perf: optimize scalar aggregate joins as semi joins by Dandandan · Pull Request #21971 · apache/datafusion

Dandandan · 2026-05-01T11:58:35Z

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.
Use logical plan functional dependencies to prove right-side uniqueness, so this works for aggregate group keys, table constraints, and future property propagation.
Preserve projection functional dependency output positions when computed projection expressions appear before projected input columns.
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_INCREMENTAL=0 cargo test -p datafusion-optimizer eliminate_join --lib
CARGO_INCREMENTAL=0 INCLUDE_TPCH=true cargo test -p datafusion-sqllogictest --test sqllogictests -- tpch
cargo fmt --all
CARGO_INCREMENTAL=0 ci/scripts/rust_clippy.sh
Earlier on this branch before the FD refactor: ./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.

Dandandan · 2026-05-01T11:59:31Z

run benchmark tpch tpch10 tpcds

adriangbot · 2026-05-01T12:02:02Z

🤖 Benchmark running (GKE) | trigger
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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Comparing scalar-subquery-semi-join (c069c6f) to d59bc72 (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-01T12:02:51Z

🤖 Benchmark running (GKE) | trigger
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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Comparing scalar-subquery-semi-join (c069c6f) to d59bc72 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-01T12:02:52Z

🤖 Benchmark running (GKE) | trigger
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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Comparing scalar-subquery-semi-join (c069c6f) to d59bc72 (merge-base) diff using: tpch10
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-01T12:17:32Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Details

Comparing HEAD and scalar-subquery-semi-join
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ scalar-subquery-semi-join ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1 │ 40.19 / 41.61 ±1.30 / 43.35 ms │ 40.97 / 42.15 ±0.89 / 43.51 ms │ no change │
│ QQuery 2 │ 21.20 / 21.32 ±0.13 / 21.49 ms │ 20.91 / 21.28 ±0.21 / 21.54 ms │ no change │
│ QQuery 3 │ 36.24 / 38.54 ±1.38 / 40.13 ms │ 36.23 / 38.69 ±1.69 / 40.95 ms │ no change │
│ QQuery 4 │ 18.19 / 18.89 ±0.57 / 19.83 ms │ 18.41 / 18.79 ±0.27 / 19.22 ms │ no change │
│ QQuery 5 │ 44.03 / 45.08 ±0.96 / 46.40 ms │ 44.04 / 46.46 ±1.25 / 47.64 ms │ no change │
│ QQuery 6 │ 17.90 / 18.43 ±0.43 / 18.97 ms │ 17.27 / 17.69 ±0.29 / 18.15 ms │ no change │
│ QQuery 7 │ 49.22 / 50.52 ±0.90 / 51.77 ms │ 50.68 / 53.27 ±1.46 / 55.23 ms │ 1.05x slower │
│ QQuery 8 │ 47.43 / 47.65 ±0.16 / 47.90 ms │ 47.18 / 47.43 ±0.25 / 47.89 ms │ no change │
│ QQuery 9 │ 51.59 / 52.20 ±0.56 / 53.08 ms │ 52.82 / 53.32 ±0.55 / 54.13 ms │ no change │
│ QQuery 10 │ 65.69 / 66.62 ±0.57 / 67.46 ms │ 65.99 / 66.66 ±0.98 / 68.60 ms │ no change │
│ QQuery 11 │ 14.26 / 14.53 ±0.28 / 15.00 ms │ 13.98 / 14.34 ±0.22 / 14.58 ms │ no change │
│ QQuery 12 │ 26.44 / 26.76 ±0.20 / 26.97 ms │ 26.56 / 27.13 ±0.50 / 27.76 ms │ no change │
│ QQuery 13 │ 35.81 / 36.66 ±0.64 / 37.63 ms │ 35.29 / 36.64 ±0.87 / 37.72 ms │ no change │
│ QQuery 14 │ 26.80 / 27.00 ±0.11 / 27.16 ms │ 27.09 / 27.73 ±1.12 / 29.96 ms │ no change │
│ QQuery 15 │ 33.29 / 34.23 ±1.45 / 37.12 ms │ 33.48 / 34.07 ±0.73 / 35.50 ms │ no change │
│ QQuery 16 │ 15.36 / 15.53 ±0.11 / 15.70 ms │ 15.37 / 15.57 ±0.13 / 15.73 ms │ no change │
│ QQuery 17 │ 78.34 / 78.67 ±0.47 / 79.57 ms │ 77.54 / 78.07 ±0.39 / 78.56 ms │ no change │
│ QQuery 18 │ 68.70 / 69.69 ±0.98 / 71.47 ms │ 69.96 / 70.76 ±0.72 / 72.03 ms │ no change │
│ QQuery 19 │ 37.90 / 38.21 ±0.53 / 39.26 ms │ 37.63 / 38.15 ±0.43 / 38.82 ms │ no change │
│ QQuery 20 │ 39.05 / 39.41 ±0.30 / 39.90 ms │ 39.21 / 39.80 ±0.53 / 40.76 ms │ no change │
│ QQuery 21 │ 61.43 / 62.05 ±0.87 / 63.73 ms │ 62.55 / 65.17 ±2.01 / 67.16 ms │ 1.05x slower │
│ QQuery 22 │ 23.66 / 24.33 ±0.56 / 25.10 ms │ 24.09 / 24.57 ±0.43 / 25.25 ms │ no change │
└───────────┴────────────────────────────────┴────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD) │ 867.95ms │
│ Total Time (scalar-subquery-semi-join) │ 877.76ms │
│ Average Time (HEAD) │ 39.45ms │
│ Average Time (scalar-subquery-semi-join) │ 39.90ms │
│ Queries Faster │ 0 │
│ Queries Slower │ 2 │
│ Queries with No Change │ 20 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴──────────┘

Resource Usage

tpch — 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 — 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

File an issue against this benchmark runner

adriangbot · 2026-05-01T12:18:48Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Details

