.02 ±0.12 / 15.13 ms │     no change │
│ QQuery 17 │ 79.12 / 81.35 ±1.81 / 84.24 ms │      78.79 / 82.23 ±4.42 / 90.58 ms │     no change │
│ QQuery 18 │ 54.87 / 56.87 ±1.56 / 59.12 ms │      55.06 / 55.74 ±0.75 / 56.96 ms │     no change │
│ QQuery 19 │ 30.38 / 31.48 ±1.29 / 33.97 ms │      30.21 / 32.39 ±1.91 / 35.27 ms │     no change │
│ QQuery 20 │ 36.10 / 36.46 ±0.32 / 37.01 ms │      36.49 / 37.00 ±0.67 / 38.32 ms │     no change │
│ QQuery 21 │ 58.16 / 59.65 ±1.54 / 62.32 ms │      58.37 / 60.09 ±2.20 / 64.25 ms │     no change │
│ QQuery 22 │ 17.13 / 19.33 ±2.88 / 24.86 ms │      16.73 / 17.82 ±1.08 / 19.19 ms │ +1.08x faster │
└───────────┴────────────────────────────────┴─────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                  ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                  │ 817.85ms │
│ Total Time (rich-T-kid_arrow-rs-12126-benchmark)   │ 820.49ms │
│ Average Time (HEAD)                                │  37.17ms │
│ Average Time (rich-T-kid_arrow-rs-12126-benchmark) │  37.29ms │
│ Queries Faster                                     │        1 │
│ Queries Slower                                     │        0 │
│ Queries with No Change                             │       21 │
│ Queries with Failure                               │        0 │
└────────────────────────────────────────────────────┴──────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 0b062c85157df1ecc618582decea1346d60db6b0 | Changed: 1293f9c59bc340cfbd91787ea6ab101698e20d1d

tpch — tpch_sf1

Query Base Changed Change
Query 1 30.1 MiB 30.1 MiB -0.1%
Query 2 23.1 MiB 23.1 MiB +0.0%
Query 3 11.5 MiB 11.5 MiB +0.0%
Query 4 32.2 MiB 32.2 MiB -0.0%
Query 5 42.8 MiB 32.8 MiB -23.4%
Query 6 832 B 832 B +0.0%
Query 7 107.6 MiB 108.5 MiB +0.8%
Query 8 39.8 MiB 39.8 MiB -0.0%
Query 9 155.6 MiB 155.6 MiB +0.0%
Query 10 53.3 MiB 53.3 MiB -0.1%
Query 11 51.2 MiB 61.9 MiB +21.0%
Query 12 12.6 MiB 22.5 MiB +78.2%
Query 13 62.5 MiB 62.7 MiB +0.4%
Query 14 7.7 MiB 7.7 MiB +0.0%
Query 15 13.0 MiB 13.1 MiB +0.8%
Query 16 34.0 MiB 43.8 MiB +28.9%
Query 17 150.3 MiB 150.3 MiB +0.0%
Query 18 127.2 MiB 127.0 MiB -0.1%
Query 19 231.0 KiB 231.0 KiB +0.0%
Query 20 60.9 MiB 60.7 MiB -0.3%
Query 21 37.7 MiB 37.7 MiB +0.0%
Query 22 41.6 MiB 31.8 MiB -23.6%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark Side Max pool peak Peak RSS Gap RSS / pool
tpch base (0b062c85157df1ecc618582decea1346d60db6b0) 155.6 MiB 1.1 GiB 1009.4 MiB 7.5×
tpch changed (1293f9c59bc340cfbd91787ea6ab101698e20d1d) 155.6 MiB 1.1 GiB 1007.6 MiB 7.5×
Resource Usage

tpch — base (merge-base)

Metric Value
Wall time 5.0s
Peak memory 1.1 GiB
Avg memory 661.7 MiB
CPU user 21.7s
CPU sys 1.7s
Peak spill 0 B

tpch — branch

Metric Value
Wall time 5.0s
Peak memory 1.1 GiB
Avg memory 660.0 MiB
CPU user 21.7s
CPU sys 1.7s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

adriangbot commented Nov 8, 2026

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing 1293f9c (1293f9c) to 0b062c8 diff

