You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "adriangb (via GitHub)" <gi...@apache.org> on 2026/05/21 18:58:49 UTC
[PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
adriangb opened a new pull request, #22439:
URL: https://github.com/apache/datafusion/pull/22439
## Summary
`repartition_file_min_size` gates how aggressively `repartitioned()` splits file groups by byte range to fan a scan out across `target_partitions` worth of cores. At 10 MiB the default leaves several SF1-sized dimension tables (TPC-H \`part\` ≈ 24 MiB, TPC-DS \`customer_address\` ≈ 7 MiB, …) on a single partition, so any CPU-bound per-batch work in the scan (filter eval, dictionary expansion, etc.) is single-threaded even when the cluster has plenty of idle cores.
At 1 MiB those same files split cleanly into \`target_partitions\` byte ranges. The cost (more \`open()\` calls, more metadata loads) is small in absolute terms (≤10 extra opens per file in the worst case, each amortised over the row-group / page-index reads) and the existing knob is still available for workloads where it matters.
## Benchmark numbers
12-core, SF1, with the existing dynamic-filter-pushdown defaults preserved:
| Suite | default (10 MiB) | with this PR (1 MiB) |
|---|---|---|
| TPC-H total | 841 ms | 776 ms |
| TPC-H Q22 | ~30 ms | ~17 ms |
| TPC-DS total | 11.0 s | 11.1 s |
| ClickBench total | 21.7 s | 19.0 s |
## Test plan
- [x] \`cargo test --test sqllogictests\` — all 472 files pass after the information_schema snapshot and a csv_files reset.
- [ ] \`run benchmarks\`
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4547774839
> @Dandandan I know you evaluated this in #19690 and then closed the issue, presumably because the benchmarks did not reproduce an improvement you ere seeing locally. Maybe it's just the new benchmark runner, but this looks like there are consistent improvements for TPCH and some minor improvements for TPC-DS
Yeah, I agree it looks good!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4511863021
🤖 Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4511840612-263-zrwps 6.12.68+ #1 SMP Wed Apr 1 02:23:28 UTC 2026 aarch64 GNU/Linux`
<details><summary>CPU Details (lscpu)</summary>
```
Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
```
</details>
Comparing lower-repartition-file-min-size (5f6b84f90eeafbdf5506d090ebace8cf43266f1d) to 50d74a7 (merge-base) [diff](https://github.com/apache/datafusion/compare/50d74a704d7f35a6cc184251a8f1236cd94de684..5f6b84f90eeafbdf5506d090ebace8cf43266f1d) using: tpch
Results will be posted here when complete
---
[File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb merged PR #22439:
URL: https://github.com/apache/datafusion/pull/22439
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4511865929
🤖 Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4511840612-261-p76vb 6.12.68+ #1 SMP Wed Apr 1 02:23:28 UTC 2026 aarch64 GNU/Linux`
<details><summary>CPU Details (lscpu)</summary>
```
Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
```
</details>