Comparing HEAD and scalar-subquery-semi-join
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ scalar-subquery-semi-join ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1 │ 327.41 / 328.62 ±0.84 / 329.47 ms │ 323.14 / 325.44 ±1.21 / 326.55 ms │ no change │
│ QQuery 2 │ 126.43 / 128.58 ±2.10 / 132.10 ms │ 124.34 / 128.33 ±2.56 / 131.90 ms │ no change │
│ QQuery 3 │ 265.04 / 269.93 ±2.98 / 273.65 ms │ 270.25 / 274.85 ±3.54 / 281.13 ms │ no change │
│ QQuery 4 │ 133.56 / 135.63 ±1.82 / 138.87 ms │ 135.08 / 136.42 ±0.87 / 137.70 ms │ no change │
│ QQuery 5 │ 388.38 / 392.76 ±2.42 / 395.42 ms │ 392.23 / 399.40 ±6.48 / 408.90 ms │ no change │
│ QQuery 6 │ 132.79 / 133.06 ±0.30 / 133.63 ms │ 132.66 / 133.39 ±0.51 / 133.89 ms │ no change │
│ QQuery 7 │ 508.12 / 514.60 ±10.43 / 535.31 ms │ 507.92 / 517.72 ±4.97 / 520.72 ms │ no change │
│ QQuery 8 │ 425.13 / 437.88 ±10.68 / 454.40 ms │ 434.33 / 440.67 ±4.52 / 446.27 ms │ no change │
│ QQuery 9 │ 604.93 / 611.04 ±4.78 / 618.39 ms │ 614.89 / 620.48 ±6.07 / 629.86 ms │ no change │
│ QQuery 10 │ 318.84 / 325.47 ±6.79 / 337.14 ms │ 325.20 / 329.83 ±4.73 / 337.23 ms │ no change │
│ QQuery 11 │ 94.59 / 96.15 ±1.27 / 97.55 ms │ 96.10 / 99.16 ±2.74 / 102.89 ms │ no change │
│ QQuery 12 │ 193.56 / 203.83 ±16.59 / 236.88 ms │ 192.81 / 197.54 ±4.98 / 204.31 ms │ no change │
│ QQuery 13 │ 294.07 / 309.33 ±16.60 / 341.48 ms │ 288.30 / 301.78 ±8.51 / 314.89 ms │ no change │
│ QQuery 14 │ 180.91 / 184.03 ±3.93 / 191.69 ms │ 180.94 / 184.40 ±4.93 / 194.17 ms │ no change │
│ QQuery 15 │ 328.53 / 333.04 ±5.77 / 344.44 ms │ 328.79 / 330.24 ±1.55 / 332.16 ms │ no change │
│ QQuery 16 │ 74.79 / 77.24 ±2.02 / 80.85 ms │ 75.20 / 76.64 ±2.32 / 81.26 ms │ no change │
│ QQuery 17 │ 684.82 / 692.90 ±6.38 / 703.77 ms │ 688.33 / 694.82 ±7.83 / 709.10 ms │ no change │
│ QQuery 18 │ 706.94 / 734.43 ±16.16 / 755.08 ms │ 711.15 / 730.66 ±14.11 / 752.64 ms │ no change │
│ QQuery 19 │ 268.81 / 279.15 ±18.04 / 315.09 ms │ 268.17 / 276.28 ±14.20 / 304.60 ms │ no change │
│ QQuery 20 │ 301.32 / 305.81 ±2.49 / 308.73 ms │ 298.24 / 304.98 ±8.76 / 321.79 ms │ no change │
│ QQuery 21 │ 733.01 / 744.85 ±7.28 / 755.12 ms │ 748.43 / 750.64 ±3.41 / 757.40 ms │ no change │
│ QQuery 22 │ 64.57 / 67.25 ±1.34 / 68.03 ms │ 64.46 / 67.99 ±1.90 / 70.06 ms │ no change │
└───────────┴────────────────────────────────────┴────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD) │ 7305.57ms │
│ Total Time (scalar-subquery-semi-join) │ 7321.65ms │
│ Average Time (HEAD) │ 332.07ms │
│ Average Time (scalar-subquery-semi-join) │ 332.80ms │
│ Queries Faster │ 0 │
│ Queries Slower │ 0 │
│ Queries with No Change │ 22 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — 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 — 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

File an issue against this benchmark runner

adriangbot · 2026-05-01T12:22:36Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Details