Run configuration 
run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"
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 rich-T-kid_arrow-rs-12126-benchmark
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_arrow-rs-12126-benchmark ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.57 ms │                             5.63 ms │     no change │
│ QQuery 2  │   77.13 ms │                            76.74 ms │     no change │
│ QQuery 3  │   28.90 ms │                            29.15 ms │     no change │
│ QQuery 4  │  542.57 ms │                           545.15 ms │     no change │
│ QQuery 5  │   51.36 ms │                            50.99 ms │     no change │
│ QQuery 6  │   34.34 ms │                            34.06 ms │     no change │
│ QQuery 7  │   71.28 ms │                            71.69 ms │     no change │
│ QQuery 8  │   38.30 ms │                            38.63 ms │     no change │
│ QQuery 9  │   50.87 ms │                            53.84 ms │  1.06x slower │
│ QQuery 10 │   59.22 ms │                            59.44 ms │     no change │
│ QQuery 11 │  342.97 ms │                           357.43 ms │     no change │
│ QQuery 12 │   26.82 ms │                            26.88 ms │     no change │
│ QQuery 13 │  112.60 ms │                           113.91 ms │     no change │
│ QQuery 14 │  380.15 ms │                           383.16 ms │     no change │
│ QQuery 15 │   53.37 ms │                            54.31 ms │     no change │
│ QQuery 16 │    6.44 ms │                             6.81 ms │  1.06x slower │
│ QQuery 17 │   77.72 ms │                            79.34 ms │     no change │
│ QQuery 18 │  106.55 ms │                           108.47 ms │     no change │
│ QQuery 19 │   42.42 ms │                            42.77 ms │     no change │
│ QQuery 20 │   32.53 ms │                            33.03 ms │     no change │
│ QQuery 21 │   17.20 ms │                            17.72 ms │     no change │
│ QQuery 22 │   66.14 ms │                            68.81 ms │     no change │
│ QQuery 23 │  296.52 ms │                           307.96 ms │     no change │
│ QQuery 24 │  197.05 ms │                           205.47 ms │     no change │
│ QQuery 25 │  106.40 ms │                           107.78 ms │     no change │
│ QQuery 26 │   45.08 ms │                            45.75 ms │     no change │
│ QQuery 27 │    6.29 ms │                             6.66 ms │  1.06x slower │
│ QQuery 28 │   57.62 ms │                            61.94 ms │  1.07x slower │
│ QQuery 29 │   96.01 ms │                            95.87 ms │     no change │
│ QQuery 30 │   35.94 ms │                            36.53 ms │     no change │
│ QQuery 31 │  110.27 ms │                           111.68 ms │     no change │
│ QQuery 32 │   20.27 ms │                            20.65 ms │     no change │
│ QQuery 33 │   37.34 ms │                            37.51 ms │     no change │
│ QQuery 34 │    9.71 ms │                             9.88 ms │     no change │
│ QQuery 35 │   70.99 ms │                            73.34 ms │     no change │
│ QQuery 36 │    6.15 ms │                             6.37 ms │     no change │
│ QQuery 37 │    7.57 ms │                             7.93 ms │     no change │
│ QQuery 38 │   62.70 ms │                            64.17 ms │     no change │
│ QQuery 39 │   73.28 ms │                            74.57 ms │     no change │
│ QQuery 40 │   23.17 ms │                            23.95 ms │     no change │
│ QQuery 41 │   14.10 ms │                            14.46 ms │     no change │
│ QQuery 42 │   24.14 ms │                            24.47 ms │     no change │
│ QQuery 43 │    4.88 ms │                             5.07 ms │     no change │
│ QQuery 44 │    8.64 ms │                             8.98 ms │     no change │
│ QQuery 45 │   36.60 ms │                            36.16 ms │     no change │
│ QQuery 46 │   11.57 ms │                            11.48 ms │     no change │
│ QQuery 47 │  209.68 ms │                           206.66 ms │     no change │
│ QQuery 48 │   91.72 ms │                            92.10 ms │     no change │
│ QQuery 49 │   69.40 ms │                            69.61 ms │     no change │
│ QQuery 50 │   57.39 ms │                            58.00 ms │     no change │
│ QQuery 51 │   93.34 ms │                            93.38 ms │     no change │
│ QQuery 52 │   24.21 ms │                            24.42 ms │     no change │
│ QQuery 53 │   29.62 ms │                            29.51 ms │     no change │
│ QQuery 54 │   53.75 ms │                            54.14 ms │     no change │
│ QQuery 55 │   24.03 ms │                            23.91 ms │     no change │
│ QQuery 56 │   38.79 ms │                            38.50 ms │     no change │
│ QQuery 57 │  153.13 ms │                           152.01 ms │     no change │
│ QQuery 58 │  104.63 ms │                           103.57 ms │     no change │
│ QQuery 59 │  108.13 ms │                           107.57 ms │     no change │
│ QQuery 60 │   38.60 ms │                            38.68 ms │     no change │
│ QQuery 61 │   10.67 ms │                            10.70 ms │     no change │
│ QQuery 62 │   38.48 ms │                            38.60 ms │     no change │
│ QQuery 63 │   29.34 ms │                            29.42 ms │     no change │
│ QQuery 64 │  368.86 ms │                           365.78 ms │     no change │
│ QQuery 65 │   77.70 ms │                            76.34 ms │     no change │
│ QQuery 66 │   73.72 ms │                            72.47 ms │     no change │
│ QQuery 67 │  223.30 ms │                           216.54 ms │     no change │
│ QQuery 68 │   11.67 ms │                            11.68 ms │     no change │
│ QQuery 69 │   54.39 ms │                            54.36 ms │     no change │
│ QQuery 70 │   97.99 ms │                            97.25 ms │     no change │
│ QQuery 71 │   35.03 ms │                            35.39 ms │     no change │
│ QQuery 72 │ 1573.84 ms │                          1411.64 ms │ +1.11x faster │
│ QQuery 73 │    9.56 ms │                             9.07 ms │ +1.05x faster │
│ QQuery 74 │  236.38 ms │                           221.98 ms │ +1.06x faster │
│ QQuery 75 │  139.91 ms │                           137.87 ms │     no change │
│ QQuery 76 │   35.03 ms │                            34.74 ms │     no change │
│ QQuery 77 │   60.22 ms │                            59.34 ms │     no change │
│ QQuery 78 │  275.78 ms │                           268.02 ms │     no change │
│ QQuery 79 │   68.71 ms │                            67.75 ms │     no change │
│ QQuery 80 │   94.90 ms │                            94.08 ms │     no change │
│ QQuery 81 │   28.90 ms │                            27.86 ms │     no change │
│ QQuery 82 │   17.97 ms │                            17.50 ms │     no change │
│ QQuery 83 │   33.31 ms │                            33.00 ms │     no change │