Comparing lower-repartition-file-min-size (5f6b84f90eeafbdf5506d090ebace8cf43266f1d) to 50d74a7 (merge-base) [diff](https://github.com/apache/datafusion/compare/50d74a704d7f35a6cc184251a8f1236cd94de684..5f6b84f90eeafbdf5506d090ebace8cf43266f1d) using: clickbench_partitioned
Results will be posted here when complete
---
[File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4511866282
🤖 Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4511840612-262-mnchx 6.12.68+ #1 SMP Wed Apr 1 02:23:28 UTC 2026 aarch64 GNU/Linux`
<details><summary>CPU Details (lscpu)</summary>
```
Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
```
</details>
Comparing lower-repartition-file-min-size (5f6b84f90eeafbdf5506d090ebace8cf43266f1d) to 50d74a7 (merge-base) [diff](https://github.com/apache/datafusion/compare/50d74a704d7f35a6cc184251a8f1236cd94de684..5f6b84f90eeafbdf5506d090ebace8cf43266f1d) using: tpcds
Results will be posted here when complete
---
[File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4512047552
🤖 Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB)
<details><summary>CPU Details (lscpu)</summary>
```
Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
```
</details>
<details><summary>Details</summary>
<p>
```
Comparing HEAD and lower-repartition-file-min-size
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ lower-repartition-file-min-size ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0 │ 1.19 / 4.94 ±7.22 / 19.39 ms │ 1.23 / 4.88 ±7.15 / 19.17 ms │ no change │
│ QQuery 1 │ 12.73 / 13.25 ±0.32 / 13.66 ms │ 12.83 / 13.29 ±0.28 / 13.68 ms │ no change │
│ QQuery 2 │ 35.90 / 36.61 ±0.56 / 37.59 ms │ 36.71 / 37.08 ±0.35 / 37.75 ms │ no change │
│ QQuery 3 │ 31.83 / 32.29 ±0.34 / 32.77 ms │ 31.52 / 32.62 ±1.03 / 34.54 ms │ no change │
│ QQuery 4 │ 240.58 / 245.88 ±3.69 / 250.55 ms │ 246.07 / 250.69 ±3.63 / 256.78 ms │ no change │
│ QQuery 5 │ 288.85 / 290.48 ±1.31 / 292.48 ms │ 290.08 / 291.80 ±1.78 / 294.92 ms │ no change │
│ QQuery 6 │ 1.19 / 1.35 ±0.23 / 1.79 ms │ 1.19 / 1.32 ±0.23 / 1.77 ms │ no change │
│ QQuery 7 │ 14.04 / 14.22 ±0.20 / 14.58 ms │ 14.09 / 14.20 ±0.08 / 14.32 ms │ no change │
│ QQuery 8 │ 344.37 / 350.28 ±4.51 / 356.06 ms │ 339.99 / 346.68 ±3.58 / 350.31 ms │ no change │
│ QQuery 9 │ 484.42 / 502.51 ±10.33 / 512.67 ms │ 486.52 / 504.91 ±13.10 / 523.97 ms │ no change │
│ QQuery 10 │ 74.35 / 77.13 ±3.85 / 84.69 ms │ 73.03 / 73.90 ±0.48 / 74.42 ms │ no change │
│ QQuery 11 │ 86.37 / 87.95 ±1.23 / 89.70 ms │ 83.42 / 84.23 ±0.56 / 84.97 ms │ no change │
│ QQuery 12 │ 300.56 / 307.78 ±5.85 / 316.17 ms │ 284.66 / 289.38 ±4.74 / 296.51 ms │ +1.06x faster │
│ QQuery 13 │ 387.84 / 410.89 ±16.72 / 439.88 ms │ 378.65 / 398.90 ±15.91 / 416.16 ms │ no change │
│ QQuery 14 │ 299.61 / 308.39 ±5.58 / 316.44 ms │ 303.29 / 306.31 ±2.88 / 311.58 ms │ no change │
│ QQuery 15 │ 299.25 / 303.11 ±2.38 / 306.73 ms │ 301.98 / 306.30 ±2.96 / 310.42 ms │ no change │
│ QQuery 16 │ 659.95 / 666.22 ±4.29 / 673.17 ms │ 674.00 / 683.33 ±7.55 / 693.26 ms │ no change │
│ QQuery 17 │ 659.51 / 671.52 ±7.70 / 680.23 ms │ 666.72 / 682.27 ±9.52 / 692.01 ms │ no change │