Comparing HEAD and scalar-subquery-semi-join
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ scalar-subquery-semi-join ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1 │ 7.15 / 7.76 ±0.91 / 9.57 ms │ 7.34 / 7.81 ±0.89 / 9.59 ms │ no change │
│ QQuery 2 │ 146.95 / 147.23 ±0.25 / 147.63 ms │ 144.80 / 146.45 ±1.16 / 147.73 ms │ no change │
│ QQuery 3 │ 114.86 / 116.08 ±0.87 / 116.94 ms │ 113.99 / 115.25 ±1.01 / 116.87 ms │ no change │
│ QQuery 4 │ 1360.59 / 1425.55 ±56.24 / 1496.72 ms │ 1224.21 / 1237.44 ±15.73 / 1268.05 ms │ +1.15x faster │
│ QQuery 5 │ 174.12 / 176.03 ±1.07 / 177.09 ms │ 174.49 / 176.00 ±1.24 / 177.61 ms │ no change │
│ QQuery 6 │ 138.20 / 144.90 ±6.23 / 156.29 ms │ 127.35 / 130.72 ±2.22 / 134.23 ms │ +1.11x faster │
│ QQuery 7 │ 322.44 / 325.84 ±1.86 / 328.15 ms │ 318.03 / 319.70 ±1.22 / 321.39 ms │ no change │
│ QQuery 8 │ 113.62 / 114.62 ±0.60 / 115.26 ms │ 113.64 / 114.01 ±0.39 / 114.73 ms │ no change │
│ QQuery 9 │ 96.95 / 100.45 ±2.45 / 102.76 ms │ 91.37 / 95.61 ±2.20 / 97.44 ms │ no change │
│ QQuery 10 │ 102.99 / 105.46 ±2.51 / 110.04 ms │ 102.37 / 104.26 ±2.33 / 108.83 ms │ no change │
│ QQuery 11 │ 884.26 / 949.49 ±77.74 / 1079.88 ms │ 884.69 / 945.01 ±60.57 / 1041.02 ms │ no change │
│ QQuery 12 │ 46.75 / 47.07 ±0.17 / 47.26 ms │ 45.95 / 46.74 ±1.36 / 49.46 ms │ no change │
│ QQuery 13 │ 391.65 / 403.75 ±10.24 / 417.92 ms │ 397.57 / 406.77 ±4.70 / 410.02 ms │ no change │
│ QQuery 14 │ 1044.67 / 1048.28 ±2.78 / 1051.59 ms │ 1025.53 / 1035.34 ±6.41 / 1043.67 ms │ no change │
│ QQuery 15 │ 14.63 / 15.06 ±0.30 / 15.50 ms │ 14.29 / 14.67 ±0.25 / 14.94 ms │ no change │
│ QQuery 16 │ 7.34 / 7.53 ±0.18 / 7.84 ms │ 7.36 / 7.48 ±0.18 / 7.85 ms │ no change │
│ QQuery 17 │ 204.69 / 205.51 ±0.86 / 207.03 ms │ 200.18 / 201.61 ±1.04 / 203.02 ms │ no change │
│ QQuery 18 │ 125.03 / 126.15 ±0.93 / 127.58 ms │ 125.22 / 128.07 ±2.24 / 130.91 ms │ no change │
│ QQuery 19 │ 153.83 / 156.51 ±2.51 / 161.11 ms │ 163.63 / 165.76 ±1.89 / 168.14 ms │ 1.06x slower │
│ QQuery 20 │ 12.97 / 13.59 ±0.42 / 14.19 ms │ 13.73 / 13.97 ±0.15 / 14.18 ms │ no change │
│ QQuery 21 │ 19.33 / 19.63 ±0.24 / 19.92 ms │ 20.13 / 20.85 ±0.68 / 22.13 ms │ 1.06x slower │
│ QQuery 22 │ 472.97 / 474.11 ±0.83 / 475.00 ms │ 472.02 / 507.41 ±19.26 / 522.02 ms │ 1.07x slower │
│ QQuery 23 │ 1032.52 / 1110.78 ±63.40 / 1176.11 ms │ 1006.38 / 1022.54 ±21.35 / 1062.40 ms │ +1.09x faster │
│ QQuery 24 │ 640.65 / 660.04 ±17.13 / 683.13 ms │ 629.91 / 633.34 ±2.99 / 637.70 ms │ no change │
│ QQuery 25 │ 307.58 / 309.64 ±2.29 / 313.37 ms │ 302.57 / 304.12 ±1.53 / 307.03 ms │ no change │
│ QQuery 26 │ 76.33 / 78.11 ±1.97 / 81.87 ms │ 74.66 / 75.28 ±0.37 / 75.79 ms │ no change │
│ QQuery 27 │ 7.08 / 7.17 ±0.12 / 7.40 ms │ 6.95 / 7.06 ±0.10 / 7.24 ms │ no change │
│ QQuery 28 │ 148.64 / 150.25 ±2.45 / 155.12 ms │ 146.81 / 148.30 ±1.18 / 150.32 ms │ no change │
│ QQuery 29 │ 245.55 / 247.80 ±2.23 / 251.93 ms │ 243.39 / 244.91 ±2.34 / 249.48 ms │ no change │
│ QQuery 30 │ 41.87 / 42.68 ±0.46 / 43.12 ms │ 42.40 / 43.15 ±0.40 / 43.54 ms │ no change │
│ QQuery 31 │ 165.66 / 167.55 ±1.42 / 169.93 ms │ 168.00 / 171.64 ±2.72 / 174.80 ms │ no change │
│ QQuery 32 │ 13.14 / 13.55 ±0.36 / 14.09 ms │ 13.49 / 13.88 ±0.30 / 14.31 ms │ no change │
│ QQuery 33 │ 138.68 / 141.14 ±2.82 / 146.06 ms │ 141.42 / 143.17 ±2.23 / 147.47 ms │ no change │