│ QQuery 84 │   30.64 ms │                            30.63 ms │     no change │
│ QQuery 85 │  101.26 ms │                           101.01 ms │     no change │
│ QQuery 86 │   24.79 ms │                            24.52 ms │     no change │
│ QQuery 87 │   64.69 ms │                            63.13 ms │     no change │
│ QQuery 88 │   60.04 ms │                            59.54 ms │     no change │
│ QQuery 89 │   36.08 ms │                            35.88 ms │     no change │
│ QQuery 90 │   16.73 ms │                            16.45 ms │     no change │
│ QQuery 91 │   41.04 ms │                            40.47 ms │     no change │
│ QQuery 92 │   29.90 ms │                            29.81 ms │     no change │
│ QQuery 93 │   50.29 ms │                            49.42 ms │     no change │
│ QQuery 94 │   38.60 ms │                            38.49 ms │     no change │
│ QQuery 95 │   80.70 ms │                            81.85 ms │     no change │
│ QQuery 96 │   24.03 ms │                            24.26 ms │     no change │
│ QQuery 97 │   51.05 ms │                            50.67 ms │     no change │
│ QQuery 98 │   38.87 ms │                            38.81 ms │     no change │
│ QQuery 99 │   56.85 ms │                            57.45 ms │     no change │
└───────────┴────────────┴─────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                  ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                  │ 9024.40ms │
│ Total Time (rich-T-kid_arrow-rs-12126-benchmark)   │ 8882.42ms │
│ Average Time (HEAD)                                │   91.16ms │
│ Average Time (rich-T-kid_arrow-rs-12126-benchmark) │   89.72ms │
│ Queries Faster                                     │         3 │
│ Queries Slower                                     │         4 │
│ Queries with No Change                             │        92 │
│ Queries with Failure                               │         0 │
└────────────────────────────────────────────────────┴───────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and rich-T-kid_arrow-rs-12126-benchmark
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃    rich-T-kid_arrow-rs-12126-benchmark ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.57 / 6.18 ±1.01 / 8.17 ms │            5.63 / 6.27 ±0.99 / 8.24 ms │     no change │
│ QQuery 2  │        77.13 / 77.40 ±0.32 / 77.97 ms │         76.74 / 76.90 ±0.11 / 77.04 ms │     no change │
│ QQuery 3  │        28.90 / 29.29 ±0.29 / 29.76 ms │         29.15 / 29.28 ±0.16 / 29.58 ms │     no change │
│ QQuery 4  │     542.57 / 546.18 ±2.64 / 549.28 ms │      545.15 / 555.22 ±5.81 / 561.20 ms │     no change │
│ QQuery 5  │        51.36 / 51.68 ±0.25 / 52.12 ms │         50.99 / 51.47 ±0.34 / 51.94 ms │     no change │
│ QQuery 6  │        34.34 / 34.52 ±0.13 / 34.66 ms │         34.06 / 34.56 ±0.70 / 35.92 ms │     no change │
│ QQuery 7  │        71.28 / 71.74 ±0.29 / 72.16 ms │         71.69 / 74.01 ±3.18 / 80.11 ms │     no change │
│ QQuery 8  │        38.30 / 38.83 ±0.50 / 39.63 ms │         38.63 / 39.31 ±0.53 / 40.00 ms │     no change │
│ QQuery 9  │        50.87 / 53.95 ±1.98 / 57.07 ms │         53.84 / 55.21 ±0.93 / 56.38 ms │     no change │
│ QQuery 10 │        59.22 / 59.44 ±0.21 / 59.70 ms │         59.44 / 59.64 ±0.21 / 60.04 ms │     no change │
│ QQuery 11 │     342.97 / 348.09 ±4.42 / 356.03 ms │      357.43 / 363.75 ±3.89 / 368.57 ms │     no change │
│ QQuery 12 │        26.82 / 26.98 ±0.13 / 27.20 ms │         26.88 / 27.24 ±0.24 / 27.61 ms │     no change │
│ QQuery 13 │     112.60 / 113.60 ±0.64 / 114.58 ms │      113.91 / 116.63 ±3.47 / 123.46 ms │     no change │
│ QQuery 14 │     380.15 / 384.47 ±3.86 / 390.21 ms │      383.16 / 389.76 ±7.18 / 403.13 ms │     no change │
│ QQuery 15 │        53.37 / 53.92 ±0.37 / 54.45 ms │         54.31 / 55.87 ±1.89 / 59.56 ms │     no change │
│ QQuery 16 │           6.44 / 6.62 ±0.17 / 6.88 ms │            6.81 / 6.93 ±0.13 / 7.18 ms │     no change │
│ QQuery 17 │        77.72 / 79.44 ±2.04 / 83.36 ms │         79.34 / 83.08 ±4.67 / 91.42 ms │     no change │
│ QQuery 18 │     106.55 / 108.74 ±3.96 / 116.67 ms │      108.47 / 109.07 ±0.37 / 109.61 ms │     no change │
│ QQuery 19 │        42.42 / 42.67 ±0.20 / 42.95 ms │         42.77 / 42.90 ±0.08 / 42.97 ms │     no change │
│ QQuery 20 │        32.53 / 32.73 ±0.21 / 33.14 ms │         33.03 / 33.99 ±1.66 / 37.30 ms │     no change │
│ QQuery 21 │        17.20 / 17.58 ±0.68 / 18.93 ms │         17.72 / 17.90 ±0.13 / 18.09 ms │     no change │
│ QQuery 22 │        66.14 / 67.09 ±0.62 / 68.02 ms │         68.81 / 69.90 ±0.68 / 70.75 ms │     no change │
│ QQuery 23 │     296.52 / 301.40 ±6.45 / 314.12 ms │      307.96 / 310.25 ±1.53 / 312.64 ms │     no change │
│ QQuery 24 │     197.05 / 201.73 ±4.34 / 208.59 ms │      205.47 / 209.79 ±5.20 / 219.86 ms │     no change │
│ QQuery 25 │     106.40 / 108.18 ±1.10 / 109.78 ms │      107.78 / 109.56 ±2.28 / 113.96 ms │     no change │
│ QQuery 26 │        45.08 / 45.30 ±0.15 / 45.46 ms │         45.75 / 46.29 ±0.40 / 46.72 ms │     no change │
│ QQuery 27 │           6.29 / 6.44 ±0.09 / 6.55 ms │            6.66 / 6.78 ±0.07 / 6.86 ms │  1.05x slower │
│ QQuery 28 │        57.62 / 60.46 ±1.50 / 61.92 ms │         61.94 / 63.99 ±2.98 / 69.86 ms │  1.06x slower │
│ QQuery 29 │       96.01 / 98.43 ±3.16 / 104.61 ms │        95.87 / 97.92 ±1.68 / 100.32 ms │     no change │
│ QQuery 30 │        35.94 / 36.38 ±0.34 / 36.87 ms │         36.53 / 37.22 ±0.45 / 37.78 ms │     no change │
│ QQuery 31 │     110.27 / 113.73 ±4.35 / 121.60 ms │      111.68 / 113.23 ±1.81 / 116.73 ms │     no change │
│ QQuery 32 │        20.27 / 20.63 ±0.25 / 20.93 ms │         20.65 / 20.85 ±0.16 / 21.11 ms │     no change │
│ QQuery 33 │        37.34 / 37.57 ±0.27 / 38.10 ms │         37.51 / 37.94 ±0.31 / 38.27 ms │     no change │
│ QQuery 34 │           9.71 / 9.76 ±0.04 / 9.81 ms │            9.88 / 9.93 ±0.04 / 9.99 ms │     no change │
│ QQuery 35 │        70.99 / 71.85 ±0.58 / 72.82 ms │         73.34 / 73.87 ±0.54 / 74.85 ms │     no change │
│ QQuery 36 │           6.15 / 6.29 ±0.22 / 6.72 ms │            6.37 / 6.53 ±0.09 / 6.61 ms │     no change │
│ QQuery 37 │           7.57 / 7.67 ±0.07 / 7.79 ms │            7.93 / 8.18 ±0.22 / 8.57 ms │  1.07x slower │
│ QQuery 38 │        62.70 / 64.59 ±1.48 / 67.09 ms │         64.17 / 65.05 ±0.51 / 65.50 ms │     no change │
│ QQuery 39 │        73.28 / 74.11 ±0.93 / 75.89 ms │         74.57 / 75.01 ±0.41 / 75.55 ms │     no change │
│ QQuery 40 │        23.17 / 23.47 ±0.24 / 23.90 ms │         23.95 / 25.07 ±1.19 / 27.21 ms │  1.07x slower │
│ QQuery 41 │        14.10 / 14.20 ±0.12 / 14.42 ms │         14.46 / 15.34 ±1.56 / 18.45 ms │  1.08x slower │
│ QQuery 42 │        24.14 / 24.45 ±0.21 / 24.74 ms │         24.47 / 25.36 ±0.76 / 26.60 ms │     no change │
│ QQuery 43 │           4.88 / 5.01 ±0.20 / 5.40 ms │            5.07 / 5.20 ±0.12 / 5.41 ms │     no change │