│ QQuery 18 │ 1340.85 / 1385.15 ±41.23 / 1441.00 ms │ 1356.50 / 1389.15 ±21.64 / 1414.20 ms │ no change │
│ QQuery 19 │ 29.17 / 38.77 ±7.92 / 50.31 ms │ 29.18 / 42.71 ±21.77 / 85.55 ms │ 1.10x slower │
│ QQuery 20 │ 524.84 / 534.44 ±7.26 / 546.97 ms │ 521.79 / 531.52 ±7.36 / 541.86 ms │ no change │
│ QQuery 21 │ 605.00 / 611.54 ±4.13 / 617.11 ms │ 601.76 / 610.35 ±4.63 / 614.27 ms │ no change │
│ QQuery 22 │ 1074.65 / 1091.83 ±14.98 / 1117.46 ms │ 1076.33 / 1098.59 ±12.17 / 1112.48 ms │ no change │
│ QQuery 23 │ 3296.90 / 3342.09 ±26.95 / 3370.82 ms │ 3322.50 / 3358.94 ±20.16 / 3377.13 ms │ no change │
│ QQuery 24 │ 44.18 / 50.32 ±10.35 / 70.95 ms │ 43.05 / 47.13 ±6.25 / 59.53 ms │ +1.07x faster │
│ QQuery 25 │ 115.89 / 118.66 ±3.13 / 124.52 ms │ 116.07 / 118.26 ±1.29 / 120.08 ms │ no change │
│ QQuery 26 │ 43.99 / 45.87 ±1.41 / 47.97 ms │ 43.94 / 45.49 ±1.91 / 49.25 ms │ no change │
│ QQuery 27 │ 692.02 / 699.59 ±10.22 / 719.58 ms │ 697.27 / 702.37 ±4.67 / 710.55 ms │ no change │
│ QQuery 28 │ 3140.38 / 3146.09 ±4.89 / 3152.80 ms │ 3139.57 / 3163.07 ±21.23 / 3198.19 ms │ no change │
│ QQuery 29 │ 41.94 / 47.71 ±6.70 / 57.98 ms │ 41.78 / 43.30 ±2.52 / 48.33 ms │ +1.10x faster │
│ QQuery 30 │ 310.54 / 334.80 ±24.72 / 381.05 ms │ 328.35 / 333.71 ±3.74 / 338.48 ms │ no change │
│ QQuery 31 │ 294.53 / 307.85 ±10.17 / 324.36 ms │ 301.20 / 308.97 ±4.02 / 312.72 ms │ no change │
│ QQuery 32 │ 1026.84 / 1036.66 ±11.58 / 1058.87 ms │ 1018.91 / 1026.91 ±8.98 / 1044.03 ms │ no change │
│ QQuery 33 │ 1568.82 / 1617.67 ±44.38 / 1691.71 ms │ 1530.15 / 1577.67 ±37.22 / 1618.80 ms │ no change │
│ QQuery 34 │ 1574.50 / 1621.70 ±42.02 / 1690.07 ms │ 1533.96 / 1618.27 ±45.85 / 1672.99 ms │ no change │
│ QQuery 35 │ 319.99 / 364.57 ±70.16 / 504.46 ms │ 312.84 / 331.56 ±24.13 / 378.35 ms │ +1.10x faster │
│ QQuery 36 │ 71.10 / 73.37 ±2.36 / 77.77 ms │ 68.23 / 74.70 ±6.27 / 83.71 ms │ no change │
│ QQuery 37 │ 36.99 / 43.20 ±5.20 / 49.35 ms │ 38.36 / 40.35 ±2.39 / 45.04 ms │ +1.07x faster │
│ QQuery 38 │ 41.82 / 45.28 ±2.33 / 48.95 ms │ 42.12 / 46.05 ±2.88 / 49.38 ms │ no change │
│ QQuery 39 │ 166.53 / 168.48 ±1.43 / 170.52 ms │ 147.94 / 159.49 ±7.27 / 170.09 ms │ +1.06x faster │
│ QQuery 40 │ 15.14 / 17.41 ±3.20 / 23.73 ms │ 14.79 / 15.35 ±0.46 / 15.95 ms │ +1.13x faster │
│ QQuery 41 │ 15.30 / 15.68 ±0.30 / 16.18 ms │ 14.97 / 16.65 ±1.91 / 19.64 ms │ 1.06x slower │
│ QQuery 42 │ 13.87 / 14.29 ±0.39 / 14.97 ms │ 13.80 / 15.50 ±2.15 / 19.71 ms │ 1.08x slower │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 21097.79ms │
│ Total Time (lower-repartition-file-min-size) │ 21038.15ms │
│ Average Time (HEAD) │ 490.65ms │
│ Average Time (lower-repartition-file-min-size) │ 489.26ms │
│ Queries Faster │ 7 │
│ Queries Slower │ 3 │
│ Queries with No Change │ 33 │
│ Queries with Failure │ 0 │
└────────────────────────────────────────────────┴────────────┘
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_partitioned — base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 110.0s |
| Peak memory | 31.0 GiB |
