You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "waynexia (via GitHub)" <gi...@apache.org> on 2026/05/13 09:37:51 UTC
[PR] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
waynexia opened a new pull request, #22145:
URL: https://github.com/apache/datafusion/pull/22145
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
- Closes #.
## Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
-->
This is an (agent made) PoC branch for clickbench q4 `SELECT COUNT(DISTINCT UserID) FROM hits;`. In my test environment it's about 25% faster than `main`.
The basic idea is to skip maintaining group-id when there is no requirement of it, e.g., for group-by cases without any accumulator.
Other findings on this experiment:
- this approach seems only work for primitive types, for other cases, like q5 `SELECT COUNT(DISTINCT SearchPhrase) FROM hits;` where the key is string type, the group calculation is not the critical path
- I tried another approach for the same q4 query, where I change the group id type from 64 bits to 32 bits, it also improves a lot. My suspection is the changed memory layout of HashTable is the source of improvement. We can consider this further for other cases where group-id is required.
## What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
## Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are they covered by existing tests)?
-->
## Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
-->
<!--
If there are any breaking changes to public APIs, please add the `api change` label.
-->
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461223378
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4461199760)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4461199760-148-2qfd9 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461222334
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4461199921)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4461199921-147-v2d9c 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4450179783
> ```
> โ QQuery 40 โ 13.86 / 14.30 ยฑ0.31 / 14.65 ms โ 13.83 / 17.12 ยฑ3.59 / 22.13 ms โ 1.20x slower โ
> โ QQuery 42 โ 13.11 / 13.56 ยฑ0.63 / 14.79 ms โ 12.73 / 16.60 ยฑ7.03 / 30.64 ms โ 1.22x slower โ
> ```
>
> Some queries regress a lot...
This is likely noise, best is to look at the lowest reported times in those cases.
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4458365838
> rerun another round and those regression disappear ๐ผ
Some of the clickbench queries can be noisy as sometimes they can prune out more IO / decoding due to tightening filters of topk / min / max being non-deterministic (sometimes it gets a better bound earlier by chance).
I think they probably are not related at all to this code.
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4457859499
rerun another round and those regression disappear ๐ผ
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448700954
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4448673356-79-zc7ct 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449086767
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4449065667)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4449065667-81-7rccx 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448524428
Benchmark for [this request](https://github.com/apache/datafusion/pull/22145#issuecomment-4448508770) failed.
Last 20 lines of output:
<details><summary>Click to expand</summary>
```
* [new tag] 48.0.0-rc1 -> 48.0.0-rc1
* [new tag] 48.0.0-rc2 -> 48.0.0-rc2
* [new tag] 5.0.0 -> 5.0.0
* [new tag] 5.0.0-rc1 -> 5.0.0-rc1
* [new tag] 5.0.0-rc3 -> 5.0.0-rc3
* [new tag] 6.0.0 -> 6.0.0
* [new tag] 6.0.0-rc0 -> 6.0.0-rc0
* [new tag] 7.0.0 -> 7.0.0
* [new tag] 7.0.0-rc2 -> 7.0.0-rc2
* [new tag] 8.0.0 -> 8.0.0
* [new tag] 8.0.0-rc1 -> 8.0.0-rc1
* [new tag] 8.0.0-rc2 -> 8.0.0-rc2
* [new tag] 9.0.0 -> 9.0.0
* [new tag] 9.0.0-rc1 -> 9.0.0-rc1
* [new tag] ballista-0.5.0 -> ballista-0.5.0
* [new tag] ballista-0.6.0 -> ballista-0.6.0
* [new tag] ballista-0.7.0 -> ballista-0.7.0
* [new tag] python-0.3.0 -> python-0.3.0
* [new tag] python-0.4.0 -> python-0.4.0
Switched to branch 'q4'
```
</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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449218936
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4449065667)
**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 q4
--------------------
Benchmark clickbench_extended.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 0 โ 786.54 / 797.82 ยฑ6.35 / 805.19 ms โ 791.70 / 806.98 ยฑ12.49 / 821.08 ms โ no change โ
โ QQuery 1 โ 186.79 / 187.19 ยฑ0.32 / 187.60 ms โ 187.98 / 190.58 ยฑ3.69 / 197.92 ms โ no change โ
โ QQuery 2 โ 457.81 / 460.77 ยฑ2.79 / 465.83 ms โ 459.38 / 461.64 ยฑ1.38 / 463.23 ms โ no change โ
โ QQuery 3 โ 307.80 / 308.83 ยฑ0.94 / 310.50 ms โ 312.75 / 314.05 ยฑ1.03 / 314.98 ms โ no change โ
โ QQuery 4 โ 666.17 / 673.37 ยฑ8.65 / 689.97 ms โ 665.11 / 675.12 ยฑ11.04 / 695.84 ms โ no change โ
โ QQuery 5 โ 10396.71 / 10611.51 ยฑ134.12 / 10771.39 ms โ 10253.91 / 10471.92 ยฑ174.88 / 10786.78 ms โ no change โ
โ QQuery 6 โ 998.24 / 1010.63 ยฑ13.40 / 1034.10 ms โ 998.57 / 1051.73 ยฑ73.05 / 1194.45 ms โ no change โ
โ QQuery 7 โ 729.77 / 749.10 ยฑ12.93 / 770.08 ms โ 737.80 / 763.53 ยฑ16.01 / 787.16 ms โ no change โ
โ QQuery 8 โ 377.14 / 401.10 ยฑ40.03 / 481.02 ms โ 374.73 / 386.89 ยฑ13.48 / 405.40 ms โ no change โ
โ QQuery 9 โ 2841.62 / 2925.22 ยฑ61.99 / 3031.73 ms โ 2805.11 / 2947.75 ยฑ76.26 / 3022.04 ms โ no change โ
โ QQuery 10 โ 632.15 / 652.21 ยฑ19.94 / 687.11 ms โ 633.44 / 666.65 ยฑ27.04 / 708.89 ms โ no change โ
โ QQuery 11 โ 2374.98 / 2393.55 ยฑ18.22 / 2419.52 ms โ 2214.87 / 2241.66 ยฑ17.95 / 2267.82 ms โ +1.07x faster โ
โ QQuery 12 โ 186.51 / 207.94 ยฑ30.34 / 268.00 ms โ 189.17 / 207.94 ยฑ32.86 / 273.60 ms โ no change โ
โ QQuery 13 โ 550.13 / 571.96 ยฑ29.38 / 629.55 ms โ 554.15 / 560.94 ยฑ6.80 / 571.02 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 21951.20ms โ
โ Total Time (q4) โ 21747.36ms โ
โ Average Time (HEAD) โ 1567.94ms โ
โ Average Time (q4) โ 1553.38ms โ
โ Queries Faster โ 1 โ
โ Queries Slower โ 0 โ
โ Queries with No Change โ 13 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_extended โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 115.0s |
| Peak memory | 31.9 GiB |
| Avg memory | 23.9 GiB |
| CPU user | 1048.2s |
| CPU sys | 43.6s |
| Peak spill | 0 B |
**clickbench_extended โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 110.0s |
| Peak memory | 31.5 GiB |
| Avg memory | 24.2 GiB |
| CPU user | 1033.4s |
| CPU sys | 43.9s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-5018182485
Thank you for your contribution. Unfortunately, this pull request is stale because it has been open 60 days with no activity. Please remove the stale label or comment or this will be closed in 7 days.
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448508770
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448556915
Hi @waynexia, thanks for the request (https://github.com/apache/datafusion/pull/22145#issuecomment-4448556762). Only whitelisted users can trigger benchmarks. Allowed users: [Dandandan](https://github.com/Dandandan), [Fokko](https://github.com/Fokko), [Jefffrey](https://github.com/Jefffrey), [Omega359](https://github.com/Omega359), [adriangb](https://github.com/adriangb), [alamb](https://github.com/alamb), [asubiotto](https://github.com/asubiotto), [brunal](https://github.com/brunal), [buraksenn](https://github.com/buraksenn), [cetra3](https://github.com/cetra3), [codephage2020](https://github.com/codephage2020), [coderfender](https://github.com/coderfender), [comphead](https://github.com/comphead), [erenavsarogullari](https://github.com/erenavsarogullari), [etseidl](https://github.com/etseidl), [friendlymatthew](https://github.com/friendlymatthew), [gabotechs](https://github.com/gabotechs), [geoffreyclaude](https://github.com/geoffreyclaude), [grtlr](https://github.com/grtlr), [h
aohuaijin](https://github.com/haohuaijin), [jonathanc-n](https://github.com/jonathanc-n), [kevinjqliu](https://github.com/kevinjqliu), [klion26](https://github.com/klion26), [kosiew](https://github.com/kosiew), [kumarUjjawal](https://github.com/kumarUjjawal), [kunalsinghdadhwal](https://github.com/kunalsinghdadhwal), [liamzwbao](https://github.com/liamzwbao), [mbutrovich](https://github.com/mbutrovich), [mkleen](https://github.com/mkleen), [mzabaluev](https://github.com/mzabaluev), [neilconway](https://github.com/neilconway), [rluvaton](https://github.com/rluvaton), [sdf-jkl](https://github.com/sdf-jkl), [timsaucer](https://github.com/timsaucer), [xudong963](https://github.com/xudong963), [zhuqi-lucas](https://github.com/zhuqi-lucas).