│ QQuery 44 │         8.64 / 10.13 ±2.44 / 15.00 ms │            8.98 / 9.07 ±0.07 / 9.17 ms │ +1.12x faster │
│ QQuery 45 │        36.60 / 37.74 ±1.55 / 40.80 ms │         36.16 / 36.42 ±0.24 / 36.82 ms │     no change │
│ QQuery 46 │        11.57 / 11.83 ±0.20 / 12.06 ms │         11.48 / 11.69 ±0.13 / 11.88 ms │     no change │
│ QQuery 47 │     209.68 / 212.31 ±2.82 / 216.96 ms │      206.66 / 211.30 ±3.59 / 217.57 ms │     no change │
│ QQuery 48 │        91.72 / 92.14 ±0.28 / 92.55 ms │         92.10 / 93.24 ±1.23 / 95.64 ms │     no change │
│ QQuery 49 │        69.40 / 73.18 ±5.59 / 84.15 ms │         69.61 / 70.94 ±1.18 / 73.13 ms │     no change │
│ QQuery 50 │        57.39 / 58.06 ±0.56 / 59.10 ms │         58.00 / 58.34 ±0.34 / 58.95 ms │     no change │
│ QQuery 51 │        93.34 / 95.77 ±1.27 / 96.88 ms │        93.38 / 96.16 ±2.53 / 100.58 ms │     no change │
│ QQuery 52 │        24.21 / 26.75 ±4.23 / 35.19 ms │         24.42 / 24.71 ±0.20 / 25.03 ms │ +1.08x faster │
│ QQuery 53 │        29.62 / 30.83 ±1.27 / 33.18 ms │         29.51 / 29.83 ±0.19 / 30.01 ms │     no change │
│ QQuery 54 │        53.75 / 54.63 ±0.64 / 55.38 ms │         54.14 / 54.98 ±1.08 / 56.91 ms │     no change │
│ QQuery 55 │        24.03 / 24.66 ±0.45 / 25.10 ms │         23.91 / 24.05 ±0.08 / 24.12 ms │     no change │
│ QQuery 56 │        38.79 / 39.06 ±0.18 / 39.28 ms │         38.50 / 38.68 ±0.25 / 39.18 ms │     no change │
│ QQuery 57 │     153.13 / 154.73 ±1.35 / 157.17 ms │      152.01 / 154.59 ±3.34 / 161.07 ms │     no change │
│ QQuery 58 │     104.63 / 105.38 ±0.59 / 106.35 ms │      103.57 / 106.15 ±3.53 / 113.08 ms │     no change │
│ QQuery 59 │     108.13 / 108.34 ±0.13 / 108.52 ms │      107.57 / 108.10 ±0.45 / 108.89 ms │     no change │
│ QQuery 60 │        38.60 / 38.96 ±0.25 / 39.38 ms │         38.68 / 38.84 ±0.16 / 39.13 ms │     no change │
│ QQuery 61 │        10.67 / 10.81 ±0.14 / 11.06 ms │         10.70 / 10.80 ±0.14 / 11.06 ms │     no change │
│ QQuery 62 │        38.48 / 39.64 ±1.82 / 43.27 ms │         38.60 / 38.78 ±0.15 / 39.02 ms │     no change │
│ QQuery 63 │        29.34 / 30.38 ±0.97 / 32.21 ms │         29.42 / 29.83 ±0.32 / 30.28 ms │     no change │
│ QQuery 64 │     368.86 / 374.49 ±4.28 / 381.48 ms │      365.78 / 372.04 ±4.36 / 378.96 ms │     no change │
│ QQuery 65 │        77.70 / 78.84 ±0.96 / 80.57 ms │         76.34 / 76.63 ±0.20 / 76.98 ms │     no change │
│ QQuery 66 │        73.72 / 74.55 ±0.52 / 75.26 ms │         72.47 / 74.19 ±2.40 / 78.96 ms │     no change │
│ QQuery 67 │     223.30 / 227.75 ±4.78 / 235.66 ms │      216.54 / 223.71 ±4.50 / 229.53 ms │     no change │
│ QQuery 68 │        11.67 / 11.84 ±0.11 / 11.94 ms │         11.68 / 11.83 ±0.18 / 12.19 ms │     no change │
│ QQuery 69 │        54.39 / 56.11 ±1.73 / 58.99 ms │         54.36 / 54.58 ±0.17 / 54.78 ms │     no change │
│ QQuery 70 │       97.99 / 98.76 ±0.77 / 100.23 ms │         97.25 / 97.57 ±0.26 / 97.93 ms │     no change │
│ QQuery 71 │        35.03 / 36.04 ±0.76 / 37.38 ms │         35.39 / 35.86 ±0.35 / 36.49 ms │     no change │
│ QQuery 72 │ 1573.84 / 1650.42 ±47.19 / 1721.02 ms │ 1411.64 / 1483.46 ±102.68 / 1683.96 ms │ +1.11x faster │
│ QQuery 73 │           9.56 / 9.75 ±0.15 / 9.97 ms │            9.07 / 9.20 ±0.20 / 9.59 ms │ +1.06x faster │
│ QQuery 74 │     236.38 / 239.15 ±2.65 / 242.91 ms │      221.98 / 224.43 ±1.96 / 227.96 ms │ +1.07x faster │
│ QQuery 75 │     139.91 / 143.46 ±4.44 / 152.23 ms │      137.87 / 140.86 ±3.38 / 147.22 ms │     no change │
│ QQuery 76 │        35.03 / 36.66 ±2.07 / 40.75 ms │         34.74 / 34.83 ±0.06 / 34.93 ms │ +1.05x faster │
│ QQuery 77 │        60.22 / 61.08 ±0.47 / 61.60 ms │         59.34 / 59.73 ±0.35 / 60.30 ms │     no change │
│ QQuery 78 │    275.78 / 288.23 ±10.75 / 306.75 ms │     268.02 / 282.06 ±12.38 / 300.83 ms │     no change │
│ QQuery 79 │        68.71 / 69.89 ±1.15 / 71.96 ms │         67.75 / 68.05 ±0.30 / 68.54 ms │     no change │
│ QQuery 80 │        94.90 / 97.12 ±1.42 / 98.98 ms │        94.08 / 97.57 ±3.81 / 104.28 ms │     no change │
│ QQuery 81 │        28.90 / 30.52 ±2.99 / 36.49 ms │         27.86 / 28.34 ±0.33 / 28.87 ms │ +1.08x faster │
│ QQuery 82 │        17.97 / 18.38 ±0.38 / 19.10 ms │         17.50 / 17.73 ±0.12 / 17.82 ms │     no change │
│ QQuery 83 │        33.31 / 34.41 ±1.60 / 37.58 ms │         33.00 / 33.10 ±0.10 / 33.30 ms │     no change │
│ QQuery 84 │        30.64 / 31.57 ±0.54 / 32.17 ms │         30.63 / 30.93 ±0.22 / 31.23 ms │     no change │
│ QQuery 85 │     101.26 / 102.41 ±0.83 / 103.80 ms │      101.01 / 103.34 ±2.78 / 107.15 ms │     no change │
│ QQuery 86 │        24.79 / 27.33 ±4.29 / 35.88 ms │         24.52 / 24.95 ±0.42 / 25.73 ms │ +1.10x faster │
│ QQuery 87 │        64.69 / 65.78 ±1.22 / 68.04 ms │         63.13 / 63.96 ±0.43 / 64.34 ms │     no change │
│ QQuery 88 │        60.04 / 60.58 ±0.35 / 60.99 ms │         59.54 / 59.81 ±0.28 / 60.21 ms │     no change │
│ QQuery 89 │        36.08 / 36.49 ±0.30 / 36.96 ms │         35.88 / 36.53 ±0.71 / 37.86 ms │     no change │
│ QQuery 90 │        16.73 / 16.93 ±0.17 / 17.20 ms │         16.45 / 17.70 ±2.28 / 22.25 ms │     no change │
│ QQuery 91 │        41.04 / 42.12 ±0.85 / 43.30 ms │         40.47 / 41.17 ±0.66 / 42.21 ms │     no change │
│ QQuery 92 │        29.90 / 30.65 ±0.73 / 31.66 ms │         29.81 / 30.43 ±0.52 / 31.36 ms │     no change │
│ QQuery 93 │        50.29 / 51.41 ±0.63 / 52.17 ms │         49.42 / 50.23 ±0.57 / 50.92 ms │     no change │
│ QQuery 94 │        38.60 / 39.41 ±0.52 / 39.95 ms │         38.49 / 39.03 ±0.30 / 39.31 ms │     no change │
│ QQuery 95 │        80.70 / 82.89 ±1.89 / 86.39 ms │         81.85 / 83.28 ±1.41 / 85.88 ms │     no change │
│ QQuery 96 │        24.03 / 24.14 ±0.09 / 24.27 ms │         24.26 / 24.47 ±0.12 / 24.64 ms │     no change │
│ QQuery 97 │        51.05 / 51.99 ±1.20 / 54.35 ms │         50.67 / 51.27 ±0.99 / 53.25 ms │     no change │
│ QQuery 98 │        38.87 / 39.43 ±0.40 / 40.10 ms │         38.81 / 39.00 ±0.21 / 39.38 ms │     no change │
│ QQuery 99 │        56.85 / 58.69 ±3.20 / 65.08 ms │         57.45 / 57.86 ±0.31 / 58.21 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                  ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                  │ 9237.44ms │
│ Total Time (rich-T-kid_arrow-rs-12126-benchmark)   │ 9088.48ms │
│ Average Time (HEAD)                                │   93.31ms │
│ Average Time (rich-T-kid_arrow-rs-12126-benchmark) │   91.80ms │
│ Queries Faster                                     │         8 │
│ Queries Slower                                     │         5 │
│ Queries with No Change                             │        86 │
│ Queries with Failure                               │         0 │
└────────────────────────────────────────────────────┴───────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 0b062c85157df1ecc618582decea1346d60db6b0 | Changed: 1293f9c59bc340cfbd91787ea6ab101698e20d1d

