You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "Dandandan (via GitHub)" <gi...@apache.org> on 2026/05/01 11:58:38 UTC
[PR] perf: optimize scalar aggregate joins as semi joins [datafusion]
Dandandan opened a new pull request, #21971:
URL: https://github.com/apache/datafusion/pull/21971
## Which issue does this PR close?
N/A.
## Rationale for this change
TPC-H Q2, Q17, and Q20 contain correlated scalar aggregate subqueries that are decorrelated into joins. In these cases the right side is unique on the correlated join keys, and no right-side columns are projected above the join. That means the inner join is only testing existence of a matching scalar aggregate row, so a left semi join expresses the plan more directly and avoids carrying inner-join output/cardinality machinery.
PR #21240 adds physical execution for uncorrelated scalar subqueries, but these cases are correlated and still need decorrelation into a set-at-a-time join plan.
## What changes are included in this PR?
- Extend `EliminateJoin` to rewrite projected inner joins to left semi joins when the right input is provably unique on the right join keys.
- Prove uniqueness through `SubqueryAlias` / `Projection` wrappers over `Aggregate` plans whose group expressions are covered by the join keys.
- Keep the existing inner join when the projection references right columns or right-side uniqueness cannot be proven.
- Update TPC-H plan snapshots for Q2, Q17, and Q20.
## Are these changes tested?
Yes.
- `cargo test -p datafusion-optimizer eliminate_join --lib`
- `INCLUDE_TPCH=true cargo test -p datafusion-sqllogictest --test sqllogictests -- tpch`
- `cargo fmt --all`
- `ci/scripts/rust_clippy.sh`
- `./dev/rust_lint.sh`
## Are there any user-facing changes?
No API changes. Eligible logical and physical plans can now show `LeftSemi` instead of `Inner` for projected joins against unique scalar aggregate subquery results.
--
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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan closed pull request #21971: perf: optimize scalar aggregate joins as semi joins
URL: https://github.com/apache/datafusion/pull/21971
--
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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359177674
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4359167733-1953-2bclk 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 scalar-subquery-semi-join (c069c6f22ce0cf662fc5ea8b58c1a63bb1e36b64) to d59bc72 (merge-base) [diff](https://github.com/apache/datafusion/compare/d59bc7236bf734492789fca3fabd77d7b8ace0c8..c069c6f22ce0cf662fc5ea8b58c1a63bb1e36b64) 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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359665730
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359646884)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4359646884-1958-kmzhz 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 scalar-subquery-semi-join (da59e3aca2557fc9ad27ea3159557d702484a57c) to d59bc72 (merge-base) [diff](https://github.com/apache/datafusion/compare/d59bc7236bf734492789fca3fabd77d7b8ace0c8..da59e3aca2557fc9ad27ea3159557d702484a57c) 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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359228210
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733)
**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 scalar-subquery-semi-join
--------------------
Benchmark tpch_sf10.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโ
โ Query โ HEAD โ scalar-subquery-semi-join โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 1 โ 327.41 / 328.62 ยฑ0.84 / 329.47 ms โ 323.14 / 325.44 ยฑ1.21 / 326.55 ms โ no change โ
โ QQuery 2 โ 126.43 / 128.58 ยฑ2.10 / 132.10 ms โ 124.34 / 128.33 ยฑ2.56 / 131.90 ms โ no change โ
โ QQuery 3 โ 265.04 / 269.93 ยฑ2.98 / 273.65 ms โ 270.25 / 274.85 ยฑ3.54 / 281.13 ms โ no change โ
โ QQuery 4 โ 133.56 / 135.63 ยฑ1.82 / 138.87 ms โ 135.08 / 136.42 ยฑ0.87 / 137.70 ms โ no change โ
โ QQuery 5 โ 388.38 / 392.76 ยฑ2.42 / 395.42 ms โ 392.23 / 399.40 ยฑ6.48 / 408.90 ms โ no change โ
โ QQuery 6 โ 132.79 / 133.06 ยฑ0.30 / 133.63 ms โ 132.66 / 133.39 ยฑ0.51 / 133.89 ms โ no change โ