---
[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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448786692
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356)
**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 q4
--------------------
Benchmark tpch_sf1.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 1 โ 39.13 / 40.83 ยฑ1.23 / 42.17 ms โ 39.11 / 39.75 ยฑ0.93 / 41.59 ms โ no change โ
โ QQuery 2 โ 20.69 / 21.03 ยฑ0.32 / 21.42 ms โ 20.40 / 20.71 ยฑ0.28 / 21.12 ms โ no change โ
โ QQuery 3 โ 35.67 / 37.57 ยฑ1.02 / 38.63 ms โ 34.95 / 35.68 ยฑ0.64 / 36.86 ms โ +1.05x faster โ
โ QQuery 4 โ 17.39 / 17.50 ยฑ0.12 / 17.73 ms โ 17.21 / 17.65 ยฑ0.48 / 18.58 ms โ no change โ
โ QQuery 5 โ 42.64 / 45.30 ยฑ1.37 / 46.57 ms โ 43.14 / 44.66 ยฑ1.26 / 46.38 ms โ no change โ
โ QQuery 6 โ 16.56 / 17.52 ยฑ0.89 / 19.12 ms โ 16.57 / 16.63 ยฑ0.07 / 16.75 ms โ +1.05x faster โ
โ QQuery 7 โ 49.69 / 50.56 ยฑ0.54 / 51.37 ms โ 50.47 / 52.59 ยฑ1.32 / 53.75 ms โ no change โ
โ QQuery 8 โ 46.10 / 46.32 ยฑ0.14 / 46.52 ms โ 46.08 / 46.45 ยฑ0.33 / 47.03 ms โ no change โ
โ QQuery 9 โ 51.16 / 52.22 ยฑ0.68 / 53.21 ms โ 51.90 / 52.10 ยฑ0.18 / 52.40 ms โ no change โ
โ QQuery 10 โ 64.31 / 64.53 ยฑ0.12 / 64.65 ms โ 64.06 / 64.29 ยฑ0.16 / 64.55 ms โ no change โ
โ QQuery 11 โ 13.47 / 14.11 ยฑ0.83 / 15.72 ms โ 13.51 / 13.97 ยฑ0.61 / 15.17 ms โ no change โ
โ QQuery 12 โ 24.86 / 25.15 ยฑ0.25 / 25.61 ms โ 25.30 / 25.62 ยฑ0.19 / 25.84 ms โ no change โ
โ QQuery 13 โ 35.89 / 36.81 ยฑ0.88 / 38.05 ms โ 35.09 / 35.82 ยฑ0.54 / 36.54 ms โ no change โ
โ QQuery 14 โ 25.87 / 26.75 ยฑ1.24 / 29.21 ms โ 25.78 / 26.05 ยฑ0.18 / 26.34 ms โ no change โ
โ QQuery 15 โ 31.76 / 32.39 ยฑ0.63 / 33.23 ms โ 31.47 / 31.93 ยฑ0.48 / 32.73 ms โ no change โ
โ QQuery 16 โ 15.01 / 15.16 ยฑ0.14 / 15.40 ms โ 15.18 / 15.31 ยฑ0.11 / 15.51 ms โ no change โ
โ QQuery 17 โ 74.55 / 76.87 ยฑ1.59 / 78.89 ms โ 75.82 / 77.14 ยฑ1.57 / 79.58 ms โ no change โ
โ QQuery 18 โ 69.33 / 71.14 ยฑ2.04 / 74.95 ms โ 69.82 / 70.70 ยฑ1.15 / 72.94 ms โ no change โ
โ QQuery 19 โ 35.88 / 36.66 ยฑ0.94 / 38.48 ms โ 35.58 / 35.92 ยฑ0.27 / 36.27 ms โ no change โ
โ QQuery 20 โ 38.25 / 38.52 ยฑ0.17 / 38.73 ms โ 38.33 / 39.23 ยฑ1.49 / 42.21 ms โ no change โ
โ QQuery 21 โ 56.75 / 59.27 ยฑ2.18 / 63.07 ms โ 57.63 / 62.20 ยฑ3.04 / 67.19 ms โ no change โ
โ QQuery 22 โ 23.26 / 23.66 ยฑ0.28 / 23.93 ms โ 23.75 / 23.92 ยฑ0.23 / 24.35 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 849.89ms โ
โ Total Time (q4) โ 848.33ms โ
โ Average Time (HEAD) โ 38.63ms โ
โ Average Time (q4) โ 38.56ms โ
โ Queries Faster โ 2 โ
โ Queries Slower โ 0 โ
โ Queries with No Change โ 20 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpch โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 5.0s |
| Peak memory | 5.9 GiB |
| Avg memory | 5.3 GiB |
| CPU user | 31.2s |
| CPU sys | 2.4s |
| Peak spill | 0 B |
**tpch โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 5.0s |
| Peak memory | 5.7 GiB |
| Avg memory | 5.2 GiB |
| CPU user | 31.2s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461222138
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4461199476)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4461199476-146-xrlgw 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449088660
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4449065667)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4449065667-82-5w7wp 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) using: clickbench_extended
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449065667
run benchmark clickbench_partitioned clickbench_extended
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461199760
run benchmark clickbench_partitioned
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461331710
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4461199760)
**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 q4
--------------------
Benchmark clickbench_partitioned.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 0 โ 1.22 / 4.78 ยฑ6.97 / 18.72 ms โ 1.26 / 4.86 ยฑ6.98 / 18.83 ms โ no change โ
โ QQuery 1 โ 12.62 / 13.14 ยฑ0.26 / 13.30 ms โ 12.62 / 12.73 ยฑ0.07 / 12.84 ms โ no change โ
โ QQuery 2 โ 35.57 / 36.19 ยฑ0.63 / 37.35 ms โ 35.36 / 35.68 ยฑ0.19 / 35.90 ms โ no change โ
โ QQuery 3 โ 31.76 / 32.70 ยฑ0.88 / 34.11 ms โ 30.57 / 30.94 ยฑ0.36 / 31.62 ms โ +1.06x faster โ
โ QQuery 4 โ 243.39 / 248.77 ยฑ3.63 / 253.57 ms โ 215.12 / 224.49 ยฑ13.17 / 250.11 ms โ +1.11x faster โ
โ QQuery 5 โ 290.34 / 292.86 ยฑ1.75 / 295.41 ms โ 276.76 / 284.27 ยฑ8.49 / 296.02 ms โ no change โ
โ QQuery 6 โ 6.80 / 7.26 ยฑ0.44 / 8.06 ms โ 7.46 / 7.80 ยฑ0.26 / 8.15 ms โ 1.07x slower โ
โ QQuery 7 โ 13.80 / 13.92 ยฑ0.08 / 14.02 ms โ 14.33 / 14.58 ยฑ0.31 / 15.08 ms โ no change โ
โ QQuery 8 โ 332.63 / 335.70 ยฑ2.83 / 340.65 ms โ 311.18 / 316.39 ยฑ8.45 / 333.16 ms โ +1.06x faster โ
โ QQuery 9 โ 464.09 / 475.89 ยฑ10.56 / 489.86 ms โ 449.41 / 467.17 ยฑ16.94 / 494.58 ms โ no change โ
โ QQuery 10 โ 71.55 / 72.44 ยฑ0.73 / 73.38 ms โ 70.73 / 71.41 ยฑ0.77 / 72.87 ms โ no change โ
โ QQuery 11 โ 80.97 / 82.35 ยฑ0.93 / 83.74 ms โ 80.91 / 84.27 ยฑ5.65 / 95.54 ms โ no change โ
โ QQuery 12 โ 279.52 / 287.73 ยฑ5.64 / 296.66 ms โ 270.69 / 283.13 ยฑ13.30 / 300.92 ms โ no change โ
โ QQuery 13 โ 397.90 / 410.36 ยฑ10.45 / 425.89 ms โ 382.53 / 405.13 ยฑ18.04 / 427.38 ms โ no change โ
โ QQuery 14 โ 281.95 / 288.38 ยฑ5.44 / 296.12 ms โ 292.10 / 297.68 ยฑ4.96 / 307.00 ms โ no change โ
โ QQuery 15 โ 281.93 / 295.65 ยฑ8.53 / 308.14 ms โ 289.62 / 299.83 ยฑ8.13 / 314.08 ms โ no change โ
โ QQuery 16 โ 601.45 / 621.16 ยฑ18.50 / 644.87 ms โ 622.74 / 632.54 ยฑ9.26 / 643.74 ms โ no change โ
โ QQuery 17 โ 639.56 / 644.88 ยฑ6.32 / 653.92 ms โ 625.00 / 637.03 ยฑ10.88 / 654.30 ms โ no change โ
โ QQuery 18 โ 1253.11 / 1280.05 ยฑ22.57 / 1313.59 ms โ 1229.67 / 1266.35 ยฑ23.39 / 1301.70 ms โ no change โ
โ QQuery 19 โ 28.89 / 29.27 ยฑ0.24 / 29.66 ms โ 28.68 / 29.53 ยฑ0.46 / 29.97 ms โ no change โ
โ QQuery 20 โ 521.63 / 532.76 ยฑ9.82 / 547.16 ms โ 523.74 / 527.83 ยฑ2.64 / 531.49 ms โ no change โ
โ QQuery 21 โ 592.41 / 611.30 ยฑ13.29 / 631.12 ms โ 598.31 / 606.97 ยฑ8.82 / 618.79 ms โ no change โ
โ QQuery 22 โ 1069.42 / 1092.39 ยฑ15.31 / 1111.34 ms โ 1082.12 / 1099.94 ยฑ16.67 / 1130.64 ms โ no change โ
โ QQuery 23 โ 3205.93 / 3262.22 ยฑ41.41 / 3325.07 ms โ 3285.10 / 3328.94 ยฑ27.47 / 3352.94 ms โ no change โ
โ QQuery 24 โ 43.18 / 44.75 ยฑ1.60 / 47.84 ms โ 41.81 / 47.88 ยฑ7.37 / 60.79 ms โ 1.07x slower โ
โ QQuery 25 โ 112.54 / 116.18 ยฑ3.26 / 121.97 ms โ 113.32 / 116.08 ยฑ2.85 / 120.15 ms โ no change โ
โ QQuery 26 โ 43.93 / 45.50 ยฑ1.80 / 48.83 ms โ 42.29 / 42.56 ยฑ0.32 / 43.20 ms โ +1.07x faster โ
โ QQuery 27 โ 677.22 / 684.38 ยฑ6.01 / 691.71 ms โ 676.29 / 686.88 ยฑ9.96 / 701.94 ms โ no change โ
โ QQuery 28 โ 3056.96 / 3094.00 ยฑ22.81 / 3116.95 ms โ 3028.88 / 3058.19 ยฑ26.83 / 3098.35 ms โ no change โ
โ QQuery 29 โ 41.84 / 51.97 ยฑ12.36 / 73.50 ms โ 41.57 / 44.75 ยฑ4.92 / 54.51 ms โ +1.16x faster โ
โ QQuery 30 โ 297.85 / 310.57 ยฑ9.82 / 324.03 ms โ 307.38 / 319.29 ยฑ7.07 / 328.45 ms โ no change โ
โ QQuery 31 โ 294.98 / 307.09 ยฑ9.91 / 320.33 ms โ 304.04 / 311.27 ยฑ7.13 / 324.66 ms โ no change โ
โ QQuery 32 โ 960.59 / 987.02 ยฑ21.64 / 1020.08 ms โ 970.69 / 990.40 ยฑ17.27 / 1021.21 ms โ no change โ
โ QQuery 33 โ 1418.85 / 1534.06 ยฑ58.96 / 1583.36 ms โ 1501.16 / 1557.77 ยฑ48.69 / 1645.68 ms โ no change โ
โ QQuery 34 โ 1478.82 / 1541.38 ยฑ48.45 / 1616.41 ms โ 1508.31 / 1553.75 ยฑ33.53 / 1605.22 ms โ no change โ
โ QQuery 35 โ 285.62 / 318.85 ยฑ59.60 / 437.99 ms โ 325.72 / 336.24 ยฑ9.68 / 354.07 ms โ 1.05x slower โ
โ QQuery 36 โ 61.13 / 69.21 ยฑ5.76 / 75.27 ms โ 68.96 / 75.32 ยฑ5.87 / 85.01 ms โ 1.09x slower โ
โ QQuery 37 โ 37.09 / 37.81 ยฑ0.52 / 38.53 ms โ 36.92 / 39.01 ยฑ2.02 / 42.71 ms โ no change โ
โ QQuery 38 โ 42.37 / 43.80 ยฑ0.94 / 45.31 ms โ 41.41 / 44.66 ยฑ2.05 / 47.27 ms โ no change โ
โ QQuery 39 โ 126.62 / 141.35 ยฑ9.43 / 155.74 ms โ 140.86 / 149.01 ยฑ8.34 / 161.59 ms โ 1.05x slower โ
โ QQuery 40 โ 13.78 / 14.13 ยฑ0.24 / 14.51 ms โ 14.81 / 15.28 ยฑ0.30 / 15.64 ms โ 1.08x slower โ
โ QQuery 41 โ 13.46 / 13.74 ยฑ0.16 / 13.90 ms โ 14.34 / 16.09 ยฑ2.90 / 21.84 ms โ 1.17x slower โ
โ QQuery 42 โ 12.77 / 13.15 ยฑ0.21 / 13.36 ms โ 13.70 / 13.97 ยฑ0.27 / 14.47 ms โ 1.06x slower โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 20341.10ms โ
โ Total Time (q4) โ 20387.89ms โ
โ Average Time (HEAD) โ 473.05ms โ
โ Average Time (q4) โ 474.14ms โ
โ Queries Faster โ 5 โ
โ Queries Slower โ 8 โ
โ Queries with No Change โ 30 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_partitioned โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 105.0s |
| Peak memory | 31.0 GiB |
| Avg memory | 23.5 GiB |
| CPU user | 1069.0s |
| CPU sys | 69.0s |
| Peak spill | 0 B |
**clickbench_partitioned โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 105.0s |
| Peak memory | 31.1 GiB |
| Avg memory | 23.3 GiB |
| CPU user | 1064.6s |
| CPU sys | 71.0s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4440470558
Benchmark for [this request](https://github.com/apache/datafusion/pull/22145#issuecomment-4440449362) failed.
Last 20 lines of output:
<details><summary>Click to expand</summary>
```
* [new tag] 48.0.0-rc1 -> 48.0.0-rc1
* [new tag] 48.0.0-rc2 -> 48.0.0-rc2
* [new tag] 5.0.0 -> 5.0.0
* [new tag] 5.0.0-rc1 -> 5.0.0-rc1
* [new tag] 5.0.0-rc3 -> 5.0.0-rc3
* [new tag] 6.0.0 -> 6.0.0
* [new tag] 6.0.0-rc0 -> 6.0.0-rc0
* [new tag] 7.0.0 -> 7.0.0
* [new tag] 7.0.0-rc2 -> 7.0.0-rc2
* [new tag] 8.0.0 -> 8.0.0
* [new tag] 8.0.0-rc1 -> 8.0.0-rc1
* [new tag] 8.0.0-rc2 -> 8.0.0-rc2
* [new tag] 9.0.0 -> 9.0.0
* [new tag] 9.0.0-rc1 -> 9.0.0-rc1
* [new tag] ballista-0.5.0 -> ballista-0.5.0
* [new tag] ballista-0.6.0 -> ballista-0.6.0
* [new tag] ballista-0.7.0 -> ballista-0.7.0
* [new tag] python-0.3.0 -> python-0.3.0
* [new tag] python-0.4.0 -> python-0.4.0
Switched to branch 'q4'
```
</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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4440476186
Benchmark for [this request](https://github.com/apache/datafusion/pull/22145#issuecomment-4440449362) failed.