tpcds — tpcds_sf1

Query Base Changed Change
Query 1 156.1 KiB 198.0 KiB +26.8%
Query 2 91.0 MiB 79.0 MiB -13.2%
Query 3 1.9 MiB 2.1 MiB +9.9%
Query 4 87.8 MiB 87.8 MiB +0.0%
Query 5 12.5 MiB 12.0 MiB -3.9%
Query 6 4.8 MiB 4.7 MiB -1.7%
Query 7 3.8 MiB 3.8 MiB +0.5%
Query 8 2.1 MiB 1.8 MiB -13.8%
Query 9 32.1 KiB 32.1 KiB +0.0%
Query 10 2.0 MiB 2.0 MiB +0.0%
Query 11 58.9 MiB 58.8 MiB -0.1%
Query 12 114.4 MiB 124.1 MiB +8.5%
Query 13 13.0 MiB 13.7 MiB +5.1%
Query 14 170.9 MiB 170.9 MiB +0.0%
Query 15 4.3 MiB 4.2 MiB -2.2%
Query 16 118.6 KiB 118.6 KiB +0.0%
Query 17 41.5 MiB 42.0 MiB +1.1%
Query 18 21.3 MiB 21.2 MiB -0.1%
Query 19 2.1 MiB 2.2 MiB +1.2%
Query 20 124.3 MiB 114.6 MiB -7.8%
Query 21 3.2 MiB 4.0 MiB +22.5%
Query 22 61.8 MiB 61.8 MiB +0.0%
Query 23 212.1 MiB 211.7 MiB -0.2%
Query 24 38.7 MiB 38.7 MiB +0.0%
Query 25 99.9 MiB 99.9 MiB +0.0%
Query 26 4.4 MiB 5.8 MiB +30.2%
Query 27 1.6 MiB 1.6 MiB +0.0%
Query 28 1.5 MiB 1.3 MiB -9.1%
Query 29 102.9 MiB 104.7 MiB +1.8%
Query 30 3.8 MiB 3.8 MiB +0.4%
Query 31 13.4 MiB 13.4 MiB +0.0%
Query 32 2.7 MiB 2.7 MiB +0.0%
Query 33 4.0 MiB 4.0 MiB +0.0%
Query 34 6.0 MiB 6.0 MiB +0.0%
Query 35 17.1 MiB 13.9 MiB -18.7%
Query 36 10.3 MiB 10.3 MiB +0.0%
Query 37 3.3 MiB 3.3 MiB +0.0%
Query 38 10.9 MiB 10.8 MiB -1.2%
Query 39 73.7 MiB 42.0 MiB -43.1%
Query 40 12.2 MiB 12.2 MiB +0.0%
Query 41 2.1 MiB 2.1 MiB +0.0%
Query 42 1.0 MiB 1.3 MiB +20.9%
Query 43 275.0 KiB 275.0 KiB +0.0%
Query 44 233.2 KiB 233.2 KiB +0.0%
Query 45 4.5 MiB 4.8 MiB +5.9%
Query 46 6.2 MiB 6.2 MiB +0.0%
Query 47 131.7 MiB 142.0 MiB +7.8%
Query 48 11.7 MiB 11.7 MiB +0.0%
Query 49 52.8 MiB 42.8 MiB -18.9%
Query 50 17.3 MiB 16.9 MiB -2.2%
Query 51 178.0 MiB 177.7 MiB -0.2%
Query 52 1.8 MiB 2.0 MiB +13.1%
Query 53 30.6 MiB 30.6 MiB +0.0%
Query 54 2.2 MiB 2.2 MiB +0.0%
Query 55 1.9 MiB 1.7 MiB -8.3%
Query 56 3.0 MiB 2.9 MiB -0.6%
Query 57 122.6 MiB 122.6 MiB +0.0%
Query 58 4.1 MiB 3.9 MiB -5.0%
Query 59 11.7 MiB 11.7 MiB +0.0%
Query 60 5.7 MiB 5.7 MiB -0.0%
Query 61 982.9 KiB 1.2 MiB +19.8%
Query 62 2.3 MiB 2.6 MiB +14.1%
Query 63 30.6 MiB 30.6 MiB +0.0%
Query 64 115.3 MiB 117.2 MiB +1.6%
Query 65 29.9 MiB 29.9 MiB +0.0%
Query 66 14.9 MiB 11.9 MiB -20.7%
Query 67 135.1 MiB 138.0 MiB +2.1%
Query 68 6.2 MiB 6.2 MiB +0.0%
Query 69 4.3 MiB 4.3 MiB +0.0%
Query 70 44.0 MiB 44.0 MiB -0.0%
Query 71 50.4 MiB 60.4 MiB +19.8%
Query 72 97.2 MiB 97.2 MiB +0.0%
Query 73 6.0 MiB 6.0 MiB +0.0%
Query 74 36.2 MiB 36.2 MiB +0.0%
Query 75 22.6 MiB 22.2 MiB -1.7%
Query 76 4.4 MiB 4.2 MiB -4.8%
Query 77 4.1 MiB 4.1 MiB -0.0%
Query 78 335.7 MiB 335.7 MiB +0.0%
Query 79 9.3 MiB 9.3 MiB +0.0%
Query 80 41.7 MiB 43.2 MiB +3.7%
Query 81 4.1 MiB 4.2 MiB +2.3%
Query 82 4.9 MiB 4.9 MiB +0.0%
Query 83 9.0 MiB 8.6 MiB -5.3%
Query 84 11.9 MiB 11.9 MiB +0.0%
Query 85 17.0 MiB 17.0 MiB +0.0%
Query 86 82.1 MiB 82.1 MiB +0.0%
Query 87 12.9 MiB 12.9 MiB +0.0%
Query 88 1.6 MiB 1.6 MiB -3.4%
Query 89 122.1 MiB 122.2 MiB +0.1%
Query 90 348.6 KiB 348.6 KiB +0.0%
Query 91 21.4 MiB 21.4 MiB -0.2%
Query 92 2.5 MiB 2.7 MiB +8.9%
Query 93 15.1 MiB 15.4 MiB +1.7%
Query 94 3.7 MiB 3.7 MiB +0.0%
Query 95 24.0 MiB 22.9 MiB -4.3%
Query 96 224.7 KiB 224.7 KiB +0.0%
Query 97 33.1 MiB 33.1 MiB -0.2%
Query 98 124.5 MiB 114.8 MiB -7.8%
Query 99 1.9 MiB 1.8 MiB -3.6%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark Side Max pool peak Peak RSS Gap RSS / pool
tpcds base (0b062c85157df1ecc618582decea1346d60db6b0) 335.7 MiB 2.1 GiB 1.7 GiB 6.3×
tpcds changed (1293f9c59bc340cfbd91787ea6ab101698e20d1d) 335.7 MiB 2.2 GiB 1.9 GiB 6.7×
Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 50.0s
Peak memory 2.1 GiB
Avg memory 1.4 GiB
CPU user 190.0s
CPU sys 5.8s
Peak spill 0 B