โ QQuery 7 โ 508.12 / 514.60 ยฑ10.43 / 535.31 ms โ 507.92 / 517.72 ยฑ4.97 / 520.72 ms โ no change โ
โ QQuery 8 โ 425.13 / 437.88 ยฑ10.68 / 454.40 ms โ 434.33 / 440.67 ยฑ4.52 / 446.27 ms โ no change โ
โ QQuery 9 โ 604.93 / 611.04 ยฑ4.78 / 618.39 ms โ 614.89 / 620.48 ยฑ6.07 / 629.86 ms โ no change โ
โ QQuery 10 โ 318.84 / 325.47 ยฑ6.79 / 337.14 ms โ 325.20 / 329.83 ยฑ4.73 / 337.23 ms โ no change โ
โ QQuery 11 โ 94.59 / 96.15 ยฑ1.27 / 97.55 ms โ 96.10 / 99.16 ยฑ2.74 / 102.89 ms โ no change โ
โ QQuery 12 โ 193.56 / 203.83 ยฑ16.59 / 236.88 ms โ 192.81 / 197.54 ยฑ4.98 / 204.31 ms โ no change โ
โ QQuery 13 โ 294.07 / 309.33 ยฑ16.60 / 341.48 ms โ 288.30 / 301.78 ยฑ8.51 / 314.89 ms โ no change โ
โ QQuery 14 โ 180.91 / 184.03 ยฑ3.93 / 191.69 ms โ 180.94 / 184.40 ยฑ4.93 / 194.17 ms โ no change โ
โ QQuery 15 โ 328.53 / 333.04 ยฑ5.77 / 344.44 ms โ 328.79 / 330.24 ยฑ1.55 / 332.16 ms โ no change โ
โ QQuery 16 โ 74.79 / 77.24 ยฑ2.02 / 80.85 ms โ 75.20 / 76.64 ยฑ2.32 / 81.26 ms โ no change โ
โ QQuery 17 โ 684.82 / 692.90 ยฑ6.38 / 703.77 ms โ 688.33 / 694.82 ยฑ7.83 / 709.10 ms โ no change โ
โ QQuery 18 โ 706.94 / 734.43 ยฑ16.16 / 755.08 ms โ 711.15 / 730.66 ยฑ14.11 / 752.64 ms โ no change โ
โ QQuery 19 โ 268.81 / 279.15 ยฑ18.04 / 315.09 ms โ 268.17 / 276.28 ยฑ14.20 / 304.60 ms โ no change โ
โ QQuery 20 โ 301.32 / 305.81 ยฑ2.49 / 308.73 ms โ 298.24 / 304.98 ยฑ8.76 / 321.79 ms โ no change โ
โ QQuery 21 โ 733.01 / 744.85 ยฑ7.28 / 755.12 ms โ 748.43 / 750.64 ยฑ3.41 / 757.40 ms โ no change โ
โ QQuery 22 โ 64.57 / 67.25 ยฑ1.34 / 68.03 ms โ 64.46 / 67.99 ยฑ1.90 / 70.06 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 7305.57ms โ
โ Total Time (scalar-subquery-semi-join) โ 7321.65ms โ
โ Average Time (HEAD) โ 332.07ms โ
โ Average Time (scalar-subquery-semi-join) โ 332.80ms โ
โ Queries Faster โ 0 โ
โ Queries Slower โ 0 โ
โ Queries with No Change โ 22 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpch10 โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 40.0s |
| Peak memory | 12.4 GiB |
| Avg memory | 8.4 GiB |
| CPU user | 391.4s |
| CPU sys | 20.4s |
| Peak spill | 0 B |
**tpch10 โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 40.0s |
| Peak memory | 12.4 GiB |
| Avg memory | 8.5 GiB |
| CPU user | 393.1s |
| CPU sys | 20.2s |
| 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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359177699
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4359167733-1952-brdrq 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 scalar-subquery-semi-join (c069c6f22ce0cf662fc5ea8b58c1a63bb1e36b64) to d59bc72 (merge-base) [diff](https://github.com/apache/datafusion/compare/d59bc7236bf734492789fca3fabd77d7b8ace0c8..c069c6f22ce0cf662fc5ea8b58c1a63bb1e36b64) using: tpch10
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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359223994
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733)
**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 scalar-subquery-semi-join
--------------------
Benchmark tpch_sf1.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโ
โ Query โ HEAD โ scalar-subquery-semi-join โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 1 โ 40.19 / 41.61 ยฑ1.30 / 43.35 ms โ 40.97 / 42.15 ยฑ0.89 / 43.51 ms โ no change โ
โ QQuery 2 โ 21.20 / 21.32 ยฑ0.13 / 21.49 ms โ 20.91 / 21.28 ยฑ0.21 / 21.54 ms โ no change โ
โ QQuery 3 โ 36.24 / 38.54 ยฑ1.38 / 40.13 ms โ 36.23 / 38.69 ยฑ1.69 / 40.95 ms โ no change โ
โ QQuery 4 โ 18.19 / 18.89 ยฑ0.57 / 19.83 ms โ 18.41 / 18.79 ยฑ0.27 / 19.22 ms โ no change โ
โ QQuery 5 โ 44.03 / 45.08 ยฑ0.96 / 46.40 ms โ 44.04 / 46.46 ยฑ1.25 / 47.64 ms โ no change โ
โ QQuery 6 โ 17.90 / 18.43 ยฑ0.43 / 18.97 ms โ 17.27 / 17.69 ยฑ0.29 / 18.15 ms โ no change โ
โ QQuery 7 โ 49.22 / 50.52 ยฑ0.90 / 51.77 ms โ 50.68 / 53.27 ยฑ1.46 / 55.23 ms โ 1.05x slower โ
โ QQuery 8 โ 47.43 / 47.65 ยฑ0.16 / 47.90 ms โ 47.18 / 47.43 ยฑ0.25 / 47.89 ms โ no change โ
โ QQuery 9 โ 51.59 / 52.20 ยฑ0.56 / 53.08 ms โ 52.82 / 53.32 ยฑ0.55 / 54.13 ms โ no change โ
โ QQuery 10 โ 65.69 / 66.62 ยฑ0.57 / 67.46 ms โ 65.99 / 66.66 ยฑ0.98 / 68.60 ms โ no change โ
