Skip to content

perf: replace adjacent rank self-joins with window functions - #27634

Draft
Dandandan wants to merge 1 commit into
apache:mainfrom
Dandandan:perf/ranked-neighbors
Draft

Dandandan wants to merge 1 commit into
apache:mainfrom
Dandandan:perf/ranked-neighbors

Conversation

@Dandandan

Copy link
Copy Markdown
Contributor

Which issue does this PR close?

No linked issue.

Rationale for this change

TPC-DS Q47 and Q57 each join the same ranked input three times to obtain the preceding and following row. One pass over the grouped input can supply all three outputs.

What changes are included in this PR?

Recognize ordinary equality joins on adjacent ranks when grouping proves the ordering has no peers. Replace the two neighbor joins with offset-one windows. Check neighbor existence independently of nullable payloads and exclude NULL partition keys according to inner-join semantics. The separate batched evaluation experiment gave mixed incremental results and is omitted.

What is the testing strategy for this PR?

Compare the full original Q47 and Q57 results with the baseline. Exercise tied ranks, NULL partition keys, NULL payloads, nonunit offsets and small record batches.

Independent release builds from 04b910869 on macOS arm64, using DuckDB 1.4.1 SF1 Parquet data, upstream query files and ten target partitions. For each arm, five fresh processes ran the original query seven times each. Values are median warm SQL times in milliseconds, excluding the first execution of each process. Each independently built patch also completed all 22 TPC-H and 99 TPC-DS queries with baseline row counts.

Query Baseline ms Candidate ms Change
TPC-DS Q47 132.55 40.15 -69.7%
TPC-DS Q57 89.60 32.05 -64.2%
TPC-H Q18 98.90 96.55 -2.4%

The executed Q47 and Q57 plans each use a single ranked producer. Both retained all 100 result rows. In a separate ablation, adding batched offset-one evaluation to the rewrite changed Q47 from 46.35 to 44.40 ms and Q57 from 31.90 to 32.20 ms. This PR contains the logical rewrite alone.

Reproduce a query from the benchmarks directory with:

dfbench tpcds -p "$TPCDS_PARQUET" -Q ../datafusion/core/tests/tpc-ds -q 47 -n 10 -i 7
dfbench tpch -p "$TPCH_DATA" -f parquet -q 18 -n 10 -i 7

Are there any user-facing changes?

Eligible ranked self-joins can be replaced by bounded windows.

Implementation and benchmark preparation were AI-assisted.

@github-actions github-actions Bot added core Core DataFusion crate sqllogictest SQL Logic Tests (.slt) labels Nov 21, 2026
@Dandandan

Copy link
Copy Markdown
Contributor Author

Run benchmarks

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c7062599694-5574-bzmrh 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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/ranked-neighbors (df3e4e2) to 04b9108 (merge-base) diff

Run configuration
run benchmark tpcds

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c7062599694-5575-78vjj 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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/ranked-neighbors (df3e4e2) to 04b9108 (merge-base) diff

Run configuration
run benchmark tpch

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c7062599694-5573-qdnrq 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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/ranked-neighbors (df3e4e2) to 04b9108 (merge-base) diff

Run configuration
run benchmark clickbench_partitioned

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing perf/ranked-neighbors (df3e4e2) to 04b9108 (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_ranked-neighbors
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ perf_ranked-neighbors ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 41.47 ms │              41.19 ms │ no change │
│ QQuery 2  │ 18.81 ms │              19.00 ms │ no change │
│ QQuery 3  │ 29.27 ms │              29.45 ms │ no change │
│ QQuery 4  │ 17.99 ms │              17.93 ms │ no change │
│ QQuery 5  │ 37.13 ms │              36.79 ms │ no change │
│ QQuery 6  │ 14.44 ms │              14.31 ms │ no change │
│ QQuery 7  │ 43.62 ms │              42.59 ms │ no change │
│ QQuery 8  │ 41.55 ms │              41.10 ms │ no change │
│ QQuery 9  │ 50.75 ms │              51.00 ms │ no change │
│ QQuery 10 │ 42.47 ms │              41.65 ms │ no change │
│ QQuery 11 │ 14.17 ms │              14.28 ms │ no change │
│ QQuery 12 │ 18.50 ms │              18.50 ms │ no change │
│ QQuery 13 │ 42.42 ms │              42.08 ms │ no change │
│ QQuery 14 │ 25.10 ms │              25.18 ms │ no change │
│ QQuery 15 │ 32.38 ms │              32.27 ms │ no change │
│ QQuery 16 │ 13.85 ms │              13.92 ms │ no change │
│ QQuery 17 │ 73.54 ms │              72.97 ms │ no change │
│ QQuery 18 │ 42.18 ms │              41.87 ms │ no change │
│ QQuery 19 │ 32.08 ms │              32.53 ms │ no change │
│ QQuery 20 │ 33.22 ms │              32.85 ms │ no change │
│ QQuery 21 │ 58.69 ms │              57.94 ms │ no change │
│ QQuery 22 │ 14.29 ms │              14.36 ms │ no change │
└───────────┴──────────┴───────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                    ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                    │ 737.90ms │
│ Total Time (perf_ranked-neighbors)   │ 733.75ms │
│ Average Time (HEAD)                  │  33.54ms │
│ Average Time (perf_ranked-neighbors) │  33.35ms │
│ Queries Faster                       │        0 │
│ Queries Slower                       │        0 │
│ Queries with No Change               │       22 │
│ Queries with Failure                 │        0 │
└──────────────────────────────────────┴──────────┘