tpcds — branch

Metric Value
Wall time 50.0s
Peak memory 2.2 GiB
Avg memory 1.4 GiB
CPU user 182.4s
CPU sys 5.7s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

adriangbot commented Nov 8, 2026

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing 1293f9c (1293f9c) to 0b062c8 diff

Run configuration 
run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "0b062c85157df1ecc618582decea1346d60db6b0"
changed:
  ref: "1293f9c59bc340cfbd91787ea6ab101698e20d1d"
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 rich-T-kid_arrow-rs-12126-benchmark
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_arrow-rs-12126-benchmark ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.17 ms │                             1.14 ms │     no change │
│ QQuery 1  │   11.45 ms │                            11.48 ms │     no change │
│ QQuery 2  │   35.88 ms │                            35.13 ms │     no change │
│ QQuery 3  │   31.34 ms │                            31.09 ms │     no change │
│ QQuery 4  │  252.15 ms │                           247.45 ms │     no change │
│ QQuery 5  │  688.59 ms │                           698.94 ms │     no change │
│ QQuery 6  │    1.19 ms │                             1.21 ms │     no change │
│ QQuery 7  │   12.75 ms │                            12.86 ms │     no change │
│ QQuery 8  │  349.79 ms │                           347.76 ms │     no change │
│ QQuery 9  │  491.11 ms │                           494.59 ms │     no change │
│ QQuery 10 │  204.43 ms │                           204.17 ms │     no change │
│ QQuery 11 │  215.36 ms │                           215.99 ms │     no change │
│ QQuery 12 │  698.43 ms │                           682.77 ms │     no change │
│ QQuery 13 │  949.59 ms │                           963.39 ms │     no change │
│ QQuery 14 │  752.71 ms │                           744.85 ms │     no change │
│ QQuery 15 │  303.71 ms │                           304.20 ms │     no change │
│ QQuery 16 │ 1268.83 ms │                          1264.39 ms │     no change │
│ QQuery 17 │ 1220.38 ms │                          1230.41 ms │     no change │
│ QQuery 18 │ 1990.70 ms │                          1983.64 ms │     no change │
│ QQuery 19 │   27.76 ms │                            27.63 ms │     no change │
│ QQuery 20 │ 1069.74 ms │                          1066.63 ms │     no change │
│ QQuery 21 │ 1384.33 ms │                          1388.23 ms │     no change │
│ QQuery 22 │ 3331.43 ms │                          3265.51 ms │     no change │
│ QQuery 23 │ 9397.85 ms │                          9307.36 ms │     no change │
│ QQuery 24 │   96.91 ms │                            96.87 ms │     no change │
│ QQuery 25 │  356.67 ms │                           358.00 ms │     no change │
│ QQuery 26 │   96.95 ms │                            97.54 ms │     no change │
│ QQuery 27 │ 1159.06 ms │                          1162.85 ms │     no change │
│ QQuery 28 │ 2545.13 ms │                          2551.77 ms │     no change │
│ QQuery 29 │   38.94 ms │                            38.57 ms │     no change │
│ QQuery 30 │  558.58 ms │                           558.85 ms │     no change │
│ QQuery 31 │  552.35 ms │                           549.93 ms │     no change │
│ QQuery 32 │ 1078.08 ms │                          1098.30 ms │     no change │
│ QQuery 33 │ 3030.26 ms │                          3066.18 ms │     no change │
│ QQuery 34 │ 3091.87 ms │                          3104.13 ms │     no change │
│ QQuery 35 │  313.21 ms │                           316.09 ms │     no change │
│ QQuery 36 │  128.73 ms │                           128.32 ms │     no change │
│ QQuery 37 │   96.99 ms │                            96.08 ms │     no change │
│ QQuery 38 │   82.18 ms │                            82.20 ms │     no change │
│ QQuery 39 │  296.02 ms │                           278.66 ms │ +1.06x faster │
│ QQuery 40 │   14.87 ms │                            14.61 ms │     no change │
│ QQuery 41 │   14.18 ms │                            14.43 ms │     no change │
│ QQuery 42 │   13.94 ms │                            13.81 ms │     no change │
└───────────┴────────────┴─────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                  ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                  │ 38255.63ms │
│ Total Time (rich-T-kid_arrow-rs-12126-benchmark)   │ 38158.00ms │
│ Average Time (HEAD)                                │   889.67ms │
│ Average Time (rich-T-kid_arrow-rs-12126-benchmark) │   887.40ms │
│ Queries Faster                                     │          1 │
│ Queries Slower                                     │          0 │
│ Queries with No Change                             │         42 │
│ Queries with Failure                               │          0 │
└────────────────────────────────────────────────────┴────────────┘

Distribution per query (min / mean ±stddev / max):