โ QQuery 11 โ 14.26 / 14.53 ยฑ0.28 / 15.00 ms โ 13.98 / 14.34 ยฑ0.22 / 14.58 ms โ no change โ
โ QQuery 12 โ 26.44 / 26.76 ยฑ0.20 / 26.97 ms โ 26.56 / 27.13 ยฑ0.50 / 27.76 ms โ no change โ
โ QQuery 13 โ 35.81 / 36.66 ยฑ0.64 / 37.63 ms โ 35.29 / 36.64 ยฑ0.87 / 37.72 ms โ no change โ
โ QQuery 14 โ 26.80 / 27.00 ยฑ0.11 / 27.16 ms โ 27.09 / 27.73 ยฑ1.12 / 29.96 ms โ no change โ
โ QQuery 15 โ 33.29 / 34.23 ยฑ1.45 / 37.12 ms โ 33.48 / 34.07 ยฑ0.73 / 35.50 ms โ no change โ
โ QQuery 16 โ 15.36 / 15.53 ยฑ0.11 / 15.70 ms โ 15.37 / 15.57 ยฑ0.13 / 15.73 ms โ no change โ
โ QQuery 17 โ 78.34 / 78.67 ยฑ0.47 / 79.57 ms โ 77.54 / 78.07 ยฑ0.39 / 78.56 ms โ no change โ
โ QQuery 18 โ 68.70 / 69.69 ยฑ0.98 / 71.47 ms โ 69.96 / 70.76 ยฑ0.72 / 72.03 ms โ no change โ
โ QQuery 19 โ 37.90 / 38.21 ยฑ0.53 / 39.26 ms โ 37.63 / 38.15 ยฑ0.43 / 38.82 ms โ no change โ
โ QQuery 20 โ 39.05 / 39.41 ยฑ0.30 / 39.90 ms โ 39.21 / 39.80 ยฑ0.53 / 40.76 ms โ no change โ
โ QQuery 21 โ 61.43 / 62.05 ยฑ0.87 / 63.73 ms โ 62.55 / 65.17 ยฑ2.01 / 67.16 ms โ 1.05x slower โ
โ QQuery 22 โ 23.66 / 24.33 ยฑ0.56 / 25.10 ms โ 24.09 / 24.57 ยฑ0.43 / 25.25 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 867.95ms โ
โ Total Time (scalar-subquery-semi-join) โ 877.76ms โ
โ Average Time (HEAD) โ 39.45ms โ
โ Average Time (scalar-subquery-semi-join) โ 39.90ms โ
โ Queries Faster โ 0 โ
โ Queries Slower โ 2 โ
โ Queries with No Change โ 20 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpch โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 5.0s |
| Peak memory | 5.4 GiB |
| Avg memory | 4.9 GiB |
| CPU user | 32.4s |
| CPU sys | 2.3s |
| Peak spill | 0 B |
**tpch โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 5.0s |
| Peak memory | 5.4 GiB |
| Avg memory | 5.0 GiB |
| CPU user | 32.7s |
| CPU sys | 2.2s |
| 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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359646884
run benchmark tpcds
--
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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733
run benchmark tpch tpch10 tpcds
--
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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359240903
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733)
**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 scalar-subquery-semi-join
--------------------
Benchmark tpcds_sf1.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโ
โ Query โ HEAD โ scalar-subquery-semi-join โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 1 โ 7.15 / 7.76 ยฑ0.91 / 9.57 ms โ 7.34 / 7.81 ยฑ0.89 / 9.59 ms โ no change โ
โ QQuery 2 โ 146.95 / 147.23 ยฑ0.25 / 147.63 ms โ 144.80 / 146.45 ยฑ1.16 / 147.73 ms โ no change โ
โ QQuery 3 โ 114.86 / 116.08 ยฑ0.87 / 116.94 ms โ 113.99 / 115.25 ยฑ1.01 / 116.87 ms โ no change โ
โ QQuery 4 โ 1360.59 / 1425.55 ยฑ56.24 / 1496.72 ms โ 1224.21 / 1237.44 ยฑ15.73 / 1268.05 ms โ +1.15x faster โ
โ QQuery 5 โ 174.12 / 176.03 ยฑ1.07 / 177.09 ms โ 174.49 / 176.00 ยฑ1.24 / 177.61 ms โ no change โ
โ QQuery 6 โ 138.20 / 144.90 ยฑ6.23 / 156.29 ms โ 127.35 / 130.72 ยฑ2.22 / 134.23 ms โ +1.11x faster โ
โ QQuery 7 โ 322.44 / 325.84 ยฑ1.86 / 328.15 ms โ 318.03 / 319.70 ยฑ1.22 / 321.39 ms โ no change โ
โ QQuery 8 โ 113.62 / 114.62 ยฑ0.60 / 115.26 ms โ 113.64 / 114.01 ยฑ0.39 / 114.73 ms โ no change โ
โ QQuery 9 โ 96.95 / 100.45 ยฑ2.45 / 102.76 ms โ 91.37 / 95.61 ยฑ2.20 / 97.44 ms โ no change โ
โ QQuery 10 โ 102.99 / 105.46 ยฑ2.51 / 110.04 ms โ 102.37 / 104.26 ยฑ2.33 / 108.83 ms โ no change โ
โ QQuery 11 โ 884.26 / 949.49 ยฑ77.74 / 1079.88 ms โ 884.69 / 945.01 ยฑ60.57 / 1041.02 ms โ no change โ
โ QQuery 12 โ 46.75 / 47.07 ยฑ0.17 / 47.26 ms โ 45.95 / 46.74 ยฑ1.36 / 49.46 ms โ no change โ
โ QQuery 13 โ 391.65 / 403.75 ยฑ10.24 / 417.92 ms โ 397.57 / 406.77 ยฑ4.70 / 410.02 ms โ no change โ
โ QQuery 14 โ 1044.67 / 1048.28 ยฑ2.78 / 1051.59 ms โ 1025.53 / 1035.34 ยฑ6.41 / 1043.67 ms โ no change โ
โ QQuery 15 โ 14.63 / 15.06 ยฑ0.30 / 15.50 ms โ 14.29 / 14.67 ยฑ0.25 / 14.94 ms โ no change โ
โ QQuery 16 โ 7.34 / 7.53 ยฑ0.18 / 7.84 ms โ 7.36 / 7.48 ยฑ0.18 / 7.85 ms โ no change โ
โ QQuery 17 โ 204.69 / 205.51 ยฑ0.86 / 207.03 ms โ 200.18 / 201.61 ยฑ1.04 / 203.02 ms โ no change โ