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

Comparing HEAD and perf_ranked-neighbors
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃          perf_ranked-neighbors ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 41.47 / 44.03 ±2.27 / 48.04 ms │ 41.19 / 43.39 ±2.58 / 47.91 ms │ no change │
│ QQuery 2  │ 18.81 / 19.23 ±0.23 / 19.50 ms │ 19.00 / 19.35 ±0.25 / 19.59 ms │ no change │
│ QQuery 3  │ 29.27 / 30.09 ±0.96 / 31.96 ms │ 29.45 / 29.83 ±0.34 / 30.25 ms │ no change │
│ QQuery 4  │ 17.99 / 18.59 ±0.56 / 19.59 ms │ 17.93 / 18.27 ±0.60 / 19.48 ms │ no change │
│ QQuery 5  │ 37.13 / 37.36 ±0.18 / 37.61 ms │ 36.79 / 38.00 ±1.42 / 40.58 ms │ no change │
│ QQuery 6  │ 14.44 / 14.62 ±0.12 / 14.79 ms │ 14.31 / 14.95 ±0.83 / 16.57 ms │ no change │
│ QQuery 7  │ 43.62 / 45.76 ±2.28 / 49.61 ms │ 42.59 / 45.09 ±4.25 / 53.57 ms │ no change │
│ QQuery 8  │ 41.55 / 42.10 ±0.89 / 43.89 ms │ 41.10 / 42.21 ±1.13 / 43.67 ms │ no change │
│ QQuery 9  │ 50.75 / 52.07 ±1.31 / 54.25 ms │ 51.00 / 51.42 ±0.42 / 52.13 ms │ no change │
│ QQuery 10 │ 42.47 / 43.43 ±1.67 / 46.77 ms │ 41.65 / 42.49 ±1.10 / 44.65 ms │ no change │
│ QQuery 11 │ 14.17 / 14.34 ±0.10 / 14.43 ms │ 14.28 / 14.60 ±0.32 / 15.20 ms │ no change │
│ QQuery 12 │ 18.50 / 19.04 ±0.52 / 19.98 ms │ 18.50 / 18.94 ±0.44 / 19.64 ms │ no change │
│ QQuery 13 │ 42.42 / 43.08 ±0.46 / 43.83 ms │ 42.08 / 42.55 ±0.32 / 43.04 ms │ no change │
│ QQuery 14 │ 25.10 / 25.22 ±0.09 / 25.34 ms │ 25.18 / 25.29 ±0.11 / 25.47 ms │ no change │
│ QQuery 15 │ 32.38 / 32.72 ±0.41 / 33.50 ms │ 32.27 / 32.59 ±0.38 / 33.25 ms │ no change │
│ QQuery 16 │ 13.85 / 14.04 ±0.20 / 14.42 ms │ 13.92 / 14.10 ±0.15 / 14.37 ms │ no change │
│ QQuery 17 │ 73.54 / 74.12 ±0.52 / 74.85 ms │ 72.97 / 73.55 ±0.73 / 74.97 ms │ no change │
│ QQuery 18 │ 42.18 / 44.35 ±2.73 / 49.57 ms │ 41.87 / 42.62 ±0.47 / 43.10 ms │ no change │
│ QQuery 19 │ 32.08 / 32.98 ±1.09 / 35.09 ms │ 32.53 / 32.73 ±0.14 / 32.94 ms │ no change │
│ QQuery 20 │ 33.22 / 33.76 ±0.54 / 34.76 ms │ 32.85 / 33.37 ±0.39 / 34.01 ms │ no change │
│ QQuery 21 │ 58.69 / 60.06 ±1.10 / 62.04 ms │ 57.94 / 58.98 ±1.23 / 61.28 ms │ no change │
│ QQuery 22 │ 14.29 / 14.49 ±0.10 / 14.58 ms │ 14.36 / 14.57 ±0.15 / 14.82 ms │ no change │
└───────────┴────────────────────────────────┴────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                    ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                    │ 755.47ms │
│ Total Time (perf_ranked-neighbors)   │ 748.89ms │
│ Average Time (HEAD)                  │  34.34ms │
│ Average Time (perf_ranked-neighbors) │  34.04ms │
│ Queries Faster                       │        0 │
│ Queries Slower                       │        0 │
│ Queries with No Change               │       22 │
│ Queries with Failure                 │        0 │
└──────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric Value
Wall time 5.0s
Peak memory 1.2 GiB
Avg memory 507.7 MiB
CPU user 20.3s
CPU sys 2.0s
Peak spill 0 B