Comparing HEAD and rich-T-kid_arrow-rs-12126-benchmark
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃    rich-T-kid_arrow-rs-12126-benchmark ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │            1.17 / 1.83 ±1.17 / 4.16 ms │            1.14 / 1.81 ±1.22 / 4.24 ms │     no change │
│ QQuery 1  │         11.45 / 11.63 ±0.12 / 11.77 ms │         11.48 / 11.94 ±0.27 / 12.28 ms │     no change │
│ QQuery 2  │         35.88 / 36.37 ±0.25 / 36.62 ms │         35.13 / 35.41 ±0.25 / 35.79 ms │     no change │
│ QQuery 3  │         31.34 / 32.50 ±0.88 / 34.04 ms │         31.09 / 31.70 ±0.54 / 32.65 ms │     no change │
│ QQuery 4  │      252.15 / 257.23 ±4.19 / 262.69 ms │      247.45 / 254.06 ±5.70 / 261.83 ms │     no change │
│ QQuery 5  │      688.59 / 702.29 ±9.92 / 718.26 ms │      698.94 / 705.18 ±3.80 / 710.29 ms │     no change │
│ QQuery 6  │            1.19 / 1.34 ±0.24 / 1.81 ms │            1.21 / 1.37 ±0.24 / 1.84 ms │     no change │
│ QQuery 7  │         12.75 / 12.99 ±0.14 / 13.16 ms │         12.86 / 13.03 ±0.12 / 13.23 ms │     no change │
│ QQuery 8  │      349.79 / 353.68 ±2.42 / 356.44 ms │      347.76 / 353.36 ±8.36 / 369.94 ms │     no change │
│ QQuery 9  │      491.11 / 505.51 ±9.18 / 519.49 ms │     494.59 / 506.67 ±15.89 / 536.54 ms │     no change │
│ QQuery 10 │      204.43 / 206.43 ±2.03 / 209.01 ms │      204.17 / 206.37 ±1.49 / 208.54 ms │     no change │
│ QQuery 11 │      215.36 / 217.31 ±1.10 / 218.53 ms │      215.99 / 216.51 ±0.45 / 217.07 ms │     no change │
│ QQuery 12 │      698.43 / 706.42 ±5.86 / 714.98 ms │     682.77 / 706.58 ±14.18 / 722.14 ms │     no change │
│ QQuery 13 │    949.59 / 973.19 ±28.32 / 1027.33 ms │    963.39 / 985.19 ±24.52 / 1024.88 ms │     no change │
│ QQuery 14 │     752.71 / 769.25 ±21.15 / 808.80 ms │     744.85 / 762.20 ±12.74 / 778.57 ms │     no change │
│ QQuery 15 │      303.71 / 308.41 ±5.35 / 318.06 ms │      304.20 / 310.46 ±8.11 / 326.20 ms │     no change │
│ QQuery 16 │   1268.83 / 1275.76 ±5.02 / 1280.63 ms │   1264.39 / 1278.53 ±9.74 / 1292.97 ms │     no change │
│ QQuery 17 │  1220.38 / 1229.92 ±11.18 / 1251.88 ms │  1230.41 / 1240.93 ±12.01 / 1264.06 ms │     no change │
│ QQuery 18 │  1990.70 / 2020.34 ±27.46 / 2062.94 ms │   1983.64 / 1998.01 ±8.79 / 2010.71 ms │     no change │
│ QQuery 19 │         27.76 / 29.69 ±2.87 / 35.36 ms │       27.63 / 44.28 ±32.24 / 108.75 ms │  1.49x slower │
│ QQuery 20 │  1069.74 / 1082.06 ±13.12 / 1098.78 ms │   1066.63 / 1070.46 ±2.66 / 1074.37 ms │     no change │
│ QQuery 21 │  1384.33 / 1397.50 ±11.33 / 1414.23 ms │   1388.23 / 1395.28 ±3.81 / 1399.56 ms │     no change │
│ QQuery 22 │  3331.43 / 3351.72 ±28.84 / 3408.06 ms │  3265.51 / 3312.11 ±37.81 / 3363.08 ms │     no change │
│ QQuery 23 │ 9397.85 / 9600.03 ±159.71 / 9774.49 ms │ 9307.36 / 9509.81 ±139.68 / 9725.05 ms │     no change │
│ QQuery 24 │       96.91 / 100.14 ±3.57 / 106.81 ms │      96.87 / 118.85 ±41.09 / 201.02 ms │  1.19x slower │
│ QQuery 25 │      356.67 / 359.34 ±2.67 / 362.86 ms │     358.00 / 368.97 ±20.21 / 409.36 ms │     no change │
│ QQuery 26 │        96.95 / 98.64 ±1.51 / 101.35 ms │        97.54 / 99.14 ±1.71 / 102.20 ms │     no change │
│ QQuery 27 │   1159.06 / 1167.46 ±6.78 / 1177.70 ms │   1162.85 / 1170.47 ±5.11 / 1176.50 ms │     no change │
│ QQuery 28 │  2545.13 / 2588.43 ±32.91 / 2645.18 ms │  2551.77 / 2587.64 ±33.60 / 2645.87 ms │     no change │
│ QQuery 29 │        38.94 / 53.32 ±19.80 / 89.39 ms │         38.57 / 42.65 ±7.65 / 57.96 ms │ +1.25x faster │
│ QQuery 30 │     558.58 / 571.33 ±18.83 / 608.57 ms │      558.85 / 567.77 ±9.86 / 586.86 ms │     no change │
│ QQuery 31 │      552.35 / 563.50 ±9.46 / 574.98 ms │     549.93 / 567.44 ±19.46 / 602.47 ms │     no change │
│ QQuery 32 │ 1078.08 / 1211.81 ±147.00 / 1481.81 ms │  1098.30 / 1141.24 ±36.13 / 1185.84 ms │ +1.06x faster │
│ QQuery 33 │  3030.26 / 3161.95 ±73.76 / 3232.27 ms │  3066.18 / 3161.56 ±93.71 / 3336.36 ms │     no change │
│ QQuery 34 │  3091.87 / 3153.31 ±56.90 / 3252.39 ms │  3104.13 / 3211.11 ±75.49 / 3308.49 ms │     no change │
│ QQuery 35 │     313.21 / 360.51 ±49.21 / 441.00 ms │     316.09 / 345.74 ±33.42 / 409.71 ms │     no change │
│ QQuery 36 │      128.73 / 136.56 ±5.52 / 145.97 ms │      128.32 / 140.82 ±9.97 / 154.89 ms │     no change │
│ QQuery 37 │        96.99 / 99.94 ±2.80 / 104.00 ms │       96.08 / 102.17 ±3.78 / 106.90 ms │     no change │
│ QQuery 38 │         82.18 / 85.04 ±1.99 / 87.92 ms │         82.20 / 85.51 ±2.69 / 90.19 ms │     no change │
│ QQuery 39 │      296.02 / 303.29 ±5.83 / 309.99 ms │     278.66 / 290.62 ±11.45 / 312.04 ms │     no change │
│ QQuery 40 │         14.87 / 18.00 ±2.98 / 22.86 ms │         14.61 / 17.27 ±3.88 / 24.97 ms │     no change │
│ QQuery 41 │         14.18 / 14.66 ±0.41 / 15.36 ms │         14.43 / 17.79 ±5.15 / 27.84 ms │  1.21x slower │
│ QQuery 42 │         13.94 / 14.10 ±0.18 / 14.37 ms │         13.81 / 14.20 ±0.32 / 14.73 ms │     no change │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                  ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                  │ 39144.74ms │
│ Total Time (rich-T-kid_arrow-rs-12126-benchmark)   │ 39002.21ms │
│ Average Time (HEAD)                                │   910.34ms │
│ Average Time (rich-T-kid_arrow-rs-12126-benchmark) │   907.03ms │
│ Queries Faster                                     │          2 │
│ Queries Slower                                     │          3 │
│ Queries with No Change                             │         38 │
│ Queries with Failure                               │          0 │
└────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 0b062c85157df1ecc618582decea1346d60db6b0 | Changed: 1293f9c59bc340cfbd91787ea6ab101698e20d1d