โ QQuery 18 โ 125.03 / 126.15 ยฑ0.93 / 127.58 ms โ 125.22 / 128.07 ยฑ2.24 / 130.91 ms โ no change โ
โ QQuery 19 โ 153.83 / 156.51 ยฑ2.51 / 161.11 ms โ 163.63 / 165.76 ยฑ1.89 / 168.14 ms โ 1.06x slower โ
โ QQuery 20 โ 12.97 / 13.59 ยฑ0.42 / 14.19 ms โ 13.73 / 13.97 ยฑ0.15 / 14.18 ms โ no change โ
โ QQuery 21 โ 19.33 / 19.63 ยฑ0.24 / 19.92 ms โ 20.13 / 20.85 ยฑ0.68 / 22.13 ms โ 1.06x slower โ
โ QQuery 22 โ 472.97 / 474.11 ยฑ0.83 / 475.00 ms โ 472.02 / 507.41 ยฑ19.26 / 522.02 ms โ 1.07x slower โ
โ QQuery 23 โ 1032.52 / 1110.78 ยฑ63.40 / 1176.11 ms โ 1006.38 / 1022.54 ยฑ21.35 / 1062.40 ms โ +1.09x faster โ
โ QQuery 24 โ 640.65 / 660.04 ยฑ17.13 / 683.13 ms โ 629.91 / 633.34 ยฑ2.99 / 637.70 ms โ no change โ
โ QQuery 25 โ 307.58 / 309.64 ยฑ2.29 / 313.37 ms โ 302.57 / 304.12 ยฑ1.53 / 307.03 ms โ no change โ
โ QQuery 26 โ 76.33 / 78.11 ยฑ1.97 / 81.87 ms โ 74.66 / 75.28 ยฑ0.37 / 75.79 ms โ no change โ
โ QQuery 27 โ 7.08 / 7.17 ยฑ0.12 / 7.40 ms โ 6.95 / 7.06 ยฑ0.10 / 7.24 ms โ no change โ
โ QQuery 28 โ 148.64 / 150.25 ยฑ2.45 / 155.12 ms โ 146.81 / 148.30 ยฑ1.18 / 150.32 ms โ no change โ
โ QQuery 29 โ 245.55 / 247.80 ยฑ2.23 / 251.93 ms โ 243.39 / 244.91 ยฑ2.34 / 249.48 ms โ no change โ
โ QQuery 30 โ 41.87 / 42.68 ยฑ0.46 / 43.12 ms โ 42.40 / 43.15 ยฑ0.40 / 43.54 ms โ no change โ
โ QQuery 31 โ 165.66 / 167.55 ยฑ1.42 / 169.93 ms โ 168.00 / 171.64 ยฑ2.72 / 174.80 ms โ no change โ
โ QQuery 32 โ 13.14 / 13.55 ยฑ0.36 / 14.09 ms โ 13.49 / 13.88 ยฑ0.30 / 14.31 ms โ no change โ
โ QQuery 33 โ 138.68 / 141.14 ยฑ2.82 / 146.06 ms โ 141.42 / 143.17 ยฑ2.23 / 147.47 ms โ no change โ
โ QQuery 34 โ 6.99 / 7.16 ยฑ0.19 / 7.52 ms โ 7.40 / 7.59 ยฑ0.14 / 7.77 ms โ 1.06x slower โ
โ QQuery 35 โ 102.15 / 102.75 ยฑ0.56 / 103.76 ms โ 102.62 / 104.80 ยฑ2.27 / 109.19 ms โ no change โ
โ QQuery 36 โ 6.55 / 6.75 ยฑ0.18 / 7.07 ms โ 7.06 / 7.24 ยฑ0.11 / 7.39 ms โ 1.07x slower โ
โ QQuery 37 โ 8.32 / 8.40 ยฑ0.06 / 8.47 ms โ 8.58 / 8.82 ยฑ0.19 / 9.10 ms โ no change โ
โ QQuery 38 โ 85.90 / 90.19 ยฑ4.50 / 98.73 ms โ 89.51 / 91.06 ยฑ0.82 / 91.87 ms โ no change โ
โ QQuery 39 โ 116.53 / 118.36 ยฑ1.33 / 120.26 ms โ 116.77 / 120.79 ยฑ5.07 / 130.70 ms โ no change โ
โ QQuery 40 โ 94.28 / 98.82 ยฑ3.91 / 105.66 ms โ 94.60 / 96.17 ยฑ1.51 / 98.88 ms โ no change โ
โ QQuery 41 โ 14.65 / 14.84 ยฑ0.14 / 15.09 ms โ 14.51 / 14.78 ยฑ0.29 / 15.32 ms โ no change โ
โ QQuery 42 โ 108.33 / 110.06 ยฑ2.15 / 113.88 ms โ 107.14 / 108.83 ยฑ1.07 / 110.48 ms โ no change โ
โ QQuery 43 โ 5.72 / 5.91 ยฑ0.12 / 6.07 ms โ 5.81 / 6.63 ยฑ1.55 / 9.72 ms โ 1.12x slower โ
โ QQuery 44 โ 10.98 / 12.16 ยฑ2.31 / 16.77 ms โ 10.86 / 10.92 ยฑ0.09 / 11.10 ms โ +1.11x faster โ
โ QQuery 45 โ 45.39 / 45.61 ยฑ0.20 / 45.98 ms โ 43.38 / 44.18 ยฑ0.56 / 44.72 ms โ no change โ
โ QQuery 46 โ 8.57 / 8.70 ยฑ0.16 / 9.01 ms โ 8.52 / 8.64 ยฑ0.20 / 9.03 ms โ no change โ
โ QQuery 47 โ 698.30 / 790.16 ยฑ66.84 / 855.78 ms โ 662.56 / 677.03 ยฑ11.62 / 697.59 ms โ +1.17x faster โ
โ QQuery 48 โ 281.08 / 291.82 ยฑ6.35 / 300.72 ms โ 272.61 / 274.36 ยฑ2.09 / 278.34 ms โ +1.06x faster โ
โ QQuery 49 โ 245.91 / 247.19 ยฑ1.46 / 250.02 ms โ 242.31 / 243.02 ยฑ0.47 / 243.72 ms โ no change โ
โ QQuery 50 โ 189.81 / 196.22 ยฑ3.77 / 200.95 ms โ 180.73 / 186.03 ยฑ3.97 / 192.49 ms โ +1.05x faster โ
โ QQuery 51 โ 172.34 / 174.83 ยฑ2.07 / 177.56 ms โ 170.57 / 173.01 ยฑ1.33 / 174.57 ms โ no change โ
โ QQuery 52 โ 108.85 / 110.99 ยฑ3.31 / 117.50 ms โ 107.34 / 109.44 ยฑ2.20 / 113.55 ms โ no change โ
โ QQuery 53 โ 103.28 / 104.12 ยฑ0.56 / 104.91 ms โ 103.50 / 104.39 ยฑ0.97 / 106.17 ms โ no change โ
โ QQuery 54 โ 145.46 / 147.52 ยฑ2.18 / 151.61 ms โ 154.40 / 156.00 ยฑ1.89 / 159.58 ms โ 1.06x slower โ
โ QQuery 55 โ 107.21 / 108.19 ยฑ0.67 / 109.14 ms โ 110.05 / 110.69 ยฑ0.62 / 111.55 ms โ no change โ
โ QQuery 56 โ 139.42 / 140.95 ยฑ0.84 / 141.76 ms โ 145.02 / 147.30 ยฑ1.92 / 150.84 ms โ no change โ
โ QQuery 57 โ 164.52 / 166.70 ยฑ2.00 / 170.40 ms โ 172.27 / 173.88 ยฑ1.68 / 176.77 ms โ no change โ
โ QQuery 58 โ 245.92 / 247.72 ยฑ1.55 / 250.17 ms โ 248.55 / 250.76 ยฑ1.57 / 252.28 ms โ no change โ
โ QQuery 59 โ 200.29 / 206.51 ยฑ3.18 / 208.60 ms โ 192.75 / 194.73 ยฑ2.86 / 200.30 ms โ +1.06x faster โ
โ QQuery 60 โ 149.13 / 150.57 ยฑ1.60 / 153.37 ms โ 141.00 / 142.08 ยฑ0.83 / 143.57 ms โ +1.06x faster โ