Last 20 lines of output:
<details><summary>Click to expand</summary>
```
* [new tag] 48.0.0-rc1 -> 48.0.0-rc1
* [new tag] 48.0.0-rc2 -> 48.0.0-rc2
* [new tag] 5.0.0 -> 5.0.0
* [new tag] 5.0.0-rc1 -> 5.0.0-rc1
* [new tag] 5.0.0-rc3 -> 5.0.0-rc3
* [new tag] 6.0.0 -> 6.0.0
* [new tag] 6.0.0-rc0 -> 6.0.0-rc0
* [new tag] 7.0.0 -> 7.0.0
* [new tag] 7.0.0-rc2 -> 7.0.0-rc2
* [new tag] 8.0.0 -> 8.0.0
* [new tag] 8.0.0-rc1 -> 8.0.0-rc1
* [new tag] 8.0.0-rc2 -> 8.0.0-rc2
* [new tag] 9.0.0 -> 9.0.0
* [new tag] 9.0.0-rc1 -> 9.0.0-rc1
* [new tag] ballista-0.5.0 -> ballista-0.5.0
* [new tag] ballista-0.6.0 -> ballista-0.6.0
* [new tag] ballista-0.7.0 -> ballista-0.7.0
* [new tag] python-0.3.0 -> python-0.3.0
* [new tag] python-0.4.0 -> python-0.4.0
Switched to branch 'q4'
```
</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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448694970
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4448673356-80-zdxf9 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448846680
Hmmm, the result does not seem good ๐ค Let me dig deeper
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448699138
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4448673356-78-lmv8g 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 q4 (65980f0637f7d62b28d8bed9770b3ce92051e666) to 7f2f78d (merge-base) [diff](https://github.com/apache/datafusion/compare/7f2f78d48b6d3d6aee2ce2fd29910bb4c11b1012..65980f0637f7d62b28d8bed9770b3ce92051e666) 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461199476
run benchmark clickbench_partitioned
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449064007
```
โ QQuery 4 โ 232.65 / 234.21 ยฑ1.59 / 236.52 ms โ 213.86 / 215.85 ยฑ1.58 / 218.49 ms โ +1.09x faster โ
```
Seems like the query is faster though (and memory is ~halved) so seems like a great optimization - I expect it for higher cardinality
Benchmark runner might have a bit more cache and thus might not hit RAM, etc.
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4496880730
@codex review
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461330085
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4461199476)
**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 q4
--------------------
Benchmark clickbench_partitioned.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 0 โ 1.20 / 4.71 ยฑ6.88 / 18.47 ms โ 1.19 / 4.72 ยฑ6.90 / 18.53 ms โ no change โ
โ QQuery 1 โ 12.38 / 12.87 ยฑ0.28 / 13.21 ms โ 12.44 / 12.86 ยฑ0.27 / 13.21 ms โ no change โ
โ QQuery 2 โ 35.67 / 35.96 ยฑ0.42 / 36.79 ms โ 35.58 / 35.79 ยฑ0.29 / 36.36 ms โ no change โ
โ QQuery 3 โ 31.27 / 31.70 ยฑ0.56 / 32.67 ms โ 30.66 / 31.03 ยฑ0.26 / 31.31 ms โ no change โ
โ QQuery 4 โ 234.24 / 239.69 ยฑ3.43 / 244.76 ms โ 216.31 / 226.03 ยฑ8.31 / 238.97 ms โ +1.06x faster โ
โ QQuery 5 โ 280.91 / 283.29 ยฑ1.59 / 285.20 ms โ 281.10 / 282.15 ยฑ1.45 / 284.99 ms โ no change โ
โ QQuery 6 โ 6.50 / 7.11 ยฑ0.50 / 7.61 ms โ 6.02 / 6.81 ยฑ0.68 / 7.82 ms โ no change โ
โ QQuery 7 โ 13.82 / 14.75 ยฑ1.61 / 17.96 ms โ 13.75 / 13.90 ยฑ0.10 / 14.04 ms โ +1.06x faster โ
โ QQuery 8 โ 317.14 / 321.46 ยฑ2.44 / 324.73 ms โ 316.39 / 319.13 ยฑ1.65 / 321.36 ms โ no change โ
โ QQuery 9 โ 451.72 / 458.89 ยฑ5.69 / 466.39 ms โ 457.81 / 463.01 ยฑ3.20 / 466.31 ms โ no change โ
โ QQuery 10 โ 70.90 / 71.34 ยฑ0.40 / 71.86 ms โ 69.99 / 71.28 ยฑ0.74 / 72.28 ms โ no change โ
โ QQuery 11 โ 82.12 / 82.83 ยฑ0.42 / 83.39 ms โ 80.82 / 82.81 ยฑ2.81 / 88.41 ms โ no change โ
โ QQuery 12 โ 276.10 / 280.50 ยฑ3.10 / 284.56 ms โ 275.38 / 283.46 ยฑ6.75 / 292.93 ms โ no change โ
โ QQuery 13 โ 386.72 / 394.42 ยฑ8.23 / 409.41 ms โ 391.69 / 396.27 ยฑ4.01 / 403.78 ms โ no change โ
โ QQuery 14 โ 281.94 / 286.41 ยฑ4.90 / 295.37 ms โ 284.65 / 292.80 ยฑ11.96 / 316.44 ms โ no change โ
โ QQuery 15 โ 278.97 / 284.57 ยฑ6.43 / 296.55 ms โ 283.48 / 292.87 ยฑ11.96 / 316.43 ms โ no change โ
โ QQuery 16 โ 602.16 / 610.19 ยฑ6.54 / 620.82 ms โ 608.83 / 617.75 ยฑ6.41 / 625.34 ms โ no change โ
โ QQuery 17 โ 603.73 / 609.21 ยฑ3.89 / 613.45 ms โ 615.71 / 618.91 ยฑ2.45 / 622.91 ms โ no change โ
โ QQuery 18 โ 1200.24 / 1211.40 ยฑ9.20 / 1223.45 ms โ 1219.37 / 1235.43 ยฑ10.63 / 1252.32 ms โ no change โ
โ QQuery 19 โ 28.25 / 33.07 ยฑ9.00 / 51.05 ms โ 28.17 / 29.61 ยฑ1.63 / 32.73 ms โ +1.12x faster โ
โ QQuery 20 โ 516.65 / 527.68 ยฑ8.72 / 539.95 ms โ 519.77 / 525.06 ยฑ6.45 / 533.75 ms โ no change โ
โ QQuery 21 โ 596.48 / 603.32 ยฑ5.93 / 611.54 ms โ 597.28 / 599.41 ยฑ1.96 / 602.83 ms โ no change โ
โ QQuery 22 โ 1059.05 / 1076.40 ยฑ11.65 / 1095.72 ms โ 1058.91 / 1075.25 ยฑ11.80 / 1091.72 ms โ no change โ
โ QQuery 23 โ 3222.14 / 3242.74 ยฑ20.33 / 3278.36 ms โ 3213.82 / 3237.32 ยฑ16.96 / 3260.07 ms โ no change โ
โ QQuery 24 โ 41.38 / 45.38 ยฑ3.99 / 52.49 ms โ 41.29 / 42.51 ยฑ1.05 / 44.12 ms โ +1.07x faster โ
โ QQuery 25 โ 112.56 / 113.89 ยฑ0.85 / 114.92 ms โ 112.58 / 115.17 ยฑ2.44 / 119.81 ms โ no change โ
โ QQuery 26 โ 42.41 / 45.23 ยฑ3.13 / 51.00 ms โ 42.52 / 45.16 ยฑ2.39 / 48.51 ms โ no change โ
โ QQuery 27 โ 679.30 / 683.22 ยฑ3.39 / 688.14 ms โ 675.39 / 680.23 ยฑ3.58 / 684.74 ms โ no change โ
โ QQuery 28 โ 3017.51 / 3031.76 ยฑ9.23 / 3046.60 ms โ 3004.34 / 3019.91 ยฑ13.49 / 3042.36 ms โ no change โ
โ QQuery 29 โ 41.83 / 50.41 ยฑ7.93 / 62.53 ms โ 41.52 / 47.57 ยฑ11.68 / 70.92 ms โ +1.06x faster โ
โ QQuery 30 โ 301.41 / 307.26 ยฑ4.00 / 312.31 ms โ 301.02 / 305.76 ยฑ4.16 / 313.24 ms โ no change โ
โ QQuery 31 โ 293.84 / 300.22 ยฑ6.62 / 310.75 ms โ 294.03 / 297.16 ยฑ1.98 / 299.90 ms โ no change โ
โ QQuery 32 โ 920.01 / 929.95 ยฑ7.58 / 939.43 ms โ 934.02 / 947.36 ยฑ16.16 / 978.74 ms โ no change โ
โ QQuery 33 โ 1440.94 / 1454.73 ยฑ9.37 / 1467.55 ms โ 1442.61 / 1461.42 ยฑ11.95 / 1477.88 ms โ no change โ
โ QQuery 34 โ 1444.77 / 1491.79 ยฑ41.51 / 1550.07 ms โ 1465.78 / 1480.08 ยฑ9.82 / 1494.47 ms โ no change โ
โ QQuery 35 โ 294.95 / 310.53 ยฑ19.96 / 349.35 ms โ 290.60 / 301.44 ยฑ8.33 / 310.54 ms โ no change โ
โ QQuery 36 โ 64.05 / 72.92 ยฑ9.47 / 89.73 ms โ 62.13 / 72.71 ยฑ6.82 / 83.06 ms โ no change โ
โ QQuery 37 โ 36.78 / 38.32 ยฑ1.49 / 40.82 ms โ 35.47 / 40.75 ยฑ5.03 / 48.57 ms โ 1.06x slower โ
โ QQuery 38 โ 40.87 / 43.72 ยฑ2.03 / 46.49 ms โ 40.48 / 43.39 ยฑ1.75 / 44.93 ms โ no change โ
โ QQuery 39 โ 130.57 / 136.90 ยฑ8.05 / 151.96 ms โ 129.18 / 138.77 ยฑ7.50 / 147.89 ms โ no change โ
โ QQuery 40 โ 14.77 / 15.16 ยฑ0.31 / 15.50 ms โ 14.03 / 14.43 ยฑ0.34 / 15.03 ms โ no change โ
โ QQuery 41 โ 13.97 / 14.93 ยฑ1.40 / 17.67 ms โ 14.05 / 14.19 ยฑ0.16 / 14.45 ms โ no change โ
โ QQuery 42 โ 13.26 / 13.57 ยฑ0.21 / 13.88 ms โ 13.46 / 13.54 ยฑ0.07 / 13.66 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 19824.34ms โ
โ Total Time (q4) โ 19835.22ms โ
โ Average Time (HEAD) โ 461.03ms โ
โ Average Time (q4) โ 461.28ms โ
โ Queries Faster โ 5 โ
โ Queries Slower โ 1 โ
โ Queries with No Change โ 37 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_partitioned โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 100.0s |
| Peak memory | 30.7 GiB |
| Avg memory | 23.5 GiB |
| CPU user | 1042.4s |
| CPU sys | 64.2s |
| Peak spill | 0 B |
**clickbench_partitioned โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 100.0s |
| Peak memory | 30.1 GiB |
| Avg memory | 23.0 GiB |
| CPU user | 1040.8s |
| CPU sys | 68.0s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448525486
Benchmark for [this request](https://github.com/apache/datafusion/pull/22145#issuecomment-4448508770) failed.