clickbench_partitioned

Query Base Changed Change
Query 0 0 B 0 B 0.0%
Query 1 104 B 104 B +0.0%
Query 2 936 B 936 B +0.0%
Query 3 312 B 312 B +0.0%
Query 4 829.6 MiB 813.5 MiB -1.9%
Query 5 1.4 GiB 1.4 GiB +2.4%
Query 6 0 B 0 B 0.0%
Query 7 50.2 MiB 60.3 MiB +20.2%
Query 8 948.4 MiB 966.6 MiB +1.9%
Query 9 593.9 MiB 594.2 MiB +0.1%
Query 10 72.9 MiB 74.1 MiB +1.7%
Query 11 74.8 MiB 75.8 MiB +1.4%
Query 12 1.6 GiB 1.6 GiB +0.6%
Query 13 1.6 GiB 1.6 GiB +0.0%
Query 14 1.6 GiB 1.6 GiB +0.9%
Query 15 1.2 GiB 1.3 GiB +5.1%
Query 16 3.3 GiB 3.1 GiB -4.2%
Query 17 3.2 GiB 3.2 GiB +2.0%
Query 18 7.1 GiB 7.1 GiB +0.3%
Query 19 0 B 0 B 0.0%
Query 20 104 B 104 B +0.0%
Query 21 1.5 MiB 1.5 MiB -0.0%
Query 22 3.0 MiB 2.7 MiB -10.7%
Query 23 12.5 MiB 11.0 MiB -12.1%
Query 24 5.8 MiB 5.5 MiB -5.5%
Query 25 5.9 MiB 6.5 MiB +9.4%
Query 26 6.9 MiB 6.1 MiB -11.7%
Query 27 2.5 MiB 2.5 MiB +0.0%
Query 28 1.9 GiB 1.9 GiB -4.4%
Query 29 624 B 624 B +0.0%
Query 30 625.0 MiB 669.7 MiB +7.1%
Query 31 1.5 GiB 1.5 GiB -1.6%
Query 32 7.1 GiB 7.1 GiB +0.3%
Query 33 10.0 GiB 10.2 GiB +2.1%
Query 34 10.0 GiB 9.6 GiB -3.2%
Query 35 668.7 MiB 668.8 MiB +0.0%
Query 36 101.6 MiB 100.4 MiB -1.1%
Query 37 7.4 MiB 7.0 MiB -5.0%
Query 38 5.2 MiB 5.1 MiB -0.0%
Query 39 213.0 MiB 213.0 MiB +0.0%
Query 40 2.2 MiB 2.3 MiB +6.0%
Query 41 3.3 MiB 3.1 MiB -8.3%
Query 42 1.8 MiB 1.8 MiB +1.5%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark Side Max pool peak Peak RSS Gap RSS / pool
clickbench_partitioned base (0b062c85157df1ecc618582decea1346d60db6b0) 10.0 GiB 13.2 GiB 3.2 GiB 1.3×
clickbench_partitioned changed (1293f9c59bc340cfbd91787ea6ab101698e20d1d) 10.2 GiB 13.4 GiB 3.3 GiB 1.3×
Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 200.0s
Peak memory 13.2 GiB
Avg memory 5.7 GiB
CPU user 1995.9s
CPU sys 131.6s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 200.0s
Peak memory 13.4 GiB
Avg memory 5.6 GiB
CPU user 1992.6s
CPU sys 132.2s
Peak spill 0 B

File an issue against this benchmark runner

This was referenced Nov 8, 2026
Closed
Merged
@Rich-T-kid Rich-T-kid closed this Nov 8, 2026
Surajsuthar pushed a commit to Surajsuthar/datafusion that referenced this pull request Nov 14, 2026
@Rich-T-kid @jayzhan211
perf(aggregate): reuse dictionary hashes and group mappings (apache#2…
8f279e1 
…6826)

## Which issue does this PR close?

Closes apache#26812.

## Rationale for this change

Dictionary batches often share one values array while only their keys
change. Partial aggregation was rehashing the same dictionary values for
every batch.

## What changes are included in this PR?

The non-streaming group table now caches dictionary value hashes while
the values `Arc` is unchanged. A single dictionary group key can also
map known keys directly to group indices.

```text
Dictionary batch
       |
       v
single dictionary column + complete cached mapping?
       | yes                              | no
       v                                  v
keys -> cached group IDs           per-column hashing
                                   |                 |
                            eligible dictionary   fallback
                                   |                 |
                          cached value hashes   hash column
                                   |                 |
                          scatter through keys -----+
                                   |
                                   v
                         hash/equality group lookup
```

New keys, changed or oversized dictionaries, and partial emission use
the normal fallback. Multi-column hashing keeps the same semantics as
`create_hashes`. The benchmark fixtures are split into apache#27246.

## What is the testing strategy for this PR?

Unit tests cover cached and uncached hash consistency, unseen keys,
changed dictionaries, null groups, partial emission, and forced
collisions.

Hosted cases from apache#27246 show [alternating
dictionary-only](apache#26826 (comment))
1.30× faster; [alternating dictionary +
Int32](apache#26826 (comment))
and [fresh
dictionaries](apache#26826 (comment))
flat; and [shared dictionary +
Int32](apache#26826 (comment))
1.14× faster at 32K values and 1.28× at 100K.

Exact-head hosted suites with RLE dictionary reads enabled on both
sides:

-
[TPC-H](apache#27249 (comment)):
796.01 → 799.79 ms (0.5% higher).
-
[TPC-DS](apache#27249 (comment)):
9.02 → 8.88 s (1.6% lower).
-
[ClickBench](apache#27249 (comment)):
38.26 → 38.16 s (0.3% lower).

## Are there any user-facing changes?

No.

---------

Co-authored-by: Jay Zhan <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@Rich-T-kid @adriangbot