| Avg memory | 23.1 GiB |
| CPU user | 1091.1s |
| CPU sys | 74.0s |
| Peak spill | 0 B |
**clickbench_partitioned — branch**
| Metric | Value |
|--------|-------|
| Wall time | 110.0s |
| Peak memory | 30.9 GiB |
| Avg memory | 23.3 GiB |
| CPU user | 1087.1s |
| CPU sys | 73.7s |
| Peak spill | 0 B |
</details>
---
[File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612
run benchmarks
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4547780568
Thanks!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4512005749
🤖 Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB)
<details><summary>CPU Details (lscpu)</summary>
```
Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
```
</details>
<details><summary>Details</summary>
<p>
```
Comparing HEAD and lower-repartition-file-min-size
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ lower-repartition-file-min-size ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1 │ 5.66 / 6.24 ±0.90 / 8.03 ms │ 5.76 / 6.30 ±0.88 / 8.05 ms │ no change │
│ QQuery 2 │ 79.73 / 79.97 ±0.21 / 80.30 ms │ 79.85 / 80.25 ±0.25 / 80.53 ms │ no change │
│ QQuery 3 │ 28.54 / 28.95 ±0.32 / 29.49 ms │ 28.55 / 28.97 ±0.26 / 29.36 ms │ no change │
│ QQuery 4 │ 495.97 / 501.25 ±3.10 / 504.69 ms │ 471.37 / 474.43 ±2.73 / 479.29 ms │ +1.06x faster │
│ QQuery 5 │ 50.65 / 50.91 ±0.26 / 51.34 ms │ 51.77 / 53.15 ±1.70 / 56.43 ms │ no change │
│ QQuery 6 │ 35.36 / 35.57 ±0.24 / 35.89 ms │ 35.77 / 36.35 ±0.33 / 36.75 ms │ no change │
│ QQuery 7 │ 106.57 / 108.32 ±1.92 / 111.98 ms │ 94.84 / 95.73 ±0.57 / 96.50 ms │ +1.13x faster │
│ QQuery 8 │ 36.11 / 36.68 ±0.51 / 37.57 ms │ 36.66 / 37.68 ±1.18 / 39.97 ms │ no change │
│ QQuery 9 │ 50.32 / 53.15 ±1.85 / 55.15 ms │ 53.15 / 54.23 ±1.11 / 56.37 ms │ no change │
│ QQuery 10 │ 79.57 / 80.51 ±0.50 / 81.04 ms │ 67.19 / 67.44 ±0.24 / 67.76 ms │ +1.19x faster │
│ QQuery 11 │ 304.26 / 307.29 ±2.19 / 310.57 ms │ 286.46 / 289.16 ±1.75 / 291.95 ms │ +1.06x faster │
│ QQuery 12 │ 28.31 / 28.75 ±0.29 / 29.08 ms │ 28.16 / 28.55 ±0.30 / 29.01 ms │ no change │
│ QQuery 13 │ 125.14 / 126.14 ±1.14 / 128.34 ms │ 119.56 / 120.15 ±0.32 / 120.50 ms │ no change │
│ QQuery 14 │ 503.04 / 505.33 ±1.85 / 507.55 ms │ 500.95 / 502.74 ±1.50 / 505.26 ms │ no change │
│ QQuery 15 │ 60.21 / 61.44 ±0.95 / 62.64 ms │ 57.58 / 58.30 ±0.47 / 58.79 ms │ +1.05x faster │
│ QQuery 16 │ 6.30 / 6.47 ±0.14 / 6.66 ms │ 6.45 / 6.67 ±0.33 / 7.33 ms │ no change │
│ QQuery 17 │ 80.44 / 80.79 ±0.30 / 81.31 ms │ 80.33 / 80.81 ±0.44 / 81.51 ms │ no change │
│ QQuery 18 │ 150.02 / 150.91 ±0.49 / 151.44 ms │ 122.63 / 124.56 ±1.19 / 126.29 ms │ +1.21x faster │
│ QQuery 19 │ 40.64 / 40.89 ±0.23 / 41.32 ms │ 41.19 / 41.74 ±0.62 / 42.94 ms │ no change │
│ QQuery 20 │ 34.96 / 35.46 ±0.27 / 35.72 ms │ 34.63 / 35.48 ±1.17 / 37.77 ms │ no change │
│ QQuery 21 │ 16.57 / 16.90 ±0.28 / 17.26 ms │ 16.68 / 16.87 ±0.11 / 17.00 ms │ no change │
│ QQuery 22 │ 61.15 / 62.39 ±0.81 / 63.35 ms │ 62.17 / 62.63 ±0.33 / 63.03 ms │ no change │
│ QQuery 23 │ 473.31 / 480.42 ±3.92 / 484.62 ms │ 467.56 / 471.60 ±2.39 / 474.10 ms │ no change │