Last 20 lines of output:
<details><summary>Click to expand</summary>
```
* [new tag] 48.0.0-rc1 -> 48.0.0-rc1
* [new tag] 48.0.0-rc2 -> 48.0.0-rc2
* [new tag] 5.0.0 -> 5.0.0
* [new tag] 5.0.0-rc1 -> 5.0.0-rc1
* [new tag] 5.0.0-rc3 -> 5.0.0-rc3
* [new tag] 6.0.0 -> 6.0.0
* [new tag] 6.0.0-rc0 -> 6.0.0-rc0
* [new tag] 7.0.0 -> 7.0.0
* [new tag] 7.0.0-rc2 -> 7.0.0-rc2
* [new tag] 8.0.0 -> 8.0.0
* [new tag] 8.0.0-rc1 -> 8.0.0-rc1
* [new tag] 8.0.0-rc2 -> 8.0.0-rc2
* [new tag] 9.0.0 -> 9.0.0
* [new tag] 9.0.0-rc1 -> 9.0.0-rc1
* [new tag] ballista-0.5.0 -> ballista-0.5.0
* [new tag] ballista-0.6.0 -> ballista-0.6.0
* [new tag] ballista-0.7.0 -> ballista-0.7.0
* [new tag] python-0.3.0 -> python-0.3.0
* [new tag] python-0.4.0 -> python-0.4.0
Switched to branch 'q4'
```
</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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449382845
```
โ QQuery 40 โ 13.86 / 14.30 ยฑ0.31 / 14.65 ms โ 13.83 / 17.12 ยฑ3.59 / 22.13 ms โ 1.20x slower โ
โ QQuery 42 โ 13.11 / 13.56 ยฑ0.63 / 14.79 ms โ 12.73 / 16.60 ยฑ7.03 / 30.64 ms โ 1.22x slower โ
```
Some queries regress a lot...
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4457644142
I rerun it use an aws c6a.4xlarge instance. Improvement on q4 is reproduced, but still has lots of regressions and some of them seem to be real. I'll verify them on my local environment next
<details>
<summary>Details</summary>
```
Comparing HEAD and q4
--------------------
Benchmark clickbench_partitioned.json
--------------------
| Query | HEAD | q4 | Change |
|-----------|--------------------------------------------|--------------------------------------------|---------------|
| QQuery 0 | 1.36 / 4.94 ยฑ7.12 / 19.19 ms | 1.34 / 4.96 ยฑ7.17 / 19.31 ms | no change |
| QQuery 1 | 16.62 / 19.50 ยฑ5.12 / 29.72 ms | 16.31 / 18.93 ยฑ4.47 / 27.86 ms | no change |
| QQuery 2 | 49.02 / 56.98 ยฑ13.18 / 83.28 ms | 49.78 / 59.22 ยฑ12.82 / 84.56 ms | no change |
| QQuery 3 | 53.17 / 413.82 ยฑ718.84 / 1851.51 ms | 59.77 / 344.50 ยฑ565.86 / 1476.22 ms | 1.12x slower |
| QQuery 4 | 554.10 / 563.18 ยฑ8.72 / 578.19 ms | 439.19 / 580.67 ยฑ129.14 / 769.30 ms | +1.26x faster |
| QQuery 5 | 516.88 / 862.42 ยฑ659.98 / 2182.05 ms | 520.97 / 985.10 ยฑ801.77 / 2584.76 ms | no change |
| QQuery 6 | 3.21 / 3.61 ยฑ0.66 / 4.94 ms | 3.22 / 3.63 ยฑ0.50 / 4.60 ms | no change |
| QQuery 7 | 17.74 / 17.93 ยฑ0.21 / 18.32 ms | 17.66 / 17.87 ยฑ0.20 / 18.11 ms | no change |
| QQuery 8 | 686.29 / 696.36 ยฑ5.83 / 703.40 ms | 754.21 / 1027.06 ยฑ161.68 / 1242.22 ms | 1.10x slower |
| QQuery 9 | 728.66 / 774.36 ยฑ46.60 / 853.92 ms | 757.98 / 773.33 ยฑ27.82 / 828.94 ms | no change |
| QQuery 10 | 131.60 / 136.60 ยฑ3.78 / 142.90 ms | 132.43 / 143.93 ยฑ16.42 / 176.37 ms | no change |
| QQuery 11 | 154.82 / 157.99 ยฑ3.68 / 164.70 ms | 154.22 / 157.11 ยฑ2.27 / 160.61 ms | no change |
| QQuery 12 | 556.54 / 580.66 ยฑ16.57 / 600.00 ms | 572.20 / 581.34 ยฑ5.28 / 585.67 ms | no change |
| QQuery 13 | 910.76 / 925.92 ยฑ11.40 / 942.69 ms | 917.47 / 1089.09 ยฑ312.06 / 1711.62 ms | no change |
| QQuery 14 | 592.23 / 607.12 ยฑ16.48 / 629.28 ms | 589.55 / 604.00 ยฑ15.24 / 632.15 ms | no change |
| QQuery 15 | 660.31 / 668.69 ยฑ7.57 / 680.85 ms | 660.65 / 698.46 ยฑ57.26 / 811.65 ms | no change |
| QQuery 16 | 1379.80 / 1539.88 ยฑ180.32 / 1779.23 ms | 1886.79 / 2404.33 ยฑ589.74 / 3522.30 ms | 1.37x slower |
| QQuery 17 | 1354.58 / 1382.99 ยฑ16.74 / 1399.85 ms | 1384.44 / 1492.35 ยฑ176.97 / 1845.19 ms | no change |
| QQuery 18 | 3025.36 / 3872.99 ยฑ884.61 / 5536.77 ms | 3199.56 / 4494.24 ยฑ1372.71 / 7031.04 ms | 1.06x slower |
| QQuery 19 | 49.55 / 81.69 ยฑ61.68 / 205.05 ms | 52.03 / 132.06 ยฑ99.55 / 287.15 ms | no change |
| QQuery 20 | 819.85 / 4346.66 ยฑ7038.47 / 18423.59 ms | 802.50 / 4260.38 ยฑ6900.87 / 18062.11 ms | no change |
| QQuery 21 | 1000.17 / 1016.61 ยฑ10.11 / 1031.33 ms | 1010.43 / 1068.85 ยฑ45.61 / 1122.61 ms | no change |
| QQuery 22 | 2156.11 / 4996.83 ยฑ5654.14 / 16305.09 ms | 2167.60 / 4980.96 ยฑ5607.84 / 16196.63 ms | no change |
| QQuery 23 | 8584.84 / 24247.23 ยฑ20717.36 / 60235.10 ms | 8519.57 / 17462.29 ยฑ13228.65 / 42692.78 ms | no change |
| QQuery 24 | 74.87 / 79.44 ยฑ3.41 / 85.18 ms | 72.72 / 81.10 ยฑ6.23 / 88.68 ms | no change |
| QQuery 25 | 178.34 / 183.73 ยฑ7.06 / 197.70 ms | 169.92 / 173.68 ยฑ2.34 / 177.19 ms | no change |
| QQuery 26 | 71.38 / 72.57 ยฑ1.21 / 74.72 ms | 73.72 / 75.13 ยฑ1.16 / 76.90 ms | no change |
| QQuery 27 | 1192.07 / 1217.48 ยฑ20.19 / 1242.67 ms | 1167.96 / 1187.36 ยฑ21.54 / 1224.20 ms | no change |
| QQuery 28 | 3712.95 / 3738.42 ยฑ30.59 / 3797.21 ms | 3625.79 / 3669.24 ยฑ67.10 / 3801.25 ms | no change |
| QQuery 29 | 53.46 / 65.66 ยฑ21.75 / 108.99 ms | 55.63 / 59.62 ยฑ5.05 / 68.61 ms | no change |
| QQuery 30 | 567.78 / 580.97 ยฑ8.15 / 589.39 ms | 574.09 / 579.27 ยฑ3.67 / 584.37 ms | no change |
| QQuery 31 | 703.23 / 716.17 ยฑ19.01 / 753.59 ms | 704.20 / 714.94 ยฑ6.34 / 721.92 ms | no change |
| QQuery 32 | 2717.58 / 3374.99 ยฑ571.57 / 4437.74 ms | 3369.42 / 3940.48 ยฑ562.65 / 4949.74 ms | 1.24x slower |
| QQuery 33 | 3161.04 / 3230.02 ยฑ69.97 / 3363.47 ms | 3168.24 / 3310.67 ยฑ178.75 / 3659.44 ms | no change |
| QQuery 34 | 3207.08 / 3409.06 ยฑ234.38 / 3788.10 ms | 3190.32 / 3281.64 ยฑ141.73 / 3562.07 ms | no change |
| QQuery 35 | 567.18 / 592.82 ยฑ22.12 / 631.29 ms | 564.60 / 610.33 ยฑ58.41 / 724.64 ms | no change |
| QQuery 36 | 83.78 / 94.23 ยฑ7.04 / 102.23 ms | 82.21 / 93.46 ยฑ12.43 / 117.60 ms | no change |
| QQuery 37 | 37.12 / 41.97 ยฑ4.49 / 47.37 ms | 37.79 / 42.43 ยฑ4.74 / 49.52 ms | no change |
| QQuery 38 | 44.28 / 55.30 ยฑ14.05 / 82.71 ms | 45.56 / 52.01 ยฑ4.13 / 58.37 ms | no change |
| QQuery 39 | 165.66 / 181.12 ยฑ13.67 / 206.30 ms | 161.79 / 179.09 ยฑ11.21 / 193.53 ms | no change |
| QQuery 40 | 18.57 / 22.00 ยฑ3.40 / 26.49 ms | 18.39 / 19.15 ยฑ0.74 / 20.51 ms | no change |
| QQuery 41 | 17.78 / 21.30 ยฑ3.95 / 28.81 ms | 17.61 / 19.87 ยฑ4.11 / 28.10 ms | no change |
| QQuery 42 | 14.75 / 19.02 ยฑ4.93 / 27.30 ms | 14.79 / 16.69 ยฑ3.20 / 23.07 ms | no change |
| Benchmark Summary | |
|---|---:|
| Total Time (HEAD) | 40592.46ms |
| Total Time (q4) | 41784.61ms |
| Average Time (HEAD) | 944.01ms |
| Average Time (q4) | 971.74ms |
| Queries Faster | 1 |
| Queries Slower | 5 |
| Queries with No Change | 37 |
| Queries with Failure | 0 |
```
</details>
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] closed pull request #22145: [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4)
URL: https://github.com/apache/datafusion/pull/22145
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4440471735
Benchmark for [this request](https://github.com/apache/datafusion/pull/22145#issuecomment-4440449362) failed.
Last 20 lines of output:
<details><summary>Click to expand</summary>
```
* [new tag] 48.0.0-rc1 -> 48.0.0-rc1
* [new tag] 48.0.0-rc2 -> 48.0.0-rc2
* [new tag] 5.0.0 -> 5.0.0
* [new tag] 5.0.0-rc1 -> 5.0.0-rc1
* [new tag] 5.0.0-rc3 -> 5.0.0-rc3
* [new tag] 6.0.0 -> 6.0.0
* [new tag] 6.0.0-rc0 -> 6.0.0-rc0
* [new tag] 7.0.0 -> 7.0.0
* [new tag] 7.0.0-rc2 -> 7.0.0-rc2
* [new tag] 8.0.0 -> 8.0.0
* [new tag] 8.0.0-rc1 -> 8.0.0-rc1
* [new tag] 8.0.0-rc2 -> 8.0.0-rc2
* [new tag] 9.0.0 -> 9.0.0
* [new tag] 9.0.0-rc1 -> 9.0.0-rc1
* [new tag] ballista-0.5.0 -> ballista-0.5.0
* [new tag] ballista-0.6.0 -> ballista-0.6.0
* [new tag] ballista-0.7.0 -> ballista-0.7.0
* [new tag] python-0.3.0 -> python-0.3.0
* [new tag] python-0.4.0 -> python-0.4.0
Switched to branch 'q4'
```
</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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4440449362
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448524681
Benchmark for [this request](https://github.com/apache/datafusion/pull/22145#issuecomment-4448508770) failed.
