Skip to content

perf: push semi joins below inner joins and restrict decorrelated aggregates - #26218

Draft
adriangb wants to merge 2 commits into
apache:mainfrom
pydantic:perf-push-down-semi-join
Draft

perf: push semi joins below inner joins and restrict decorrelated aggregates#26218
adriangb wants to merge 2 commits into
apache:mainfrom
pydantic:perf-push-down-semi-join

Conversation

@adriangb

@adriangb adriangb commented Oct 12, 2026

Copy link
Copy Markdown
Contributor

Add a PushDownSemiJoin logical optimizer rule
(datafusion.optimizer.enable_semi_join_pushdown, default true):

  • Push LeftSemi/LeftAnti joins below inner joins when all referenced left columns come from one input and the other input is not filtered, so the semi join removes rows before the larger inner join (TPC-H Q18).
  • For LeftSemi(L, Aggregate(group by k)) joined on L.x = k, semi join the aggregate input with the cheap filtered scan that L.x originates from, so groups that can never match are not computed (TPC-H Q2, Q17)

@claude
perf: push semi joins below inner joins and restrict decorrelated agg…
80fcd0f 
…regates

Add a PushDownSemiJoin logical optimizer rule
(datafusion.optimizer.enable_semi_join_pushdown, default true):

- Push LeftSemi/LeftAnti joins below inner joins when all referenced left
  columns come from one input and the other input is not filtered, so the
  semi join removes rows before the larger inner join (TPC-H Q18).
- For LeftSemi(L, Aggregate(group by k)) joined on L.x = k, semi join the
  aggregate input with the cheap filtered scan that L.x originates from,
  so groups that can never match are not computed (TPC-H Q2, Q17).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01YJhYrzTysgqiqrdXy8uM2U
@adriangb

adriangb commented Oct 12, 2026

Copy link
Copy Markdown
Contributor Author

run benchmarks

@github-actions github-actions Bot added documentation Improvements or additions to documentation optimizer Optimizer rules sqllogictest SQL Logic Tests (.slt) common Related to common crate labels Oct 12, 2026
@adriangbot

adriangbot commented Oct 12, 2026

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6117532024-3326-x8n7f 6.12.94+ #1 SMP Fri Aug 21 08:00:16 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 perf-push-down-semi-join (80fcd0f) to b9c8b3d (merge-base) diff

Run configuration 
run benchmark clickbench_partitioned

Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot

adriangbot commented Oct 12, 2026

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6117532024-3328-7txcq 6.12.94+ #1 SMP Fri Aug 21 08:00:16 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 perf-push-down-semi-join (80fcd0f) to b9c8b3d (merge-base) diff

Run configuration 
run benchmark tpch

Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot

adriangbot commented Oct 12, 2026

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6117532024-3327-ncmpn 6.12.94+ #1 SMP Fri Aug 21 08:00:16 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 perf-push-down-semi-join (80fcd0f) to b9c8b3d (merge-base) diff

Run configuration 
run benchmark tpcds

Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot

adriangbot commented Oct 12, 2026

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing perf-push-down-semi-join (80fcd0f) to b9c8b3d (merge-base) diff

Run configuration 
run benchmark tpch
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 perf-push-down-semi-join
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ perf-push-down-semi-join ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │ 41.30 ms │                 40.94 ms │     no change │
│ QQuery 2  │ 18.99 ms │                 18.90 ms │     no change │
│ QQuery 3  │ 28.86 ms │                 29.06 ms │     no change │
│ QQuery 4  │ 17.78 ms │                 17.64 ms │     no change │
│ QQuery 5  │ 36.30 ms │                 36.16 ms │     no change │
│ QQuery 6  │ 16.52 ms │                 16.51 ms │     no change │
│ QQuery 7  │ 41.66 ms │                 42.90 ms │     no change │
│ QQuery 8  │ 41.53 ms │                 41.27 ms │     no change │
│ QQuery 9  │ 50.26 ms │                 50.52 ms │     no change │
│ QQuery 10 │ 41.69 ms │                 41.08 ms │     no change │
│ QQuery 11 │ 14.16 ms │                 14.14 ms │     no change │
│ QQuery 12 │ 21.47 ms │                 21.50 ms │     no change │
│ QQuery 13 │ 41.94 ms │                 41.11 ms │     no change │
│ QQuery 14 │ 25.00 ms │                 24.88 ms │     no change │
│ QQuery 15 │ 31.48 ms │                 31.51 ms │     no change │
│ QQuery 16 │ 14.47 ms │                 14.36 ms │     no change │
│ QQuery 17 │ 75.84 ms │                 46.46 ms │ +1.63x faster │
│ QQuery 18 │ 62.09 ms │                 45.20 ms │ +1.37x faster │
│ QQuery 19 │ 32.36 ms │                 32.40 ms │     no change │
│ QQuery 20 │ 32.60 ms │                 32.84 ms │     no change │
│ QQuery 21 │ 56.88 ms │                 52.58 ms │ +1.08x faster │
│ QQuery 22 │ 14.53 ms │                 14.75 ms │     no change │
└───────────┴──────────┴──────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                       ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 757.73ms │
│ Total Time (perf-push-down-semi-join)   │ 706.69ms │
│ Average Time (HEAD)                     │  34.44ms │
│ Average Time (perf-push-down-semi-join) │  32.12ms │
│ Queries Faster                          │        3 │
│ Queries Slower                          │        0 │
│ Queries with No Change                  │       19 │
│ Queries with Failure                    │        0 │
└─────────────────────────────────────────┴──────────┘

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

Comparing HEAD and perf-push-down-semi-join
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃       perf-push-down-semi-join ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │ 41.30 / 42.19 ±1.37 / 44.91 ms │ 40.94 / 42.07 ±1.10 / 44.15 ms │     no change │
│ QQuery 2  │ 18.99 / 19.27 ±0.29 / 19.77 ms │ 18.90 / 19.16 ±0.26 / 19.56 ms │     no change │
│ QQuery 3  │ 28.86 / 29.09 ±0.22 / 29.52 ms │ 29.06 / 29.18 ±0.15 / 29.46 ms │     no change │
│ QQuery 4  │ 17.78 / 17.85 ±0.06 / 17.95 ms │ 17.64 / 18.04 ±0.50 / 19.01 ms │     no change │
│ QQuery 5  │ 36.30 / 37.36 ±1.04 / 38.83 ms │ 36.16 / 36.29 ±0.12 / 36.52 ms │     no change │
│ QQuery 6  │ 16.52 / 16.75 ±0.16 / 16.91 ms │ 16.51 / 16.67 ±0.16 / 16.97 ms │     no change │
│ QQuery 7  │ 41.66 / 44.76 ±2.47 / 48.96 ms │ 42.90 / 44.89 ±1.94 / 48.27 ms │     no change │
│ QQuery 8  │ 41.53 / 42.52 ±1.43 / 45.33 ms │ 41.27 / 42.58 ±1.96 / 46.46 ms │     no change │
│ QQuery 9  │ 50.26 / 51.53 ±0.98 / 53.05 ms │ 50.52 / 52.10 ±1.21 / 54.17 ms │     no change │
│ QQuery 10 │ 41.69 / 41.85 ±0.10 / 41.98 ms │ 41.08 / 41.53 ±0.39 / 42.26 ms │     no change │
│ QQuery 11 │ 14.16 / 14.34 ±0.22 / 14.78 ms │ 14.14 / 14.31 ±0.21 / 14.68 ms │     no change │
│ QQuery 12 │ 21.47 / 21.85 ±0.37 / 22.39 ms │ 21.50 / 21.89 ±0.34 / 22.43 ms │     no change │
│ QQuery 13 │ 41.94 / 42.27 ±0.23 / 42.57 ms │ 41.11 / 41.89 ±0.66 / 43.08 ms │     no change │
│ QQuery 14 │ 25.00 / 25.15 ±0.10 / 25.29 ms │ 24.88 / 25.09 ±0.16 / 25.30 ms │     no change │
│ QQuery 15 │ 31.48 / 32.18 ±0.59 / 33.05 ms │ 31.51 / 31.88 ±0.54 / 32.93 ms │     no change │
│ QQuery 16 │ 14.47 / 14.58 ±0.09 / 14.75 ms │ 14.36 / 14.67 ±0.29 / 15.16 ms │     no change │
│ QQuery 17 │ 75.84 / 76.81 ±1.14 / 79.04 ms │ 46.46 / 48.42 ±2.38 / 52.79 ms │ +1.59x faster │
│ QQuery 18 │ 62.09 / 64.83 ±2.00 / 67.48 ms │ 45.20 / 45.60 ±0.52 / 46.64 ms │ +1.42x faster │
│ QQuery 19 │ 32.36 / 32.81 ±0.57 / 33.93 ms │ 32.40 / 33.41 ±1.50 / 36.36 ms │     no change │
│ QQuery 20 │ 32.60 / 33.20 ±0.60 / 34.32 ms │ 32.84 / 33.28 ±0.58 / 34.38 ms │     no change │
│ QQuery 21 │ 56.88 / 57.62 ±0.64 / 58.79 ms │ 52.58 / 54.24 ±1.63 / 56.35 ms │ +1.06x faster │
│ QQuery 22 │ 14.53 / 14.69 ±0.09 / 14.79 ms │ 14.75 / 15.03 ±0.33 / 15.67 ms │     no change │
└───────────┴────────────────────────────────┴────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                       ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 773.52ms │
│ Total Time (perf-push-down-semi-join)   │ 722.23ms │
│ Average Time (HEAD)                     │  35.16ms │
│ Average Time (perf-push-down-semi-join) │  32.83ms │
│ Queries Faster                          │        3 │
│ Queries Slower                          │        0 │
│ Queries with No Change                  │       19 │
│ Queries with Failure                    │        0 │
└─────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric Value
Wall time 5.0s
Peak memory 1.3 GiB
Avg memory 513.4 MiB
CPU user 21.8s
CPU sys 1.9s
Peak spill 0 B

tpch — branch

Metric Value
Wall time 5.0s
Peak memory 993.2 MiB
Avg memory 450.1 MiB
CPU user 18.6s
CPU sys 1.9s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

adriangbot commented Oct 12, 2026

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing perf-push-down-semi-join (80fcd0f) to b9c8b3d (merge-base) diff

Run configuration 
run benchmark tpcds
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 perf-push-down-semi-join
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ perf-push-down-semi-join ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.49 ms │                  5.58 ms │      no change │
│ QQuery 2  │   80.44 ms │                 80.31 ms │      no change │
│ QQuery 3  │   28.95 ms │                 28.94 ms │      no change │
│ QQuery 4  │  449.28 ms │                451.87 ms │      no change │
│ QQuery 5  │   50.74 ms │                 52.01 ms │      no change │
│ QQuery 6  │   35.43 ms │                 27.46 ms │  +1.29x faster │
│ QQuery 7  │   74.82 ms │                 74.68 ms │      no change │
│ QQuery 8  │   36.26 ms │                 36.15 ms │      no change │
│ QQuery 9  │   49.82 ms │                 54.12 ms │   1.09x slower │
│ QQuery 10 │   62.61 ms │                 64.14 ms │      no change │
│ QQuery 11 │  281.28 ms │                277.53 ms │      no change │
│ QQuery 12 │   29.18 ms │                 29.37 ms │      no change │
│ QQuery 13 │  117.79 ms │                117.60 ms │      no change │
│ QQuery 14 │  393.98 ms │                262.72 ms │  +1.50x faster │
│ QQuery 15 │   54.74 ms │                 21.61 ms │  +2.53x faster │
│ QQuery 16 │    6.49 ms │                  6.53 ms │      no change │
│ QQuery 17 │   78.11 ms │                 64.56 ms │  +1.21x faster │
│ QQuery 18 │  104.82 ms │                104.74 ms │      no change │
│ QQuery 19 │   42.20 ms │                 41.77 ms │      no change │
│ QQuery 20 │   35.98 ms │                 36.02 ms │      no change │
│ QQuery 21 │   16.99 ms │                 17.22 ms │      no change │
│ QQuery 22 │   63.52 ms │                 64.34 ms │      no change │
│ QQuery 23 │  314.24 ms │                310.46 ms │      no change │
│ QQuery 24 │  196.18 ms │                198.36 ms │      no change │
│ QQuery 25 │  107.95 ms │                 81.51 ms │  +1.32x faster │
│ QQuery 26 │   48.58 ms │                 48.68 ms │      no change │
│ QQuery 27 │    5.86 ms │                  6.31 ms │   1.08x slower │
│ QQuery 28 │   57.12 ms │                 56.60 ms │      no change │
│ QQuery 29 │   95.80 ms │                 66.95 ms │  +1.43x faster │
│ QQuery 30 │   32.97 ms │                 33.05 ms │      no change │
│ QQuery 31 │  109.95 ms │                110.09 ms │      no change │
│ QQuery 32 │   20.50 ms │                 20.58 ms │      no change │
│ QQuery 33 │   37.81 ms │                 38.20 ms │      no change │
│ QQuery 34 │   10.19 ms │                 10.08 ms │      no change │
│ QQuery 35 │   73.54 ms │                 69.59 ms │  +1.06x faster │
│ QQuery 36 │    5.77 ms │                  6.49 ms │   1.12x slower │
│ QQuery 37 │    6.87 ms │                  6.87 ms │      no change │
│ QQuery 38 │   62.42 ms │                 62.16 ms │      no change │
│ QQuery 39 │   73.66 ms │                 75.47 ms │      no change │
│ QQuery 40 │   23.36 ms │                 23.77 ms │      no change │
│ QQuery 41 │   10.98 ms │                 11.29 ms │      no change │
│ QQuery 42 │   24.04 ms │                 24.08 ms │      no change │
│ QQuery 43 │    4.77 ms │                  4.80 ms │      no change │
│ QQuery 44 │    8.95 ms │                  8.75 ms │      no change │
│ QQuery 45 │   38.27 ms │                 21.75 ms │  +1.76x faster │
│ QQuery 46 │   11.94 ms │                 11.89 ms │      no change │
│ QQuery 47 │  222.82 ms │                223.83 ms │      no change │
│ QQuery 48 │   94.81 ms │                 95.05 ms │      no change │
│ QQuery 49 │   71.45 ms │                 69.95 ms │      no change │
│ QQuery 50 │   57.98 ms │                 65.61 ms │   1.13x slower │
│ QQuery 51 │   92.59 ms │                 93.02 ms │      no change │
│ QQuery 52 │   24.40 ms │                 24.32 ms │      no change │
│ QQuery 53 │   29.26 ms │                 29.21 ms │      no change │
│ QQuery 54 │   54.13 ms │                 54.82 ms │      no change │
│ QQuery 55 │   23.42 ms │                 23.74 ms │      no change │
│ QQuery 56 │   38.47 ms │                 38.61 ms │      no change │
│ QQuery 57 │  168.95 ms │                169.95 ms │      no change │
│ QQuery 58 │  108.21 ms │                 53.17 ms │  +2.04x faster │
│ QQuery 59 │  116.17 ms │                115.42 ms │      no change │
│ QQuery 60 │   39.02 ms │                 38.86 ms │      no change │
│ QQuery 61 │   11.58 ms │                 11.69 ms │      no change │
│ QQuery 62 │   44.77 ms │                 31.00 ms │  +1.44x faster │
│ QQuery 63 │   29.52 ms │                 29.37 ms │      no change │
│ QQuery 64 │  361.44 ms │                361.86 ms │      no change │
│ QQuery 65 │  129.39 ms │                 66.22 ms │  +1.95x faster │
│ QQuery 66 │   77.49 ms │                 77.90 ms │      no change │
│ QQuery 67 │  248.39 ms │                252.17 ms │      no change │
│ QQuery 68 │   12.10 ms │                 12.14 ms │      no change │
│ QQuery 69 │   57.32 ms │                 57.70 ms │      no change │
│ QQuery 70 │  104.65 ms │                 98.69 ms │  +1.06x faster │
│ QQuery 71 │   35.55 ms │                 34.85 ms │      no change │
│ QQuery 72 │ 1726.75 ms │                112.25 ms │ +15.38x faster │
│ QQuery 73 │   10.09 ms │                  9.86 ms │      no change │
│ QQuery 74 │  162.38 ms │                161.13 ms │      no change │
│ QQuery 75 │  141.37 ms │                140.76 ms │      no change │
│ QQuery 76 │   34.34 ms │                 34.24 ms │      no change │
│ QQuery 77 │   61.37 ms │                 60.58 ms │      no change │
│ QQuery 78 │  162.78 ms │                160.79 ms │      no change │
│ QQuery 79 │   65.99 ms │                 66.19 ms │      no change │
│ QQuery 80 │   96.61 ms │                 95.93 ms │      no change │
│ QQuery 81 │   26.12 ms │                 26.14 ms │      no change │
│ QQuery 82 │   16.41 ms │                 17.06 ms │      no change │
│ QQuery 83 │   33.69 ms │                 27.51 ms │  +1.22x faster │
│ QQuery 84 │   29.81 ms │                 29.84 ms │      no change │
│ QQuery 85 │  103.56 ms │                103.23 ms │      no change │
│ QQuery 86 │   25.88 ms │                 26.05 ms │      no change │
│ QQuery 87 │   61.91 ms │                 61.82 ms │      no change │
│ QQuery 88 │   61.66 ms │                 62.04 ms │      no change │
│ QQuery 89 │   35.02 ms │                 35.16 ms │      no change │
│ QQuery 90 │   16.66 ms │                 16.85 ms │      no change │
│ QQuery 91 │   44.63 ms │                 42.95 ms │      no change │
│ QQuery 92 │   29.78 ms │                 29.90 ms │      no change │
│ QQuery 93 │   49.03 ms │                 49.74 ms │      no change │
│ QQuery 94 │   38.75 ms │                 38.53 ms │      no change │
│ QQuery 95 │   79.75 ms │                 79.80 ms │      no change │
│ QQuery 96 │   24.00 ms │                 23.97 ms │      no change │
│ QQuery 97 │   51.91 ms │                 51.77 ms │      no change │
│ QQuery 98 │   43.05 ms │                 42.93 ms │      no change │
│ QQuery 99 │   65.86 ms │                 45.28 ms │  +1.45x faster │
└───────────┴────────────┴──────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 9033.66ms │
│ Total Time (perf-push-down-semi-join)   │ 7008.53ms │
│ Average Time (HEAD)                     │   91.25ms │
│ Average Time (perf-push-down-semi-join) │   70.79ms │
│ Queries Faster                          │        15 │
│ Queries Slower                          │         4 │
│ Queries with No Change                  │        80 │
│ Queries with Failure                    │         0 │
└─────────────────────────────────────────┴───────────┘