│ QQuery 24 │ 231.90 / 235.61 ±3.89 / 242.89 ms │ 222.95 / 224.48 ±1.80 / 227.13 ms │ no change │
│ QQuery 25 │ 113.00 / 113.63 ±0.52 / 114.47 ms │ 112.08 / 113.43 ±1.08 / 115.34 ms │ no change │
│ QQuery 26 │ 69.61 / 70.22 ±0.57 / 71.01 ms │ 57.62 / 57.95 ±0.26 / 58.26 ms │ +1.21x faster │
│ QQuery 27 │ 6.40 / 6.51 ±0.19 / 6.88 ms │ 6.53 / 6.67 ±0.14 / 6.93 ms │ no change │
│ QQuery 28 │ 56.57 / 61.12 ±2.28 / 62.50 ms │ 61.76 / 62.57 ±0.54 / 63.17 ms │ no change │
│ QQuery 29 │ 97.65 / 99.05 ±1.03 / 100.43 ms │ 97.60 / 97.81 ±0.16 / 98.07 ms │ no change │
│ QQuery 30 │ 29.72 / 30.15 ±0.43 / 30.87 ms │ 32.06 / 33.16 ±1.83 / 36.81 ms │ 1.10x slower │
│ QQuery 31 │ 110.57 / 111.76 ±1.69 / 115.06 ms │ 111.62 / 112.14 ±0.43 / 112.85 ms │ no change │
│ QQuery 32 │ 20.05 / 20.28 ±0.23 / 20.69 ms │ 19.78 / 20.03 ±0.35 / 20.71 ms │ no change │
│ QQuery 33 │ 38.16 / 38.43 ±0.20 / 38.67 ms │ 38.07 / 38.54 ±0.30 / 38.96 ms │ no change │
│ QQuery 34 │ 9.05 / 9.46 ±0.35 / 9.88 ms │ 9.13 / 9.47 ±0.22 / 9.82 ms │ no change │
│ QQuery 35 │ 79.55 / 79.80 ±0.20 / 80.00 ms │ 75.17 / 76.36 ±1.11 / 77.84 ms │ no change │
│ QQuery 36 │ 5.79 / 5.95 ±0.14 / 6.18 ms │ 5.84 / 5.98 ±0.15 / 6.26 ms │ no change │
│ QQuery 37 │ 6.74 / 6.84 ±0.06 / 6.92 ms │ 6.72 / 6.89 ±0.12 / 7.06 ms │ no change │
│ QQuery 38 │ 68.43 / 69.31 ±0.66 / 70.45 ms │ 62.33 / 62.99 ±0.50 / 63.78 ms │ +1.10x faster │
│ QQuery 39 │ 97.36 / 98.03 ±0.54 / 98.97 ms │ 97.55 / 98.93 ±1.23 / 100.89 ms │ no change │
│ QQuery 40 │ 22.27 / 22.60 ±0.26 / 22.98 ms │ 22.72 / 23.54 ±0.64 / 24.45 ms │ no change │
│ QQuery 41 │ 11.02 / 11.25 ±0.15 / 11.48 ms │ 10.93 / 11.20 ±0.18 / 11.48 ms │ no change │
│ QQuery 42 │ 23.60 / 23.90 ±0.31 / 24.46 ms │ 23.48 / 23.86 ±0.34 / 24.46 ms │ no change │
│ QQuery 43 │ 4.55 / 4.68 ±0.13 / 4.93 ms │ 4.70 / 4.76 ±0.08 / 4.91 ms │ no change │
│ QQuery 44 │ 10.19 / 11.43 ±1.99 / 15.40 ms │ 10.40 / 10.51 ±0.10 / 10.67 ms │ +1.09x faster │
│ QQuery 45 │ 39.59 / 40.57 ±0.68 / 41.66 ms │ 37.79 / 38.36 ±0.42 / 38.86 ms │ +1.06x faster │
│ QQuery 46 │ 12.75 / 12.98 ±0.19 / 13.17 ms │ 11.24 / 11.38 ±0.17 / 11.71 ms │ +1.14x faster │
│ QQuery 47 │ 223.52 / 226.65 ±2.77 / 231.79 ms │ 225.54 / 227.94 ±1.94 / 230.59 ms │ no change │
│ QQuery 48 │ 102.38 / 102.89 ±0.44 / 103.59 ms │ 96.68 / 98.18 ±1.00 / 99.67 ms │ no change │
│ QQuery 49 │ 78.14 / 79.56 ±1.30 / 81.67 ms │ 79.12 / 80.00 ±0.65 / 80.89 ms │ no change │
│ QQuery 50 │ 59.64 / 60.23 ±0.34 / 60.64 ms │ 59.29 / 60.51 ±1.37 / 63.05 ms │ no change │
│ QQuery 51 │ 92.17 / 94.01 ±1.23 / 95.96 ms │ 94.11 / 95.09 ±1.50 / 98.07 ms │ no change │
│ QQuery 52 │ 23.74 / 24.05 ±0.35 / 24.73 ms │ 23.87 / 24.20 ±0.30 / 24.75 ms │ no change │
│ QQuery 53 │ 29.12 / 29.43 ±0.23 / 29.75 ms │ 29.33 / 29.57 ±0.21 / 29.88 ms │ no change │
│ QQuery 54 │ 53.75 / 54.14 ±0.31 / 54.53 ms │ 54.60 / 56.05 ±1.62 / 59.22 ms │ no change │
│ QQuery 55 │ 23.43 / 23.56 ±0.11 / 23.74 ms │ 23.32 / 24.14 ±0.93 / 25.95 ms │ no change │
│ QQuery 56 │ 38.33 / 38.73 ±0.37 / 39.21 ms │ 38.50 / 38.93 ±0.39 / 39.55 ms │ no change │