│ QQuery 34 │ 6.99 / 7.16 ±0.19 / 7.52 ms │ 7.40 / 7.59 ±0.14 / 7.77 ms │ 1.06x slower │
│ QQuery 35 │ 102.15 / 102.75 ±0.56 / 103.76 ms │ 102.62 / 104.80 ±2.27 / 109.19 ms │ no change │
│ QQuery 36 │ 6.55 / 6.75 ±0.18 / 7.07 ms │ 7.06 / 7.24 ±0.11 / 7.39 ms │ 1.07x slower │
│ QQuery 37 │ 8.32 / 8.40 ±0.06 / 8.47 ms │ 8.58 / 8.82 ±0.19 / 9.10 ms │ no change │
│ QQuery 38 │ 85.90 / 90.19 ±4.50 / 98.73 ms │ 89.51 / 91.06 ±0.82 / 91.87 ms │ no change │
│ QQuery 39 │ 116.53 / 118.36 ±1.33 / 120.26 ms │ 116.77 / 120.79 ±5.07 / 130.70 ms │ no change │
│ QQuery 40 │ 94.28 / 98.82 ±3.91 / 105.66 ms │ 94.60 / 96.17 ±1.51 / 98.88 ms │ no change │
│ QQuery 41 │ 14.65 / 14.84 ±0.14 / 15.09 ms │ 14.51 / 14.78 ±0.29 / 15.32 ms │ no change │
│ QQuery 42 │ 108.33 / 110.06 ±2.15 / 113.88 ms │ 107.14 / 108.83 ±1.07 / 110.48 ms │ no change │
│ QQuery 43 │ 5.72 / 5.91 ±0.12 / 6.07 ms │ 5.81 / 6.63 ±1.55 / 9.72 ms │ 1.12x slower │
│ QQuery 44 │ 10.98 / 12.16 ±2.31 / 16.77 ms │ 10.86 / 10.92 ±0.09 / 11.10 ms │ +1.11x faster │
│ QQuery 45 │ 45.39 / 45.61 ±0.20 / 45.98 ms │ 43.38 / 44.18 ±0.56 / 44.72 ms │ no change │
│ QQuery 46 │ 8.57 / 8.70 ±0.16 / 9.01 ms │ 8.52 / 8.64 ±0.20 / 9.03 ms │ no change │
│ QQuery 47 │ 698.30 / 790.16 ±66.84 / 855.78 ms │ 662.56 / 677.03 ±11.62 / 697.59 ms │ +1.17x faster │
│ QQuery 48 │ 281.08 / 291.82 ±6.35 / 300.72 ms │ 272.61 / 274.36 ±2.09 / 278.34 ms │ +1.06x faster │
│ QQuery 49 │ 245.91 / 247.19 ±1.46 / 250.02 ms │ 242.31 / 243.02 ±0.47 / 243.72 ms │ no change │
│ QQuery 50 │ 189.81 / 196.22 ±3.77 / 200.95 ms │ 180.73 / 186.03 ±3.97 / 192.49 ms │ +1.05x faster │
│ QQuery 51 │ 172.34 / 174.83 ±2.07 / 177.56 ms │ 170.57 / 173.01 ±1.33 / 174.57 ms │ no change │
│ QQuery 52 │ 108.85 / 110.99 ±3.31 / 117.50 ms │ 107.34 / 109.44 ±2.20 / 113.55 ms │ no change │
│ QQuery 53 │ 103.28 / 104.12 ±0.56 / 104.91 ms │ 103.50 / 104.39 ±0.97 / 106.17 ms │ no change │
│ QQuery 54 │ 145.46 / 147.52 ±2.18 / 151.61 ms │ 154.40 / 156.00 ±1.89 / 159.58 ms │ 1.06x slower │
│ QQuery 55 │ 107.21 / 108.19 ±0.67 / 109.14 ms │ 110.05 / 110.69 ±0.62 / 111.55 ms │ no change │
│ QQuery 56 │ 139.42 / 140.95 ±0.84 / 141.76 ms │ 145.02 / 147.30 ±1.92 / 150.84 ms │ no change │
│ QQuery 57 │ 164.52 / 166.70 ±2.00 / 170.40 ms │ 172.27 / 173.88 ±1.68 / 176.77 ms │ no change │
│ QQuery 58 │ 245.92 / 247.72 ±1.55 / 250.17 ms │ 248.55 / 250.76 ±1.57 / 252.28 ms │ no change │
│ QQuery 59 │ 200.29 / 206.51 ±3.18 / 208.60 ms │ 192.75 / 194.73 ±2.86 / 200.30 ms │ +1.06x faster │
│ QQuery 60 │ 149.13 / 150.57 ±1.60 / 153.37 ms │ 141.00 / 142.08 ±0.83 / 143.57 ms │ +1.06x faster │
│ QQuery 61 │ 14.42 / 14.60 ±0.19 / 14.92 ms │ 13.21 / 13.37 ±0.19 / 13.73 ms │ +1.09x faster │
│ QQuery 62 │ 890.00 / 913.85 ±23.56 / 945.63 ms │ 852.52 / 855.93 ±2.89 / 859.81 ms │ +1.07x faster │
│ QQuery 63 │ 103.52 / 104.26 ±0.58 / 105.18 ms │ 102.16 / 104.25 ±1.92 / 107.73 ms │ no change │
│ QQuery 64 │ 630.33 / 636.71 ±3.24 / 638.97 ms │ 624.45 / 627.27 ±2.24 / 630.74 ms │ no change │
│ QQuery 65 │ 244.23 / 248.28 ±3.31 / 252.93 ms │ 242.41 / 247.68 ±2.96 / 251.51 ms │ no change │
│ QQuery 66 │ 216.26 / 226.99 ±11.28 / 242.92 ms │ 216.27 / 226.74 ±10.17 / 241.57 ms │ no change │
│ QQuery 67 │ 296.32 / 315.43 ±13.06 / 331.18 ms │ 288.52 / 300.56 ±12.69 / 319.17 ms │ no change │
│ QQuery 68 │ 9.57 / 9.76 ±0.11 / 9.89 ms │ 8.67 / 8.80 ±0.15 / 9.08 ms │ +1.11x faster │
│ QQuery 69 │ 102.09 / 104.32 ±1.63 / 106.29 ms │ 97.25 / 98.22 ±0.85 / 99.64 ms │ +1.06x faster │