tpch — branch

Metric Value
Wall time 5.0s
Peak memory 1.2 GiB
Avg memory 483.3 MiB
CPU user 20.4s
CPU sys 1.8s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing perf/ranked-neighbors (df3e4e2) to 04b9108 (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_ranked-neighbors
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ perf_ranked-neighbors ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.24 ms │               5.33 ms │     no change │
│ QQuery 2  │   79.04 ms │              78.91 ms │     no change │
│ QQuery 3  │   28.39 ms │              28.62 ms │     no change │
│ QQuery 4  │  439.84 ms │             441.26 ms │     no change │
│ QQuery 5  │   48.99 ms │              49.74 ms │     no change │
│ QQuery 6  │   33.93 ms │              35.36 ms │     no change │
│ QQuery 7  │   70.62 ms │              71.90 ms │     no change │
│ QQuery 8  │   37.06 ms │              37.50 ms │     no change │
│ QQuery 9  │   52.21 ms │              51.74 ms │     no change │
│ QQuery 10 │   59.54 ms │              60.49 ms │     no change │
│ QQuery 11 │  264.63 ms │             280.05 ms │  1.06x slower │
│ QQuery 12 │   28.58 ms │              29.08 ms │     no change │
│ QQuery 13 │  111.96 ms │             112.61 ms │     no change │
│ QQuery 14 │  376.75 ms │             376.96 ms │     no change │
│ QQuery 15 │   53.39 ms │              54.62 ms │     no change │
│ QQuery 16 │    5.86 ms │               5.97 ms │     no change │
│ QQuery 17 │   74.89 ms │              75.70 ms │     no change │
│ QQuery 18 │   91.82 ms │              93.57 ms │     no change │
│ QQuery 19 │   39.83 ms │              40.83 ms │     no change │
│ QQuery 20 │   35.27 ms │              35.53 ms │     no change │
│ QQuery 21 │   16.79 ms │              16.99 ms │     no change │
│ QQuery 22 │   61.80 ms │              62.45 ms │     no change │
│ QQuery 23 │  289.59 ms │             293.65 ms │     no change │
│ QQuery 24 │  188.04 ms │             195.15 ms │     no change │
│ QQuery 25 │  103.27 ms │             103.89 ms │     no change │
│ QQuery 26 │   45.14 ms │              45.80 ms │     no change │
│ QQuery 27 │    4.95 ms │               5.03 ms │     no change │
│ QQuery 28 │   57.50 ms │              57.19 ms │     no change │
│ QQuery 29 │   90.41 ms │              90.77 ms │     no change │
│ QQuery 30 │   31.80 ms │              32.42 ms │     no change │
│ QQuery 31 │  105.40 ms │             107.97 ms │     no change │
│ QQuery 32 │   19.25 ms │              20.33 ms │  1.06x slower │
│ QQuery 33 │   36.22 ms │              36.89 ms │     no change │
│ QQuery 34 │    5.30 ms │               5.53 ms │     no change │
│ QQuery 35 │   69.75 ms │              71.94 ms │     no change │
│ QQuery 36 │    5.33 ms │               5.60 ms │     no change │
│ QQuery 37 │    6.69 ms │               7.04 ms │  1.05x slower │
│ QQuery 38 │   59.54 ms │              61.58 ms │     no change │
│ QQuery 39 │   72.77 ms │              73.75 ms │     no change │
│ QQuery 40 │   22.14 ms │              22.92 ms │     no change │
│ QQuery 41 │   12.52 ms │              12.81 ms │     no change │
│ QQuery 42 │   23.48 ms │              23.76 ms │     no change │
│ QQuery 43 │    4.54 ms │               4.51 ms │     no change │
│ QQuery 44 │    5.49 ms │               5.53 ms │     no change │
│ QQuery 45 │   36.58 ms │              36.85 ms │     no change │
│ QQuery 46 │   11.80 ms │              12.32 ms │     no change │
│ QQuery 47 │  215.25 ms │              73.19 ms │ +2.94x faster │
│ QQuery 48 │   90.04 ms │              91.45 ms │     no change │
│ QQuery 49 │   69.69 ms │              69.80 ms │     no change │
│ QQuery 50 │   55.51 ms │              56.44 ms │     no change │
│ QQuery 51 │   93.72 ms │              93.33 ms │     no change │
│ QQuery 52 │   23.25 ms │              24.02 ms │     no change │
│ QQuery 53 │   29.05 ms │              29.24 ms │     no change │
│ QQuery 54 │   53.15 ms │              53.89 ms │     no change │
│ QQuery 55 │   23.31 ms │              23.74 ms │     no change │
│ QQuery 56 │   37.43 ms │              37.85 ms │     no change │
│ QQuery 57 │  165.56 ms │              55.35 ms │ +2.99x faster │
│ QQuery 58 │  108.02 ms │             106.58 ms │     no change │
│ QQuery 59 │  114.27 ms │             115.94 ms │     no change │
│ QQuery 60 │   37.23 ms │              38.06 ms │     no change │
│ QQuery 61 │    8.80 ms │               9.13 ms │     no change │
│ QQuery 62 │   43.46 ms │              44.73 ms │     no change │
│ QQuery 63 │   28.69 ms │              29.89 ms │     no change │
│ QQuery 64 │  344.10 ms │             357.40 ms │     no change │
│ QQuery 65 │   66.05 ms │              67.87 ms │     no change │
│ QQuery 66 │   75.32 ms │              74.68 ms │     no change │
│ QQuery 67 │  212.21 ms │             221.06 ms │     no change │
│ QQuery 68 │    6.40 ms │               6.44 ms │     no change │
│ QQuery 69 │   55.46 ms │              56.76 ms │     no change │
│ QQuery 70 │   77.35 ms │              79.35 ms │     no change │
│ QQuery 71 │   34.06 ms │              35.26 ms │     no change │
│ QQuery 72 │ 1593.55 ms │            1661.03 ms │     no change │
│ QQuery 73 │    5.15 ms │               5.27 ms │     no change │
│ QQuery 74 │  192.61 ms │             190.39 ms │     no change │
│ QQuery 75 │  138.00 ms │             136.85 ms │     no change │
│ QQuery 76 │   33.95 ms │              34.03 ms │     no change │
│ QQuery 77 │   60.27 ms │              60.06 ms │     no change │
│ QQuery 78 │  271.83 ms │             266.45 ms │     no change │
│ QQuery 79 │   66.83 ms │              66.52 ms │     no change │
│ QQuery 80 │   93.43 ms │              93.48 ms │     no change │
│ QQuery 81 │   25.70 ms │              25.98 ms │     no change │
│ QQuery 82 │   16.27 ms │              16.15 ms │     no change │
│ QQuery 83 │   32.64 ms │              32.97 ms │     no change │
│ QQuery 84 │   29.55 ms │              29.34 ms │     no change │
│ QQuery 85 │   96.57 ms │              96.38 ms │     no change │
│ QQuery 86 │   25.22 ms │              25.14 ms │     no change │
│ QQuery 87 │   60.28 ms │              60.48 ms │     no change │
│ QQuery 88 │   59.46 ms │              59.56 ms │     no change │
│ QQuery 89 │   34.85 ms │              34.62 ms │     no change │
│ QQuery 90 │   16.05 ms │              16.31 ms │     no change │
│ QQuery 91 │   40.22 ms │              40.76 ms │     no change │
│ QQuery 92 │   28.58 ms │              29.52 ms │     no change │
│ QQuery 93 │   49.30 ms │              47.95 ms │     no change │
│ QQuery 94 │   38.03 ms │              38.25 ms │     no change │
│ QQuery 95 │   77.86 ms │              77.98 ms │     no change │
│ QQuery 96 │   23.64 ms │              24.16 ms │     no change │
│ QQuery 97 │   50.28 ms │              50.50 ms │     no change │
│ QQuery 98 │   42.36 ms │              42.96 ms │     no change │
│ QQuery 99 │   65.47 ms │              64.37 ms │     no change │
└───────────┴────────────┴───────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                    ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                    │ 8659.30ms │
│ Total Time (perf_ranked-neighbors)   │ 8557.31ms │
│ Average Time (HEAD)                  │   87.47ms │
│ Average Time (perf_ranked-neighbors) │   86.44ms │
│ Queries Faster                       │         2 │
│ Queries Slower                       │         3 │
│ Queries with No Change               │        94 │
│ Queries with Failure                 │         0 │
└──────────────────────────────────────┴───────────┘