โ QQuery 61 โ 14.42 / 14.60 ยฑ0.19 / 14.92 ms โ 13.21 / 13.37 ยฑ0.19 / 13.73 ms โ +1.09x faster โ
โ QQuery 62 โ 890.00 / 913.85 ยฑ23.56 / 945.63 ms โ 852.52 / 855.93 ยฑ2.89 / 859.81 ms โ +1.07x faster โ
โ QQuery 63 โ 103.52 / 104.26 ยฑ0.58 / 105.18 ms โ 102.16 / 104.25 ยฑ1.92 / 107.73 ms โ no change โ
โ QQuery 64 โ 630.33 / 636.71 ยฑ3.24 / 638.97 ms โ 624.45 / 627.27 ยฑ2.24 / 630.74 ms โ no change โ
โ QQuery 65 โ 244.23 / 248.28 ยฑ3.31 / 252.93 ms โ 242.41 / 247.68 ยฑ2.96 / 251.51 ms โ no change โ
โ QQuery 66 โ 216.26 / 226.99 ยฑ11.28 / 242.92 ms โ 216.27 / 226.74 ยฑ10.17 / 241.57 ms โ no change โ
โ QQuery 67 โ 296.32 / 315.43 ยฑ13.06 / 331.18 ms โ 288.52 / 300.56 ยฑ12.69 / 319.17 ms โ no change โ
โ QQuery 68 โ 9.57 / 9.76 ยฑ0.11 / 9.89 ms โ 8.67 / 8.80 ยฑ0.15 / 9.08 ms โ +1.11x faster โ
โ QQuery 69 โ 102.09 / 104.32 ยฑ1.63 / 106.29 ms โ 97.25 / 98.22 ยฑ0.85 / 99.64 ms โ +1.06x faster โ
โ QQuery 70 โ 317.97 / 332.36 ยฑ9.41 / 344.42 ms โ 327.36 / 338.42 ยฑ7.68 / 348.67 ms โ no change โ
โ QQuery 71 โ 134.43 / 139.03 ยฑ4.56 / 147.58 ms โ 137.40 / 139.53 ยฑ1.31 / 141.29 ms โ no change โ
โ QQuery 72 โ 536.59 / 548.38 ยฑ8.25 / 559.75 ms โ 553.50 / 564.49 ยฑ8.29 / 577.06 ms โ no change โ
โ QQuery 73 โ 6.79 / 6.95 ยฑ0.17 / 7.28 ms โ 7.01 / 7.15 ยฑ0.11 / 7.35 ms โ no change โ
โ QQuery 74 โ 560.74 / 568.89 ยฑ7.44 / 581.91 ms โ 569.49 / 590.48 ยฑ16.21 / 617.96 ms โ no change โ
โ QQuery 75 โ 267.67 / 269.40 ยฑ1.17 / 271.04 ms โ 265.36 / 267.72 ยฑ2.46 / 272.48 ms โ no change โ
โ QQuery 76 โ 130.87 / 131.64 ยฑ0.42 / 132.05 ms โ 129.57 / 129.96 ยฑ0.37 / 130.65 ms โ no change โ
โ QQuery 77 โ 188.78 / 190.57 ยฑ2.31 / 195.05 ms โ 187.50 / 188.82 ยฑ1.28 / 191.12 ms โ no change โ
โ QQuery 78 โ 310.11 / 312.04 ยฑ1.25 / 313.82 ms โ 305.59 / 308.57 ยฑ2.12 / 311.71 ms โ no change โ
โ QQuery 79 โ 228.01 / 232.26 ยฑ3.36 / 237.61 ms โ 228.38 / 231.44 ยฑ1.99 / 233.72 ms โ no change โ
โ QQuery 80 โ 306.57 / 308.79 ยฑ1.84 / 312.11 ms โ 299.71 / 302.09 ยฑ1.76 / 305.06 ms โ no change โ
โ QQuery 81 โ 26.93 / 27.83 ยฑ0.57 / 28.65 ms โ 25.84 / 27.03 ยฑ2.29 / 31.62 ms โ no change โ
โ QQuery 82 โ 42.08 / 42.34 ยฑ0.21 / 42.69 ms โ 39.09 / 39.73 ยฑ0.56 / 40.47 ms โ +1.07x faster โ
โ QQuery 83 โ 37.82 / 38.24 ยฑ0.32 / 38.74 ms โ 34.90 / 35.59 ยฑ0.47 / 36.33 ms โ +1.07x faster โ
โ QQuery 84 โ 48.35 / 49.58 ยฑ1.97 / 53.51 ms โ 46.38 / 47.13 ยฑ0.69 / 48.07 ms โ no change โ
โ QQuery 85 โ 148.57 / 150.17 ยฑ1.31 / 151.79 ms โ 142.14 / 143.93 ยฑ1.88 / 147.44 ms โ no change โ
โ QQuery 86 โ 39.59 / 39.96 ยฑ0.29 / 40.35 ms โ 36.91 / 37.56 ยฑ0.40 / 37.97 ms โ +1.06x faster โ
โ QQuery 87 โ 4.10 / 4.20 ยฑ0.14 / 4.47 ms โ 3.49 / 3.60 ยฑ0.16 / 3.92 ms โ +1.17x faster โ
โ QQuery 88 โ 107.22 / 108.18 ยฑ0.82 / 109.67 ms โ 99.62 / 101.11 ยฑ2.04 / 105.11 ms โ +1.07x faster โ
โ QQuery 89 โ 122.75 / 127.25 ยฑ5.92 / 138.82 ms โ 117.15 / 118.24 ยฑ1.36 / 120.73 ms โ +1.08x faster โ
โ QQuery 90 โ 24.08 / 24.84 ยฑ0.43 / 25.27 ms โ 23.42 / 23.80 ยฑ0.29 / 24.22 ms โ no change โ
โ QQuery 91 โ 62.56 / 63.50 ยฑ0.79 / 64.49 ms โ 60.12 / 62.00 ยฑ2.66 / 67.27 ms โ no change โ
โ QQuery 92 โ 60.10 / 61.40 ยฑ1.94 / 65.24 ms โ 58.81 / 59.25 ยฑ0.52 / 60.26 ms โ no change โ
โ QQuery 93 โ 174.90 / 176.10 ยฑ0.84 / 177.03 ms โ 169.41 / 170.87 ยฑ1.09 / 172.66 ms โ no change โ
โ QQuery 94 โ 62.97 / 65.72 ยฑ3.27 / 72.04 ms โ 63.36 / 63.69 ยฑ0.38 / 64.24 ms โ no change โ
โ QQuery 95 โ 113.97 / 115.09 ยฑ0.77 / 115.92 ms โ 115.47 / 116.53 ยฑ1.32 / 119.09 ms โ no change โ
โ QQuery 96 โ 71.33 / 72.68 ยฑ0.83 / 73.55 ms โ 70.07 / 71.90 ยฑ1.24 / 73.08 ms โ no change โ
โ QQuery 97 โ 114.32 / 116.04 ยฑ1.02 / 117.39 ms โ 114.37 / 120.31 ยฑ3.21 / 124.00 ms โ no change โ
โ QQuery 98 โ 153.21 / 155.46 ยฑ1.17 / 156.38 ms โ 149.66 / 150.99 ยฑ0.88 / 152.07 ms โ no change โ
โ QQuery 99 โ 10707.19 / 10852.52 ยฑ131.42 / 11031.81 ms โ 10793.69 / 10916.26 ยฑ99.32 / 11079.54 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 30678.11ms โ
โ Total Time (scalar-subquery-semi-join) โ 30182.55ms โ
โ Average Time (HEAD) โ 309.88ms โ
โ Average Time (scalar-subquery-semi-join) โ 304.87ms โ
โ Queries Faster โ 19 โ
โ Queries Slower โ 7 โ