│ QQuery 70 │ 317.97 / 332.36 ±9.41 / 344.42 ms │ 327.36 / 338.42 ±7.68 / 348.67 ms │ no change │
│ QQuery 71 │ 134.43 / 139.03 ±4.56 / 147.58 ms │ 137.40 / 139.53 ±1.31 / 141.29 ms │ no change │
│ QQuery 72 │ 536.59 / 548.38 ±8.25 / 559.75 ms │ 553.50 / 564.49 ±8.29 / 577.06 ms │ no change │
│ QQuery 73 │ 6.79 / 6.95 ±0.17 / 7.28 ms │ 7.01 / 7.15 ±0.11 / 7.35 ms │ no change │
│ QQuery 74 │ 560.74 / 568.89 ±7.44 / 581.91 ms │ 569.49 / 590.48 ±16.21 / 617.96 ms │ no change │
│ QQuery 75 │ 267.67 / 269.40 ±1.17 / 271.04 ms │ 265.36 / 267.72 ±2.46 / 272.48 ms │ no change │
│ QQuery 76 │ 130.87 / 131.64 ±0.42 / 132.05 ms │ 129.57 / 129.96 ±0.37 / 130.65 ms │ no change │
│ QQuery 77 │ 188.78 / 190.57 ±2.31 / 195.05 ms │ 187.50 / 188.82 ±1.28 / 191.12 ms │ no change │
│ QQuery 78 │ 310.11 / 312.04 ±1.25 / 313.82 ms │ 305.59 / 308.57 ±2.12 / 311.71 ms │ no change │
│ QQuery 79 │ 228.01 / 232.26 ±3.36 / 237.61 ms │ 228.38 / 231.44 ±1.99 / 233.72 ms │ no change │
│ QQuery 80 │ 306.57 / 308.79 ±1.84 / 312.11 ms │ 299.71 / 302.09 ±1.76 / 305.06 ms │ no change │
│ QQuery 81 │ 26.93 / 27.83 ±0.57 / 28.65 ms │ 25.84 / 27.03 ±2.29 / 31.62 ms │ no change │
│ QQuery 82 │ 42.08 / 42.34 ±0.21 / 42.69 ms │ 39.09 / 39.73 ±0.56 / 40.47 ms │ +1.07x faster │
│ QQuery 83 │ 37.82 / 38.24 ±0.32 / 38.74 ms │ 34.90 / 35.59 ±0.47 / 36.33 ms │ +1.07x faster │
│ QQuery 84 │ 48.35 / 49.58 ±1.97 / 53.51 ms │ 46.38 / 47.13 ±0.69 / 48.07 ms │ no change │
│ QQuery 85 │ 148.57 / 150.17 ±1.31 / 151.79 ms │ 142.14 / 143.93 ±1.88 / 147.44 ms │ no change │
│ QQuery 86 │ 39.59 / 39.96 ±0.29 / 40.35 ms │ 36.91 / 37.56 ±0.40 / 37.97 ms │ +1.06x faster │
│ QQuery 87 │ 4.10 / 4.20 ±0.14 / 4.47 ms │ 3.49 / 3.60 ±0.16 / 3.92 ms │ +1.17x faster │
│ QQuery 88 │ 107.22 / 108.18 ±0.82 / 109.67 ms │ 99.62 / 101.11 ±2.04 / 105.11 ms │ +1.07x faster │
│ QQuery 89 │ 122.75 / 127.25 ±5.92 / 138.82 ms │ 117.15 / 118.24 ±1.36 / 120.73 ms │ +1.08x faster │
│ QQuery 90 │ 24.08 / 24.84 ±0.43 / 25.27 ms │ 23.42 / 23.80 ±0.29 / 24.22 ms │ no change │
│ QQuery 91 │ 62.56 / 63.50 ±0.79 / 64.49 ms │ 60.12 / 62.00 ±2.66 / 67.27 ms │ no change │
│ QQuery 92 │ 60.10 / 61.40 ±1.94 / 65.24 ms │ 58.81 / 59.25 ±0.52 / 60.26 ms │ no change │
│ QQuery 93 │ 174.90 / 176.10 ±0.84 / 177.03 ms │ 169.41 / 170.87 ±1.09 / 172.66 ms │ no change │
│ QQuery 94 │ 62.97 / 65.72 ±3.27 / 72.04 ms │ 63.36 / 63.69 ±0.38 / 64.24 ms │ no change │
│ QQuery 95 │ 113.97 / 115.09 ±0.77 / 115.92 ms │ 115.47 / 116.53 ±1.32 / 119.09 ms │ no change │
│ QQuery 96 │ 71.33 / 72.68 ±0.83 / 73.55 ms │ 70.07 / 71.90 ±1.24 / 73.08 ms │ no change │
│ QQuery 97 │ 114.32 / 116.04 ±1.02 / 117.39 ms │ 114.37 / 120.31 ±3.21 / 124.00 ms │ no change │
│ QQuery 98 │ 153.21 / 155.46 ±1.17 / 156.38 ms │ 149.66 / 150.99 ±0.88 / 152.07 ms │ no change │
│ QQuery 99 │ 10707.19 / 10852.52 ±131.42 / 11031.81 ms │ 10793.69 / 10916.26 ±99.32 / 11079.54 ms │ no change │
└───────────┴───────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 30678.11ms │
│ Total Time (scalar-subquery-semi-join) │ 30182.55ms │
│ Average Time (HEAD) │ 309.88ms │
│ Average Time (scalar-subquery-semi-join) │ 304.87ms │
│ Queries Faster │ 19 │
│ Queries Slower │ 7 │
│ Queries with No Change │ 73 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — 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 — 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