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

Comparing HEAD and perf_ranked-neighbors
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                 perf_ranked-neighbors ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.24 / 5.74 ±0.94 / 7.61 ms │           5.33 / 5.88 ±0.96 / 7.78 ms │     no change │
│ QQuery 2  │        79.04 / 79.36 ±0.34 / 79.96 ms │        78.91 / 79.09 ±0.13 / 79.24 ms │     no change │
│ QQuery 3  │        28.39 / 28.57 ±0.13 / 28.76 ms │        28.62 / 28.94 ±0.30 / 29.42 ms │     no change │
│ QQuery 4  │    439.84 / 532.51 ±58.64 / 582.06 ms │     441.26 / 444.65 ±3.17 / 448.81 ms │ +1.20x faster │
│ QQuery 5  │        48.99 / 49.59 ±0.39 / 50.19 ms │        49.74 / 51.92 ±4.08 / 60.08 ms │     no change │
│ QQuery 6  │        33.93 / 34.31 ±0.38 / 34.88 ms │        35.36 / 35.86 ±0.41 / 36.51 ms │     no change │
│ QQuery 7  │        70.62 / 71.07 ±0.27 / 71.42 ms │        71.90 / 72.33 ±0.42 / 73.08 ms │     no change │
│ QQuery 8  │        37.06 / 38.71 ±3.15 / 45.01 ms │        37.50 / 37.95 ±0.25 / 38.20 ms │     no change │
│ QQuery 9  │        52.21 / 54.05 ±2.07 / 57.87 ms │        51.74 / 53.62 ±1.27 / 55.05 ms │     no change │
│ QQuery 10 │        59.54 / 59.89 ±0.26 / 60.29 ms │        60.49 / 60.66 ±0.21 / 61.07 ms │     no change │
│ QQuery 11 │     264.63 / 266.56 ±2.25 / 270.49 ms │     280.05 / 282.02 ±2.00 / 285.62 ms │  1.06x slower │
│ QQuery 12 │        28.58 / 28.95 ±0.44 / 29.82 ms │        29.08 / 29.34 ±0.18 / 29.59 ms │     no change │
│ QQuery 13 │     111.96 / 112.73 ±0.54 / 113.45 ms │     112.61 / 113.51 ±0.95 / 115.22 ms │     no change │
│ QQuery 14 │     376.75 / 380.88 ±2.25 / 383.21 ms │     376.96 / 382.46 ±3.71 / 388.42 ms │     no change │
│ QQuery 15 │        53.39 / 55.12 ±1.65 / 58.20 ms │        54.62 / 55.09 ±0.44 / 55.69 ms │     no change │
│ QQuery 16 │           5.86 / 6.00 ±0.18 / 6.37 ms │           5.97 / 6.17 ±0.26 / 6.67 ms │     no change │
│ QQuery 17 │        74.89 / 77.33 ±3.60 / 84.46 ms │        75.70 / 77.74 ±3.62 / 84.96 ms │     no change │
│ QQuery 18 │        91.82 / 92.85 ±0.93 / 94.56 ms │        93.57 / 94.65 ±1.31 / 97.20 ms │     no change │
│ QQuery 19 │        39.83 / 41.13 ±1.24 / 43.49 ms │        40.83 / 41.33 ±0.34 / 41.91 ms │     no change │
│ QQuery 20 │        35.27 / 35.71 ±0.39 / 36.22 ms │        35.53 / 35.97 ±0.38 / 36.52 ms │     no change │
│ QQuery 21 │        16.79 / 17.00 ±0.18 / 17.30 ms │        16.99 / 17.65 ±0.99 / 19.63 ms │     no change │
│ QQuery 22 │        61.80 / 62.18 ±0.32 / 62.64 ms │        62.45 / 62.89 ±0.51 / 63.78 ms │     no change │
│ QQuery 23 │     289.59 / 295.88 ±6.88 / 307.72 ms │     293.65 / 297.70 ±3.91 / 305.12 ms │     no change │
│ QQuery 24 │     188.04 / 191.37 ±4.44 / 200.06 ms │     195.15 / 198.37 ±2.93 / 203.75 ms │     no change │
│ QQuery 25 │     103.27 / 104.87 ±1.93 / 107.71 ms │     103.89 / 104.46 ±0.47 / 105.28 ms │     no change │
│ QQuery 26 │        45.14 / 45.41 ±0.20 / 45.71 ms │        45.80 / 46.93 ±1.60 / 50.09 ms │     no change │
│ QQuery 27 │          4.95 / 6.70 ±3.10 / 12.89 ms │           5.03 / 5.21 ±0.18 / 5.55 ms │ +1.28x faster │
│ QQuery 28 │        57.50 / 60.80 ±1.78 / 62.66 ms │        57.19 / 59.89 ±2.20 / 62.42 ms │     no change │
│ QQuery 29 │        90.41 / 91.80 ±1.21 / 93.38 ms │        90.77 / 92.71 ±2.25 / 97.06 ms │     no change │
│ QQuery 30 │        31.80 / 32.05 ±0.18 / 32.29 ms │        32.42 / 33.45 ±1.48 / 36.35 ms │     no change │
│ QQuery 31 │     105.40 / 106.49 ±0.90 / 107.76 ms │     107.97 / 108.98 ±0.79 / 109.95 ms │     no change │
│ QQuery 32 │        19.25 / 19.48 ±0.29 / 20.05 ms │        20.33 / 20.47 ±0.13 / 20.64 ms │  1.05x slower │
│ QQuery 33 │        36.22 / 36.84 ±0.45 / 37.51 ms │        36.89 / 37.30 ±0.46 / 38.12 ms │     no change │