โ Queries with No Change โ 73 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpcds โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 155.0s |
| Peak memory | 6.4 GiB |
| Avg memory | 5.7 GiB |
| CPU user | 243.1s |
| CPU sys | 8.3s |
| Peak spill | 0 B |
**tpcds โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 155.0s |
| Peak memory | 6.4 GiB |
| Avg memory | 5.7 GiB |
| CPU user | 238.8s |
| CPU sys | 8.1s |
| 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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359753236
๐ค Benchmark completed (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359646884)
**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 scalar-subquery-semi-join
--------------------
Benchmark tpcds_sf1.json
--------------------
โโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโโโ
โ Query โ HEAD โ scalar-subquery-semi-join โ Change โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ QQuery 1 โ 6.90 / 7.37 ยฑ0.85 / 9.06 ms โ 6.79 / 7.25 ยฑ0.85 / 8.94 ms โ no change โ
โ QQuery 2 โ 144.20 / 144.70 ยฑ0.48 / 145.60 ms โ 144.27 / 144.70 ยฑ0.34 / 145.14 ms โ no change โ
โ QQuery 3 โ 113.76 / 115.75 ยฑ2.96 / 121.60 ms โ 114.11 / 115.13 ยฑ1.21 / 117.46 ms โ no change โ
โ QQuery 4 โ 1296.12 / 1298.54 ยฑ2.39 / 1302.86 ms โ 1292.19 / 1296.36 ยฑ3.84 / 1302.41 ms โ no change โ
โ QQuery 5 โ 174.49 / 175.31 ยฑ1.27 / 177.83 ms โ 174.27 / 175.63 ยฑ1.37 / 177.66 ms โ no change โ
โ QQuery 6 โ 133.07 / 137.47 ยฑ2.34 / 139.56 ms โ 133.89 / 139.97 ยฑ5.64 / 149.65 ms โ no change โ
โ QQuery 7 โ 321.56 / 323.32 ยฑ1.51 / 325.32 ms โ 319.26 / 322.02 ยฑ2.45 / 325.02 ms โ no change โ
โ QQuery 8 โ 113.07 / 114.94 ยฑ1.43 / 117.28 ms โ 114.15 / 115.36 ยฑ0.96 / 116.82 ms โ no change โ
โ QQuery 9 โ 91.57 / 98.43 ยฑ4.13 / 102.19 ms โ 95.92 / 101.86 ยฑ4.73 / 110.45 ms โ no change โ
โ QQuery 10 โ 102.53 / 103.15 ยฑ0.61 / 104.31 ms โ 102.84 / 103.32 ยฑ0.44 / 104.04 ms โ no change โ
โ QQuery 11 โ 888.73 / 897.16 ยฑ5.85 / 905.97 ms โ 889.08 / 903.70 ยฑ8.00 / 912.77 ms โ no change โ
โ QQuery 12 โ 43.57 / 43.85 ยฑ0.21 / 44.10 ms โ 43.26 / 43.57 ยฑ0.27 / 43.93 ms โ no change โ
โ QQuery 13 โ 388.78 / 391.98 ยฑ2.33 / 395.57 ms โ 388.40 / 390.20 ยฑ2.00 / 393.55 ms โ no change โ
โ QQuery 14 โ 1024.92 / 1031.10 ยฑ4.85 / 1038.87 ms โ 1176.64 / 1184.12 ยฑ8.59 / 1200.33 ms โ 1.15x slower โ
โ QQuery 15 โ 14.39 / 14.67 ยฑ0.16 / 14.82 ms โ 14.23 / 14.67 ยฑ0.35 / 15.22 ms โ no change โ
โ QQuery 16 โ 7.43 / 7.57 ยฑ0.23 / 8.03 ms โ 7.28 / 7.42 ยฑ0.08 / 7.53 ms โ no change โ
โ QQuery 17 โ 202.74 / 203.77 ยฑ0.54 / 204.30 ms โ 202.07 / 204.00 ยฑ2.08 / 207.98 ms โ no change โ
โ QQuery 18 โ 124.09 / 125.01 ยฑ0.79 / 126.47 ms โ 123.14 / 123.85 ยฑ0.56 / 124.61 ms โ no change โ
โ QQuery 19 โ 153.38 / 154.12 ยฑ0.61 / 155.23 ms โ 153.54 / 156.31 ยฑ1.57 / 158.07 ms โ no change โ
โ QQuery 20 โ 13.04 / 13.49 ยฑ0.27 / 13.85 ms โ 12.85 / 13.34 ยฑ0.29 / 13.72 ms โ no change โ
โ QQuery 21 โ 18.87 / 19.52 ยฑ0.88 / 21.27 ms โ 18.84 / 19.25 ยฑ0.24 / 19.60 ms โ no change โ
โ QQuery 22 โ 469.58 / 476.02 ยฑ4.43 / 482.38 ms โ 468.21 / 472.81 ยฑ3.84 / 478.09 ms โ no change โ
โ QQuery 23 โ 1018.74 / 1022.42 ยฑ3.62 / 1028.28 ms โ 1012.22 / 1019.60 ยฑ4.90 / 1027.63 ms โ no change โ
โ QQuery 24 โ 635.37 / 639.05 ยฑ3.83 / 645.87 ms โ 635.86 / 641.76 ยฑ3.31 / 645.96 ms โ no change โ
โ QQuery 25 โ 305.20 / 307.55 ยฑ1.69 / 310.00 ms โ 306.60 / 308.42 ยฑ1.60 / 310.90 ms โ no change โ
โ QQuery 26 โ 76.69 / 77.09 ยฑ0.35 / 77.63 ms โ 76.23 / 76.59 ยฑ0.20 / 76.83 ms โ no change โ
โ QQuery 27 โ 7.05 / 7.13 ยฑ0.12 / 7.36 ms โ 7.19 / 7.27 ยฑ0.10 / 7.47 ms โ no change โ
โ QQuery 28 โ 147.30 / 149.23 ยฑ2.99 / 155.16 ms โ 147.78 / 148.95 ยฑ1.52 / 151.95 ms โ no change โ
โ QQuery 29 โ 246.40 / 248.06 ยฑ1.68 / 251.11 ms โ 245.96 / 247.84 ยฑ2.88 / 253.55 ms โ no change โ
โ QQuery 30 โ 41.59 / 42.26 ยฑ0.53 / 43.02 ms โ 41.63 / 41.84 ยฑ0.20 / 42.11 ms โ no change โ
โ QQuery 31 โ 165.45 / 166.87 ยฑ1.46 / 169.50 ms โ 165.87 / 167.00 ยฑ0.76 / 168.18 ms โ no change โ
โ QQuery 32 โ 13.35 / 13.53 ยฑ0.17 / 13.79 ms โ 13.19 / 13.48 ยฑ0.35 / 14.14 ms โ no change โ
โ QQuery 33 โ 137.65 / 141.42 ยฑ3.09 / 146.58 ms โ 139.36 / 140.43 ยฑ0.61 / 141.15 ms โ no change โ