File an issue against this benchmark runner

Dandandan · 2026-05-01T14:02:34Z

run benchmark tpcds

adriangbot · 2026-05-01T14:06:42Z

🤖 Benchmark running (GKE) | trigger
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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Comparing scalar-subquery-semi-join (da59e3a) to d59bc72 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-01T14:26:06Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

Architecture: aarch64
CPU op-mode(s): 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: ARM
Model name: Neoverse-V2
Model: 1
Thread(s) per core: 1
Core(s) per cluster: 16
Socket(s): -
Cluster(s): 1
Stepping: r0p1
BogoMIPS: 2000.00
Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache: 1 MiB (16 instances)
L1i cache: 1 MiB (16 instances)
L2 cache: 32 MiB (16 instances)
L3 cache: 80 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-15
Vulnerability Gather data sampling: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; __user pointer sanitization
Vulnerability Spectre v2: Mitigation; CSV2, BHB
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Not affected
Vulnerability Vmscape: Not affected

Details

Comparing HEAD and scalar-subquery-semi-join
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ scalar-subquery-semi-join ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1 │ 6.90 / 7.37 ±0.85 / 9.06 ms │ 6.79 / 7.25 ±0.85 / 8.94 ms │ no change │
│ QQuery 2 │ 144.20 / 144.70 ±0.48 / 145.60 ms │ 144.27 / 144.70 ±0.34 / 145.14 ms │ no change │
│ QQuery 3 │ 113.76 / 115.75 ±2.96 / 121.60 ms │ 114.11 / 115.13 ±1.21 / 117.46 ms │ no change │
│ QQuery 4 │ 1296.12 / 1298.54 ±2.39 / 1302.86 ms │ 1292.19 / 1296.36 ±3.84 / 1302.41 ms │ no change │
│ QQuery 5 │ 174.49 / 175.31 ±1.27 / 177.83 ms │ 174.27 / 175.63 ±1.37 / 177.66 ms │ no change │
│ QQuery 6 │ 133.07 / 137.47 ±2.34 / 139.56 ms │ 133.89 / 139.97 ±5.64 / 149.65 ms │ no change │
│ QQuery 7 │ 321.56 / 323.32 ±1.51 / 325.32 ms │ 319.26 / 322.02 ±2.45 / 325.02 ms │ no change │
│ QQuery 8 │ 113.07 / 114.94 ±1.43 / 117.28 ms │ 114.15 / 115.36 ±0.96 / 116.82 ms │ no change │
│ QQuery 9 │ 91.57 / 98.43 ±4.13 / 102.19 ms │ 95.92 / 101.86 ±4.73 / 110.45 ms │ no change │
│ QQuery 10 │ 102.53 / 103.15 ±0.61 / 104.31 ms │ 102.84 / 103.32 ±0.44 / 104.04 ms │ no change │
│ QQuery 11 │ 888.73 / 897.16 ±5.85 / 905.97 ms │ 889.08 / 903.70 ±8.00 / 912.77 ms │ no change │
│ QQuery 12 │ 43.57 / 43.85 ±0.21 / 44.10 ms │ 43.26 / 43.57 ±0.27 / 43.93 ms │ no change │
│ QQuery 13 │ 388.78 / 391.98 ±2.33 / 395.57 ms │ 388.40 / 390.20 ±2.00 / 393.55 ms │ no change │
│ QQuery 14 │ 1024.92 / 1031.10 ±4.85 / 1038.87 ms │ 1176.64 / 1184.12 ±8.59 / 1200.33 ms │ 1.15x slower │
│ QQuery 15 │ 14.39 / 14.67 ±0.16 / 14.82 ms │ 14.23 / 14.67 ±0.35 / 15.22 ms │ no change │
│ QQuery 16 │ 7.43 / 7.57 ±0.23 / 8.03 ms │ 7.28 / 7.42 ±0.08 / 7.53 ms │ no change │
│ QQuery 17 │ 202.74 / 203.77 ±0.54 / 204.30 ms │ 202.07 / 204.00 ±2.08 / 207.98 ms │ no change │
│ QQuery 18 │ 124.09 / 125.01 ±0.79 / 126.47 ms │ 123.14 / 123.85 ±0.56 / 124.61 ms │ no change │
│ QQuery 19 │ 153.38 / 154.12 ±0.61 / 155.23 ms │ 153.54 / 156.31 ±1.57 / 158.07 ms │ no change │
│ QQuery 20 │ 13.04 / 13.49 ±0.27 / 13.85 ms │ 12.85 / 13.34 ±0.29 / 13.72 ms │ no change │
│ QQuery 21 │ 18.87 / 19.52 ±0.88 / 21.27 ms │ 18.84 / 19.25 ±0.24 / 19.60 ms │ no change │
│ QQuery 22 │ 469.58 / 476.02 ±4.43 / 482.38 ms │ 468.21 / 472.81 ±3.84 / 478.09 ms │ no change │
│ QQuery 23 │ 1018.74 / 1022.42 ±3.62 / 1028.28 ms │ 1012.22 / 1019.60 ±4.90 / 1027.63 ms │ no change │
│ QQuery 24 │ 635.37 / 639.05 ±3.83 / 645.87 ms │ 635.86 / 641.76 ±3.31 / 645.96 ms │ no change │
│ QQuery 25 │ 305.20 / 307.55 ±1.69 / 310.00 ms │ 306.60 / 308.42 ±1.60 / 310.90 ms │ no change │
│ QQuery 26 │ 76.69 / 77.09 ±0.35 / 77.63 ms │ 76.23 / 76.59 ±0.20 / 76.83 ms │ no change │
│ QQuery 27 │ 7.05 / 7.13 ±0.12 / 7.36 ms │ 7.19 / 7.27 ±0.10 / 7.47 ms │ no change │
│ QQuery 28 │ 147.30 / 149.23 ±2.99 / 155.16 ms │ 147.78 / 148.95 ±1.52 / 151.95 ms │ no change │
│ QQuery 29 │ 246.40 / 248.06 ±1.68 / 251.11 ms │ 245.96 / 247.84 ±2.88 / 253.55 ms │ no change │
│ QQuery 30 │ 41.59 / 42.26 ±0.53 / 43.02 ms │ 41.63 / 41.84 ±0.20 / 42.11 ms │ no change │