│ QQuery 34 │           5.30 / 5.45 ±0.24 / 5.92 ms │           5.53 / 5.69 ±0.18 / 6.05 ms │     no change │
│ QQuery 35 │        69.75 / 72.18 ±2.57 / 76.75 ms │        71.94 / 73.19 ±1.24 / 75.54 ms │     no change │
│ QQuery 36 │           5.33 / 5.53 ±0.32 / 6.16 ms │           5.60 / 5.78 ±0.26 / 6.29 ms │     no change │
│ QQuery 37 │           6.69 / 7.01 ±0.40 / 7.78 ms │           7.04 / 7.08 ±0.03 / 7.11 ms │     no change │
│ QQuery 38 │        59.54 / 60.32 ±0.86 / 61.99 ms │        61.58 / 62.08 ±0.58 / 63.19 ms │     no change │
│ QQuery 39 │        72.77 / 75.52 ±4.44 / 84.37 ms │        73.75 / 74.42 ±0.58 / 75.29 ms │     no change │
│ QQuery 40 │        22.14 / 22.33 ±0.13 / 22.54 ms │        22.92 / 24.32 ±1.83 / 27.93 ms │  1.09x slower │
│ QQuery 41 │        12.52 / 13.64 ±1.91 / 17.46 ms │        12.81 / 13.27 ±0.32 / 13.78 ms │     no change │
│ QQuery 42 │        23.48 / 23.85 ±0.24 / 24.16 ms │        23.76 / 24.17 ±0.23 / 24.45 ms │     no change │
│ QQuery 43 │           4.54 / 5.17 ±0.95 / 7.03 ms │           4.51 / 4.68 ±0.22 / 5.11 ms │ +1.10x faster │
│ QQuery 44 │           5.49 / 5.61 ±0.14 / 5.88 ms │           5.53 / 5.62 ±0.07 / 5.74 ms │     no change │
│ QQuery 45 │        36.58 / 37.23 ±0.53 / 37.96 ms │        36.85 / 37.79 ±0.74 / 38.73 ms │     no change │
│ QQuery 46 │        11.80 / 12.08 ±0.20 / 12.32 ms │        12.32 / 12.65 ±0.23 / 12.96 ms │     no change │
│ QQuery 47 │     215.25 / 220.56 ±4.67 / 229.32 ms │        73.19 / 74.21 ±1.03 / 76.10 ms │ +2.97x faster │
│ QQuery 48 │        90.04 / 92.13 ±2.57 / 97.08 ms │        91.45 / 92.55 ±0.99 / 94.25 ms │     no change │
│ QQuery 49 │        69.69 / 70.31 ±0.62 / 71.32 ms │        69.80 / 70.93 ±0.92 / 72.01 ms │     no change │
│ QQuery 50 │        55.51 / 55.88 ±0.26 / 56.17 ms │        56.44 / 57.89 ±2.17 / 62.18 ms │     no change │
│ QQuery 51 │       93.72 / 95.84 ±2.37 / 100.32 ms │        93.33 / 95.90 ±1.38 / 97.48 ms │     no change │
│ QQuery 52 │        23.25 / 23.96 ±0.44 / 24.59 ms │        24.02 / 24.26 ±0.27 / 24.77 ms │     no change │
│ QQuery 53 │        29.05 / 29.26 ±0.11 / 29.36 ms │        29.24 / 29.42 ±0.12 / 29.62 ms │     no change │
│ QQuery 54 │        53.15 / 53.40 ±0.28 / 53.93 ms │        53.89 / 54.38 ±0.53 / 55.39 ms │     no change │
│ QQuery 55 │        23.31 / 23.54 ±0.12 / 23.64 ms │        23.74 / 23.86 ±0.08 / 23.95 ms │     no change │
│ QQuery 56 │        37.43 / 37.78 ±0.28 / 38.22 ms │        37.85 / 38.40 ±0.44 / 39.09 ms │     no change │
│ QQuery 57 │     165.56 / 170.35 ±4.93 / 179.54 ms │        55.35 / 56.32 ±1.02 / 58.29 ms │ +3.02x faster │
│ QQuery 58 │     108.02 / 109.93 ±1.92 / 113.23 ms │     106.58 / 108.64 ±2.17 / 112.76 ms │     no change │
│ QQuery 59 │     114.27 / 114.86 ±0.33 / 115.19 ms │     115.94 / 116.43 ±0.60 / 117.56 ms │     no change │
│ QQuery 60 │        37.23 / 37.96 ±0.73 / 39.36 ms │        38.06 / 38.78 ±0.40 / 39.27 ms │     no change │
│ QQuery 61 │           8.80 / 8.97 ±0.16 / 9.28 ms │           9.13 / 9.40 ±0.25 / 9.85 ms │     no change │
│ QQuery 62 │        43.46 / 44.08 ±0.77 / 45.54 ms │        44.73 / 46.05 ±2.13 / 50.29 ms │     no change │
│ QQuery 63 │        28.69 / 29.08 ±0.36 / 29.70 ms │        29.89 / 30.12 ±0.18 / 30.34 ms │     no change │
│ QQuery 64 │     344.10 / 348.42 ±2.38 / 350.69 ms │     357.40 / 358.95 ±1.79 / 362.34 ms │     no change │
│ QQuery 65 │        66.05 / 67.14 ±0.62 / 67.96 ms │        67.87 / 68.22 ±0.23 / 68.48 ms │     no change │
│ QQuery 66 │        75.32 / 77.77 ±3.44 / 84.58 ms │        74.68 / 75.49 ±0.44 / 76.01 ms │     no change │
│ QQuery 67 │     212.21 / 218.12 ±6.48 / 230.71 ms │     221.06 / 226.10 ±3.62 / 232.36 ms │     no change │
│ QQuery 68 │           6.40 / 6.56 ±0.21 / 6.98 ms │          6.44 / 9.13 ±4.94 / 19.00 ms │  1.39x slower │
│ QQuery 69 │        55.46 / 55.81 ±0.36 / 56.42 ms │        56.76 / 57.44 ±0.57 / 58.41 ms │     no change │