│ QQuery 57 │ 173.55 / 175.20 ±1.24 / 177.09 ms │ 175.69 / 177.16 ±1.41 / 179.64 ms │ no change │
│ QQuery 58 │ 116.01 / 117.70 ±0.98 / 118.76 ms │ 115.24 / 115.75 ±0.47 / 116.60 ms │ no change │
│ QQuery 59 │ 116.61 / 117.05 ±0.36 / 117.38 ms │ 117.47 / 118.16 ±0.74 / 119.56 ms │ no change │
│ QQuery 60 │ 39.07 / 39.90 ±1.04 / 41.92 ms │ 38.90 / 39.40 ±0.33 / 39.82 ms │ no change │
│ QQuery 61 │ 12.49 / 12.65 ±0.20 / 13.03 ms │ 12.60 / 12.75 ±0.15 / 13.04 ms │ no change │
│ QQuery 62 │ 45.68 / 45.90 ±0.17 / 46.11 ms │ 45.67 / 46.53 ±1.39 / 49.31 ms │ no change │
│ QQuery 63 │ 29.28 / 29.58 ±0.26 / 29.98 ms │ 30.00 / 30.13 ±0.15 / 30.36 ms │ no change │
│ QQuery 64 │ 457.68 / 461.65 ±3.21 / 466.81 ms │ 393.46 / 399.25 ±4.63 / 406.84 ms │ +1.16x faster │
│ QQuery 65 │ 145.65 / 148.32 ±1.66 / 150.51 ms │ 145.68 / 147.18 ±1.12 / 148.45 ms │ no change │
│ QQuery 66 │ 77.84 / 80.26 ±2.71 / 85.36 ms │ 77.46 / 79.47 ±3.33 / 86.11 ms │ no change │
│ QQuery 67 │ 241.81 / 246.80 ±4.56 / 255.36 ms │ 245.55 / 249.62 ±3.39 / 253.78 ms │ no change │
│ QQuery 68 │ 12.65 / 12.86 ±0.16 / 13.12 ms │ 11.39 / 11.55 ±0.19 / 11.90 ms │ +1.11x faster │
│ QQuery 69 │ 75.12 / 75.52 ±0.21 / 75.67 ms │ 61.48 / 61.87 ±0.45 / 62.58 ms │ +1.22x faster │
│ QQuery 70 │ 104.13 / 108.16 ±3.51 / 114.68 ms │ 104.37 / 110.93 ±7.90 / 126.24 ms │ no change │
│ QQuery 71 │ 35.06 / 35.59 ±0.37 / 36.19 ms │ 34.93 / 35.61 ±0.39 / 36.09 ms │ no change │
│ QQuery 72 │ 2084.20 / 2225.30 ±109.92 / 2394.98 ms │ 2064.93 / 2123.91 ±44.83 / 2182.04 ms │ no change │
│ QQuery 73 │ 8.78 / 9.10 ±0.31 / 9.67 ms │ 8.94 / 9.62 ±0.47 / 10.30 ms │ 1.06x slower │
│ QQuery 74 │ 176.17 / 179.86 ±3.84 / 187.27 ms │ 165.45 / 167.50 ±1.89 / 170.65 ms │ +1.07x faster │
│ QQuery 75 │ 145.18 / 147.24 ±2.86 / 152.89 ms │ 147.10 / 148.65 ±2.29 / 153.20 ms │ no change │
│ QQuery 76 │ 35.11 / 35.29 ±0.13 / 35.50 ms │ 35.53 / 35.61 ±0.07 / 35.73 ms │ no change │
│ QQuery 77 │ 59.67 / 59.99 ±0.20 / 60.18 ms │ 60.51 / 61.88 ±1.59 / 64.99 ms │ no change │
│ QQuery 78 │ 186.39 / 189.44 ±1.71 / 190.97 ms │ 186.07 / 187.86 ±1.29 / 190.07 ms │ no change │
│ QQuery 79 │ 67.10 / 67.67 ±0.78 / 69.22 ms │ 66.88 / 67.43 ±0.34 / 67.85 ms │ no change │
│ QQuery 80 │ 97.99 / 100.54 ±3.61 / 107.69 ms │ 99.60 / 100.77 ±0.69 / 101.72 ms │ no change │
│ QQuery 81 │ 23.62 / 23.84 ±0.24 / 24.28 ms │ 25.18 / 25.82 ±0.53 / 26.50 ms │ 1.08x slower │
│ QQuery 82 │ 16.02 / 16.51 ±0.54 / 17.52 ms │ 16.15 / 17.37 ±1.61 / 20.53 ms │ 1.05x slower │
│ QQuery 83 │ 35.67 / 35.95 ±0.18 / 36.20 ms │ 39.54 / 40.15 ±0.53 / 41.14 ms │ 1.12x slower │
│ QQuery 84 │ 42.63 / 43.96 ±1.47 / 45.85 ms │ 34.88 / 35.05 ±0.11 / 35.21 ms │ +1.25x faster │
│ QQuery 85 │ 134.67 / 135.55 ±0.52 / 136.15 ms │ 107.49 / 110.01 ±3.10 / 115.77 ms │ +1.23x faster │
│ QQuery 86 │ 24.46 / 24.81 ±0.28 / 25.28 ms │ 24.73 / 25.26 ±0.49 / 26.15 ms │ no change │
│ QQuery 87 │ 68.65 / 70.23 ±1.66 / 73.40 ms │ 63.36 / 64.31 ±0.69 / 65.35 ms │ +1.09x faster │
│ QQuery 88 │ 61.62 / 62.05 ±0.37 / 62.57 ms │ 62.13 / 62.54 ±0.36 / 62.99 ms │ no change │