│ QQuery 31 │ 165.45 / 166.87 ±1.46 / 169.50 ms │ 165.87 / 167.00 ±0.76 / 168.18 ms │ no change │
│ QQuery 32 │ 13.35 / 13.53 ±0.17 / 13.79 ms │ 13.19 / 13.48 ±0.35 / 14.14 ms │ no change │
│ QQuery 33 │ 137.65 / 141.42 ±3.09 / 146.58 ms │ 139.36 / 140.43 ±0.61 / 141.15 ms │ no change │
│ QQuery 34 │ 6.92 / 7.06 ±0.23 / 7.53 ms │ 6.93 / 7.09 ±0.17 / 7.42 ms │ no change │
│ QQuery 35 │ 100.83 / 102.10 ±1.02 / 103.64 ms │ 100.61 / 102.52 ±1.57 / 104.15 ms │ no change │
│ QQuery 36 │ 6.53 / 6.89 ±0.23 / 7.21 ms │ 6.51 / 6.84 ±0.17 / 7.00 ms │ no change │
│ QQuery 37 │ 8.51 / 8.75 ±0.12 / 8.87 ms │ 8.30 / 8.38 ±0.07 / 8.51 ms │ no change │
│ QQuery 38 │ 85.70 / 88.35 ±2.21 / 90.77 ms │ 86.16 / 88.28 ±2.82 / 93.85 ms │ no change │
│ QQuery 39 │ 115.85 / 119.92 ±5.64 / 130.96 ms │ 116.87 / 117.58 ±0.63 / 118.58 ms │ no change │
│ QQuery 40 │ 94.66 / 96.81 ±1.97 / 100.12 ms │ 93.30 / 94.85 ±1.28 / 96.49 ms │ no change │
│ QQuery 41 │ 14.40 / 14.59 ±0.25 / 15.08 ms │ 14.38 / 14.61 ±0.29 / 15.19 ms │ no change │
│ QQuery 42 │ 107.60 / 108.67 ±1.23 / 111.05 ms │ 107.49 / 108.55 ±0.63 / 109.18 ms │ no change │
│ QQuery 43 │ 5.62 / 5.83 ±0.18 / 6.17 ms │ 5.82 / 6.12 ±0.50 / 7.12 ms │ no change │
│ QQuery 44 │ 11.08 / 11.17 ±0.13 / 11.43 ms │ 10.83 / 12.08 ±2.10 / 16.26 ms │ 1.08x slower │
│ QQuery 45 │ 45.05 / 45.47 ±0.26 / 45.83 ms │ 44.60 / 45.19 ±0.31 / 45.49 ms │ no change │
│ QQuery 46 │ 8.47 / 8.66 ±0.24 / 9.11 ms │ 8.47 / 8.63 ±0.18 / 8.98 ms │ no change │
│ QQuery 47 │ 687.27 / 698.27 ±7.47 / 706.05 ms │ 689.14 / 694.08 ±2.85 / 697.89 ms │ no change │
│ QQuery 48 │ 270.72 / 278.25 ±5.60 / 287.72 ms │ 270.48 / 274.96 ±3.79 / 280.64 ms │ no change │
│ QQuery 49 │ 244.32 / 245.63 ±1.09 / 247.49 ms │ 243.61 / 244.34 ±0.73 / 245.42 ms │ no change │
│ QQuery 50 │ 185.18 / 186.90 ±1.65 / 189.75 ms │ 183.17 / 187.21 ±4.84 / 196.60 ms │ no change │
│ QQuery 51 │ 171.38 / 173.03 ±0.92 / 174.16 ms │ 171.66 / 174.78 ±2.70 / 178.32 ms │ no change │
│ QQuery 52 │ 107.62 / 108.32 ±0.51 / 109.07 ms │ 107.31 / 108.09 ±0.42 / 108.52 ms │ no change │
│ QQuery 53 │ 102.79 / 105.15 ±3.39 / 111.87 ms │ 102.62 / 104.78 ±2.47 / 109.40 ms │ no change │
│ QQuery 54 │ 145.32 / 147.00 ±1.49 / 149.68 ms │ 146.46 / 147.47 ±0.52 / 147.90 ms │ no change │
│ QQuery 55 │ 106.50 / 107.54 ±0.81 / 108.88 ms │ 107.59 / 108.94 ±1.70 / 112.21 ms │ no change │
│ QQuery 56 │ 139.45 / 140.72 ±1.28 / 143.15 ms │ 139.32 / 141.81 ±1.70 / 143.65 ms │ no change │
│ QQuery 57 │ 163.78 / 166.56 ±3.07 / 172.35 ms │ 164.06 / 165.76 ±1.42 / 168.38 ms │ no change │
│ QQuery 58 │ 244.78 / 245.97 ±1.41 / 248.69 ms │ 242.46 / 244.24 ±1.26 / 246.13 ms │ no change │
│ QQuery 59 │ 192.21 / 194.74 ±1.89 / 197.59 ms │ 192.45 / 195.71 ±3.18 / 201.10 ms │ no change │
│ QQuery 60 │ 140.76 / 141.85 ±1.17 / 144.02 ms │ 141.02 / 142.72 ±0.94 / 143.72 ms │ no change │
│ QQuery 61 │ 13.41 / 13.51 ±0.13 / 13.76 ms │ 13.44 / 14.56 ±1.86 / 18.26 ms │ 1.08x slower │
│ QQuery 62 │ 861.00 / 865.49 ±3.52 / 871.57 ms │ 855.62 / 862.02 ±5.51 / 869.03 ms │ no change │
│ QQuery 63 │ 103.05 / 104.66 ±1.93 / 108.42 ms │ 102.59 / 103.47 ±0.59 / 104.35 ms │ no change │
│ QQuery 64 │ 627.36 / 632.25 ±4.66 / 639.41 ms │ 630.95 / 634.40 ±2.68 / 637.43 ms │ no change │
│ QQuery 65 │ 244.89 / 247.50 ±2.03 / 250.57 ms │ 245.93 / 252.56 ±6.29 / 263.15 ms │ no change │
│ QQuery 66 │ 215.69 / 230.75 ±13.51 / 254.01 ms │ 213.79 / 223.71 ±7.55 / 232.27 ms │ no change │
│ QQuery 67 │ 290.81 / 293.75 ±2.41 / 298.02 ms │ 289.95 / 295.27 ±5.55 / 305.97 ms │ no change │
│ QQuery 68 │ 8.45 / 8.79 ±0.24 / 9.19 ms │ 8.47 / 8.70 ±0.19 / 8.99 ms │ no change │
│ QQuery 69 │ 98.36 / 102.01 ±5.67 / 113.27 ms │ 98.40 / 100.93 ±4.15 / 109.20 ms │ no change │
│ QQuery 70 │ 308.96 / 324.77 ±8.35 / 332.49 ms │ 313.55 / 319.28 ±4.74 / 326.19 ms │ no change │
│ QQuery 71 │ 133.47 / 134.88 ±1.31 / 137.01 ms │ 132.94 / 137.17 ±5.09 / 145.83 ms │ no change │
│ QQuery 72 │ 547.50 / 549.90 ±1.82 / 552.15 ms │ 531.11 / 540.55 ±8.23 / 550.76 ms │ no change │
│ QQuery 73 │ 6.71 / 6.93 ±0.22 / 7.35 ms │ 6.74 / 6.90 ±0.22 / 7.34 ms │ no change │
│ QQuery 74 │ 563.67 / 570.29 ±4.14 / 576.20 ms │ 566.79 / 575.15 ±4.71 / 580.89 ms │ no change │
│ QQuery 75 │ 266.02 / 268.58 ±3.16 / 274.73 ms │ 265.03 / 267.15 ±1.38 / 269.00 ms │ no change │
│ QQuery 76 │ 130.02 / 132.10 ±2.52 / 137.03 ms │ 129.46 / 131.65 ±2.63 / 136.77 ms │ no change │
│ QQuery 77 │ 188.28 / 189.72 ±2.20 / 194.09 ms │ 188.28 / 189.58 ±1.14 / 191.24 ms │ no change │
│ QQuery 78 │ 309.35 / 312.34 ±2.39 / 316.64 ms │ 308.54 / 311.79 ±2.50 / 315.63 ms │ no change │
│ QQuery 79 │ 229.93 / 232.89 ±3.43 / 239.52 ms │ 228.66 / 231.40 ±1.75 / 233.60 ms │ no change │
│ QQuery 80 │ 300.50 / 303.37 ±1.61 / 304.97 ms │ 299.99 / 303.34 ±3.23 / 307.59 ms │ no change │
│ QQuery 81 │ 25.94 / 26.22 ±0.16 / 26.44 ms │ 25.51 / 26.22 ±0.69 / 27.45 ms │ no change │
│ QQuery 82 │ 39.79 / 39.99 ±0.16 / 40.17 ms │ 39.24 / 39.60 ±0.30 / 40.12 ms │ no change │
│ QQuery 83 │ 35.47 / 36.28 ±1.22 / 38.70 ms │ 35.55 / 35.73 ±0.16 / 35.98 ms │ no change │
│ QQuery 84 │ 46.25 / 46.49 ±0.14 / 46.63 ms │ 46.26 / 47.20 ±1.42 / 50.01 ms │ no change │
│ QQuery 85 │ 141.68 / 142.15 ±0.40 / 142.77 ms │ 141.18 / 142.06 ±0.69 / 143.13 ms │ no change │
│ QQuery 86 │ 37.32 / 38.34 ±1.16 / 40.46 ms │ 36.84 / 37.51 ±0.42 / 38.13 ms │ no change │
│ QQuery 87 │ 3.47 / 3.58 ±0.17 / 3.92 ms │ 3.49 / 3.65 ±0.25 / 4.14 ms │ no change │
│ QQuery 88 │ 99.17 / 99.84 ±0.55 / 100.76 ms │ 99.48 / 100.98 ±1.67 / 104.19 ms │ no change │
│ QQuery 89 │ 116.34 / 120.10 ±5.75 / 131.48 ms │ 116.61 / 119.31 ±4.48 / 128.24 ms │ no change │
│ QQuery 90 │ 22.40 / 22.98 ±0.62 / 24.11 ms │ 22.34 / 23.11 ±0.58 / 24.14 ms │ no change │
│ QQuery 91 │ 58.98 / 60.16 ±1.07 / 62.04 ms │ 58.90 / 59.49 ±0.57 / 60.25 ms │ no change │
│ QQuery 92 │ 56.56 / 57.16 ±0.45 / 57.81 ms │ 56.48 / 57.10 ±0.37 / 57.65 ms │ no change │
│ QQuery 93 │ 162.56 / 163.69 ±0.92 / 165.17 ms │ 161.66 / 163.05 ±0.86 / 164.29 ms │ no change │
│ QQuery 94 │ 60.92 / 61.84 ±1.25 / 64.30 ms │ 60.39 / 60.97 ±0.48 / 61.72 ms │ no change │
│ QQuery 95 │ 111.28 / 112.50 ±0.82 / 113.74 ms │ 111.45 / 113.16 ±1.63 / 116.15 ms │ no change │
│ QQuery 96 │ 68.16 / 69.39 ±0.68 / 70.07 ms │ 68.14 / 69.13 ±0.56 / 69.74 ms │ no change │
│ QQuery 97 │ 112.77 / 113.85 ±1.02 / 115.69 ms │ 113.17 / 116.01 ±3.54 / 122.83 ms │ no change │
│ QQuery 98 │ 149.04 / 150.23 ±1.08 / 151.98 ms │ 149.92 / 151.60 ±1.65 / 153.70 ms │ no change │
│ QQuery 99 │ 10694.40 / 10817.56 ±102.26 / 10960.22 ms │ 10707.55 / 10795.51 ±57.87 / 10843.63 ms │ no change │
└───────────┴───────────────────────────────────────────┴──────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 29982.91ms │
│ Total Time (scalar-subquery-semi-join) │ 30094.64ms │
│ Average Time (HEAD) │ 302.86ms │
│ Average Time (scalar-subquery-semi-join) │ 303.99ms │
│ Queries Faster │ 0 │
│ Queries Slower │ 3 │
│ Queries with No Change │ 96 │
│ Queries with Failure │ 0 │
└──────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — 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 — 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

File an issue against this benchmark runner

github-actionsBot added optimizer Optimizer rules sqllogictest SQL Logic Tests (.slt) labels May 1, 2026

perf: optimize scalar aggregate joins as semi joins
da59e3a

Dandandan force-pushed the scalar-subquery-semi-join branch from c069c6f to da59e3aCompare May 1, 2026 12:19

github-actionsBot added the logical-expr Logical plan and expressions label May 1, 2026

Dandandan closed this May 1, 2026

Dandandan reopened this May 1, 2026

Dandandan closed this May 1, 2026

Navigation Menu

Uh oh!

perf: optimize scalar aggregate joins as semi joins - #21971

perf: optimize scalar aggregate joins as semi joins#21971
Dandandan wants to merge 1 commit into
apache:mainfrom
Dandandan:scalar-subquery-semi-join

Dandandan commented May 1, 2026•
edited
Loading

Uh oh!

Dandandan commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

Dandandan commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

Dandandan commented May 1, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

Dandandan commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

Dandandan commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

adriangbot commented May 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Dandandan commented May 1, 2026•
edited
Loading