โ QQuery 34 โ 6.92 / 7.06 ยฑ0.23 / 7.53 ms โ 6.93 / 7.09 ยฑ0.17 / 7.42 ms โ no change โ
โ QQuery 35 โ 100.83 / 102.10 ยฑ1.02 / 103.64 ms โ 100.61 / 102.52 ยฑ1.57 / 104.15 ms โ no change โ
โ QQuery 36 โ 6.53 / 6.89 ยฑ0.23 / 7.21 ms โ 6.51 / 6.84 ยฑ0.17 / 7.00 ms โ no change โ
โ QQuery 37 โ 8.51 / 8.75 ยฑ0.12 / 8.87 ms โ 8.30 / 8.38 ยฑ0.07 / 8.51 ms โ no change โ
โ QQuery 38 โ 85.70 / 88.35 ยฑ2.21 / 90.77 ms โ 86.16 / 88.28 ยฑ2.82 / 93.85 ms โ no change โ
โ QQuery 39 โ 115.85 / 119.92 ยฑ5.64 / 130.96 ms โ 116.87 / 117.58 ยฑ0.63 / 118.58 ms โ no change โ
โ QQuery 40 โ 94.66 / 96.81 ยฑ1.97 / 100.12 ms โ 93.30 / 94.85 ยฑ1.28 / 96.49 ms โ no change โ
โ QQuery 41 โ 14.40 / 14.59 ยฑ0.25 / 15.08 ms โ 14.38 / 14.61 ยฑ0.29 / 15.19 ms โ no change โ
โ QQuery 42 โ 107.60 / 108.67 ยฑ1.23 / 111.05 ms โ 107.49 / 108.55 ยฑ0.63 / 109.18 ms โ no change โ
โ QQuery 43 โ 5.62 / 5.83 ยฑ0.18 / 6.17 ms โ 5.82 / 6.12 ยฑ0.50 / 7.12 ms โ no change โ
โ QQuery 44 โ 11.08 / 11.17 ยฑ0.13 / 11.43 ms โ 10.83 / 12.08 ยฑ2.10 / 16.26 ms โ 1.08x slower โ
โ QQuery 45 โ 45.05 / 45.47 ยฑ0.26 / 45.83 ms โ 44.60 / 45.19 ยฑ0.31 / 45.49 ms โ no change โ
โ QQuery 46 โ 8.47 / 8.66 ยฑ0.24 / 9.11 ms โ 8.47 / 8.63 ยฑ0.18 / 8.98 ms โ no change โ
โ QQuery 47 โ 687.27 / 698.27 ยฑ7.47 / 706.05 ms โ 689.14 / 694.08 ยฑ2.85 / 697.89 ms โ no change โ
โ QQuery 48 โ 270.72 / 278.25 ยฑ5.60 / 287.72 ms โ 270.48 / 274.96 ยฑ3.79 / 280.64 ms โ no change โ
โ QQuery 49 โ 244.32 / 245.63 ยฑ1.09 / 247.49 ms โ 243.61 / 244.34 ยฑ0.73 / 245.42 ms โ no change โ
โ QQuery 50 โ 185.18 / 186.90 ยฑ1.65 / 189.75 ms โ 183.17 / 187.21 ยฑ4.84 / 196.60 ms โ no change โ
โ QQuery 51 โ 171.38 / 173.03 ยฑ0.92 / 174.16 ms โ 171.66 / 174.78 ยฑ2.70 / 178.32 ms โ no change โ
โ QQuery 52 โ 107.62 / 108.32 ยฑ0.51 / 109.07 ms โ 107.31 / 108.09 ยฑ0.42 / 108.52 ms โ no change โ
โ QQuery 53 โ 102.79 / 105.15 ยฑ3.39 / 111.87 ms โ 102.62 / 104.78 ยฑ2.47 / 109.40 ms โ no change โ
โ QQuery 54 โ 145.32 / 147.00 ยฑ1.49 / 149.68 ms โ 146.46 / 147.47 ยฑ0.52 / 147.90 ms โ no change โ
โ QQuery 55 โ 106.50 / 107.54 ยฑ0.81 / 108.88 ms โ 107.59 / 108.94 ยฑ1.70 / 112.21 ms โ no change โ
โ QQuery 56 โ 139.45 / 140.72 ยฑ1.28 / 143.15 ms โ 139.32 / 141.81 ยฑ1.70 / 143.65 ms โ no change โ
โ QQuery 57 โ 163.78 / 166.56 ยฑ3.07 / 172.35 ms โ 164.06 / 165.76 ยฑ1.42 / 168.38 ms โ no change โ
โ QQuery 58 โ 244.78 / 245.97 ยฑ1.41 / 248.69 ms โ 242.46 / 244.24 ยฑ1.26 / 246.13 ms โ no change โ
โ QQuery 59 โ 192.21 / 194.74 ยฑ1.89 / 197.59 ms โ 192.45 / 195.71 ยฑ3.18 / 201.10 ms โ no change โ
โ QQuery 60 โ 140.76 / 141.85 ยฑ1.17 / 144.02 ms โ 141.02 / 142.72 ยฑ0.94 / 143.72 ms โ no change โ
โ QQuery 61 โ 13.41 / 13.51 ยฑ0.13 / 13.76 ms โ 13.44 / 14.56 ยฑ1.86 / 18.26 ms โ 1.08x slower โ
โ QQuery 62 โ 861.00 / 865.49 ยฑ3.52 / 871.57 ms โ 855.62 / 862.02 ยฑ5.51 / 869.03 ms โ no change โ
โ QQuery 63 โ 103.05 / 104.66 ยฑ1.93 / 108.42 ms โ 102.59 / 103.47 ยฑ0.59 / 104.35 ms โ no change โ
โ QQuery 64 โ 627.36 / 632.25 ยฑ4.66 / 639.41 ms โ 630.95 / 634.40 ยฑ2.68 / 637.43 ms โ no change โ
โ QQuery 65 โ 244.89 / 247.50 ยฑ2.03 / 250.57 ms โ 245.93 / 252.56 ยฑ6.29 / 263.15 ms โ no change โ
โ QQuery 66 โ 215.69 / 230.75 ยฑ13.51 / 254.01 ms โ 213.79 / 223.71 ยฑ7.55 / 232.27 ms โ no change โ
โ QQuery 67 โ 290.81 / 293.75 ยฑ2.41 / 298.02 ms โ 289.95 / 295.27 ยฑ5.55 / 305.97 ms โ no change โ
โ QQuery 68 โ 8.45 / 8.79 ยฑ0.24 / 9.19 ms โ 8.47 / 8.70 ยฑ0.19 / 8.99 ms โ no change โ
โ QQuery 69 โ 98.36 / 102.01 ยฑ5.67 / 113.27 ms โ 98.40 / 100.93 ยฑ4.15 / 109.20 ms โ no change โ
โ QQuery 70 โ 308.96 / 324.77 ยฑ8.35 / 332.49 ms โ 313.55 / 319.28 ยฑ4.74 / 326.19 ms โ no change โ
โ QQuery 71 โ 133.47 / 134.88 ยฑ1.31 / 137.01 ms โ 132.94 / 137.17 ยฑ5.09 / 145.83 ms โ no change โ
โ QQuery 72 โ 547.50 / 549.90 ยฑ1.82 / 552.15 ms โ 531.11 / 540.55 ยฑ8.23 / 550.76 ms โ no change โ
โ QQuery 73 โ 6.71 / 6.93 ยฑ0.22 / 7.35 ms โ 6.74 / 6.90 ยฑ0.22 / 7.34 ms โ no change โ
โ QQuery 74 โ 563.67 / 570.29 ยฑ4.14 / 576.20 ms โ 566.79 / 575.15 ยฑ4.71 / 580.89 ms โ no change โ
โ QQuery 75 โ 266.02 / 268.58 ยฑ3.16 / 274.73 ms โ 265.03 / 267.15 ยฑ1.38 / 269.00 ms โ no change โ
โ QQuery 76 โ 130.02 / 132.10 ยฑ2.52 / 137.03 ms โ 129.46 / 131.65 ยฑ2.63 / 136.77 ms โ no change โ