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

Comparing HEAD and perf-push-down-semi-join
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃          perf-push-down-semi-join ┃         Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.49 / 6.18 ±1.06 / 8.27 ms │       5.58 / 6.18 ±1.05 / 8.26 ms │      no change │
│ QQuery 2  │        80.44 / 81.19 ±0.58 / 81.86 ms │    80.31 / 80.55 ±0.18 / 80.78 ms │      no change │
│ QQuery 3  │        28.95 / 29.18 ±0.21 / 29.52 ms │    28.94 / 29.21 ±0.19 / 29.45 ms │      no change │
│ QQuery 4  │     449.28 / 453.97 ±2.96 / 457.06 ms │ 451.87 / 454.25 ±1.22 / 455.34 ms │      no change │
│ QQuery 5  │        50.74 / 52.88 ±2.05 / 56.53 ms │    52.01 / 53.40 ±2.23 / 57.81 ms │      no change │
│ QQuery 6  │        35.43 / 35.93 ±0.38 / 36.57 ms │    27.46 / 27.56 ±0.07 / 27.64 ms │  +1.30x faster │
│ QQuery 7  │        74.82 / 75.59 ±0.56 / 76.41 ms │    74.68 / 75.36 ±0.56 / 76.11 ms │      no change │
│ QQuery 8  │        36.26 / 36.65 ±0.27 / 36.95 ms │    36.15 / 36.51 ±0.27 / 36.95 ms │      no change │
│ QQuery 9  │        49.82 / 54.76 ±3.44 / 59.59 ms │    54.12 / 54.91 ±0.79 / 56.13 ms │      no change │
│ QQuery 10 │        62.61 / 62.88 ±0.24 / 63.25 ms │    64.14 / 64.32 ±0.16 / 64.60 ms │      no change │
│ QQuery 11 │     281.28 / 284.21 ±1.93 / 286.76 ms │ 277.53 / 281.49 ±2.86 / 285.84 ms │      no change │
│ QQuery 12 │        29.18 / 30.69 ±2.43 / 35.51 ms │    29.37 / 29.61 ±0.18 / 29.92 ms │      no change │
│ QQuery 13 │     117.79 / 118.41 ±0.41 / 118.88 ms │ 117.60 / 119.69 ±1.54 / 121.94 ms │      no change │
│ QQuery 14 │     393.98 / 398.21 ±5.18 / 408.34 ms │ 262.72 / 264.21 ±1.07 / 265.87 ms │  +1.51x faster │
│ QQuery 15 │        54.74 / 55.29 ±0.44 / 56.00 ms │    21.61 / 21.77 ±0.11 / 21.93 ms │  +2.54x faster │
│ QQuery 16 │           6.49 / 6.67 ±0.23 / 7.11 ms │      6.53 / 7.85 ±2.26 / 12.34 ms │   1.18x slower │
│ QQuery 17 │        78.11 / 79.16 ±1.06 / 81.19 ms │    64.56 / 66.32 ±1.53 / 69.16 ms │  +1.19x faster │
│ QQuery 18 │     104.82 / 106.40 ±1.69 / 109.47 ms │ 104.74 / 106.56 ±2.20 / 110.87 ms │      no change │
│ QQuery 19 │        42.20 / 42.58 ±0.52 / 43.60 ms │    41.77 / 42.63 ±1.28 / 45.18 ms │      no change │
│ QQuery 20 │        35.98 / 36.35 ±0.33 / 36.83 ms │    36.02 / 36.38 ±0.20 / 36.60 ms │      no change │
│ QQuery 21 │        16.99 / 17.32 ±0.23 / 17.60 ms │    17.22 / 17.51 ±0.23 / 17.82 ms │      no change │
│ QQuery 22 │        63.52 / 65.30 ±1.65 / 68.05 ms │    64.34 / 65.22 ±0.51 / 65.66 ms │      no change │
│ QQuery 23 │     314.24 / 315.82 ±1.65 / 318.99 ms │ 310.46 / 313.98 ±3.85 / 320.81 ms │      no change │
│ QQuery 24 │     196.18 / 197.48 ±1.22 / 199.66 ms │ 198.36 / 199.64 ±1.00 / 201.26 ms │      no change │
│ QQuery 25 │     107.95 / 109.08 ±1.09 / 110.54 ms │    81.51 / 83.65 ±2.29 / 87.79 ms │  +1.30x faster │
│ QQuery 26 │        48.58 / 49.01 ±0.30 / 49.40 ms │    48.68 / 49.21 ±0.38 / 49.77 ms │      no change │
│ QQuery 27 │           5.86 / 6.45 ±0.75 / 7.89 ms │       6.31 / 6.46 ±0.16 / 6.75 ms │      no change │
│ QQuery 28 │        57.12 / 59.40 ±1.64 / 61.16 ms │    56.60 / 59.13 ±2.89 / 64.10 ms │      no change │
│ QQuery 29 │       95.80 / 98.32 ±2.31 / 102.06 ms │    66.95 / 68.19 ±1.80 / 71.70 ms │  +1.44x faster │
│ QQuery 30 │        32.97 / 33.22 ±0.19 / 33.44 ms │    33.05 / 33.58 ±0.46 / 34.35 ms │      no change │
│ QQuery 31 │     109.95 / 111.08 ±1.47 / 113.87 ms │ 110.09 / 110.56 ±0.38 / 111.16 ms │      no change │
│ QQuery 32 │        20.50 / 20.99 ±0.28 / 21.36 ms │    20.58 / 20.97 ±0.27 / 21.31 ms │      no change │
│ QQuery 33 │        37.81 / 38.02 ±0.21 / 38.28 ms │    38.20 / 39.76 ±2.20 / 44.06 ms │      no change │
│ QQuery 34 │        10.19 / 10.27 ±0.10 / 10.43 ms │    10.08 / 10.31 ±0.17 / 10.56 ms │      no change │
│ QQuery 35 │        73.54 / 74.27 ±0.76 / 75.69 ms │    69.59 / 70.51 ±0.70 / 71.62 ms │  +1.05x faster │
│ QQuery 36 │           5.77 / 5.90 ±0.21 / 6.33 ms │       6.49 / 6.69 ±0.28 / 7.23 ms │   1.13x slower │
│ QQuery 37 │           6.87 / 6.98 ±0.09 / 7.11 ms │       6.87 / 7.08 ±0.15 / 7.34 ms │      no change │
│ QQuery 38 │        62.42 / 64.11 ±2.70 / 69.41 ms │    62.16 / 62.39 ±0.23 / 62.75 ms │      no change │
│ QQuery 39 │        73.66 / 74.40 ±0.59 / 75.12 ms │    75.47 / 77.17 ±2.47 / 81.96 ms │      no change │
│ QQuery 40 │        23.36 / 24.43 ±0.98 / 26.26 ms │    23.77 / 24.74 ±0.96 / 26.48 ms │      no change │
│ QQuery 41 │        10.98 / 11.20 ±0.27 / 11.72 ms │    11.29 / 11.42 ±0.14 / 11.69 ms │      no change │
│ QQuery 42 │        24.04 / 24.34 ±0.42 / 25.15 ms │    24.08 / 25.09 ±0.89 / 26.44 ms │      no change │
│ QQuery 43 │           4.77 / 4.90 ±0.21 / 5.32 ms │       4.80 / 4.96 ±0.21 / 5.37 ms │      no change │
│ QQuery 44 │           8.95 / 9.01 ±0.05 / 9.10 ms │       8.75 / 8.98 ±0.13 / 9.11 ms │      no change │
│ QQuery 45 │        38.27 / 39.61 ±1.79 / 43.16 ms │    21.75 / 21.97 ±0.17 / 22.20 ms │  +1.80x faster │
│ QQuery 46 │        11.94 / 12.32 ±0.34 / 12.84 ms │    11.89 / 12.23 ±0.36 / 12.91 ms │      no change │
│ QQuery 47 │     222.82 / 229.95 ±4.87 / 235.91 ms │ 223.83 / 229.13 ±3.78 / 234.65 ms │      no change │
│ QQuery 48 │        94.81 / 95.51 ±0.38 / 95.85 ms │    95.05 / 95.97 ±0.82 / 97.47 ms │      no change │
│ QQuery 49 │        71.45 / 72.83 ±1.48 / 75.63 ms │    69.95 / 71.47 ±1.02 / 73.16 ms │      no change │
│ QQuery 50 │        57.98 / 58.64 ±0.90 / 60.40 ms │    65.61 / 67.23 ±2.03 / 71.02 ms │   1.15x slower │
│ QQuery 51 │        92.59 / 93.47 ±0.72 / 94.48 ms │    93.02 / 95.52 ±1.74 / 98.31 ms │      no change │
│ QQuery 52 │        24.40 / 24.69 ±0.16 / 24.91 ms │    24.32 / 24.47 ±0.12 / 24.68 ms │      no change │
│ QQuery 53 │        29.26 / 29.62 ±0.24 / 29.87 ms │    29.21 / 29.54 ±0.26 / 29.93 ms │      no change │
│ QQuery 54 │        54.13 / 55.10 ±1.14 / 57.32 ms │    54.82 / 55.28 ±0.47 / 56.12 ms │      no change │
│ QQuery 55 │        23.42 / 23.88 ±0.24 / 24.12 ms │    23.74 / 24.28 ±0.51 / 25.22 ms │      no change │
│ QQuery 56 │        38.47 / 39.22 ±0.59 / 40.23 ms │    38.61 / 39.37 ±0.65 / 40.55 ms │      no change │
│ QQuery 57 │     168.95 / 172.10 ±2.33 / 175.92 ms │ 169.95 / 172.43 ±2.87 / 177.93 ms │      no change │
│ QQuery 58 │     108.21 / 110.27 ±1.45 / 112.11 ms │    53.17 / 54.48 ±1.12 / 56.34 ms │  +2.02x faster │
│ QQuery 59 │     116.17 / 117.26 ±0.86 / 118.70 ms │ 115.42 / 116.52 ±0.84 / 117.59 ms │      no change │
│ QQuery 60 │        39.02 / 39.48 ±0.30 / 39.94 ms │    38.86 / 39.95 ±0.72 / 41.11 ms │      no change │
│ QQuery 61 │        11.58 / 11.83 ±0.20 / 12.16 ms │    11.69 / 11.88 ±0.13 / 12.07 ms │      no change │
│ QQuery 62 │        44.77 / 45.28 ±0.55 / 46.13 ms │    31.00 / 32.07 ±1.17 / 34.24 ms │  +1.41x faster │
│ QQuery 63 │        29.52 / 29.94 ±0.35 / 30.38 ms │    29.37 / 29.54 ±0.12 / 29.72 ms │      no change │
│ QQuery 64 │     361.44 / 363.72 ±1.50 / 365.93 ms │ 361.86 / 366.06 ±4.47 / 372.25 ms │      no change │
│ QQuery 65 │     129.39 / 131.27 ±2.12 / 134.73 ms │    66.22 / 67.14 ±0.49 / 67.64 ms │  +1.96x faster │
│ QQuery 66 │        77.49 / 79.79 ±2.80 / 85.24 ms │    77.90 / 80.62 ±4.12 / 88.61 ms │      no change │
│ QQuery 67 │     248.39 / 253.05 ±2.73 / 256.43 ms │ 252.17 / 254.69 ±2.34 / 258.37 ms │      no change │
│ QQuery 68 │        12.10 / 12.24 ±0.15 / 12.51 ms │    12.14 / 12.29 ±0.10 / 12.40 ms │      no change │
│ QQuery 69 │        57.32 / 57.94 ±0.78 / 59.46 ms │    57.70 / 57.90 ±0.16 / 58.18 ms │      no change │
│ QQuery 70 │     104.65 / 109.06 ±5.84 / 120.26 ms │  98.69 / 102.56 ±6.51 / 115.54 ms │  +1.06x faster │
│ QQuery 71 │        35.55 / 36.11 ±0.53 / 37.09 ms │    34.85 / 36.19 ±1.09 / 37.96 ms │      no change │
│ QQuery 72 │ 1726.75 / 1804.81 ±57.20 / 1884.98 ms │ 112.25 / 115.15 ±3.13 / 120.89 ms │ +15.67x faster │
│ QQuery 73 │        10.09 / 10.59 ±0.52 / 11.56 ms │      9.86 / 9.99 ±0.13 / 10.22 ms │  +1.06x faster │
│ QQuery 74 │     162.38 / 164.28 ±1.11 / 165.37 ms │ 161.13 / 163.84 ±2.12 / 167.47 ms │      no change │
│ QQuery 75 │     141.37 / 144.00 ±2.20 / 147.00 ms │ 140.76 / 142.05 ±1.31 / 144.53 ms │      no change │
│ QQuery 76 │        34.34 / 34.58 ±0.26 / 34.90 ms │    34.24 / 34.49 ±0.17 / 34.74 ms │      no change │
│ QQuery 77 │        61.37 / 62.94 ±2.25 / 67.25 ms │    60.58 / 62.95 ±3.12 / 69.11 ms │      no change │
│ QQuery 78 │     162.78 / 166.08 ±2.88 / 170.77 ms │ 160.79 / 166.01 ±4.36 / 171.94 ms │      no change │
│ QQuery 79 │        65.99 / 67.96 ±2.52 / 72.82 ms │    66.19 / 66.79 ±0.46 / 67.31 ms │      no change │
│ QQuery 80 │        96.61 / 97.96 ±1.07 / 99.79 ms │    95.93 / 97.00 ±1.13 / 99.10 ms │      no change │
│ QQuery 81 │        26.12 / 26.30 ±0.11 / 26.41 ms │    26.14 / 26.49 ±0.31 / 26.94 ms │      no change │
│ QQuery 82 │        16.41 / 16.74 ±0.26 / 17.13 ms │    17.06 / 19.97 ±4.91 / 29.76 ms │   1.19x slower │
│ QQuery 83 │        33.69 / 35.84 ±2.76 / 41.11 ms │    27.51 / 27.88 ±0.21 / 28.12 ms │  +1.29x faster │
│ QQuery 84 │        29.81 / 30.28 ±0.32 / 30.72 ms │    29.84 / 29.98 ±0.14 / 30.23 ms │      no change │
│ QQuery 85 │     103.56 / 105.47 ±2.09 / 109.23 ms │ 103.23 / 104.65 ±1.71 / 107.70 ms │      no change │
│ QQuery 86 │        25.88 / 26.02 ±0.12 / 26.22 ms │    26.05 / 27.87 ±3.39 / 34.63 ms │   1.07x slower │
│ QQuery 87 │        61.91 / 62.71 ±0.64 / 63.44 ms │    61.82 / 62.55 ±0.69 / 63.82 ms │      no change │
│ QQuery 88 │        61.66 / 63.25 ±2.50 / 68.20 ms │    62.04 / 62.51 ±0.52 / 63.47 ms │      no change │
│ QQuery 89 │        35.02 / 35.87 ±0.62 / 36.91 ms │    35.16 / 35.47 ±0.23 / 35.82 ms │      no change │
│ QQuery 90 │        16.66 / 16.92 ±0.24 / 17.34 ms │    16.85 / 17.06 ±0.21 / 17.46 ms │      no change │
│ QQuery 91 │        44.63 / 46.15 ±1.46 / 48.23 ms │    42.95 / 44.94 ±1.98 / 48.68 ms │      no change │
│ QQuery 92 │        29.78 / 30.24 ±0.50 / 31.20 ms │    29.90 / 30.74 ±0.63 / 31.75 ms │      no change │
│ QQuery 93 │        49.03 / 50.35 ±0.94 / 51.63 ms │    49.74 / 50.49 ±0.47 / 50.95 ms │      no change │
│ QQuery 94 │        38.75 / 39.13 ±0.30 / 39.47 ms │    38.53 / 38.92 ±0.27 / 39.30 ms │      no change │
│ QQuery 95 │        79.75 / 80.95 ±1.20 / 83.08 ms │    79.80 / 81.50 ±2.36 / 86.10 ms │      no change │
│ QQuery 96 │        24.00 / 24.10 ±0.06 / 24.16 ms │    23.97 / 24.14 ±0.18 / 24.40 ms │      no change │
│ QQuery 97 │        51.91 / 54.07 ±3.18 / 60.38 ms │    51.77 / 52.80 ±0.85 / 53.92 ms │      no change │
│ QQuery 98 │        43.05 / 43.80 ±0.38 / 44.10 ms │    42.93 / 43.22 ±0.36 / 43.93 ms │      no change │
│ QQuery 99 │        65.86 / 66.11 ±0.21 / 66.44 ms │    45.28 / 45.46 ±0.15 / 45.66 ms │  +1.45x faster │
└───────────┴───────────────────────────────────────┴───────────────────────────────────┴────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 9230.20ms │
│ Total Time (perf-push-down-semi-join)   │ 7121.63ms │
│ Average Time (HEAD)                     │   93.23ms │
│ Average Time (perf-push-down-semi-join) │   71.94ms │
│ Queries Faster                          │        16 │
│ Queries Slower                          │         5 │
│ Queries with No Change                  │        78 │
│ Queries with Failure                    │         0 │
└─────────────────────────────────────────┴───────────┘

Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 50.0s
Peak memory 2.4 GiB
Avg memory 1.6 GiB
CPU user 198.3s
CPU sys 5.7s
Peak spill 0 B

tpcds — branch

Metric Value
Wall time 40.0s
Peak memory 2.1 GiB
Avg memory 1.3 GiB
CPU user 101.4s
CPU sys 5.2s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

adriangbot commented Oct 12, 2026

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing perf-push-down-semi-join (80fcd0f) to b9c8b3d (merge-base) diff

Run configuration 
run benchmark clickbench_partitioned
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 perf-push-down-semi-join
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ perf-push-down-semi-join ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.25 ms │                  1.27 ms │     no change │
│ QQuery 1  │   12.05 ms │                 11.83 ms │     no change │
│ QQuery 2  │   37.02 ms │                 37.01 ms │     no change │
│ QQuery 3  │   32.43 ms │                 32.15 ms │     no change │
│ QQuery 4  │  237.57 ms │                243.68 ms │     no change │
│ QQuery 5  │  275.93 ms │                273.33 ms │     no change │
│ QQuery 6  │    1.35 ms │                  1.27 ms │ +1.06x faster │
│ QQuery 7  │   12.97 ms │                 13.14 ms │     no change │
│ QQuery 8  │  328.33 ms │                336.29 ms │     no change │
│ QQuery 9  │  465.68 ms │                465.87 ms │     no change │
│ QQuery 10 │   65.84 ms │                 64.27 ms │     no change │
│ QQuery 11 │   76.31 ms │                 74.39 ms │     no change │
│ QQuery 12 │  276.27 ms │                263.89 ms │     no change │
│ QQuery 13 │  368.19 ms │                376.61 ms │     no change │
│ QQuery 14 │  288.97 ms │                288.11 ms │     no change │
│ QQuery 15 │  303.25 ms │                304.95 ms │     no change │
│ QQuery 16 │  632.66 ms │                648.85 ms │     no change │
│ QQuery 17 │  641.63 ms │                645.10 ms │     no change │
│ QQuery 18 │ 1254.90 ms │               1299.43 ms │     no change │
│ QQuery 19 │   27.84 ms │                 28.13 ms │     no change │
│ QQuery 20 │  530.67 ms │                528.51 ms │     no change │
│ QQuery 21 │  522.38 ms │                523.12 ms │     no change │
│ QQuery 22 │ 1029.84 ms │               1021.96 ms │     no change │
│ QQuery 23 │ 3125.71 ms │               3121.74 ms │     no change │
│ QQuery 24 │   41.44 ms │                 44.91 ms │  1.08x slower │
│ QQuery 25 │  106.33 ms │                108.38 ms │     no change │
│ QQuery 26 │   41.36 ms │                 41.48 ms │     no change │
│ QQuery 27 │  523.60 ms │                518.55 ms │     no change │
│ QQuery 28 │ 2836.12 ms │               2871.20 ms │     no change │
│ QQuery 29 │   42.39 ms │                 41.92 ms │     no change │
│ QQuery 30 │  314.74 ms │                322.48 ms │     no change │
│ QQuery 31 │  282.67 ms │                280.10 ms │     no change │
│ QQuery 32 │ 1069.71 ms │               1042.71 ms │     no change │
│ QQuery 33 │ 1660.76 ms │               1590.79 ms │     no change │
│ QQuery 34 │ 1601.98 ms │               1604.19 ms │     no change │
│ QQuery 35 │  311.64 ms │                294.42 ms │ +1.06x faster │
│ QQuery 36 │   67.00 ms │                 68.88 ms │     no change │
│ QQuery 37 │   35.21 ms │                 35.42 ms │     no change │
│ QQuery 38 │   40.96 ms │                 44.79 ms │  1.09x slower │
│ QQuery 39 │  141.29 ms │                150.36 ms │  1.06x slower │
│ QQuery 40 │   14.35 ms │                 14.30 ms │     no change │
│ QQuery 41 │   13.95 ms │                 13.71 ms │     no change │
│ QQuery 42 │   14.12 ms │                 13.03 ms │ +1.08x faster │
└───────────┴────────────┴──────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 19708.67ms │
│ Total Time (perf-push-down-semi-join)   │ 19706.51ms │
│ Average Time (HEAD)                     │   458.34ms │
│ Average Time (perf-push-down-semi-join) │   458.29ms │
│ Queries Faster                          │          3 │
│ Queries Slower                          │          3 │
│ Queries with No Change                  │         37 │
│ Queries with Failure                    │          0 │
└─────────────────────────────────────────┴────────────┘

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

Comparing HEAD and perf-push-down-semi-join
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃               perf-push-down-semi-join ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.25 / 4.17 ±5.69 / 15.54 ms │           1.27 / 4.25 ±5.79 / 15.83 ms │     no change │
│ QQuery 1  │         12.05 / 12.19 ±0.09 / 12.32 ms │         11.83 / 13.01 ±1.10 / 15.09 ms │  1.07x slower │
│ QQuery 2  │         37.02 / 37.39 ±0.34 / 37.82 ms │         37.01 / 37.51 ±0.35 / 38.12 ms │     no change │
│ QQuery 3  │         32.43 / 33.11 ±0.69 / 34.32 ms │         32.15 / 32.82 ±0.64 / 34.01 ms │     no change │
│ QQuery 4  │      237.57 / 241.37 ±3.39 / 247.74 ms │      243.68 / 247.05 ±2.46 / 250.42 ms │     no change │
│ QQuery 5  │      275.93 / 279.91 ±2.92 / 283.62 ms │      273.33 / 277.58 ±2.84 / 281.94 ms │     no change │
│ QQuery 6  │            1.35 / 1.59 ±0.24 / 1.94 ms │            1.27 / 1.42 ±0.24 / 1.90 ms │ +1.11x faster │
│ QQuery 7  │         12.97 / 13.13 ±0.09 / 13.24 ms │         13.14 / 13.25 ±0.13 / 13.42 ms │     no change │
│ QQuery 8  │      328.33 / 335.46 ±4.54 / 340.92 ms │      336.29 / 340.91 ±5.43 / 351.45 ms │     no change │
│ QQuery 9  │      465.68 / 475.12 ±6.09 / 482.51 ms │     465.87 / 478.28 ±12.08 / 496.55 ms │     no change │
│ QQuery 10 │         65.84 / 66.72 ±0.62 / 67.40 ms │         64.27 / 67.24 ±2.89 / 72.49 ms │     no change │
│ QQuery 11 │         76.31 / 80.20 ±5.21 / 90.17 ms │         74.39 / 75.41 ±0.76 / 76.69 ms │ +1.06x faster │
│ QQuery 12 │      276.27 / 282.05 ±5.25 / 289.58 ms │      263.89 / 272.28 ±5.70 / 279.99 ms │     no change │
│ QQuery 13 │      368.19 / 375.96 ±5.35 / 384.14 ms │     376.61 / 397.28 ±17.16 / 421.25 ms │  1.06x slower │
│ QQuery 14 │      288.97 / 295.76 ±7.22 / 309.30 ms │      288.11 / 294.53 ±7.56 / 308.86 ms │     no change │
│ QQuery 15 │      303.25 / 312.15 ±6.75 / 321.78 ms │      304.95 / 308.77 ±2.67 / 312.60 ms │     no change │
│ QQuery 16 │     632.66 / 650.95 ±10.80 / 665.02 ms │      648.85 / 655.43 ±5.67 / 664.96 ms │     no change │
│ QQuery 17 │      641.63 / 656.23 ±9.02 / 669.99 ms │     645.10 / 661.57 ±18.60 / 697.79 ms │     no change │
│ QQuery 18 │  1254.90 / 1312.23 ±46.35 / 1372.50 ms │  1299.43 / 1312.39 ±10.17 / 1325.30 ms │     no change │
│ QQuery 19 │         27.84 / 28.66 ±0.82 / 30.12 ms │        28.13 / 36.38 ±15.61 / 67.59 ms │  1.27x slower │
│ QQuery 20 │      530.67 / 538.91 ±6.22 / 546.45 ms │      528.51 / 534.40 ±6.50 / 542.66 ms │     no change │
│ QQuery 21 │      522.38 / 526.34 ±2.60 / 530.18 ms │      523.12 / 529.03 ±6.32 / 540.92 ms │     no change │
│ QQuery 22 │  1029.84 / 1039.39 ±12.44 / 1063.85 ms │   1021.96 / 1036.36 ±8.12 / 1046.86 ms │     no change │
│ QQuery 23 │ 3125.71 / 3376.31 ±157.14 / 3584.15 ms │ 3121.74 / 3351.03 ±115.29 / 3420.30 ms │     no change │
│ QQuery 24 │      41.44 / 123.13 ±66.35 / 212.75 ms │     44.91 / 187.89 ±127.44 / 419.97 ms │  1.53x slower │
│ QQuery 25 │     106.33 / 130.90 ±46.77 / 224.41 ms │      108.38 / 108.79 ±0.45 / 109.67 ms │ +1.20x faster │
│ QQuery 26 │         41.36 / 42.62 ±1.73 / 46.02 ms │         41.48 / 42.63 ±0.96 / 44.40 ms │     no change │
│ QQuery 27 │      523.60 / 529.31 ±4.64 / 537.30 ms │      518.55 / 528.91 ±6.47 / 536.72 ms │     no change │
│ QQuery 28 │  2836.12 / 2914.84 ±40.32 / 2947.19 ms │  2871.20 / 2931.66 ±44.63 / 2994.45 ms │     no change │
│ QQuery 29 │         42.39 / 49.91 ±8.71 / 63.24 ms │       41.92 / 63.53 ±34.69 / 131.63 ms │  1.27x slower │
│ QQuery 30 │      314.74 / 319.96 ±5.21 / 326.74 ms │     322.48 / 331.39 ±10.53 / 351.55 ms │     no change │
│ QQuery 31 │     282.67 / 297.66 ±13.53 / 319.12 ms │     280.10 / 300.60 ±15.45 / 321.32 ms │     no change │
│ QQuery 32 │ 1069.71 / 1196.97 ±103.67 / 1362.35 ms │  1042.71 / 1082.97 ±43.43 / 1165.07 ms │ +1.11x faster │
│ QQuery 33 │ 1660.76 / 1743.30 ±109.60 / 1951.64 ms │ 1590.79 / 1735.21 ±102.30 / 1870.79 ms │     no change │
│ QQuery 34 │  1601.98 / 1662.48 ±47.09 / 1720.03 ms │ 1604.19 / 1699.72 ±107.03 / 1899.19 ms │     no change │
│ QQuery 35 │    311.64 / 447.91 ±128.30 / 634.29 ms │     294.42 / 320.11 ±28.22 / 373.62 ms │ +1.40x faster │
│ QQuery 36 │         67.00 / 70.14 ±3.69 / 77.36 ms │         68.88 / 75.66 ±5.16 / 84.44 ms │  1.08x slower │
│ QQuery 37 │         35.21 / 36.25 ±0.69 / 37.33 ms │         35.42 / 40.54 ±5.40 / 49.25 ms │  1.12x slower │
│ QQuery 38 │        40.96 / 47.38 ±10.94 / 69.14 ms │         44.79 / 46.99 ±2.62 / 51.88 ms │     no change │
│ QQuery 39 │     141.29 / 159.30 ±13.17 / 179.43 ms │      150.36 / 152.65 ±1.75 / 154.71 ms │     no change │
│ QQuery 40 │         14.35 / 14.70 ±0.28 / 15.12 ms │         14.30 / 18.95 ±5.39 / 25.61 ms │  1.29x slower │
│ QQuery 41 │        13.95 / 28.99 ±27.97 / 84.88 ms │         13.71 / 14.17 ±0.49 / 15.05 ms │ +2.05x faster │
│ QQuery 42 │         14.12 / 14.31 ±0.17 / 14.52 ms │         13.03 / 13.52 ±0.31 / 13.84 ms │ +1.06x faster │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 20805.33ms │
│ Total Time (perf-push-down-semi-join)   │ 20674.11ms │
│ Average Time (HEAD)                     │   483.84ms │
│ Average Time (perf-push-down-semi-join) │   480.79ms │
│ Queries Faster                          │          7 │
│ Queries Slower                          │          8 │
│ Queries with No Change                  │         28 │
│ Queries with Failure                    │          0 │
└─────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 105.0s
Peak memory 17.8 GiB
Avg memory 6.0 GiB
CPU user 1016.2s
CPU sys 104.2s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 105.0s
Peak memory 17.7 GiB
Avg memory 6.3 GiB
CPU user 1009.7s
CPU sys 102.8s
Peak spill 0 B

File an issue against this benchmark runner

@github-actions

github-actions Bot commented Oct 12, 2026 •
edited
Loading

Copy link
Copy Markdown

Thank you for opening this pull request!

Reviewer note: cargo-semver-checks reported the current version number is not SemVer-compatible with the changes in this pull request (compared against the base branch).

Details 
     Cloning apache/main
    Building datafusion v55.1.0 (current)
       Built [  54.337s] (current)
     Parsing datafusion v55.1.0 (current)
      Parsed [   0.031s] (current)
    Building datafusion v55.1.0 (baseline)
       Built [  53.744s] (baseline)
     Parsing datafusion v55.1.0 (baseline)
      Parsed [   0.032s] (baseline)
    Checking datafusion v55.1.0 -> v55.1.0 (no change; assume patch)
     Checked [   0.594s] 223 checks: 223 pass, 31 skip
     Summary no semver update required
    Finished [ 110.548s] datafusion
    Building datafusion-common v55.1.0 (current)
       Built [  31.691s] (current)
     Parsing datafusion-common v55.1.0 (current)
      Parsed [   0.062s] (current)
    Building datafusion-common v55.1.0 (baseline)
       Built [  31.652s] (baseline)
     Parsing datafusion-common v55.1.0 (baseline)
      Parsed [   0.068s] (baseline)
    Checking datafusion-common v55.1.0 -> v55.1.0 (no change; assume patch)
     Checked [   0.718s] 223 checks: 222 pass, 1 fail, 0 warn, 31 skip

--- failure constructible_struct_adds_field: struct exhaustively constructible through public API adds field ---

Description:
A pub struct that could be exhaustively constructed with a literal using only public API has a new pub field, breaking existing exhaustive literals.
        ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.50.0/src/lints/constructible_struct_adds_field.ron

Failed in:
  field OptimizerOptions.enable_semi_join_pushdown in /home/runner/work/datafusion/datafusion/datafusion/common/src/config.rs:1643

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [  65.115s] datafusion-common
    Building datafusion-optimizer v55.1.0 (current)
       Built [  24.655s] (current)
     Parsing datafusion-optimizer v55.1.0 (current)
      Parsed [   0.031s] (current)
    Building datafusion-optimizer v55.1.0 (baseline)
       Built [  25.315s] (baseline)
     Parsing datafusion-optimizer v55.1.0 (baseline)
      Parsed [   0.033s] (baseline)
    Checking datafusion-optimizer v55.1.0 -> v55.1.0 (no change; assume patch)
     Checked [   0.149s] 223 checks: 223 pass, 31 skip
     Summary no semver update required
    Finished [  50.945s] datafusion-optimizer
    Building datafusion-sqllogictest v55.1.0 (current)
       Built [  93.716s] (current)
     Parsing datafusion-sqllogictest v55.1.0 (current)
      Parsed [   0.015s] (current)
    Building datafusion-sqllogictest v55.1.0 (baseline)
       Built [  93.975s] (baseline)
     Parsing datafusion-sqllogictest v55.1.0 (baseline)
      Parsed [   0.016s] (baseline)
    Checking datafusion-sqllogictest v55.1.0 -> v55.1.0 (no change; assume patch)
     Checked [   0.095s] 223 checks: 223 pass, 31 skip
     Summary no semver update required
    Finished [ 190.896s] datafusion-sqllogictest

@github-actions github-actions Bot added the auto detected api change Auto detected API change label Oct 12, 2026
@github-actions github-actions Bot added the core Core DataFusion crate label Oct 12, 2026
@codecov-commenter

codecov-commenter commented Oct 12, 2026

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 82.43902% with 72 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.80%. Comparing base (b9c8b3d) to head (93b7c6d).
⚠️ Report is 159 commits behind head on main.

Files with missing lines Patch % Lines
datafusion/optimizer/src/push_down_semi_join.rs 82.43% 29 Missing and 43 partials ⚠️
Additional details and impacted files 
@@            Coverage Diff             @@
##             main   #26218      +/-   ##
==========================================
+ Coverage   82.57%   82.80%   +0.23%     
==========================================
  Files        1142     1150       +8     
  Lines      441215   452645   +11430     
  Branches   441215   452645   +11430     
==========================================
+ Hits       364332   374822   +10490     
- Misses      54841    54981     +140     
- Partials    22042    22842     +800

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
  • 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

auto detected api change Auto detected API change common Related to common crate core Core DataFusion crate documentation Improvements or additions to documentation optimizer Optimizer rules sqllogictest SQL Logic Tests (.slt)

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@adriangb @adriangbot @codecov-commenter @claude