│ QQuery 89 │ 35.62 / 35.75 ±0.10 / 35.87 ms │ 35.49 / 36.38 ±1.13 / 38.56 ms │ no change │
│ QQuery 90 │ 16.72 / 16.96 ±0.29 / 17.54 ms │ 16.64 / 17.17 ±0.73 / 18.61 ms │ no change │
│ QQuery 91 │ 51.31 / 51.62 ±0.36 / 52.31 ms │ 44.53 / 45.12 ±0.58 / 46.22 ms │ +1.14x faster │
│ QQuery 92 │ 28.68 / 29.08 ±0.38 / 29.61 ms │ 28.83 / 29.08 ±0.24 / 29.49 ms │ no change │
│ QQuery 93 │ 49.88 / 50.59 ±0.46 / 51.13 ms │ 50.08 / 50.65 ±0.44 / 51.25 ms │ no change │
│ QQuery 94 │ 37.05 / 37.34 ±0.34 / 38.00 ms │ 38.12 / 39.02 ±0.99 / 40.86 ms │ no change │
│ QQuery 95 │ 83.37 / 86.12 ±2.36 / 90.34 ms │ 83.77 / 84.45 ±0.59 / 85.37 ms │ no change │
│ QQuery 96 │ 23.83 / 24.17 ±0.42 / 24.97 ms │ 24.02 / 24.16 ±0.21 / 24.56 ms │ no change │
│ QQuery 97 │ 45.70 / 46.12 ±0.22 / 46.30 ms │ 45.67 / 46.03 ±0.19 / 46.18 ms │ no change │
│ QQuery 98 │ 41.68 / 42.79 ±0.59 / 43.37 ms │ 41.87 / 42.78 ±0.90 / 44.31 ms │ no change │
│ QQuery 99 │ 69.39 / 69.93 ±0.34 / 70.34 ms │ 68.56 / 69.74 ±0.99 / 71.43 ms │ no change │
└───────────┴────────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 10426.49ms │
│ Total Time (lower-repartition-file-min-size) │ 10063.12ms │
│ Average Time (HEAD) │ 105.32ms │
│ Average Time (lower-repartition-file-min-size) │ 101.65ms │
│ Queries Faster │ 19 │
│ Queries Slower │ 5 │
│ Queries with No Change │ 75 │
│ Queries with Failure │ 0 │
└────────────────────────────────────────────────┴────────────┘
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpcds — base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 55.0s |
| Peak memory | 7.0 GiB |
| Avg memory | 6.3 GiB |
| CPU user | 234.1s |
| CPU sys | 5.5s |
| Peak spill | 0 B |
**tpcds — branch**
| Metric | Value |
|--------|-------|
| Wall time | 55.0s |
| Peak memory | 6.9 GiB |
| Avg memory | 6.1 GiB |
| CPU user | 233.1s |
| CPU sys | 5.5s |
| Peak spill | 0 B |
</details>
---
[File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4511995613
🤖 Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22439#issuecomment-4511840612)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB)
<details><summary>CPU Details (lscpu)</summary>
```
Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected
```
</details>
<details><summary>Details</summary>
<p>
```
Comparing HEAD and lower-repartition-file-min-size
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ lower-repartition-file-min-size ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1 │ 38.42 / 40.10 ±1.42 / 42.17 ms │ 38.50 / 39.01 ±0.96 / 40.94 ms │ no change │
│ QQuery 2 │ 19.74 / 20.59 ±1.16 / 22.86 ms │ 18.70 / 19.30 ±0.67 / 20.60 ms │ +1.07x faster │
│ QQuery 3 │ 32.53 / 34.11 ±1.56 / 36.92 ms │ 31.24 / 32.68 ±1.56 / 34.63 ms │ no change │
│ QQuery 4 │ 17.56 / 17.75 ±0.12 / 17.92 ms │ 17.57 / 17.92 ±0.34 / 18.56 ms │ no change │
│ QQuery 5 │ 42.37 / 43.42 ±1.56 / 46.52 ms │ 37.74 / 39.75 ±1.26 / 41.31 ms │ +1.09x faster │
│ QQuery 6 │ 16.43 / 16.54 ±0.09 / 16.67 ms │ 16.38 / 16.53 ±0.11 / 16.68 ms │ no change │
│ QQuery 7 │ 46.05 / 48.40 ±1.66 / 50.28 ms │ 43.98 / 46.77 ±1.68 / 49.22 ms │ no change │