Last 20 lines of output:
<details><summary>Click to expand</summary>
```
* [new tag] 48.0.0-rc1 -> 48.0.0-rc1
* [new tag] 48.0.0-rc2 -> 48.0.0-rc2
* [new tag] 5.0.0 -> 5.0.0
* [new tag] 5.0.0-rc1 -> 5.0.0-rc1
* [new tag] 5.0.0-rc3 -> 5.0.0-rc3
* [new tag] 6.0.0 -> 6.0.0
* [new tag] 6.0.0-rc0 -> 6.0.0-rc0
* [new tag] 7.0.0 -> 7.0.0
* [new tag] 7.0.0-rc2 -> 7.0.0-rc2
* [new tag] 8.0.0 -> 8.0.0
* [new tag] 8.0.0-rc1 -> 8.0.0-rc1
* [new tag] 8.0.0-rc2 -> 8.0.0-rc2
* [new tag] 9.0.0 -> 9.0.0
* [new tag] 9.0.0-rc1 -> 9.0.0-rc1
* [new tag] ballista-0.5.0 -> ballista-0.5.0
* [new tag] ballista-0.6.0 -> ballista-0.6.0
* [new tag] ballista-0.7.0 -> ballista-0.7.0
* [new tag] python-0.3.0 -> python-0.3.0
* [new tag] python-0.4.0 -> python-0.4.0
Switched to branch 'q4'
```
</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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448556762
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448801710
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356)
**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 q4
--------------------
Benchmark tpcds_sf1.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 1 โ 6.24 / 6.75 ยฑ0.84 / 8.44 ms โ 6.28 / 6.76 ยฑ0.83 / 8.42 ms โ no change โ
โ QQuery 2 โ 81.98 / 82.54 ยฑ0.29 / 82.77 ms โ 82.42 / 83.36 ยฑ0.93 / 84.67 ms โ no change โ
โ QQuery 3 โ 28.90 / 29.39 ยฑ0.29 / 29.78 ms โ 29.19 / 29.56 ยฑ0.26 / 29.96 ms โ no change โ
โ QQuery 4 โ 531.79 / 540.08 ยฑ11.18 / 562.05 ms โ 516.91 / 542.94 ยฑ25.90 / 575.22 ms โ no change โ
โ QQuery 5 โ 54.13 / 54.60 ยฑ0.36 / 55.22 ms โ 54.69 / 55.45 ยฑ0.73 / 56.83 ms โ no change โ
โ QQuery 6 โ 37.33 / 37.59 ยฑ0.18 / 37.82 ms โ 36.73 / 37.51 ยฑ0.53 / 38.15 ms โ no change โ
โ QQuery 7 โ 112.42 / 114.08 ยฑ1.51 / 115.91 ms โ 110.35 / 113.40 ยฑ2.12 / 115.56 ms โ no change โ
โ QQuery 8 โ 40.61 / 40.92 ยฑ0.27 / 41.34 ms โ 40.53 / 41.24 ยฑ0.38 / 41.64 ms โ no change โ
โ QQuery 9 โ 52.35 / 54.97 ยฑ1.71 / 57.63 ms โ 54.01 / 55.36 ยฑ0.85 / 56.61 ms โ no change โ
โ QQuery 10 โ 81.69 / 83.30 ยฑ1.46 / 85.78 ms โ 81.74 / 83.38 ยฑ1.77 / 86.31 ms โ no change โ
โ QQuery 11 โ 321.59 / 353.39 ยฑ17.77 / 371.79 ms โ 315.49 / 349.39 ยฑ19.12 / 369.07 ms โ no change โ
โ QQuery 12 โ 28.80 / 29.29 ยฑ0.62 / 30.48 ms โ 29.16 / 29.39 ยฑ0.23 / 29.77 ms โ no change โ
โ QQuery 13 โ 132.08 / 133.43 ยฑ1.05 / 134.89 ms โ 132.51 / 133.41 ยฑ0.67 / 134.28 ms โ no change โ
โ QQuery 14 โ 521.23 / 523.63 ยฑ1.79 / 525.73 ms โ 521.41 / 524.74 ยฑ2.07 / 527.16 ms โ no change โ
โ QQuery 15 โ 65.87 / 67.23 ยฑ0.84 / 68.19 ms โ 66.33 / 67.08 ยฑ0.92 / 68.75 ms โ no change โ
โ QQuery 16 โ 7.38 / 7.49 ยฑ0.07 / 7.57 ms โ 7.43 / 7.57 ยฑ0.11 / 7.73 ms โ no change โ
โ QQuery 17 โ 83.78 / 85.87 ยฑ1.27 / 87.13 ms โ 85.27 / 87.33 ยฑ2.00 / 89.80 ms โ no change โ
โ QQuery 18 โ 155.82 / 160.18 ยฑ2.57 / 162.31 ms โ 157.22 / 159.99 ยฑ1.55 / 161.40 ms โ no change โ
โ QQuery 19 โ 41.70 / 42.19 ยฑ0.43 / 43.00 ms โ 41.86 / 42.47 ยฑ0.71 / 43.87 ms โ no change โ
โ QQuery 20 โ 36.85 / 37.80 ยฑ0.66 / 38.82 ms โ 37.13 / 37.59 ยฑ0.29 / 38.01 ms โ no change โ
โ QQuery 21 โ 18.80 / 19.02 ยฑ0.15 / 19.22 ms โ 18.93 / 19.22 ยฑ0.17 / 19.40 ms โ no change โ
โ QQuery 22 โ 64.78 / 65.61 ยฑ0.54 / 66.27 ms โ 65.74 / 66.66 ยฑ0.85 / 68.01 ms โ no change โ
โ QQuery 23 โ 498.96 / 507.43 ยฑ6.18 / 517.09 ms โ 502.76 / 508.71 ยฑ6.71 / 521.61 ms โ no change โ
โ QQuery 24 โ 249.29 / 252.78 ยฑ3.36 / 258.91 ms โ 246.44 / 251.28 ยฑ3.67 / 257.64 ms โ no change โ
โ QQuery 25 โ 119.28 / 121.44 ยฑ1.45 / 123.67 ms โ 120.34 / 121.46 ยฑ0.76 / 122.71 ms โ no change โ
โ QQuery 26 โ 73.99 / 74.68 ยฑ1.14 / 76.97 ms โ 73.39 / 74.52 ยฑ1.24 / 76.92 ms โ no change โ
โ QQuery 27 โ 7.45 / 7.53 ยฑ0.05 / 7.58 ms โ 7.37 / 7.46 ยฑ0.06 / 7.55 ms โ no change โ
โ QQuery 28 โ 59.52 / 62.10 ยฑ2.09 / 64.24 ms โ 60.37 / 63.43 ยฑ1.73 / 65.71 ms โ no change โ
โ QQuery 29 โ 103.58 / 106.14 ยฑ2.46 / 110.49 ms โ 104.76 / 106.97 ยฑ1.38 / 108.30 ms โ no change โ
โ QQuery 30 โ 31.67 / 32.57 ยฑ0.69 / 33.49 ms โ 31.48 / 31.80 ยฑ0.27 / 32.25 ms โ no change โ
โ QQuery 31 โ 112.90 / 115.61 ยฑ1.99 / 119.04 ms โ 115.06 / 116.72 ยฑ1.21 / 118.61 ms โ no change โ
โ QQuery 32 โ 21.65 / 23.25 ยฑ2.36 / 27.94 ms โ 21.89 / 22.83 ยฑ1.61 / 26.05 ms โ no change โ
โ QQuery 33 โ 40.54 / 41.06 ยฑ0.30 / 41.46 ms โ 40.64 / 41.13 ยฑ0.45 / 41.97 ms โ no change โ
โ QQuery 34 โ 10.37 / 10.84 ยฑ0.36 / 11.45 ms โ 10.35 / 10.70 ยฑ0.35 / 11.34 ms โ no change โ
โ QQuery 35 โ 83.25 / 84.00 ยฑ0.80 / 85.56 ms โ 83.72 / 84.51 ยฑ0.79 / 85.96 ms โ no change โ
โ QQuery 36 โ 6.89 / 6.96 ยฑ0.07 / 7.08 ms โ 6.85 / 6.91 ยฑ0.08 / 7.08 ms โ no change โ
โ QQuery 37 โ 7.69 / 7.83 ยฑ0.09 / 7.94 ms โ 7.64 / 7.78 ยฑ0.11 / 7.93 ms โ no change โ
โ QQuery 38 โ 69.93 / 72.14 ยฑ1.61 / 74.94 ms โ 69.79 / 72.85 ยฑ1.89 / 75.50 ms โ no change โ
โ QQuery 39 โ 99.81 / 102.07 ยฑ1.34 / 103.64 ms โ 101.00 / 101.87 ยฑ0.44 / 102.21 ms โ no change โ
โ QQuery 40 โ 24.03 / 24.48 ยฑ0.50 / 25.37 ms โ 23.52 / 24.23 ยฑ1.10 / 26.41 ms โ no change โ
โ QQuery 41 โ 14.07 / 14.34 ยฑ0.22 / 14.63 ms โ 14.27 / 14.42 ยฑ0.08 / 14.51 ms โ no change โ
โ QQuery 42 โ 24.45 / 24.91 ยฑ0.35 / 25.44 ms โ 24.76 / 25.08 ยฑ0.28 / 25.59 ms โ no change โ
โ QQuery 43 โ 5.39 / 5.52 ยฑ0.17 / 5.85 ms โ 5.35 / 5.45 ยฑ0.16 / 5.78 ms โ no change โ
โ QQuery 44 โ 11.00 / 11.49 ยฑ0.80 / 13.08 ms โ 10.94 / 11.68 ยฑ1.21 / 14.08 ms โ no change โ
โ QQuery 45 โ 41.60 / 42.55 ยฑ0.51 / 43.13 ms โ 41.00 / 42.33 ยฑ1.05 / 43.90 ms โ no change โ
โ QQuery 46 โ 13.61 / 14.10 ยฑ0.30 / 14.40 ms โ 13.96 / 14.49 ยฑ0.37 / 15.00 ms โ no change โ
โ QQuery 47 โ 234.69 / 237.59 ยฑ2.62 / 242.50 ms โ 237.67 / 240.04 ยฑ3.69 / 247.36 ms โ no change โ
โ QQuery 48 โ 106.96 / 107.62 ยฑ0.55 / 108.25 ms โ 107.62 / 108.84 ยฑ1.69 / 112.16 ms โ no change โ
โ QQuery 49 โ 83.60 / 86.43 ยฑ4.12 / 94.34 ms โ 83.56 / 84.39 ยฑ0.65 / 85.34 ms โ no change โ
โ QQuery 50 โ 62.99 / 64.80 ยฑ1.94 / 68.55 ms โ 63.18 / 64.11 ยฑ1.13 / 66.17 ms โ no change โ
โ QQuery 51 โ 92.39 / 95.75 ยฑ2.65 / 99.48 ms โ 93.26 / 97.99 ยฑ3.25 / 103.27 ms โ no change โ
โ QQuery 52 โ 24.63 / 24.85 ยฑ0.13 / 25.01 ms โ 24.78 / 24.99 ยฑ0.22 / 25.40 ms โ no change โ
โ QQuery 53 โ 30.43 / 30.69 ยฑ0.23 / 31.09 ms โ 31.11 / 31.23 ยฑ0.11 / 31.42 ms โ no change โ
โ QQuery 54 โ 55.20 / 57.10 ยฑ2.23 / 61.44 ms โ 54.60 / 55.40 ยฑ0.88 / 57.00 ms โ no change โ
โ QQuery 55 โ 24.98 / 25.58 ยฑ0.51 / 26.51 ms โ 24.68 / 24.89 ยฑ0.23 / 25.34 ms โ no change โ
โ QQuery 56 โ 42.08 / 43.72 ยฑ1.86 / 47.20 ms โ 42.17 / 43.74 ยฑ1.45 / 46.24 ms โ no change โ
โ QQuery 57 โ 185.94 / 187.90 ยฑ1.12 / 189.39 ms โ 184.53 / 187.42 ยฑ1.74 / 189.05 ms โ no change โ
โ QQuery 58 โ 120.35 / 121.87 ยฑ0.85 / 122.74 ms โ 121.35 / 122.23 ยฑ0.77 / 123.62 ms โ no change โ
โ QQuery 59 โ 121.43 / 122.31 ยฑ0.66 / 123.28 ms โ 120.95 / 121.95 ยฑ0.76 / 123.03 ms โ no change โ
โ QQuery 60 โ 41.13 / 41.38 ยฑ0.21 / 41.61 ms โ 40.78 / 41.73 ยฑ0.55 / 42.44 ms โ no change โ
โ QQuery 61 โ 14.27 / 14.47 ยฑ0.13 / 14.60 ms โ 14.53 / 14.56 ยฑ0.03 / 14.61 ms โ no change โ
โ QQuery 62 โ 48.39 / 49.98 ยฑ1.57 / 52.96 ms โ 47.53 / 48.77 ยฑ1.52 / 51.77 ms โ no change โ
โ QQuery 63 โ 31.33 / 31.61 ยฑ0.25 / 31.96 ms โ 32.13 / 33.25 ยฑ2.15 / 37.54 ms โ 1.05x slower โ
โ QQuery 64 โ 487.96 / 493.27 ยฑ3.79 / 499.22 ms โ 488.83 / 493.08 ยฑ2.46 / 496.44 ms โ no change โ
โ QQuery 65 โ 149.80 / 156.36 ยฑ4.04 / 162.42 ms โ 151.19 / 154.81 ยฑ2.60 / 158.38 ms โ no change โ
โ QQuery 66 โ 86.89 / 88.24 ยฑ0.83 / 89.14 ms โ 87.16 / 87.84 ยฑ0.60 / 88.63 ms โ no change โ
โ QQuery 67 โ 272.11 / 273.22 ยฑ1.04 / 274.97 ms โ 269.30 / 278.41 ยฑ5.63 / 284.10 ms โ no change โ
โ QQuery 68 โ 14.71 / 16.84 ยฑ3.42 / 23.64 ms โ 14.39 / 14.63 ยฑ0.23 / 14.93 ms โ +1.15x faster โ
โ QQuery 69 โ 79.66 / 80.17 ยฑ0.41 / 80.62 ms โ 79.73 / 80.26 ยฑ0.62 / 81.46 ms โ no change โ
โ QQuery 70 โ 108.80 / 113.34 ยฑ4.03 / 119.75 ms โ 110.57 / 114.86 ยฑ5.01 / 124.02 ms โ no change โ
โ QQuery 71 โ 35.84 / 36.50 ยฑ0.47 / 37.29 ms โ 36.76 / 37.53 ยฑ0.48 / 38.19 ms โ no change โ
โ QQuery 72 โ 2287.44 / 2374.17 ยฑ64.43 / 2470.10 ms โ 2249.42 / 2367.83 ยฑ103.12 / 2537.52 ms โ no change โ
โ QQuery 73 โ 9.70 / 10.02 ยฑ0.26 / 10.34 ms โ 9.95 / 10.51 ยฑ0.48 / 11.19 ms โ no change โ
โ QQuery 74 โ 186.49 / 193.15 ยฑ7.26 / 204.40 ms โ 186.76 / 192.94 ยฑ8.95 / 210.74 ms โ no change โ
โ QQuery 75 โ 150.66 / 151.99 ยฑ1.65 / 155.11 ms โ 151.60 / 154.55 ยฑ1.95 / 157.73 ms โ no change โ
โ QQuery 76 โ 36.72 / 37.91 ยฑ0.84 / 39.21 ms โ 37.74 / 38.08 ยฑ0.42 / 38.83 ms โ no change โ
โ QQuery 77 โ 62.30 / 62.77 ยฑ0.26 / 63.05 ms โ 62.90 / 64.08 ยฑ1.16 / 65.49 ms โ no change โ
โ QQuery 78 โ 193.74 / 203.02 ยฑ5.75 / 208.59 ms โ 196.05 / 201.70 ยฑ5.27 / 211.15 ms โ no change โ
โ QQuery 79 โ 68.38 / 68.92 ยฑ0.68 / 70.20 ms โ 68.20 / 68.78 ยฑ0.96 / 70.70 ms โ no change โ
โ QQuery 80 โ 105.09 / 108.39 ยฑ3.66 / 113.60 ms โ 103.98 / 106.38 ยฑ2.52 / 111.07 ms โ no change โ
โ QQuery 81 โ 24.65 / 25.07 ยฑ0.53 / 26.06 ms โ 25.87 / 26.48 ยฑ0.47 / 27.05 ms โ 1.06x slower โ
โ QQuery 82 โ 16.61 / 17.11 ยฑ0.29 / 17.49 ms โ 17.22 / 17.57 ยฑ0.53 / 18.59 ms โ no change โ
โ QQuery 83 โ 37.32 / 38.46 ยฑ1.36 / 41.04 ms โ 37.57 / 38.25 ยฑ0.88 / 39.97 ms โ no change โ
โ QQuery 84 โ 43.30 / 44.07 ยฑ0.83 / 45.46 ms โ 43.17 / 43.44 ยฑ0.26 / 43.91 ms โ no change โ
โ QQuery 85 โ 138.00 / 139.95 ยฑ2.21 / 144.05 ms โ 137.52 / 139.53 ยฑ2.00 / 143.15 ms โ no change โ
โ QQuery 86 โ 25.18 / 25.82 ยฑ0.62 / 26.89 ms โ 25.27 / 25.50 ยฑ0.31 / 26.11 ms โ no change โ
โ QQuery 87 โ 69.03 / 72.09 ยฑ2.57 / 76.51 ms โ 69.52 / 71.05 ยฑ1.70 / 74.00 ms โ no change โ
โ QQuery 88 โ 65.49 / 65.76 ยฑ0.25 / 66.21 ms โ 66.34 / 67.93 ยฑ0.82 / 68.55 ms โ no change โ
โ QQuery 89 โ 36.39 / 36.57 ยฑ0.10 / 36.66 ms โ 36.92 / 37.34 ยฑ0.30 / 37.77 ms โ no change โ
โ QQuery 90 โ 17.56 / 18.61 ยฑ1.75 / 22.09 ms โ 17.66 / 17.86 ยฑ0.15 / 18.14 ms โ no change โ
โ QQuery 91 โ 53.13 / 54.24 ยฑ0.94 / 55.83 ms โ 52.56 / 52.87 ยฑ0.22 / 53.15 ms โ no change โ
โ QQuery 92 โ 29.55 / 31.32 ยฑ1.45 / 33.39 ms โ 29.48 / 30.80 ยฑ1.68 / 34.05 ms โ no change โ
โ QQuery 93 โ 53.54 / 54.33 ยฑ0.60 / 55.35 ms โ 54.80 / 56.76 ยฑ1.60 / 58.85 ms โ no change โ
โ QQuery 94 โ 37.84 / 38.28 ยฑ0.22 / 38.46 ms โ 37.90 / 38.41 ยฑ0.45 / 39.04 ms โ no change โ
โ QQuery 95 โ 88.45 / 91.47 ยฑ2.80 / 96.58 ms โ 89.20 / 92.04 ยฑ2.12 / 95.72 ms โ no change โ
โ QQuery 96 โ 24.49 / 24.94 ยฑ0.40 / 25.63 ms โ 24.63 / 24.98 ยฑ0.27 / 25.32 ms โ no change โ
โ QQuery 97 โ 46.75 / 47.55 ยฑ0.45 / 48.03 ms โ 47.01 / 48.15 ยฑ1.00 / 49.65 ms โ no change โ
โ QQuery 98 โ 42.56 / 42.99 ยฑ0.46 / 43.83 ms โ 43.52 / 43.98 ยฑ0.58 / 45.10 ms โ no change โ
โ QQuery 99 โ 70.83 / 72.33 ยฑ1.36 / 74.42 ms โ 70.67 / 71.09 ยฑ0.34 / 71.65 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 11088.07ms โ
โ Total Time (q4) โ 11098.27ms โ
โ Average Time (HEAD) โ 112.00ms โ
โ Average Time (q4) โ 112.10ms โ
โ Queries Faster โ 1 โ
โ Queries Slower โ 2 โ
โ Queries with No Change โ 96 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpcds โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 60.0s |
| Peak memory | 7.1 GiB |
| Avg memory | 6.3 GiB |
| CPU user | 254.4s |
| CPU sys | 6.5s |
| Peak spill | 0 B |
**tpcds โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 60.0s |
| Peak memory | 7.1 GiB |
| Avg memory | 6.2 GiB |
| CPU user | 257.0s |
| CPU sys | 6.8s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4448817345
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4448673356)
**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 q4
--------------------
Benchmark clickbench_partitioned.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 0 โ 1.20 / 4.69 ยฑ6.86 / 18.40 ms โ 1.21 / 4.71 ยฑ6.87 / 18.45 ms โ no change โ
โ QQuery 1 โ 12.48 / 12.86 ยฑ0.25 / 13.26 ms โ 12.45 / 12.95 ยฑ0.26 / 13.23 ms โ no change โ
โ QQuery 2 โ 35.38 / 35.85 ยฑ0.33 / 36.13 ms โ 35.75 / 36.02 ยฑ0.27 / 36.53 ms โ no change โ
โ QQuery 3 โ 30.36 / 30.99 ยฑ0.81 / 32.58 ms โ 30.72 / 30.88 ยฑ0.19 / 31.23 ms โ no change โ
โ QQuery 4 โ 232.65 / 234.21 ยฑ1.59 / 236.52 ms โ 213.86 / 215.85 ยฑ1.58 / 218.49 ms โ +1.09x faster โ
โ QQuery 5 โ 279.97 / 283.25 ยฑ2.42 / 286.26 ms โ 277.39 / 279.05 ยฑ1.95 / 281.47 ms โ no change โ
โ QQuery 6 โ 6.59 / 7.11 ยฑ0.33 / 7.56 ms โ 7.02 / 7.25 ยฑ0.22 / 7.61 ms โ no change โ
โ QQuery 7 โ 13.73 / 13.78 ยฑ0.05 / 13.87 ms โ 13.73 / 13.81 ยฑ0.12 / 14.04 ms โ no change โ
โ QQuery 8 โ 316.64 / 321.16 ยฑ4.22 / 328.38 ms โ 313.53 / 318.65 ยฑ4.98 / 327.97 ms โ no change โ
โ QQuery 9 โ 451.14 / 456.79 ยฑ5.49 / 467.00 ms โ 451.46 / 456.03 ยฑ3.45 / 460.95 ms โ no change โ
โ QQuery 10 โ 69.19 / 69.88 ยฑ0.41 / 70.33 ms โ 69.58 / 70.36 ยฑ0.68 / 71.55 ms โ no change โ
โ QQuery 11 โ 81.29 / 81.89 ยฑ0.63 / 82.76 ms โ 80.36 / 81.29 ยฑ0.59 / 81.84 ms โ no change โ
โ QQuery 12 โ 273.58 / 276.74 ยฑ3.33 / 282.70 ms โ 271.47 / 277.61 ยฑ5.58 / 287.44 ms โ no change โ
โ QQuery 13 โ 380.50 / 390.57 ยฑ9.15 / 407.47 ms โ 376.60 / 385.16 ยฑ8.29 / 399.18 ms โ no change โ
โ QQuery 14 โ 279.39 / 281.34 ยฑ1.95 / 284.38 ms โ 275.76 / 280.05 ยฑ3.48 / 285.77 ms โ no change โ
โ QQuery 15 โ 277.20 / 283.55 ยฑ6.20 / 291.70 ms โ 276.22 / 279.30 ยฑ3.00 / 285.05 ms โ no change โ
โ QQuery 16 โ 602.01 / 609.34 ยฑ5.16 / 616.48 ms โ 597.18 / 609.15 ยฑ10.81 / 625.50 ms โ no change โ
โ QQuery 17 โ 601.14 / 607.67 ยฑ5.24 / 615.38 ms โ 600.22 / 606.81 ยฑ6.36 / 617.50 ms โ no change โ
โ QQuery 18 โ 1192.63 / 1209.70 ยฑ10.06 / 1222.79 ms โ 1185.08 / 1203.70 ยฑ11.24 / 1216.19 ms โ no change โ
โ QQuery 19 โ 27.75 / 31.34 ยฑ6.83 / 44.99 ms โ 27.65 / 30.09 ยฑ4.22 / 38.53 ms โ no change โ
โ QQuery 20 โ 517.06 / 521.72 ยฑ4.73 / 530.00 ms โ 514.19 / 518.53 ยฑ5.85 / 529.96 ms โ no change โ
โ QQuery 21 โ 597.15 / 601.30 ยฑ3.44 / 606.52 ms โ 590.95 / 596.78 ยฑ4.57 / 603.50 ms โ no change โ
โ QQuery 22 โ 1056.81 / 1059.52 ยฑ2.86 / 1064.66 ms โ 1047.76 / 1057.07 ยฑ6.84 / 1068.13 ms โ no change โ
โ QQuery 23 โ 3164.94 / 3193.47 ยฑ27.53 / 3244.35 ms โ 3185.53 / 3200.48 ยฑ12.46 / 3215.52 ms โ no change โ
โ QQuery 24 โ 41.30 / 41.81 ยฑ0.42 / 42.47 ms โ 42.28 / 46.86 ยฑ4.50 / 52.96 ms โ 1.12x slower โ
โ QQuery 25 โ 111.63 / 112.86 ยฑ0.85 / 113.87 ms โ 111.98 / 113.91 ยฑ1.57 / 116.37 ms โ no change โ
โ QQuery 26 โ 42.26 / 44.24 ยฑ1.91 / 47.73 ms โ 42.26 / 43.37 ยฑ0.82 / 44.40 ms โ no change โ
โ QQuery 27 โ 664.44 / 670.72 ยฑ4.23 / 676.55 ms โ 665.17 / 674.24 ยฑ5.49 / 680.11 ms โ no change โ
โ QQuery 28 โ 3023.60 / 3033.90 ยฑ8.64 / 3044.41 ms โ 2997.37 / 3023.53 ยฑ22.07 / 3060.18 ms โ no change โ
โ QQuery 29 โ 41.61 / 46.09 ยฑ8.58 / 63.24 ms โ 41.22 / 44.45 ยฑ5.73 / 55.89 ms โ no change โ
โ QQuery 30 โ 300.57 / 303.48 ยฑ4.37 / 312.09 ms โ 297.94 / 302.28 ยฑ2.99 / 306.12 ms โ no change โ
โ QQuery 31 โ 287.54 / 291.53 ยฑ3.93 / 296.49 ms โ 287.15 / 297.07 ยฑ8.76 / 311.87 ms โ no change โ
โ QQuery 32 โ 903.66 / 923.37 ยฑ27.52 / 977.51 ms โ 902.05 / 947.70 ยฑ28.19 / 987.82 ms โ no change โ
โ QQuery 33 โ 1428.20 / 1454.73 ยฑ24.71 / 1499.14 ms โ 1438.07 / 1458.38 ยฑ14.02 / 1476.53 ms โ no change โ
โ QQuery 34 โ 1444.46 / 1470.21 ยฑ20.84 / 1498.60 ms โ 1449.58 / 1466.42 ยฑ16.32 / 1497.54 ms โ no change โ
โ QQuery 35 โ 289.45 / 303.91 ยฑ18.47 / 339.04 ms โ 289.11 / 310.86 ยฑ25.68 / 346.40 ms โ no change โ
โ QQuery 36 โ 62.40 / 70.38 ยฑ5.84 / 80.46 ms โ 64.59 / 67.46 ยฑ2.75 / 71.81 ms โ no change โ
โ QQuery 37 โ 35.92 / 37.45 ยฑ1.77 / 40.86 ms โ 35.35 / 38.43 ยฑ3.07 / 42.37 ms โ no change โ
โ QQuery 38 โ 40.81 / 45.57 ยฑ4.26 / 53.52 ms โ 40.22 / 42.88 ยฑ2.28 / 46.43 ms โ +1.06x faster โ
โ QQuery 39 โ 132.92 / 136.43 ยฑ3.27 / 142.64 ms โ 120.20 / 129.46 ยฑ5.80 / 136.17 ms โ +1.05x faster โ
โ QQuery 40 โ 14.34 / 17.06 ยฑ4.62 / 26.25 ms โ 14.27 / 16.21 ยฑ3.58 / 23.36 ms โ no change โ
โ QQuery 41 โ 14.05 / 14.77 ยฑ0.57 / 15.58 ms โ 13.92 / 14.93 ยฑ1.58 / 18.07 ms โ no change โ
โ QQuery 42 โ 13.28 / 15.32 ยฑ3.47 / 22.25 ms โ 13.21 / 13.49 ยฑ0.19 / 13.70 ms โ +1.14x faster โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 19652.53ms โ
โ Total Time (q4) โ 19623.50ms โ
โ Average Time (HEAD) โ 457.04ms โ
โ Average Time (q4) โ 456.36ms โ
โ Queries Faster โ 4 โ
โ Queries Slower โ 1 โ
โ Queries with No Change โ 38 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_partitioned โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 100.0s |
| Peak memory | 30.4 GiB |
| Avg memory | 23.4 GiB |
| CPU user | 1036.9s |
| CPU sys | 60.9s |
| Peak spill | 0 B |
**clickbench_partitioned โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 100.0s |
| Peak memory | 30.4 GiB |
| Avg memory | 23.3 GiB |
| CPU user | 1033.9s |
| CPU sys | 61.6s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on code in PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#discussion_r3240189863
##########
datafusion/physical-plan/src/aggregates/group_values/single_group_by/primitive.rs:
##########
@@ -79,147 +79,334 @@ hash_float!(f16, f32, f64);
///
/// This specialization is significantly faster than using the more general
/// purpose `Row`s format
+enum GroupValuesPrimitiveState<T: ArrowPrimitiveType> {
+ GroupIds {
+ /// Stores the `(group_index, hash)` based on the hash of its value
+ ///
+ /// We also store `hash` is for reducing cost of rehashing. Such cost
+ /// is obvious in high cardinality group by situation.
+ /// More details can see:
+ /// <https://github.com/apache/datafusion/issues/15961>
+ map: HashTable<(usize, u64)>,
Review Comment:
I found `HashTable<(usize, T::Native)>` also to be a bit faster than this current approach (and also saving memory). We can build the `Vec<T::Native>` when emitting data.
Could be part of some future PR.
--
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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4449184157
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4449065667)
**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 q4
--------------------
Benchmark clickbench_partitioned.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 0 โ 1.20 / 4.57 ยฑ6.68 / 17.93 ms โ 1.22 / 4.55 ยฑ6.62 / 17.79 ms โ no change โ
โ QQuery 1 โ 12.20 / 12.51 ยฑ0.23 / 12.79 ms โ 12.53 / 12.73 ยฑ0.12 / 12.89 ms โ no change โ
โ QQuery 2 โ 35.18 / 36.25 ยฑ1.07 / 38.21 ms โ 35.34 / 35.60 ยฑ0.23 / 35.96 ms โ no change โ
โ QQuery 3 โ 30.34 / 31.24 ยฑ0.82 / 32.75 ms โ 30.33 / 30.57 ยฑ0.23 / 30.95 ms โ no change โ
โ QQuery 4 โ 228.33 / 231.92 ยฑ2.33 / 235.39 ms โ 219.18 / 228.03 ยฑ6.79 / 235.14 ms โ no change โ
โ QQuery 5 โ 276.24 / 279.14 ยฑ2.32 / 282.16 ms โ 274.67 / 276.65 ยฑ1.31 / 278.13 ms โ no change โ
โ QQuery 6 โ 5.92 / 6.70 ยฑ0.64 / 7.73 ms โ 6.39 / 7.06 ยฑ0.64 / 7.89 ms โ 1.05x slower โ
โ QQuery 7 โ 13.69 / 13.90 ยฑ0.29 / 14.45 ms โ 13.86 / 14.07 ยฑ0.19 / 14.34 ms โ no change โ
โ QQuery 8 โ 308.78 / 311.83 ยฑ1.84 / 314.29 ms โ 308.36 / 311.51 ยฑ1.96 / 314.23 ms โ no change โ
โ QQuery 9 โ 442.03 / 449.85 ยฑ6.87 / 461.74 ms โ 440.53 / 447.78 ยฑ8.15 / 461.74 ms โ no change โ
โ QQuery 10 โ 70.02 / 70.75 ยฑ0.61 / 71.48 ms โ 69.82 / 70.15 ยฑ0.21 / 70.43 ms โ no change โ
โ QQuery 11 โ 80.95 / 81.82 ยฑ0.49 / 82.29 ms โ 80.31 / 82.69 ยฑ3.62 / 89.88 ms โ no change โ
โ QQuery 12 โ 267.99 / 272.24 ยฑ3.01 / 277.36 ms โ 268.09 / 271.76 ยฑ2.52 / 275.04 ms โ no change โ
โ QQuery 13 โ 382.56 / 389.75 ยฑ8.05 / 405.28 ms โ 376.60 / 380.50 ยฑ2.63 / 384.07 ms โ no change โ
โ QQuery 14 โ 276.50 / 279.68 ยฑ5.51 / 290.64 ms โ 276.23 / 279.34 ยฑ2.19 / 282.87 ms โ no change โ
โ QQuery 15 โ 272.62 / 277.86 ยฑ2.96 / 281.13 ms โ 270.23 / 277.20 ยฑ4.57 / 284.56 ms โ no change โ
โ QQuery 16 โ 594.09 / 601.25 ยฑ6.45 / 611.75 ms โ 598.12 / 605.52 ยฑ7.40 / 619.02 ms โ no change โ
โ QQuery 17 โ 592.66 / 602.78 ยฑ8.92 / 617.21 ms โ 591.67 / 602.68 ยฑ5.94 / 608.05 ms โ no change โ
โ QQuery 18 โ 1161.89 / 1192.05 ยฑ21.20 / 1218.29 ms โ 1175.28 / 1185.75 ยฑ8.70 / 1200.66 ms โ no change โ
โ QQuery 19 โ 27.70 / 28.47 ยฑ0.82 / 29.81 ms โ 27.89 / 31.05 ยฑ5.76 / 42.55 ms โ 1.09x slower โ
โ QQuery 20 โ 516.14 / 521.36 ยฑ5.98 / 532.84 ms โ 514.99 / 519.81 ยฑ5.29 / 527.90 ms โ no change โ
โ QQuery 21 โ 591.72 / 596.51 ยฑ3.44 / 601.38 ms โ 594.12 / 597.25 ยฑ2.22 / 601.05 ms โ no change โ
โ QQuery 22 โ 1054.30 / 1067.87 ยฑ7.52 / 1077.40 ms โ 1048.92 / 1062.54 ยฑ9.80 / 1076.10 ms โ no change โ
โ QQuery 23 โ 3150.22 / 3177.28 ยฑ23.67 / 3220.50 ms โ 3164.77 / 3187.59 ยฑ13.20 / 3200.42 ms โ no change โ
โ QQuery 24 โ 41.78 / 45.60 ยฑ5.33 / 56.02 ms โ 41.43 / 46.69 ยฑ4.31 / 51.73 ms โ no change โ
โ QQuery 25 โ 111.75 / 114.21 ยฑ1.90 / 117.42 ms โ 112.10 / 114.16 ยฑ2.25 / 118.32 ms โ no change โ
โ QQuery 26 โ 41.83 / 42.91 ยฑ1.86 / 46.62 ms โ 41.75 / 43.95 ยฑ2.85 / 49.46 ms โ no change โ
โ QQuery 27 โ 672.84 / 677.26 ยฑ3.10 / 682.25 ms โ 665.25 / 673.45 ยฑ5.71 / 679.45 ms โ no change โ
โ QQuery 28 โ 2993.40 / 3002.88 ยฑ5.27 / 3007.90 ms โ 2992.50 / 3004.51 ยฑ10.82 / 3023.98 ms โ no change โ
โ QQuery 29 โ 41.56 / 45.46 ยฑ4.68 / 54.06 ms โ 41.30 / 46.46 ยฑ9.85 / 66.14 ms โ no change โ
โ QQuery 30 โ 297.14 / 302.40 ยฑ4.43 / 309.36 ms โ 295.32 / 301.74 ยฑ4.37 / 306.48 ms โ no change โ
โ QQuery 31 โ 286.99 / 296.01 ยฑ6.12 / 306.09 ms โ 290.55 / 294.17 ยฑ2.21 / 296.33 ms โ no change โ
โ QQuery 32 โ 908.85 / 921.08 ยฑ9.66 / 932.84 ms โ 911.08 / 930.44 ยฑ18.75 / 962.61 ms โ no change โ
โ QQuery 33 โ 1426.03 / 1445.82 ยฑ16.34 / 1469.47 ms โ 1422.49 / 1450.02 ยฑ27.52 / 1500.95 ms โ no change โ
โ QQuery 34 โ 1434.50 / 1439.89 ยฑ3.17 / 1442.92 ms โ 1436.97 / 1445.87 ยฑ6.12 / 1455.35 ms โ no change โ
โ QQuery 35 โ 280.45 / 290.12 ยฑ8.52 / 305.58 ms โ 286.88 / 294.22 ยฑ7.28 / 306.10 ms โ no change โ
โ QQuery 36 โ 65.15 / 69.76 ยฑ4.40 / 77.25 ms โ 65.77 / 67.81 ยฑ1.55 / 70.48 ms โ no change โ
โ QQuery 37 โ 35.40 / 39.91 ยฑ3.88 / 46.27 ms โ 35.17 / 39.02 ยฑ3.54 / 45.20 ms โ no change โ
โ QQuery 38 โ 42.97 / 45.38 ยฑ2.57 / 49.12 ms โ 40.94 / 43.22 ยฑ1.66 / 45.95 ms โ no change โ
โ QQuery 39 โ 131.80 / 137.55 ยฑ3.95 / 141.75 ms โ 127.72 / 136.03 ยฑ8.27 / 151.54 ms โ no change โ
โ QQuery 40 โ 13.86 / 14.30 ยฑ0.31 / 14.65 ms โ 13.83 / 17.12 ยฑ3.59 / 22.13 ms โ 1.20x slower โ
โ QQuery 41 โ 13.59 / 15.21 ยฑ2.82 / 20.85 ms โ 13.38 / 13.53 ยฑ0.12 / 13.71 ms โ +1.12x faster โ
โ QQuery 42 โ 13.11 / 13.56 ยฑ0.63 / 14.79 ms โ 12.73 / 16.60 ยฑ7.03 / 30.64 ms โ 1.22x slower โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 19496.91ms โ
โ Total Time (q4) โ 19501.39ms โ
โ Average Time (HEAD) โ 453.42ms โ
โ Average Time (q4) โ 453.52ms โ
โ Queries Faster โ 1 โ
โ Queries Slower โ 4 โ
โ Queries with No Change โ 38 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_partitioned โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 100.0s |
| Peak memory | 31.0 GiB |
| Avg memory | 23.6 GiB |
| CPU user | 1029.8s |
| CPU sys | 61.2s |
| Peak spill | 0 B |
**clickbench_partitioned โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 100.0s |
| Peak memory | 31.9 GiB |
| Avg memory | 23.6 GiB |
| CPU user | 1028.6s |
| CPU sys | 61.8s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461331751
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/22145#issuecomment-4461199921)
**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 q4
--------------------
Benchmark clickbench_partitioned.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ q4 โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 0 โ 1.23 / 4.76 ยฑ6.92 / 18.59 ms โ 1.25 / 4.82 ยฑ6.98 / 18.78 ms โ no change โ
โ QQuery 1 โ 12.89 / 13.07 ยฑ0.11 / 13.18 ms โ 12.70 / 13.12 ยฑ0.24 / 13.37 ms โ no change โ
โ QQuery 2 โ 36.34 / 36.86 ยฑ0.36 / 37.29 ms โ 35.87 / 36.27 ยฑ0.29 / 36.73 ms โ no change โ
โ QQuery 3 โ 31.56 / 32.94 ยฑ1.11 / 34.87 ms โ 31.81 / 32.26 ยฑ0.38 / 32.77 ms โ no change โ
โ QQuery 4 โ 240.65 / 245.56 ยฑ2.93 / 249.81 ms โ 216.21 / 230.63 ยฑ12.53 / 247.06 ms โ +1.06x faster โ
โ QQuery 5 โ 287.84 / 290.17 ยฑ1.33 / 291.97 ms โ 285.30 / 288.68 ยฑ2.04 / 290.65 ms โ no change โ
โ QQuery 6 โ 7.36 / 8.72 ยฑ2.32 / 13.35 ms โ 7.48 / 7.96 ยฑ0.32 / 8.38 ms โ +1.10x faster โ
โ QQuery 7 โ 13.89 / 14.07 ยฑ0.11 / 14.21 ms โ 14.14 / 14.36 ยฑ0.20 / 14.71 ms โ no change โ
โ QQuery 8 โ 325.36 / 330.01 ยฑ2.78 / 332.85 ms โ 330.90 / 334.63 ยฑ2.43 / 337.57 ms โ no change โ
โ QQuery 9 โ 468.59 / 474.20 ยฑ4.86 / 482.09 ms โ 461.37 / 471.01 ยฑ5.91 / 478.72 ms โ no change โ
โ QQuery 10 โ 71.79 / 72.31 ยฑ0.49 / 72.93 ms โ 71.33 / 73.37 ยฑ1.21 / 74.68 ms โ no change โ
โ QQuery 11 โ 82.86 / 84.07 ยฑ1.48 / 86.81 ms โ 81.88 / 83.69 ยฑ2.48 / 88.56 ms โ no change โ
โ QQuery 12 โ 286.67 / 290.60 ยฑ3.29 / 295.31 ms โ 282.99 / 286.23 ยฑ3.33 / 292.55 ms โ no change โ
โ QQuery 13 โ 396.94 / 404.33 ยฑ5.35 / 412.23 ms โ 392.21 / 404.07 ยฑ11.13 / 423.66 ms โ no change โ
โ QQuery 14 โ 288.16 / 297.60 ยฑ11.73 / 319.92 ms โ 289.71 / 294.47 ยฑ2.88 / 297.46 ms โ no change โ
โ QQuery 15 โ 290.10 / 293.69 ยฑ2.87 / 298.71 ms โ 293.96 / 297.60 ยฑ3.73 / 303.94 ms โ no change โ
โ QQuery 16 โ 630.07 / 632.64 ยฑ2.42 / 637.12 ms โ 616.46 / 632.94 ยฑ10.09 / 647.72 ms โ no change โ
โ QQuery 17 โ 623.20 / 629.19 ยฑ4.42 / 634.24 ms โ 624.09 / 641.87 ยฑ20.59 / 680.83 ms โ no change โ
โ QQuery 18 โ 1222.28 / 1246.12 ยฑ18.64 / 1264.45 ms โ 1218.89 / 1245.56 ยฑ15.87 / 1258.65 ms โ no change โ
โ QQuery 19 โ 28.93 / 29.37 ยฑ0.24 / 29.65 ms โ 28.95 / 29.23 ยฑ0.19 / 29.52 ms โ no change โ
โ QQuery 20 โ 522.96 / 542.12 ยฑ20.96 / 577.75 ms โ 524.56 / 530.65 ยฑ3.63 / 534.13 ms โ no change โ
โ QQuery 21 โ 600.32 / 603.13 ยฑ2.08 / 606.21 ms โ 606.92 / 609.67 ยฑ2.26 / 612.24 ms โ no change โ
โ QQuery 22 โ 1069.37 / 1081.80 ยฑ7.96 / 1093.04 ms โ 1080.11 / 1090.70 ยฑ6.73 / 1098.29 ms โ no change โ
โ QQuery 23 โ 3261.51 / 3304.43 ยฑ35.63 / 3351.47 ms โ 3264.87 / 3300.21 ยฑ26.49 / 3340.27 ms โ no change โ
โ QQuery 24 โ 42.71 / 47.66 ยฑ8.87 / 65.39 ms โ 43.82 / 48.13 ยฑ5.63 / 58.89 ms โ no change โ
โ QQuery 25 โ 114.38 / 117.61 ยฑ4.06 / 125.44 ms โ 114.66 / 115.57 ยฑ1.33 / 118.23 ms โ no change โ
โ QQuery 26 โ 43.54 / 48.05 ยฑ6.10 / 60.03 ms โ 43.69 / 44.04 ยฑ0.39 / 44.80 ms โ +1.09x faster โ
โ QQuery 27 โ 673.98 / 681.57 ยฑ4.94 / 687.81 ms โ 680.07 / 683.35 ยฑ3.76 / 690.38 ms โ no change โ
โ QQuery 28 โ 3055.30 / 3078.95 ยฑ23.81 / 3120.68 ms โ 3041.05 / 3061.79 ยฑ15.33 / 3088.03 ms โ no change โ
โ QQuery 29 โ 41.93 / 51.53 ยฑ18.40 / 88.33 ms โ 42.70 / 45.95 ยฑ6.06 / 58.07 ms โ +1.12x faster โ
โ QQuery 30 โ 308.64 / 317.45 ยฑ7.61 / 329.92 ms โ 309.46 / 317.19 ยฑ6.40 / 327.02 ms โ no change โ
โ QQuery 31 โ 297.94 / 301.87 ยฑ2.34 / 305.27 ms โ 296.66 / 308.77 ยฑ8.61 / 323.29 ms โ no change โ
โ QQuery 32 โ 964.69 / 985.76 ยฑ18.58 / 1018.71 ms โ 978.90 / 999.70 ยฑ25.33 / 1037.83 ms โ no change โ
โ QQuery 33 โ 1487.31 / 1515.95 ยฑ24.58 / 1558.69 ms โ 1473.12 / 1502.25 ยฑ15.93 / 1516.24 ms โ no change โ
โ QQuery 34 โ 1492.43 / 1522.68 ยฑ23.28 / 1564.14 ms โ 1491.67 / 1514.82 ยฑ13.77 / 1533.18 ms โ no change โ
โ QQuery 35 โ 302.10 / 314.81 ยฑ12.38 / 330.04 ms โ 305.66 / 329.99 ยฑ33.39 / 395.27 ms โ no change โ
โ QQuery 36 โ 63.69 / 67.31 ยฑ3.13 / 72.73 ms โ 63.56 / 71.47 ยฑ10.36 / 91.64 ms โ 1.06x slower โ
โ QQuery 37 โ 36.77 / 43.63 ยฑ4.20 / 49.56 ms โ 37.75 / 45.98 ยฑ7.47 / 58.50 ms โ 1.05x slower โ
โ QQuery 38 โ 41.08 / 45.27 ยฑ3.62 / 51.78 ms โ 42.27 / 44.31 ยฑ2.17 / 48.52 ms โ no change โ
โ QQuery 39 โ 124.79 / 137.84 ยฑ9.34 / 153.35 ms โ 130.91 / 138.22 ยฑ4.99 / 145.07 ms โ no change โ
โ QQuery 40 โ 14.75 / 17.42 ยฑ3.72 / 24.77 ms โ 14.95 / 18.43 ยฑ4.73 / 27.46 ms โ 1.06x slower โ
โ QQuery 41 โ 14.22 / 14.41 ยฑ0.18 / 14.68 ms โ 14.64 / 14.76 ยฑ0.10 / 14.94 ms โ no change โ
โ QQuery 42 โ 13.69 / 13.88 ยฑ0.12 / 14.07 ms โ 13.85 / 14.14 ยฑ0.19 / 14.42 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 20285.41ms โ
โ Total Time (q4) โ 20272.84ms โ
โ Average Time (HEAD) โ 471.75ms โ
โ Average Time (q4) โ 471.46ms โ
โ Queries Faster โ 4 โ
โ Queries Slower โ 3 โ
โ Queries with No Change โ 36 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**clickbench_partitioned โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 105.0s |
| Peak memory | 30.1 GiB |
| Avg memory | 23.2 GiB |
| CPU user | 1062.1s |
| CPU sys | 69.2s |
| Peak spill | 0 B |
**clickbench_partitioned โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 105.0s |
| Peak memory | 30.8 GiB |
| Avg memory | 23.7 GiB |
| CPU user | 1060.3s |
| CPU sys | 70.8s |
| 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] [PoC] perf: optimize group-only group-by case for primitive cases (clickbench q4) [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #22145:
URL: https://github.com/apache/datafusion/pull/22145#issuecomment-4461199921
run benchmark clickbench_partitioned
--
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