│ QQuery 70 │        77.35 / 78.87 ±2.21 / 83.23 ms │        79.35 / 79.77 ±0.27 / 80.12 ms │     no change │
│ QQuery 71 │        34.06 / 34.50 ±0.29 / 34.88 ms │        35.26 / 35.48 ±0.16 / 35.65 ms │     no change │
│ QQuery 72 │ 1593.55 / 1664.88 ±58.39 / 1743.64 ms │ 1661.03 / 1702.60 ±35.33 / 1766.48 ms │     no change │
│ QQuery 73 │           5.15 / 5.31 ±0.20 / 5.70 ms │           5.27 / 5.39 ±0.22 / 5.82 ms │     no change │
│ QQuery 74 │     192.61 / 197.04 ±3.00 / 201.72 ms │     190.39 / 194.26 ±3.54 / 199.19 ms │     no change │
│ QQuery 75 │     138.00 / 138.55 ±0.97 / 140.48 ms │     136.85 / 138.17 ±1.53 / 140.94 ms │     no change │
│ QQuery 76 │        33.95 / 36.19 ±2.48 / 40.93 ms │        34.03 / 34.34 ±0.18 / 34.58 ms │ +1.05x faster │
│ QQuery 77 │        60.27 / 60.83 ±0.41 / 61.51 ms │        60.06 / 60.66 ±1.08 / 62.82 ms │     no change │
│ QQuery 78 │     271.83 / 280.00 ±6.30 / 287.88 ms │     266.45 / 277.99 ±8.91 / 292.24 ms │     no change │
│ QQuery 79 │        66.83 / 68.54 ±2.71 / 73.94 ms │        66.52 / 69.82 ±5.05 / 79.79 ms │     no change │
│ QQuery 80 │        93.43 / 95.24 ±1.86 / 97.52 ms │        93.48 / 95.03 ±1.41 / 96.95 ms │     no change │
│ QQuery 81 │        25.70 / 25.99 ±0.21 / 26.33 ms │        25.98 / 26.23 ±0.18 / 26.50 ms │     no change │
│ QQuery 82 │        16.27 / 16.34 ±0.09 / 16.52 ms │        16.15 / 16.39 ±0.16 / 16.64 ms │     no change │
│ QQuery 83 │        32.64 / 34.73 ±3.35 / 41.42 ms │        32.97 / 35.16 ±3.48 / 42.10 ms │     no change │
│ QQuery 84 │        29.55 / 30.23 ±0.64 / 31.39 ms │        29.34 / 29.86 ±0.30 / 30.25 ms │     no change │
│ QQuery 85 │        96.57 / 97.90 ±0.99 / 99.55 ms │       96.38 / 98.15 ±1.50 / 100.65 ms │     no change │
│ QQuery 86 │        25.22 / 25.63 ±0.40 / 26.37 ms │        25.14 / 25.47 ±0.20 / 25.76 ms │     no change │
│ QQuery 87 │        60.28 / 62.60 ±2.54 / 67.36 ms │        60.48 / 61.42 ±0.80 / 62.49 ms │     no change │
│ QQuery 88 │        59.46 / 59.91 ±0.58 / 61.06 ms │        59.56 / 60.39 ±1.08 / 62.51 ms │     no change │
│ QQuery 89 │        34.85 / 35.22 ±0.22 / 35.50 ms │        34.62 / 35.10 ±0.46 / 35.87 ms │     no change │
│ QQuery 90 │        16.05 / 16.31 ±0.22 / 16.70 ms │        16.31 / 16.49 ±0.20 / 16.75 ms │     no change │
│ QQuery 91 │        40.22 / 40.55 ±0.19 / 40.77 ms │        40.76 / 41.16 ±0.39 / 41.84 ms │     no change │
│ QQuery 92 │        28.58 / 29.44 ±0.53 / 30.23 ms │        29.52 / 31.70 ±2.95 / 37.56 ms │  1.08x slower │
│ QQuery 93 │        49.30 / 50.09 ±0.67 / 50.95 ms │        47.95 / 49.12 ±0.95 / 50.74 ms │     no change │
│ QQuery 94 │        38.03 / 38.35 ±0.28 / 38.69 ms │        38.25 / 39.13 ±0.90 / 40.73 ms │     no change │
│ QQuery 95 │        77.86 / 78.38 ±0.44 / 79.17 ms │        77.98 / 81.48 ±3.79 / 88.77 ms │     no change │
│ QQuery 96 │        23.64 / 23.80 ±0.23 / 24.23 ms │        24.16 / 25.91 ±2.92 / 31.73 ms │  1.09x slower │
│ QQuery 97 │        50.28 / 52.24 ±2.36 / 56.69 ms │        50.50 / 51.35 ±0.51 / 51.92 ms │     no change │
│ QQuery 98 │        42.36 / 43.28 ±0.65 / 44.23 ms │        42.96 / 43.53 ±0.63 / 44.66 ms │     no change │
│ QQuery 99 │        65.47 / 65.85 ±0.29 / 66.30 ms │        64.37 / 64.77 ±0.28 / 65.07 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                    ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                    │ 8946.33ms │
│ Total Time (perf_ranked-neighbors)   │ 8715.13ms │
│ Average Time (HEAD)                  │   90.37ms │
│ Average Time (perf_ranked-neighbors) │   88.03ms │
│ Queries Faster                       │         6 │
│ Queries Slower                       │         6 │
│ Queries with No Change               │        87 │
│ Queries with Failure                 │         0 │
└──────────────────────────────────────┴───────────┘

Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 50.0s
Peak memory 2.1 GiB
Avg memory 1.4 GiB
CPU user 186.1s
CPU sys 4.9s
Peak spill 0 B

tpcds — branch

Metric Value
Wall time 45.0s
Peak memory 1.8 GiB
Avg memory 1.3 GiB
CPU user 185.9s
CPU sys 4.9s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing perf/ranked-neighbors (df3e4e2) to 04b9108 (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_ranked-neighbors
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ perf_ranked-neighbors ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.15 ms │               1.20 ms │     no change │
│ QQuery 1  │   11.78 ms │              11.90 ms │     no change │
│ QQuery 2  │   34.90 ms │              34.79 ms │     no change │
│ QQuery 3  │   30.72 ms │              30.57 ms │     no change │
│ QQuery 4  │  235.70 ms │             235.22 ms │     no change │
│ QQuery 5  │  278.79 ms │             279.36 ms │     no change │
│ QQuery 6  │    1.19 ms │               1.19 ms │     no change │
│ QQuery 7  │   13.10 ms │              12.88 ms │     no change │
│ QQuery 8  │  326.91 ms │             321.96 ms │     no change │
│ QQuery 9  │  465.34 ms │             465.10 ms │     no change │
│ QQuery 10 │   65.05 ms │              64.89 ms │     no change │
│ QQuery 11 │   76.32 ms │              75.60 ms │     no change │
│ QQuery 12 │  261.52 ms │             261.72 ms │     no change │
│ QQuery 13 │  363.75 ms │             358.47 ms │     no change │
│ QQuery 14 │  278.21 ms │             275.44 ms │     no change │
│ QQuery 15 │  280.81 ms │             285.43 ms │     no change │
│ QQuery 16 │  611.77 ms │             610.67 ms │     no change │
│ QQuery 17 │  620.46 ms │             612.21 ms │     no change │
│ QQuery 18 │ 1221.77 ms │            1193.72 ms │     no change │
│ QQuery 19 │   27.61 ms │              27.47 ms │     no change │
│ QQuery 20 │  518.65 ms │             519.79 ms │     no change │
│ QQuery 21 │  509.72 ms │             510.95 ms │     no change │
│ QQuery 22 │  980.35 ms │             972.87 ms │     no change │
│ QQuery 23 │ 3029.41 ms │            3060.84 ms │     no change │
│ QQuery 24 │   43.10 ms │              41.32 ms │     no change │
│ QQuery 25 │  105.39 ms │             105.97 ms │     no change │
│ QQuery 26 │   41.70 ms │              41.90 ms │     no change │
│ QQuery 27 │  508.46 ms │             511.14 ms │     no change │
│ QQuery 28 │ 2848.19 ms │            2903.19 ms │     no change │
│ QQuery 29 │   37.70 ms │              38.47 ms │     no change │
│ QQuery 30 │  300.98 ms │             303.96 ms │     no change │
│ QQuery 31 │  272.02 ms │             273.88 ms │     no change │
│ QQuery 32 │  990.64 ms │             922.30 ms │ +1.07x faster │
│ QQuery 33 │ 1421.44 ms │            1443.80 ms │     no change │
│ QQuery 34 │ 1425.41 ms │            1441.00 ms │     no change │
│ QQuery 35 │  293.42 ms │             294.02 ms │     no change │
│ QQuery 36 │   65.40 ms │              68.35 ms │     no change │
│ QQuery 37 │   34.59 ms │              34.53 ms │     no change │
│ QQuery 38 │   38.61 ms │              38.79 ms │     no change │
│ QQuery 39 │  135.95 ms │             139.16 ms │     no change │
│ QQuery 40 │   13.17 ms │              13.26 ms │     no change │
│ QQuery 41 │   12.26 ms │              12.89 ms │  1.05x slower │
│ QQuery 42 │   11.95 ms │              12.49 ms │     no change │
└───────────┴────────────┴───────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                    ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                    │ 18845.35ms │
│ Total Time (perf_ranked-neighbors)   │ 18864.65ms │
│ Average Time (HEAD)                  │   438.26ms │
│ Average Time (perf_ranked-neighbors) │   438.71ms │
│ Queries Faster                       │          1 │
│ Queries Slower                       │          1 │
│ Queries with No Change               │         41 │
│ Queries with Failure                 │          0 │
└──────────────────────────────────────┴────────────┘