│ QQuery 8 │ 44.92 / 45.18 ±0.34 / 45.83 ms │ 43.15 / 43.96 ±1.12 / 46.19 ms │ no change │
│ QQuery 9 │ 49.80 / 50.99 ±0.76 / 51.82 ms │ 49.29 / 50.19 ±0.70 / 50.97 ms │ no change │
│ QQuery 10 │ 64.06 / 64.77 ±0.56 / 65.41 ms │ 42.78 / 43.18 ±0.55 / 44.24 ms │ +1.50x faster │
│ QQuery 11 │ 13.53 / 13.80 ±0.15 / 13.96 ms │ 13.46 / 13.65 ±0.13 / 13.79 ms │ no change │
│ QQuery 12 │ 24.26 / 25.29 ±0.96 / 26.63 ms │ 24.10 / 24.33 ±0.18 / 24.53 ms │ no change │
│ QQuery 13 │ 33.51 / 34.79 ±0.97 / 35.96 ms │ 33.66 / 34.31 ±0.57 / 35.07 ms │ no change │
│ QQuery 14 │ 25.53 / 25.84 ±0.18 / 26.04 ms │ 23.95 / 25.60 ±1.83 / 28.17 ms │ no change │
│ QQuery 15 │ 31.73 / 31.94 ±0.17 / 32.16 ms │ 31.68 / 32.34 ±0.81 / 33.94 ms │ no change │
│ QQuery 16 │ 14.89 / 15.00 ±0.11 / 15.14 ms │ 14.23 / 14.45 ±0.19 / 14.74 ms │ no change │
│ QQuery 17 │ 75.49 / 76.47 ±1.26 / 78.91 ms │ 74.70 / 75.89 ±1.18 / 77.60 ms │ no change │
│ QQuery 18 │ 62.44 / 63.45 ±1.33 / 66.03 ms │ 59.48 / 60.42 ±0.85 / 61.93 ms │ no change │
│ QQuery 19 │ 33.93 / 34.78 ±1.12 / 36.95 ms │ 33.17 / 33.67 ±0.33 / 33.96 ms │ no change │
│ QQuery 20 │ 38.08 / 38.23 ±0.14 / 38.46 ms │ 32.59 / 32.83 ±0.21 / 33.17 ms │ +1.16x faster │
│ QQuery 21 │ 57.62 / 58.90 ±1.71 / 62.29 ms │ 55.93 / 57.16 ±1.08 / 58.85 ms │ no change │
│ QQuery 22 │ 23.57 / 24.21 ±0.77 / 25.58 ms │ 14.17 / 14.42 ±0.18 / 14.72 ms │ +1.68x faster │
└───────────┴────────────────────────────────┴─────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD) │ 824.54ms │
│ Total Time (lower-repartition-file-min-size) │ 768.37ms │
│ Average Time (HEAD) │ 37.48ms │
│ Average Time (lower-repartition-file-min-size) │ 34.93ms │
│ Queries Faster │ 5 │
│ Queries Slower │ 0 │
│ Queries with No Change │ 17 │
│ Queries with Failure │ 0 │
└────────────────────────────────────────────────┴──────────┘
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpch — base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 5.0s |
| Peak memory | 5.6 GiB |
| Avg memory | 5.1 GiB |
| CPU user | 29.9s |
| CPU sys | 2.3s |
| Peak spill | 0 B |
**tpch — branch**
| Metric | Value |
|--------|-------|
| Wall time | 5.0s |
| Peak memory | 5.7 GiB |
| Avg memory | 4.9 GiB |
| CPU user | 30.0s |
| CPU sys | 2.4s |
| Peak spill | 0 B |
</details>
---
[File an issue](https://github.com/adriangb/datafusion-benchmarking/issues) against this benchmark runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: lower repartition_file_min_size default from 10 MiB to 1 MiB [datafusion]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb commented on PR #22439:
URL: https://github.com/apache/datafusion/pull/22439#issuecomment-4547508699
@Dandandan I know you evaluated this in https://github.com/apache/datafusion/pull/19690 and then closed the issue, presumably because the benchmarks did not reproduce an improvement you ere seeing locally. Maybe it's just the new benchmark runner, but this looks like there are consistent improvements for TPCH.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org