โ QQuery 77 โ 188.28 / 189.72 ยฑ2.20 / 194.09 ms โ 188.28 / 189.58 ยฑ1.14 / 191.24 ms โ no change โ
โ QQuery 78 โ 309.35 / 312.34 ยฑ2.39 / 316.64 ms โ 308.54 / 311.79 ยฑ2.50 / 315.63 ms โ no change โ
โ QQuery 79 โ 229.93 / 232.89 ยฑ3.43 / 239.52 ms โ 228.66 / 231.40 ยฑ1.75 / 233.60 ms โ no change โ
โ QQuery 80 โ 300.50 / 303.37 ยฑ1.61 / 304.97 ms โ 299.99 / 303.34 ยฑ3.23 / 307.59 ms โ no change โ
โ QQuery 81 โ 25.94 / 26.22 ยฑ0.16 / 26.44 ms โ 25.51 / 26.22 ยฑ0.69 / 27.45 ms โ no change โ
โ QQuery 82 โ 39.79 / 39.99 ยฑ0.16 / 40.17 ms โ 39.24 / 39.60 ยฑ0.30 / 40.12 ms โ no change โ
โ QQuery 83 โ 35.47 / 36.28 ยฑ1.22 / 38.70 ms โ 35.55 / 35.73 ยฑ0.16 / 35.98 ms โ no change โ
โ QQuery 84 โ 46.25 / 46.49 ยฑ0.14 / 46.63 ms โ 46.26 / 47.20 ยฑ1.42 / 50.01 ms โ no change โ
โ QQuery 85 โ 141.68 / 142.15 ยฑ0.40 / 142.77 ms โ 141.18 / 142.06 ยฑ0.69 / 143.13 ms โ no change โ
โ QQuery 86 โ 37.32 / 38.34 ยฑ1.16 / 40.46 ms โ 36.84 / 37.51 ยฑ0.42 / 38.13 ms โ no change โ
โ QQuery 87 โ 3.47 / 3.58 ยฑ0.17 / 3.92 ms โ 3.49 / 3.65 ยฑ0.25 / 4.14 ms โ no change โ
โ QQuery 88 โ 99.17 / 99.84 ยฑ0.55 / 100.76 ms โ 99.48 / 100.98 ยฑ1.67 / 104.19 ms โ no change โ
โ QQuery 89 โ 116.34 / 120.10 ยฑ5.75 / 131.48 ms โ 116.61 / 119.31 ยฑ4.48 / 128.24 ms โ no change โ
โ QQuery 90 โ 22.40 / 22.98 ยฑ0.62 / 24.11 ms โ 22.34 / 23.11 ยฑ0.58 / 24.14 ms โ no change โ
โ QQuery 91 โ 58.98 / 60.16 ยฑ1.07 / 62.04 ms โ 58.90 / 59.49 ยฑ0.57 / 60.25 ms โ no change โ
โ QQuery 92 โ 56.56 / 57.16 ยฑ0.45 / 57.81 ms โ 56.48 / 57.10 ยฑ0.37 / 57.65 ms โ no change โ
โ QQuery 93 โ 162.56 / 163.69 ยฑ0.92 / 165.17 ms โ 161.66 / 163.05 ยฑ0.86 / 164.29 ms โ no change โ
โ QQuery 94 โ 60.92 / 61.84 ยฑ1.25 / 64.30 ms โ 60.39 / 60.97 ยฑ0.48 / 61.72 ms โ no change โ
โ QQuery 95 โ 111.28 / 112.50 ยฑ0.82 / 113.74 ms โ 111.45 / 113.16 ยฑ1.63 / 116.15 ms โ no change โ
โ QQuery 96 โ 68.16 / 69.39 ยฑ0.68 / 70.07 ms โ 68.14 / 69.13 ยฑ0.56 / 69.74 ms โ no change โ
โ QQuery 97 โ 112.77 / 113.85 ยฑ1.02 / 115.69 ms โ 113.17 / 116.01 ยฑ3.54 / 122.83 ms โ no change โ
โ QQuery 98 โ 149.04 / 150.23 ยฑ1.08 / 151.98 ms โ 149.92 / 151.60 ยฑ1.65 / 153.70 ms โ no change โ
โ QQuery 99 โ 10694.40 / 10817.56 ยฑ102.26 / 10960.22 ms โ 10707.55 / 10795.51 ยฑ57.87 / 10843.63 ms โ no change โ
โโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโณโโโโโโโโโโโโโ
โ Benchmark Summary โ โ
โกโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฉ
โ Total Time (HEAD) โ 29982.91ms โ
โ Total Time (scalar-subquery-semi-join) โ 30094.64ms โ
โ Average Time (HEAD) โ 302.86ms โ
โ Average Time (scalar-subquery-semi-join) โ 303.99ms โ
โ Queries Faster โ 0 โ
โ Queries Slower โ 3 โ
โ Queries with No Change โ 96 โ
โ Queries with Failure โ 0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโ
```
</p>
</details>
<details><summary>Resource Usage</summary>
**tpcds โ base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 155.0s |
| Peak memory | 6.3 GiB |
| Avg memory | 5.6 GiB |
| CPU user | 236.6s |
| CPU sys | 8.2s |
| Peak spill | 0 B |
**tpcds โ branch**
| Metric | Value |
|--------|-------|
| Wall time | 155.0s |
| Peak memory | 6.3 GiB |
| Avg memory | 5.6 GiB |
| CPU user | 236.8s |
| CPU sys | 8.0s |
| 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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan closed pull request #21971: perf: optimize scalar aggregate joins as semi joins
URL: https://github.com/apache/datafusion/pull/21971
--
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] perf: optimize scalar aggregate joins as semi joins [datafusion]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #21971:
URL: https://github.com/apache/datafusion/pull/21971#issuecomment-4359175271
๐ค Benchmark running (GKE) | [trigger](https://github.com/apache/datafusion/pull/21971#issuecomment-4359167733)
**Instance:** `c4a-highmem-16` (12 vCPU / 65 GiB) | `Linux bench-c4359167733-1951-7q2dw 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 scalar-subquery-semi-join (c069c6f22ce0cf662fc5ea8b58c1a63bb1e36b64) to d59bc72 (merge-base) [diff](https://github.com/apache/datafusion/compare/d59bc7236bf734492789fca3fabd77d7b8ace0c8..c069c6f22ce0cf662fc5ea8b58c1a63bb1e36b64) 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