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

Comparing HEAD and perf_ranked-neighbors
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                 perf_ranked-neighbors ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.15 / 1.75 ±1.11 / 3.97 ms │           1.20 / 1.76 ±1.03 / 3.81 ms │     no change │
│ QQuery 1  │        11.78 / 12.03 ±0.13 / 12.14 ms │        11.90 / 12.02 ±0.16 / 12.33 ms │     no change │
│ QQuery 2  │        34.90 / 35.07 ±0.15 / 35.30 ms │        34.79 / 35.19 ±0.36 / 35.77 ms │     no change │
│ QQuery 3  │        30.72 / 31.42 ±0.72 / 32.43 ms │        30.57 / 30.69 ±0.13 / 30.90 ms │     no change │
│ QQuery 4  │     235.70 / 240.25 ±2.82 / 244.18 ms │     235.22 / 239.65 ±3.29 / 244.63 ms │     no change │
│ QQuery 5  │     278.79 / 282.07 ±1.72 / 283.48 ms │     279.36 / 280.29 ±0.78 / 281.57 ms │     no change │
│ QQuery 6  │           1.19 / 1.35 ±0.24 / 1.83 ms │           1.19 / 1.37 ±0.25 / 1.86 ms │     no change │
│ QQuery 7  │        13.10 / 13.39 ±0.20 / 13.67 ms │        12.88 / 13.04 ±0.12 / 13.18 ms │     no change │
│ QQuery 8  │     326.91 / 332.10 ±4.04 / 337.59 ms │     321.96 / 329.29 ±5.86 / 335.16 ms │     no change │
│ QQuery 9  │     465.34 / 474.98 ±6.31 / 483.54 ms │     465.10 / 472.98 ±4.82 / 478.98 ms │     no change │
│ QQuery 10 │        65.05 / 65.67 ±0.53 / 66.39 ms │        64.89 / 65.34 ±0.35 / 65.91 ms │     no change │
│ QQuery 11 │        76.32 / 76.94 ±0.48 / 77.64 ms │      75.60 / 85.54 ±12.00 / 107.29 ms │  1.11x slower │
│ QQuery 12 │     261.52 / 266.62 ±5.18 / 276.33 ms │     261.72 / 266.70 ±4.51 / 272.75 ms │     no change │
│ QQuery 13 │     363.75 / 375.81 ±9.36 / 388.37 ms │     358.47 / 368.31 ±7.17 / 380.06 ms │     no change │
│ QQuery 14 │     278.21 / 286.39 ±7.35 / 297.51 ms │     275.44 / 279.62 ±3.80 / 286.39 ms │     no change │
│ QQuery 15 │     280.81 / 284.35 ±3.70 / 291.26 ms │     285.43 / 294.58 ±5.69 / 301.34 ms │     no change │
│ QQuery 16 │     611.77 / 622.34 ±7.68 / 630.41 ms │     610.67 / 619.31 ±8.80 / 634.76 ms │     no change │
│ QQuery 17 │     620.46 / 631.48 ±8.58 / 640.89 ms │    612.21 / 632.01 ±13.98 / 649.82 ms │     no change │
│ QQuery 18 │ 1221.77 / 1257.34 ±18.16 / 1271.15 ms │ 1193.72 / 1222.39 ±15.93 / 1242.48 ms │     no change │
│ QQuery 19 │        27.61 / 34.77 ±8.98 / 49.45 ms │        27.47 / 28.16 ±0.82 / 29.69 ms │ +1.23x faster │
│ QQuery 20 │     518.65 / 523.87 ±6.77 / 536.18 ms │     519.79 / 526.32 ±9.20 / 544.34 ms │     no change │
│ QQuery 21 │     509.72 / 515.02 ±3.17 / 519.08 ms │     510.95 / 514.88 ±4.76 / 523.93 ms │     no change │
│ QQuery 22 │     980.35 / 986.65 ±4.56 / 993.41 ms │   972.87 / 989.81 ±17.80 / 1023.69 ms │     no change │
│ QQuery 23 │ 3029.41 / 3085.44 ±41.65 / 3147.89 ms │ 3060.84 / 3103.44 ±33.10 / 3147.34 ms │     no change │
│ QQuery 24 │        43.10 / 46.51 ±3.58 / 52.30 ms │        41.32 / 43.45 ±2.70 / 48.67 ms │ +1.07x faster │
│ QQuery 25 │     105.39 / 107.20 ±2.39 / 111.89 ms │     105.97 / 108.98 ±4.43 / 117.73 ms │     no change │
│ QQuery 26 │        41.70 / 42.17 ±0.70 / 43.56 ms │        41.90 / 43.05 ±0.90 / 44.06 ms │     no change │
│ QQuery 27 │     508.46 / 517.25 ±6.46 / 527.04 ms │     511.14 / 520.72 ±6.98 / 530.40 ms │     no change │
│ QQuery 28 │ 2848.19 / 2891.79 ±26.95 / 2932.57 ms │ 2903.19 / 2921.74 ±24.70 / 2968.89 ms │     no change │
│ QQuery 29 │      37.70 / 63.76 ±27.38 / 102.49 ms │        38.47 / 44.86 ±6.99 / 55.77 ms │ +1.42x faster │
│ QQuery 30 │     300.98 / 311.82 ±9.54 / 323.96 ms │     303.96 / 310.28 ±4.64 / 315.67 ms │     no change │
│ QQuery 31 │    272.02 / 287.78 ±18.10 / 321.31 ms │     273.88 / 286.40 ±7.12 / 295.96 ms │     no change │
│ QQuery 32 │  990.64 / 1038.06 ±39.27 / 1097.85 ms │    922.30 / 959.84 ±30.56 / 995.96 ms │ +1.08x faster │
│ QQuery 33 │ 1421.44 / 1473.52 ±32.42 / 1518.50 ms │ 1443.80 / 1466.02 ±18.78 / 1489.80 ms │     no change │
│ QQuery 34 │ 1425.41 / 1461.95 ±30.52 / 1504.19 ms │ 1441.00 / 1468.54 ±21.00 / 1497.31 ms │     no change │
│ QQuery 35 │   293.42 / 375.80 ±140.86 / 655.86 ms │    294.02 / 315.36 ±30.53 / 375.13 ms │ +1.19x faster │
│ QQuery 36 │        65.40 / 71.29 ±6.11 / 82.73 ms │        68.35 / 74.32 ±7.88 / 89.22 ms │     no change │
│ QQuery 37 │        34.59 / 36.49 ±1.95 / 39.94 ms │        34.53 / 36.85 ±3.26 / 43.28 ms │     no change │
│ QQuery 38 │        38.61 / 40.29 ±1.44 / 42.14 ms │        38.79 / 43.24 ±2.67 / 46.71 ms │  1.07x slower │
│ QQuery 39 │     135.95 / 142.99 ±4.32 / 147.69 ms │     139.16 / 146.88 ±5.88 / 154.73 ms │     no change │
│ QQuery 40 │        13.17 / 13.60 ±0.55 / 14.66 ms │        13.26 / 13.96 ±0.49 / 14.65 ms │     no change │
│ QQuery 41 │        12.26 / 12.42 ±0.14 / 12.63 ms │        12.89 / 13.07 ±0.21 / 13.45 ms │  1.05x slower │
│ QQuery 42 │        11.95 / 12.23 ±0.21 / 12.59 ms │        12.49 / 13.93 ±2.49 / 18.89 ms │  1.14x slower │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                    ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                    │ 19384.00ms │
│ Total Time (perf_ranked-neighbors)   │ 19244.16ms │
│ Average Time (HEAD)                  │   450.79ms │
│ Average Time (perf_ranked-neighbors) │   447.54ms │
│ Queries Faster                       │          5 │
│ Queries Slower                       │          4 │
│ Queries with No Change               │         34 │
│ Queries with Failure                 │          0 │
└──────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 100.0s
Peak memory 11.3 GiB
Avg memory 4.5 GiB
CPU user 991.7s
CPU sys 67.8s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 100.0s
Peak memory 12.5 GiB
Avg memory 4.6 GiB
CPU user 987.3s
CPU sys 67.7s
Peak spill 0 B

File an issue against this benchmark runner

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

Labels

core Core DataFusion crate sqllogictest SQL Logic Tests (.slt)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants