fix: propagate column statistics through CAST in projections by Dandandan · Pull Request #21500 · apache/datafusion

Dandandan · 2026-04-09T09:31:53Z

Which issue does this PR close?

Closes #.

Rationale for this change

When join keys involve CAST expressions (e.g. CAST(w_warehouse_sk AS Float64) = cs_warehouse_sk), column statistics were lost in ProjectionExec because project_statistics only handled plain Column and Literal expressions. This caused the join cardinality estimator to fall back to num_rows as distinct count, severely underestimating FK joins and producing poor join ordering.

This commonly occurs with TPC-DS data where fact table FK columns are Float64 (from pandas/pyarrow generators) and dimension PKs are Int32, requiring a CAST in the join key projection.

What changes are included in this PR?

Propagate distinct_count and null_count through numeric CAST expressions in ProjectionExec::project_statistics. The distinct count is made Inexact since casting can reduce (but never increase) distinct values. min/max/sum are cleared since they may not be valid after type conversion.

TPC-DS Q99 (local, prefer_hash_join=true): 10.4s → ~60ms.

Are these changes tested?

Added test_project_statistics_through_cast unit test.

Are there any user-facing changes?

Join plans may improve for queries with CAST expressions in join keys.

🤖 Generated with Claude Code

Dandandan · 2026-04-09T09:51:28Z

run benchmarks

adriangbot · 2026-04-09T09:54:02Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213222926-988-kp48v 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (991084b) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T09:54:04Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213222926-987-tdn6d 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (991084b) to 5ba06ac (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T09:54:19Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213222926-986-nbdt5 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (991084b) to 5ba06ac (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

Dandandan · 2026-04-09T09:59:52Z

run benchmarks

adriangbot · 2026-04-09T10:02:13Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213270390-990-92mrf 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (ad81053) to 5ba06ac (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:02:17Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213270390-991-pqrbh 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (ad81053) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:02:18Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213270390-989-zqkhv 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (ad81053) to 5ba06ac (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:09:38Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃       fix-join-cardinality-estimation ┃          Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.56 / 7.03 ±0.81 / 8.64 ms │        10.53 / 10.93 ±0.68 / 12.28 ms │    1.55x slower │
│ QQuery 2  │        143.31 / 144.29 ±0.83 / 145.36 ms │     144.59 / 145.17 ±0.61 / 146.33 ms │       no change │
│ QQuery 3  │        112.69 / 113.32 ±0.89 / 115.08 ms │     113.42 / 114.54 ±0.85 / 115.56 ms │       no change │
│ QQuery 4  │    1276.10 / 1332.08 ±71.03 / 1471.61 ms │ 1530.38 / 1553.49 ±21.76 / 1590.97 ms │    1.17x slower │
│ QQuery 5  │        171.10 / 172.91 ±1.21 / 174.76 ms │     170.50 / 172.88 ±1.48 / 174.10 ms │       no change │
│ QQuery 6  │       826.46 / 843.50 ±20.56 / 883.48 ms │     181.47 / 183.17 ±1.46 / 185.09 ms │   +4.60x faster │
│ QQuery 7  │        336.81 / 341.30 ±3.99 / 348.06 ms │     335.45 / 337.25 ±1.82 / 340.41 ms │       no change │
│ QQuery 8  │        114.38 / 116.11 ±1.11 / 117.85 ms │     113.22 / 114.01 ±0.62 / 115.11 ms │       no change │
│ QQuery 9  │        101.54 / 102.88 ±1.95 / 106.75 ms │     104.88 / 105.68 ±0.63 / 106.30 ms │       no change │
│ QQuery 10 │        103.64 / 104.56 ±0.77 / 105.59 ms │     129.25 / 131.39 ±1.48 / 133.76 ms │    1.26x slower │
│ QQuery 11 │       849.01 / 879.71 ±21.98 / 907.54 ms │  984.93 / 1025.03 ±27.33 / 1057.92 ms │    1.17x slower │
│ QQuery 12 │           44.54 / 46.20 ±1.27 / 48.16 ms │        45.13 / 46.76 ±0.99 / 47.71 ms │       no change │
│ QQuery 13 │        393.43 / 395.50 ±1.28 / 397.20 ms │     397.06 / 400.18 ±2.92 / 404.87 ms │       no change │
│ QQuery 14 │        978.67 / 988.52 ±6.36 / 997.78 ms │  1584.42 / 1588.07 ±3.44 / 1592.20 ms │    1.61x slower │
│ QQuery 15 │           15.46 / 16.12 ±0.75 / 17.52 ms │        15.37 / 16.75 ±1.04 / 17.79 ms │       no change │
│ QQuery 16 │              6.90 / 8.03 ±0.85 / 9.36 ms │        46.45 / 47.13 ±0.67 / 48.40 ms │    5.87x slower │
│ QQuery 17 │        226.41 / 228.68 ±2.33 / 232.86 ms │     224.10 / 226.45 ±1.77 / 229.10 ms │       no change │
│ QQuery 18 │        125.04 / 125.86 ±0.69 / 126.83 ms │     118.42 / 122.04 ±1.92 / 124.02 ms │       no change │
│ QQuery 19 │        152.60 / 154.43 ±1.77 / 157.47 ms │     152.88 / 155.30 ±1.88 / 157.90 ms │       no change │
│ QQuery 20 │           13.67 / 14.16 ±0.44 / 14.68 ms │        13.72 / 14.06 ±0.39 / 14.76 ms │       no change │
│ QQuery 21 │           19.01 / 19.18 ±0.14 / 19.38 ms │        19.22 / 19.83 ±0.33 / 20.16 ms │       no change │
│ QQuery 22 │        485.52 / 487.27 ±2.88 / 493.02 ms │     489.32 / 495.12 ±3.49 / 499.49 ms │       no change │
│ QQuery 23 │        855.72 / 866.94 ±6.84 / 876.02 ms │     964.92 / 978.03 ±7.53 / 984.96 ms │    1.13x slower │
│ QQuery 24 │        377.43 / 378.37 ±0.80 / 379.46 ms │     760.72 / 764.99 ±4.29 / 772.74 ms │    2.02x slower │
│ QQuery 25 │        336.27 / 338.03 ±1.08 / 339.65 ms │     269.15 / 272.09 ±1.69 / 274.29 ms │   +1.24x faster │
│ QQuery 26 │           78.76 / 81.89 ±1.69 / 83.87 ms │        79.43 / 80.68 ±0.98 / 81.85 ms │       no change │
│ QQuery 27 │              7.01 / 7.43 ±0.38 / 7.93 ms │           7.46 / 7.85 ±0.46 / 8.75 ms │    1.06x slower │
│ QQuery 28 │        147.76 / 148.63 ±1.07 / 150.64 ms │     148.43 / 150.33 ±1.87 / 153.59 ms │       no change │
│ QQuery 29 │        278.09 / 280.15 ±1.13 / 281.52 ms │     257.46 / 261.24 ±2.17 / 263.18 ms │   +1.07x faster │
│ QQuery 30 │           42.36 / 44.25 ±1.17 / 45.96 ms │        38.25 / 40.09 ±0.94 / 40.97 ms │   +1.10x faster │
│ QQuery 31 │        168.12 / 170.03 ±1.33 / 171.53 ms │     422.27 / 427.45 ±3.41 / 431.59 ms │    2.51x slower │
│ QQuery 32 │           55.85 / 57.21 ±1.29 / 59.62 ms │        14.32 / 15.47 ±0.82 / 16.28 ms │   +3.70x faster │
│ QQuery 33 │        138.53 / 139.21 ±0.84 / 140.82 ms │     141.09 / 142.99 ±0.97 / 143.68 ms │       no change │
│ QQuery 34 │              6.90 / 7.19 ±0.20 / 7.47 ms │        15.00 / 15.72 ±0.67 / 16.57 ms │    2.19x slower │
│ QQuery 35 │        105.05 / 106.90 ±1.25 / 108.90 ms │     126.20 / 128.29 ±1.24 / 129.53 ms │    1.20x slower │
│ QQuery 36 │              6.42 / 6.68 ±0.17 / 6.91 ms │           6.76 / 7.01 ±0.18 / 7.23 ms │       no change │
│ QQuery 37 │              8.44 / 8.71 ±0.40 / 9.50 ms │           9.10 / 9.42 ±0.28 / 9.93 ms │    1.08x slower │
│ QQuery 38 │           81.51 / 84.77 ±3.45 / 91.37 ms │     143.83 / 148.14 ±4.52 / 156.65 ms │    1.75x slower │
│ QQuery 39 │        121.53 / 123.73 ±2.31 / 127.79 ms │     126.07 / 128.48 ±2.21 / 131.31 ms │       no change │
│ QQuery 40 │        109.09 / 114.59 ±6.80 / 128.00 ms │        21.44 / 22.40 ±0.86 / 23.92 ms │   +5.11x faster │
│ QQuery 41 │           14.59 / 15.57 ±0.68 / 16.72 ms │        15.22 / 15.49 ±0.31 / 16.09 ms │       no change │
│ QQuery 42 │        106.58 / 108.09 ±1.21 / 110.18 ms │     110.66 / 111.41 ±0.54 / 112.34 ms │       no change │
│ QQuery 43 │              5.87 / 6.14 ±0.26 / 6.60 ms │           6.22 / 6.31 ±0.11 / 6.52 ms │       no change │
│ QQuery 44 │           11.53 / 11.88 ±0.36 / 12.50 ms │        11.86 / 12.24 ±0.20 / 12.39 ms │       no change │
│ QQuery 45 │           49.44 / 51.04 ±1.35 / 53.30 ms │        48.51 / 49.51 ±0.65 / 50.39 ms │       no change │
│ QQuery 46 │              8.22 / 8.58 ±0.38 / 9.27 ms │        15.97 / 17.42 ±0.86 / 18.60 ms │    2.03x slower │
│ QQuery 47 │        668.25 / 678.21 ±5.97 / 685.66 ms │    701.24 / 728.52 ±14.64 / 742.22 ms │    1.07x slower │
│ QQuery 48 │        286.90 / 291.10 ±2.70 / 294.44 ms │     279.17 / 279.97 ±0.45 / 280.54 ms │       no change │
│ QQuery 49 │        250.43 / 254.48 ±3.04 / 259.46 ms │     249.53 / 251.92 ±1.77 / 254.17 ms │       no change │
│ QQuery 50 │        222.63 / 226.73 ±2.28 / 229.59 ms │     183.34 / 185.50 ±2.07 / 188.09 ms │   +1.22x faster │
│ QQuery 51 │        177.76 / 180.16 ±2.16 / 183.79 ms │     178.38 / 180.22 ±1.64 / 182.37 ms │       no change │
│ QQuery 52 │        106.26 / 107.65 ±0.96 / 108.96 ms │     107.43 / 109.36 ±1.65 / 112.05 ms │       no change │
│ QQuery 53 │        101.08 / 102.23 ±1.18 / 104.22 ms │     101.29 / 103.42 ±1.39 / 105.47 ms │       no change │
│ QQuery 54 │        145.61 / 147.37 ±1.76 / 150.29 ms │     185.94 / 187.09 ±1.52 / 190.03 ms │    1.27x slower │
│ QQuery 55 │        107.10 / 108.61 ±1.23 / 110.32 ms │     105.68 / 106.76 ±0.65 / 107.64 ms │       no change │
│ QQuery 56 │        140.10 / 142.28 ±1.40 / 144.21 ms │     138.30 / 139.52 ±0.89 / 140.54 ms │       no change │
│ QQuery 57 │        176.34 / 177.37 ±0.71 / 178.47 ms │     170.56 / 175.32 ±3.65 / 181.66 ms │       no change │
│ QQuery 58 │        283.70 / 295.71 ±8.37 / 304.21 ms │     311.75 / 317.11 ±3.23 / 321.18 ms │    1.07x slower │
│ QQuery 59 │        197.71 / 199.96 ±2.01 / 202.74 ms │     365.33 / 369.82 ±3.21 / 374.60 ms │    1.85x slower │
│ QQuery 60 │        142.93 / 144.05 ±1.18 / 145.79 ms │     140.69 / 142.13 ±1.38 / 144.15 ms │       no change │
│ QQuery 61 │           13.07 / 13.35 ±0.25 / 13.69 ms │        18.67 / 19.23 ±0.47 / 19.92 ms │    1.44x slower │
│ QQuery 62 │      910.00 / 959.21 ±68.31 / 1094.22 ms │        62.84 / 64.11 ±0.99 / 65.55 ms │  +14.96x faster │
│ QQuery 63 │        102.64 / 105.26 ±2.23 / 108.64 ms │     102.26 / 103.64 ±1.64 / 106.61 ms │       no change │
│ QQuery 64 │        683.45 / 688.22 ±5.17 / 697.45 ms │ 1233.40 / 1247.45 ±20.52 / 1287.95 ms │    1.81x slower │
│ QQuery 65 │        248.48 / 251.87 ±3.79 / 259.14 ms │     260.55 / 262.74 ±1.68 / 265.32 ms │       no change │
│ QQuery 66 │        242.90 / 251.12 ±5.29 / 257.86 ms │     105.87 / 107.67 ±2.00 / 111.43 ms │   +2.33x faster │
│ QQuery 67 │        313.35 / 318.67 ±3.29 / 322.08 ms │    640.48 / 657.56 ±16.42 / 687.91 ms │    2.06x slower │
│ QQuery 68 │            9.65 / 10.99 ±1.19 / 13.09 ms │        14.79 / 17.06 ±2.68 / 22.21 ms │    1.55x slower │
│ QQuery 69 │        101.50 / 103.59 ±1.76 / 106.47 ms │     124.62 / 126.32 ±1.47 / 128.10 ms │    1.22x slower │
│ QQuery 70 │        326.23 / 343.25 ±9.20 / 352.49 ms │     321.23 / 327.25 ±4.43 / 333.78 ms │       no change │
│ QQuery 71 │        134.88 / 136.62 ±1.60 / 139.14 ms │     133.23 / 134.98 ±1.68 / 137.83 ms │       no change │
│ QQuery 72 │        614.06 / 623.01 ±5.33 / 628.81 ms │     606.06 / 611.18 ±4.24 / 618.20 ms │       no change │
│ QQuery 73 │             7.23 / 9.30 ±1.38 / 10.73 ms │        12.32 / 13.93 ±2.07 / 17.85 ms │    1.50x slower │
│ QQuery 74 │       584.80 / 610.07 ±15.22 / 625.44 ms │    628.13 / 661.10 ±20.26 / 687.90 ms │    1.08x slower │
│ QQuery 75 │        275.03 / 276.94 ±2.06 / 280.36 ms │     437.19 / 439.76 ±1.61 / 441.77 ms │    1.59x slower │
│ QQuery 76 │        131.87 / 134.10 ±1.45 / 136.10 ms │     130.97 / 132.32 ±0.96 / 133.75 ms │       no change │
│ QQuery 77 │        187.03 / 189.03 ±1.23 / 190.42 ms │     188.63 / 191.37 ±2.57 / 195.88 ms │       no change │
│ QQuery 78 │        336.27 / 340.40 ±2.94 / 345.36 ms │     367.06 / 370.42 ±2.86 / 374.71 ms │    1.09x slower │
│ QQuery 79 │        229.91 / 233.06 ±2.21 / 235.93 ms │     234.83 / 238.65 ±3.57 / 245.41 ms │       no change │
│ QQuery 80 │        319.06 / 319.93 ±0.85 / 321.36 ms │     319.38 / 321.47 ±1.44 / 323.48 ms │       no change │
│ QQuery 81 │           26.04 / 26.92 ±1.23 / 29.35 ms │        30.96 / 31.97 ±0.91 / 33.38 ms │    1.19x slower │
│ QQuery 82 │        194.31 / 197.56 ±2.48 / 200.77 ms │        70.83 / 72.82 ±1.59 / 74.91 ms │   +2.71x faster │
│ QQuery 83 │           38.58 / 38.95 ±0.37 / 39.57 ms │        44.46 / 45.84 ±1.03 / 47.56 ms │    1.18x slower │
│ QQuery 84 │           47.06 / 48.66 ±1.68 / 51.82 ms │        39.94 / 40.91 ±0.93 / 42.61 ms │   +1.19x faster │
│ QQuery 85 │        148.07 / 149.98 ±1.97 / 152.71 ms │     149.39 / 150.65 ±0.96 / 151.88 ms │       no change │
│ QQuery 86 │           39.46 / 40.41 ±0.98 / 42.19 ms │        40.90 / 41.32 ±0.50 / 42.20 ms │       no change │
│ QQuery 87 │           87.69 / 89.11 ±1.86 / 92.50 ms │     138.69 / 144.36 ±3.61 / 149.20 ms │    1.62x slower │
│ QQuery 88 │          98.86 / 99.77 ±0.93 / 101.32 ms │      99.84 / 101.49 ±1.58 / 104.44 ms │       no change │
│ QQuery 89 │        117.87 / 118.81 ±0.77 / 119.95 ms │     117.15 / 119.02 ±2.06 / 122.25 ms │       no change │
│ QQuery 90 │           23.62 / 24.04 ±0.30 / 24.50 ms │        22.29 / 22.55 ±0.32 / 23.18 ms │   +1.07x faster │
│ QQuery 91 │           62.16 / 63.25 ±0.89 / 64.48 ms │        54.94 / 55.85 ±0.77 / 56.95 ms │   +1.13x faster │
│ QQuery 92 │           56.64 / 57.41 ±0.52 / 58.22 ms │        57.48 / 58.20 ±0.63 / 59.13 ms │       no change │
│ QQuery 93 │        182.45 / 184.58 ±1.44 / 186.62 ms │     184.75 / 186.23 ±1.30 / 187.79 ms │       no change │
│ QQuery 94 │           61.25 / 61.86 ±0.68 / 63.02 ms │        67.14 / 68.01 ±0.56 / 68.81 ms │    1.10x slower │
│ QQuery 95 │        127.47 / 128.06 ±0.53 / 128.83 ms │     124.35 / 126.02 ±1.18 / 127.64 ms │       no change │
│ QQuery 96 │           69.83 / 72.45 ±1.42 / 73.91 ms │        67.97 / 69.16 ±0.91 / 70.61 ms │       no change │
│ QQuery 97 │        125.04 / 126.40 ±1.00 / 127.44 ms │     115.20 / 116.46 ±0.71 / 117.33 ms │   +1.09x faster │
│ QQuery 98 │        149.40 / 153.31 ±2.25 / 156.15 ms │     150.99 / 154.38 ±2.20 / 157.61 ms │       no change │
│ QQuery 99 │ 10750.24 / 10804.13 ±61.43 / 10917.05 ms │        60.48 / 61.40 ±1.30 / 63.94 ms │ +175.96x faster │
└───────────┴──────────────────────────────────────────┴───────────────────────────────────────┴─────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 31267.22ms │
│ Total Time (fix-join-cardinality-estimation)   │ 21829.81ms │
│ Average Time (HEAD)                            │   315.83ms │
│ Average Time (fix-join-cardinality-estimation) │   220.50ms │
│ Queries Faster                                 │         15 │
│ Queries Slower                                 │         32 │
│ Queries with No Change                         │         52 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	156.7s
Peak memory	5.7 GiB
Avg memory	4.7 GiB
CPU user	258.3s
CPU sys	16.3s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	109.4s
Peak memory	4.7 GiB
Avg memory	4.1 GiB
CPU user	199.2s
CPU sys	14.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:11:44Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.23 / 4.55 ±6.48 / 17.52 ms │          1.19 / 4.44 ±6.35 / 17.15 ms │     no change │
│ QQuery 1  │        14.49 / 14.84 ±0.20 / 15.04 ms │        14.68 / 14.80 ±0.13 / 15.04 ms │     no change │
│ QQuery 2  │        44.71 / 44.85 ±0.09 / 44.98 ms │        43.88 / 44.29 ±0.29 / 44.79 ms │     no change │
│ QQuery 3  │        41.39 / 43.65 ±2.41 / 47.56 ms │        43.10 / 45.54 ±2.13 / 48.36 ms │     no change │
│ QQuery 4  │     296.40 / 299.96 ±2.05 / 302.63 ms │    288.27 / 299.50 ±10.63 / 318.57 ms │     no change │
│ QQuery 5  │     346.33 / 351.83 ±4.20 / 357.43 ms │     342.82 / 346.63 ±3.12 / 350.28 ms │     no change │
│ QQuery 6  │           5.37 / 6.41 ±1.16 / 8.60 ms │          5.61 / 7.52 ±1.98 / 10.94 ms │  1.17x slower │
│ QQuery 7  │        16.81 / 17.96 ±1.29 / 19.87 ms │        16.64 / 18.39 ±2.00 / 22.16 ms │     no change │
│ QQuery 8  │    442.06 / 457.38 ±17.18 / 489.58 ms │     414.30 / 423.94 ±6.32 / 433.05 ms │ +1.08x faster │
│ QQuery 9  │     644.92 / 658.21 ±8.19 / 668.38 ms │    630.44 / 649.39 ±11.37 / 660.43 ms │     no change │
│ QQuery 10 │       91.97 / 94.00 ±3.02 / 100.00 ms │       92.59 / 96.27 ±2.93 / 101.60 ms │     no change │
│ QQuery 11 │     103.21 / 103.98 ±0.81 / 104.99 ms │     103.15 / 104.46 ±1.11 / 106.50 ms │     no change │
│ QQuery 12 │     339.16 / 342.51 ±3.73 / 349.54 ms │     337.13 / 339.57 ±2.17 / 342.79 ms │     no change │
│ QQuery 13 │    466.77 / 481.18 ±11.81 / 501.82 ms │    445.92 / 462.85 ±13.59 / 484.92 ms │     no change │
│ QQuery 14 │     344.13 / 346.12 ±1.84 / 349.35 ms │     343.09 / 347.38 ±3.57 / 353.39 ms │     no change │
│ QQuery 15 │     343.41 / 355.94 ±8.88 / 367.60 ms │    347.81 / 365.77 ±20.88 / 406.39 ms │     no change │
│ QQuery 16 │     713.61 / 720.42 ±5.05 / 726.05 ms │    714.22 / 726.07 ±13.51 / 751.32 ms │     no change │
│ QQuery 17 │     701.73 / 710.86 ±5.28 / 717.89 ms │     709.05 / 713.22 ±5.08 / 723.09 ms │     no change │
│ QQuery 18 │ 1411.88 / 1477.27 ±45.14 / 1529.23 ms │ 1453.47 / 1486.71 ±37.24 / 1556.34 ms │     no change │
│ QQuery 19 │        34.76 / 36.35 ±1.46 / 38.82 ms │        35.19 / 37.57 ±2.27 / 41.72 ms │     no change │
│ QQuery 20 │    718.38 / 738.80 ±18.38 / 765.09 ms │    714.53 / 729.96 ±14.55 / 757.34 ms │     no change │
│ QQuery 21 │     762.34 / 766.07 ±3.42 / 770.88 ms │     758.42 / 764.28 ±4.10 / 770.06 ms │     no change │
│ QQuery 22 │  1132.99 / 1142.54 ±5.91 / 1148.91 ms │  1132.01 / 1135.41 ±1.95 / 1137.25 ms │     no change │
│ QQuery 23 │  3094.03 / 3102.67 ±5.53 / 3109.51 ms │ 3071.56 / 3099.39 ±22.13 / 3130.36 ms │     no change │
│ QQuery 24 │     100.44 / 102.12 ±1.49 / 104.69 ms │     102.11 / 103.90 ±1.20 / 105.52 ms │     no change │
│ QQuery 25 │     139.27 / 140.94 ±1.45 / 143.17 ms │     139.19 / 140.92 ±1.24 / 142.82 ms │     no change │
│ QQuery 26 │      96.82 / 100.21 ±2.74 / 104.17 ms │     100.38 / 103.21 ±3.21 / 108.87 ms │     no change │
│ QQuery 27 │     850.99 / 858.24 ±4.96 / 863.88 ms │     852.63 / 861.68 ±9.91 / 879.52 ms │     no change │
│ QQuery 28 │ 7689.50 / 7759.69 ±43.76 / 7807.38 ms │ 7722.65 / 7743.02 ±19.65 / 7777.16 ms │     no change │
│ QQuery 29 │        50.12 / 55.12 ±6.10 / 66.19 ms │        50.54 / 54.94 ±2.48 / 57.92 ms │     no change │
│ QQuery 30 │     353.30 / 368.00 ±7.92 / 375.46 ms │     357.78 / 367.65 ±8.33 / 378.66 ms │     no change │
│ QQuery 31 │     371.33 / 378.09 ±4.08 / 384.00 ms │     369.94 / 381.54 ±8.30 / 389.08 ms │     no change │
│ QQuery 32 │ 1228.55 / 1285.53 ±41.24 / 1355.30 ms │ 1124.34 / 1252.33 ±71.39 / 1332.17 ms │     no change │
│ QQuery 33 │ 1522.61 / 1571.23 ±40.55 / 1642.44 ms │ 1440.28 / 1464.01 ±20.82 / 1500.73 ms │ +1.07x faster │
│ QQuery 34 │ 1492.06 / 1545.44 ±46.43 / 1624.03 ms │ 1463.03 / 1485.03 ±15.03 / 1507.90 ms │     no change │
│ QQuery 35 │     386.29 / 393.28 ±5.61 / 399.37 ms │     383.59 / 390.09 ±5.57 / 398.98 ms │     no change │
│ QQuery 36 │     116.23 / 120.87 ±2.44 / 122.98 ms │     110.80 / 116.85 ±4.49 / 122.90 ms │     no change │
│ QQuery 37 │        47.65 / 50.80 ±2.05 / 53.91 ms │        49.24 / 50.32 ±1.29 / 52.49 ms │     no change │
│ QQuery 38 │        75.19 / 77.12 ±1.64 / 79.77 ms │        75.12 / 77.69 ±2.56 / 82.35 ms │     no change │
│ QQuery 39 │     204.14 / 218.71 ±8.74 / 230.61 ms │     201.27 / 220.09 ±9.79 / 229.31 ms │     no change │
│ QQuery 40 │        23.80 / 26.61 ±1.98 / 29.91 ms │        22.08 / 24.52 ±2.31 / 27.63 ms │ +1.09x faster │
│ QQuery 41 │        19.75 / 21.39 ±1.06 / 22.63 ms │        19.08 / 21.08 ±1.80 / 23.38 ms │     no change │
│ QQuery 42 │        19.01 / 19.68 ±0.64 / 20.82 ms │        18.91 / 19.53 ±0.36 / 19.97 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 27411.45ms │
│ Total Time (fix-join-cardinality-estimation)   │ 27141.79ms │
│ Average Time (HEAD)                            │   637.48ms │
│ Average Time (fix-join-cardinality-estimation) │   631.20ms │
│ Queries Faster                                 │          3 │
│ Queries Slower                                 │          1 │
│ Queries with No Change                         │         39 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	138.1s
Peak memory	37.2 GiB
Avg memory	27.1 GiB
CPU user	1294.0s
CPU sys	103.4s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	136.8s
Peak memory	39.2 GiB
Avg memory	29.3 GiB
CPU user	1287.9s
CPU sys	97.4s
Peak spill	0 B

File an issue against this benchmark runner

Dandandan · 2026-04-09T10:13:15Z

run benchmark tpch tpch10

adriangbot · 2026-04-09T10:15:25Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213348828-993-qjk28 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (775906d) to 5ba06ac (merge-base) diff using: tpch10
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:16:11Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213348828-992-h7v5g 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (775906d) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:18:15Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃       fix-join-cardinality-estimation ┃          Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.72 / 7.13 ±0.76 / 8.64 ms │        10.23 / 10.78 ±0.74 / 12.24 ms │    1.51x slower │
│ QQuery 2  │        146.74 / 147.71 ±0.85 / 149.26 ms │     144.31 / 145.17 ±0.84 / 146.55 ms │       no change │
│ QQuery 3  │        114.38 / 115.72 ±0.77 / 116.72 ms │     112.99 / 113.40 ±0.67 / 114.73 ms │       no change │
│ QQuery 4  │     1288.03 / 1296.29 ±8.10 / 1311.08 ms │ 1470.12 / 1494.88 ±23.98 / 1527.98 ms │    1.15x slower │
│ QQuery 5  │        172.91 / 174.37 ±1.44 / 176.34 ms │     169.55 / 172.80 ±1.98 / 175.10 ms │       no change │
│ QQuery 6  │        840.52 / 846.65 ±3.88 / 851.61 ms │     182.29 / 184.95 ±2.99 / 190.37 ms │   +4.58x faster │
│ QQuery 7  │        342.01 / 345.38 ±1.87 / 347.59 ms │     334.87 / 338.11 ±2.13 / 341.42 ms │       no change │
│ QQuery 8  │        114.74 / 117.80 ±2.54 / 122.31 ms │     113.69 / 114.59 ±0.85 / 115.98 ms │       no change │
│ QQuery 9  │        102.27 / 108.98 ±5.94 / 120.06 ms │      99.49 / 108.22 ±7.46 / 121.98 ms │       no change │
│ QQuery 10 │        106.21 / 106.97 ±0.49 / 107.51 ms │     129.43 / 134.64 ±7.53 / 149.28 ms │    1.26x slower │
│ QQuery 11 │        881.68 / 886.21 ±3.77 / 891.36 ms │    941.28 / 958.09 ±11.93 / 971.63 ms │    1.08x slower │
│ QQuery 12 │           46.19 / 46.68 ±0.35 / 47.18 ms │        43.45 / 44.84 ±1.29 / 46.80 ms │       no change │
│ QQuery 13 │        401.81 / 404.30 ±1.81 / 406.37 ms │     394.90 / 398.35 ±1.82 / 400.11 ms │       no change │
│ QQuery 14 │      999.98 / 1007.76 ±4.10 / 1010.98 ms │  1583.05 / 1587.53 ±3.10 / 1591.93 ms │    1.58x slower │
│ QQuery 15 │           15.67 / 16.21 ±0.49 / 17.00 ms │        15.05 / 16.12 ±1.05 / 17.43 ms │       no change │
│ QQuery 16 │              7.12 / 7.72 ±0.57 / 8.72 ms │        46.01 / 47.01 ±0.60 / 47.83 ms │    6.09x slower │
│ QQuery 17 │        226.68 / 230.75 ±2.80 / 234.57 ms │     225.74 / 227.15 ±1.16 / 229.17 ms │       no change │
│ QQuery 18 │        126.49 / 127.90 ±1.53 / 130.45 ms │     121.66 / 123.02 ±1.20 / 124.82 ms │       no change │
│ QQuery 19 │        153.96 / 154.84 ±0.60 / 155.83 ms │     153.60 / 155.26 ±1.20 / 156.53 ms │       no change │
│ QQuery 20 │           13.41 / 14.23 ±0.96 / 16.03 ms │        13.20 / 13.91 ±0.83 / 15.51 ms │       no change │
│ QQuery 21 │           19.32 / 19.58 ±0.18 / 19.79 ms │        18.82 / 19.07 ±0.22 / 19.42 ms │       no change │
│ QQuery 22 │        486.95 / 488.43 ±1.21 / 490.38 ms │     486.91 / 492.72 ±4.06 / 498.56 ms │       no change │
│ QQuery 23 │        857.24 / 863.94 ±7.31 / 873.24 ms │  1083.65 / 1089.77 ±6.37 / 1102.02 ms │    1.26x slower │
│ QQuery 24 │        379.84 / 382.14 ±1.80 / 384.71 ms │     754.55 / 759.16 ±3.24 / 762.72 ms │    1.99x slower │
│ QQuery 25 │        339.90 / 341.58 ±1.36 / 343.38 ms │     268.61 / 271.89 ±2.48 / 276.20 ms │   +1.26x faster │
│ QQuery 26 │           79.82 / 80.80 ±0.72 / 81.96 ms │        78.95 / 80.85 ±1.05 / 82.07 ms │       no change │
│ QQuery 27 │              6.90 / 7.32 ±0.45 / 8.14 ms │           7.42 / 8.15 ±0.85 / 9.68 ms │    1.11x slower │
│ QQuery 28 │        149.54 / 151.60 ±1.29 / 153.36 ms │     147.74 / 149.18 ±0.99 / 150.81 ms │       no change │
│ QQuery 29 │        279.88 / 281.44 ±1.14 / 283.19 ms │     255.19 / 257.72 ±1.97 / 259.95 ms │   +1.09x faster │
│ QQuery 30 │           42.25 / 44.46 ±1.31 / 46.26 ms │        38.79 / 40.00 ±1.20 / 42.14 ms │   +1.11x faster │
│ QQuery 31 │        167.66 / 170.65 ±2.18 / 173.98 ms │     415.23 / 418.09 ±2.27 / 421.68 ms │    2.45x slower │
│ QQuery 32 │           57.09 / 57.93 ±0.96 / 59.69 ms │        13.79 / 14.48 ±0.64 / 15.59 ms │   +4.00x faster │
│ QQuery 33 │        138.90 / 140.17 ±0.66 / 140.83 ms │     137.16 / 138.67 ±1.02 / 140.06 ms │       no change │
│ QQuery 34 │              7.01 / 7.23 ±0.17 / 7.47 ms │        12.06 / 15.07 ±1.51 / 16.05 ms │    2.08x slower │
│ QQuery 35 │        105.18 / 107.82 ±1.67 / 110.28 ms │     123.55 / 125.48 ±1.01 / 126.26 ms │    1.16x slower │
│ QQuery 36 │              6.58 / 7.14 ±0.51 / 8.08 ms │           6.62 / 6.87 ±0.24 / 7.23 ms │       no change │
│ QQuery 37 │             8.15 / 9.49 ±1.53 / 12.13 ms │          8.05 / 8.73 ±0.76 / 10.18 ms │   +1.09x faster │
│ QQuery 38 │           83.45 / 87.44 ±4.76 / 96.65 ms │     135.92 / 137.78 ±1.78 / 140.54 ms │    1.58x slower │
│ QQuery 39 │        125.48 / 125.81 ±0.24 / 126.20 ms │     122.88 / 125.59 ±1.79 / 128.13 ms │       no change │
│ QQuery 40 │        110.16 / 114.43 ±5.88 / 125.89 ms │        20.24 / 21.26 ±0.75 / 22.34 ms │   +5.38x faster │
│ QQuery 41 │           14.20 / 15.24 ±0.91 / 16.53 ms │        14.26 / 14.59 ±0.25 / 14.96 ms │       no change │
│ QQuery 42 │        105.50 / 109.12 ±2.39 / 112.33 ms │     105.98 / 108.25 ±1.23 / 109.55 ms │       no change │
│ QQuery 43 │              6.10 / 6.21 ±0.16 / 6.53 ms │           5.89 / 6.00 ±0.11 / 6.20 ms │       no change │
│ QQuery 44 │           11.66 / 11.88 ±0.20 / 12.17 ms │        11.50 / 11.96 ±0.63 / 13.21 ms │       no change │
│ QQuery 45 │           49.88 / 50.67 ±0.54 / 51.30 ms │        46.78 / 48.07 ±0.86 / 49.48 ms │   +1.05x faster │
│ QQuery 46 │             8.40 / 8.95 ±0.59 / 10.05 ms │        16.47 / 17.18 ±0.46 / 17.67 ms │    1.92x slower │
│ QQuery 47 │        680.98 / 690.57 ±5.70 / 697.63 ms │     678.16 / 683.88 ±3.49 / 688.00 ms │       no change │
│ QQuery 48 │        289.31 / 293.59 ±3.54 / 298.09 ms │     279.82 / 282.03 ±1.52 / 284.41 ms │       no change │
│ QQuery 49 │        252.90 / 256.36 ±2.57 / 260.89 ms │     251.89 / 253.65 ±1.49 / 256.22 ms │       no change │
│ QQuery 50 │        220.73 / 225.81 ±3.61 / 230.89 ms │     183.68 / 185.61 ±2.01 / 188.98 ms │   +1.22x faster │
│ QQuery 51 │        181.87 / 183.91 ±2.22 / 188.08 ms │     178.19 / 179.83 ±1.69 / 182.88 ms │       no change │
│ QQuery 52 │        107.32 / 108.34 ±0.82 / 109.37 ms │     106.95 / 108.90 ±2.34 / 113.20 ms │       no change │
│ QQuery 53 │        102.25 / 103.47 ±1.05 / 104.87 ms │     102.45 / 103.14 ±0.44 / 103.59 ms │       no change │
│ QQuery 54 │        146.23 / 147.83 ±1.16 / 149.22 ms │     188.89 / 191.00 ±1.68 / 193.28 ms │    1.29x slower │
│ QQuery 55 │        106.81 / 109.42 ±1.57 / 111.62 ms │     105.83 / 107.54 ±1.74 / 110.86 ms │       no change │
│ QQuery 56 │        140.77 / 142.58 ±1.84 / 145.74 ms │     138.63 / 139.66 ±1.06 / 141.53 ms │       no change │
│ QQuery 57 │        171.56 / 173.61 ±1.59 / 175.15 ms │     174.85 / 176.96 ±1.50 / 179.00 ms │       no change │
│ QQuery 58 │        293.98 / 306.05 ±9.31 / 322.21 ms │     318.09 / 319.73 ±1.50 / 321.49 ms │       no change │
│ QQuery 59 │        197.15 / 200.21 ±1.63 / 201.79 ms │     370.67 / 375.06 ±2.65 / 377.62 ms │    1.87x slower │
│ QQuery 60 │        143.22 / 144.56 ±1.30 / 147.02 ms │     140.60 / 142.10 ±1.03 / 143.04 ms │       no change │
│ QQuery 61 │           12.95 / 13.15 ±0.35 / 13.85 ms │        18.76 / 19.27 ±0.83 / 20.92 ms │    1.47x slower │
│ QQuery 62 │       871.58 / 926.73 ±39.18 / 981.23 ms │        63.97 / 64.61 ±0.48 / 65.33 ms │  +14.34x faster │
│ QQuery 63 │        102.46 / 105.89 ±3.24 / 111.79 ms │     102.88 / 104.02 ±0.89 / 105.37 ms │       no change │
│ QQuery 64 │        675.35 / 681.86 ±3.30 / 684.52 ms │  1242.00 / 1250.15 ±7.36 / 1263.28 ms │    1.83x slower │
│ QQuery 65 │        245.59 / 253.15 ±5.50 / 261.10 ms │     256.08 / 260.22 ±2.63 / 262.95 ms │       no change │
│ QQuery 66 │       244.36 / 264.30 ±20.01 / 298.43 ms │     106.28 / 107.33 ±0.76 / 108.48 ms │   +2.46x faster │
│ QQuery 67 │        313.60 / 317.29 ±3.00 / 322.49 ms │    598.88 / 624.35 ±19.30 / 648.29 ms │    1.97x slower │
│ QQuery 68 │            9.50 / 10.54 ±0.91 / 12.01 ms │        14.65 / 16.04 ±1.18 / 17.92 ms │    1.52x slower │
│ QQuery 69 │        101.57 / 103.26 ±1.26 / 104.83 ms │     125.45 / 127.90 ±2.14 / 131.50 ms │    1.24x slower │
│ QQuery 70 │       331.14 / 344.43 ±11.20 / 360.06 ms │     316.81 / 321.30 ±4.28 / 328.86 ms │   +1.07x faster │
│ QQuery 71 │        134.30 / 135.10 ±0.81 / 136.63 ms │     134.59 / 135.50 ±0.66 / 136.35 ms │       no change │
│ QQuery 72 │        604.89 / 616.25 ±7.52 / 624.30 ms │     594.87 / 608.78 ±9.84 / 621.97 ms │       no change │
│ QQuery 73 │              7.15 / 8.15 ±0.55 / 8.72 ms │        12.38 / 14.71 ±2.18 / 18.67 ms │    1.80x slower │
│ QQuery 74 │        539.10 / 547.47 ±7.35 / 560.10 ms │     592.44 / 596.67 ±2.50 / 599.28 ms │    1.09x slower │
│ QQuery 75 │        276.98 / 278.71 ±1.69 / 281.46 ms │     434.14 / 437.36 ±2.63 / 440.18 ms │    1.57x slower │
│ QQuery 76 │        131.82 / 133.89 ±1.47 / 136.07 ms │     133.37 / 134.58 ±0.89 / 136.05 ms │       no change │
│ QQuery 77 │        188.10 / 190.16 ±2.05 / 193.27 ms │     187.20 / 189.54 ±1.62 / 191.87 ms │       no change │
│ QQuery 78 │        336.64 / 341.17 ±5.02 / 350.18 ms │     366.07 / 368.80 ±2.09 / 372.25 ms │    1.08x slower │
│ QQuery 79 │        230.61 / 233.25 ±2.00 / 235.14 ms │     228.43 / 231.39 ±2.18 / 235.10 ms │       no change │
│ QQuery 80 │        322.51 / 325.40 ±2.37 / 329.13 ms │     320.73 / 325.12 ±2.91 / 329.19 ms │       no change │
│ QQuery 81 │           25.72 / 27.10 ±0.85 / 27.88 ms │        30.44 / 31.34 ±0.70 / 32.45 ms │    1.16x slower │
│ QQuery 82 │        198.55 / 200.54 ±1.61 / 203.27 ms │        40.64 / 42.17 ±0.99 / 43.75 ms │   +4.76x faster │
│ QQuery 83 │           37.88 / 39.06 ±1.03 / 40.80 ms │        45.89 / 47.22 ±1.43 / 49.76 ms │    1.21x slower │
│ QQuery 84 │           47.54 / 49.25 ±1.96 / 53.02 ms │        39.73 / 40.41 ±0.61 / 41.53 ms │   +1.22x faster │
│ QQuery 85 │        148.30 / 149.71 ±0.83 / 150.83 ms │     150.75 / 151.83 ±1.23 / 153.91 ms │       no change │
│ QQuery 86 │           38.68 / 39.94 ±1.44 / 42.61 ms │        39.49 / 40.18 ±0.81 / 41.41 ms │       no change │
│ QQuery 87 │           84.11 / 87.43 ±2.04 / 90.16 ms │     135.69 / 137.73 ±1.31 / 139.63 ms │    1.58x slower │
│ QQuery 88 │         99.67 / 100.63 ±0.80 / 101.61 ms │      99.26 / 104.43 ±8.54 / 121.39 ms │       no change │
│ QQuery 89 │        118.00 / 120.11 ±1.41 / 121.71 ms │     115.95 / 117.68 ±1.41 / 119.73 ms │       no change │
│ QQuery 90 │           23.79 / 24.08 ±0.23 / 24.46 ms │        21.73 / 22.23 ±0.27 / 22.50 ms │   +1.08x faster │
│ QQuery 91 │           61.49 / 63.75 ±1.58 / 65.22 ms │        54.49 / 56.34 ±1.47 / 58.74 ms │   +1.13x faster │
│ QQuery 92 │           56.79 / 57.61 ±1.09 / 59.76 ms │        56.35 / 57.21 ±0.72 / 58.20 ms │       no change │
│ QQuery 93 │        185.49 / 187.82 ±1.37 / 189.77 ms │     184.43 / 186.81 ±1.43 / 188.76 ms │       no change │
│ QQuery 94 │           60.12 / 61.57 ±1.03 / 62.93 ms │        66.74 / 67.26 ±0.57 / 68.18 ms │    1.09x slower │
│ QQuery 95 │        127.73 / 129.49 ±1.27 / 131.20 ms │     125.11 / 126.23 ±1.03 / 127.85 ms │       no change │
│ QQuery 96 │           72.32 / 73.53 ±1.03 / 75.14 ms │        66.35 / 67.72 ±0.89 / 68.78 ms │   +1.09x faster │
│ QQuery 97 │        127.07 / 129.15 ±1.95 / 132.04 ms │     114.37 / 115.34 ±0.86 / 116.75 ms │   +1.12x faster │
│ QQuery 98 │        155.03 / 156.36 ±0.82 / 157.21 ms │     150.49 / 152.78 ±1.64 / 154.98 ms │       no change │
│ QQuery 99 │ 10707.93 / 10748.93 ±36.10 / 10798.10 ms │        59.85 / 60.62 ±0.49 / 61.07 ms │ +177.30x faster │
└───────────┴──────────────────────────────────────────┴───────────────────────────────────────┴─────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 31217.62ms │
│ Total Time (fix-join-cardinality-estimation)   │ 21587.67ms │
│ Average Time (HEAD)                            │   315.33ms │
│ Average Time (fix-join-cardinality-estimation) │   218.06ms │
│ Queries Faster                                 │         19 │
│ Queries Slower                                 │         29 │
│ Queries with No Change                         │         51 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	156.4s
Peak memory	5.3 GiB
Avg memory	4.4 GiB
CPU user	257.3s
CPU sys	17.4s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	108.2s
Peak memory	4.6 GiB
Avg memory	4.1 GiB
CPU user	196.0s
CPU sys	14.9s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:19:10Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.19 / 4.35 ±6.26 / 16.86 ms │          1.21 / 4.38 ±6.30 / 16.97 ms │     no change │
│ QQuery 1  │        14.16 / 14.38 ±0.14 / 14.57 ms │        13.99 / 14.29 ±0.17 / 14.47 ms │     no change │
│ QQuery 2  │        43.89 / 44.34 ±0.39 / 45.05 ms │        43.02 / 43.33 ±0.47 / 44.26 ms │     no change │
│ QQuery 3  │        42.31 / 45.38 ±1.95 / 47.87 ms │        43.62 / 48.43 ±3.51 / 54.31 ms │  1.07x slower │
│ QQuery 4  │    278.33 / 287.51 ±10.92 / 308.41 ms │     285.19 / 290.64 ±4.72 / 297.47 ms │     no change │
│ QQuery 5  │     335.27 / 340.32 ±3.73 / 346.54 ms │     331.90 / 337.67 ±5.05 / 344.82 ms │     no change │
│ QQuery 6  │          5.87 / 8.23 ±2.76 / 13.64 ms │           5.77 / 6.89 ±0.82 / 8.08 ms │ +1.20x faster │
│ QQuery 7  │        16.44 / 16.52 ±0.08 / 16.64 ms │        16.38 / 17.88 ±1.71 / 20.77 ms │  1.08x slower │
│ QQuery 8  │    398.44 / 413.21 ±12.64 / 436.20 ms │     408.53 / 420.92 ±7.46 / 430.65 ms │     no change │
│ QQuery 9  │    620.67 / 633.72 ±11.44 / 653.51 ms │     621.30 / 631.22 ±6.67 / 641.62 ms │     no change │
│ QQuery 10 │        89.10 / 91.00 ±1.60 / 93.36 ms │       92.59 / 95.12 ±2.92 / 100.43 ms │     no change │
│ QQuery 11 │     102.81 / 103.83 ±0.75 / 105.11 ms │     104.17 / 105.43 ±1.01 / 107.07 ms │     no change │
│ QQuery 12 │     334.11 / 340.87 ±6.49 / 348.75 ms │     335.23 / 337.74 ±1.90 / 340.27 ms │     no change │
│ QQuery 13 │    458.71 / 470.69 ±12.44 / 489.43 ms │    456.62 / 466.51 ±12.05 / 489.56 ms │     no change │
│ QQuery 14 │     332.83 / 336.39 ±2.81 / 339.17 ms │     336.60 / 344.22 ±5.22 / 350.67 ms │     no change │
│ QQuery 15 │    349.32 / 361.77 ±12.61 / 384.90 ms │    343.73 / 355.74 ±10.33 / 373.39 ms │     no change │
│ QQuery 16 │    697.82 / 716.08 ±17.26 / 746.73 ms │    717.87 / 740.04 ±19.01 / 772.08 ms │     no change │
│ QQuery 17 │    695.83 / 713.41 ±19.97 / 751.19 ms │     703.18 / 712.65 ±6.09 / 719.43 ms │     no change │
│ QQuery 18 │ 1376.79 / 1409.45 ±40.25 / 1479.75 ms │ 1360.08 / 1425.01 ±46.72 / 1476.88 ms │     no change │
│ QQuery 19 │      34.69 / 59.89 ±47.19 / 154.25 ms │        37.07 / 37.71 ±0.56 / 38.72 ms │ +1.59x faster │
│ QQuery 20 │    710.63 / 736.10 ±24.26 / 774.61 ms │     708.77 / 718.69 ±8.71 / 734.44 ms │     no change │
│ QQuery 21 │     757.02 / 764.86 ±6.05 / 771.82 ms │     757.22 / 761.77 ±5.24 / 771.22 ms │     no change │
│ QQuery 22 │  1127.64 / 1131.88 ±4.31 / 1139.70 ms │  1125.05 / 1132.82 ±6.93 / 1141.89 ms │     no change │
│ QQuery 23 │ 3031.55 / 3056.32 ±21.02 / 3093.69 ms │  3084.61 / 3092.59 ±7.43 / 3106.70 ms │     no change │
│ QQuery 24 │       99.14 / 99.95 ±0.87 / 101.13 ms │     100.53 / 101.59 ±0.99 / 103.27 ms │     no change │
│ QQuery 25 │     136.27 / 139.35 ±1.94 / 141.94 ms │     138.74 / 142.29 ±2.44 / 145.78 ms │     no change │
│ QQuery 26 │       95.65 / 99.21 ±1.98 / 101.57 ms │       96.03 / 99.83 ±2.96 / 104.71 ms │     no change │
│ QQuery 27 │     848.35 / 855.67 ±6.94 / 867.17 ms │    847.56 / 855.22 ±11.17 / 876.79 ms │     no change │
│ QQuery 28 │ 7644.06 / 7736.30 ±46.74 / 7773.32 ms │ 7697.04 / 7734.17 ±30.24 / 7778.47 ms │     no change │
│ QQuery 29 │        49.91 / 54.75 ±6.72 / 67.90 ms │        49.71 / 52.85 ±5.03 / 62.89 ms │     no change │
│ QQuery 30 │     359.29 / 365.52 ±5.29 / 374.50 ms │     352.05 / 357.75 ±4.05 / 361.94 ms │     no change │
│ QQuery 31 │    361.72 / 376.02 ±10.37 / 389.42 ms │     372.17 / 379.07 ±4.80 / 385.83 ms │     no change │
│ QQuery 32 │ 1237.50 / 1267.72 ±30.97 / 1312.99 ms │ 1016.02 / 1030.12 ±16.81 / 1063.11 ms │ +1.23x faster │
│ QQuery 33 │ 1496.21 / 1542.41 ±25.50 / 1569.66 ms │  1425.15 / 1439.36 ±9.59 / 1453.56 ms │ +1.07x faster │
│ QQuery 34 │ 1446.17 / 1514.08 ±50.86 / 1591.77 ms │ 1459.12 / 1478.28 ±14.42 / 1503.30 ms │     no change │
│ QQuery 35 │     377.85 / 385.03 ±5.31 / 391.56 ms │     374.61 / 380.51 ±6.60 / 391.90 ms │     no change │
│ QQuery 36 │     112.42 / 117.90 ±3.48 / 122.96 ms │     118.24 / 120.79 ±1.81 / 123.41 ms │     no change │
│ QQuery 37 │        46.88 / 48.75 ±1.80 / 51.95 ms │        47.12 / 49.44 ±1.36 / 51.32 ms │     no change │
│ QQuery 38 │        74.50 / 76.38 ±1.55 / 79.05 ms │        75.08 / 76.74 ±1.26 / 78.92 ms │     no change │
│ QQuery 39 │     204.71 / 214.15 ±6.30 / 222.75 ms │     207.71 / 214.44 ±6.60 / 225.21 ms │     no change │
│ QQuery 40 │        21.72 / 24.73 ±2.44 / 27.35 ms │        20.59 / 23.45 ±1.55 / 24.96 ms │ +1.05x faster │
│ QQuery 41 │        19.99 / 21.13 ±1.05 / 23.02 ms │        19.92 / 21.75 ±2.30 / 25.45 ms │     no change │
│ QQuery 42 │        18.55 / 19.42 ±0.76 / 20.57 ms │        18.95 / 19.53 ±0.43 / 20.26 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 27058.99ms │
│ Total Time (fix-join-cardinality-estimation)   │ 26718.87ms │
│ Average Time (HEAD)                            │   629.28ms │
│ Average Time (fix-join-cardinality-estimation) │   621.37ms │
│ Queries Faster                                 │          5 │
│ Queries Slower                                 │          2 │
│ Queries with No Change                         │         36 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	136.3s
Peak memory	39.4 GiB
Avg memory	28.6 GiB
CPU user	1275.7s
CPU sys	103.1s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	134.6s
Peak memory	38.9 GiB
Avg memory	32.4 GiB
CPU user	1278.2s
CPU sys	84.8s
Peak spill	0 B

File an issue against this benchmark runner

Dandandan · 2026-04-09T10:20:08Z

run benchmarks

Dandandan · 2026-04-09T10:20:25Z

run benchmark tpch tpch10

adriangbot · 2026-04-09T10:20:28Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213388053-994-bcsvk 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (3e0ab51) to 5ba06ac (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:22:04Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213388053-996-tpffn 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (3e0ab51) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:22:33Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213389643-998-l6vlw 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (3e0ab51) to 5ba06ac (merge-base) diff using: tpch10
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:22:53Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213389643-997-dlqlp 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (3e0ab51) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:22:55Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213388053-995-479j2 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (3e0ab51) to 5ba06ac (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:37:22Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.21 / 4.40 ±6.31 / 17.03 ms │          1.17 / 4.39 ±6.33 / 17.05 ms │     no change │
│ QQuery 1  │        14.35 / 14.59 ±0.13 / 14.71 ms │        14.09 / 14.29 ±0.17 / 14.53 ms │     no change │
│ QQuery 2  │        43.49 / 44.13 ±0.38 / 44.60 ms │        43.38 / 43.86 ±0.38 / 44.40 ms │     no change │
│ QQuery 3  │        40.44 / 44.22 ±2.80 / 48.36 ms │        44.13 / 46.06 ±1.74 / 48.38 ms │     no change │
│ QQuery 4  │     288.69 / 295.35 ±7.46 / 306.59 ms │     280.71 / 296.28 ±9.43 / 306.00 ms │     no change │
│ QQuery 5  │     342.52 / 347.95 ±4.67 / 356.66 ms │     337.38 / 340.64 ±3.63 / 346.57 ms │     no change │
│ QQuery 6  │           5.46 / 6.17 ±0.51 / 7.03 ms │           5.66 / 6.55 ±0.60 / 7.23 ms │  1.06x slower │
│ QQuery 7  │        16.41 / 17.20 ±1.13 / 19.43 ms │        16.58 / 17.52 ±1.43 / 20.36 ms │     no change │
│ QQuery 8  │     427.37 / 429.57 ±2.53 / 434.39 ms │     405.81 / 411.19 ±4.54 / 417.84 ms │     no change │
│ QQuery 9  │     627.39 / 635.69 ±5.75 / 645.34 ms │     623.50 / 634.04 ±7.36 / 643.31 ms │     no change │
│ QQuery 10 │       90.01 / 93.15 ±4.41 / 101.87 ms │       91.54 / 94.80 ±3.18 / 100.72 ms │     no change │
│ QQuery 11 │     102.75 / 103.68 ±0.83 / 105.17 ms │     104.08 / 105.51 ±1.27 / 107.45 ms │     no change │
│ QQuery 12 │     333.66 / 339.00 ±4.07 / 345.87 ms │     335.08 / 341.97 ±6.97 / 355.03 ms │     no change │
│ QQuery 13 │     438.44 / 456.16 ±9.21 / 464.96 ms │     456.11 / 471.00 ±9.22 / 482.12 ms │     no change │
│ QQuery 14 │     337.00 / 343.21 ±5.59 / 353.59 ms │     339.53 / 345.47 ±4.46 / 351.62 ms │     no change │
│ QQuery 15 │    341.85 / 353.89 ±11.40 / 374.36 ms │    342.65 / 365.76 ±19.03 / 392.66 ms │     no change │
│ QQuery 16 │     694.58 / 707.11 ±6.65 / 713.64 ms │    707.33 / 742.31 ±51.96 / 845.19 ms │     no change │
│ QQuery 17 │    697.85 / 737.92 ±30.96 / 775.59 ms │     696.61 / 701.00 ±4.86 / 710.00 ms │ +1.05x faster │
│ QQuery 18 │ 1386.55 / 1461.40 ±39.16 / 1496.61 ms │ 1441.91 / 1473.93 ±21.57 / 1499.21 ms │     no change │
│ QQuery 19 │        33.92 / 36.56 ±3.59 / 43.65 ms │        37.07 / 39.65 ±1.75 / 42.43 ms │  1.08x slower │
│ QQuery 20 │    721.54 / 734.47 ±12.96 / 752.33 ms │    723.30 / 735.59 ±17.89 / 770.70 ms │     no change │
│ QQuery 21 │     764.84 / 766.57 ±1.05 / 767.86 ms │     762.59 / 768.10 ±4.13 / 772.90 ms │     no change │
│ QQuery 22 │  1121.69 / 1130.61 ±7.61 / 1144.23 ms │  1125.34 / 1130.15 ±4.61 / 1136.51 ms │     no change │
│ QQuery 23 │ 3048.41 / 3081.19 ±28.08 / 3132.21 ms │ 3083.14 / 3103.77 ±11.32 / 3114.53 ms │     no change │
│ QQuery 24 │      99.19 / 102.33 ±2.19 / 104.86 ms │      97.82 / 101.99 ±3.10 / 106.58 ms │     no change │
│ QQuery 25 │     136.73 / 138.60 ±1.72 / 141.07 ms │     138.58 / 140.69 ±1.75 / 143.68 ms │     no change │
│ QQuery 26 │       97.85 / 99.73 ±1.39 / 101.35 ms │      99.50 / 101.51 ±1.50 / 103.15 ms │     no change │
│ QQuery 27 │     850.52 / 854.83 ±2.71 / 857.38 ms │     851.02 / 855.38 ±5.22 / 865.59 ms │     no change │
│ QQuery 28 │ 7680.92 / 7725.04 ±30.19 / 7772.41 ms │ 7648.42 / 7739.34 ±47.13 / 7777.45 ms │     no change │
│ QQuery 29 │        49.85 / 54.71 ±5.25 / 63.55 ms │        50.07 / 56.83 ±7.63 / 71.59 ms │     no change │
│ QQuery 30 │     350.93 / 357.47 ±4.71 / 365.65 ms │     352.22 / 360.90 ±7.01 / 373.28 ms │     no change │
│ QQuery 31 │     358.34 / 369.83 ±9.68 / 381.42 ms │     370.24 / 383.47 ±9.81 / 394.73 ms │     no change │
│ QQuery 32 │ 1273.30 / 1296.68 ±12.66 / 1310.56 ms │ 1231.57 / 1297.24 ±49.83 / 1386.19 ms │     no change │
│ QQuery 33 │ 1432.54 / 1472.49 ±39.70 / 1547.72 ms │ 1494.52 / 1563.55 ±52.25 / 1646.68 ms │  1.06x slower │
│ QQuery 34 │ 1432.00 / 1449.60 ±17.18 / 1477.26 ms │ 1493.77 / 1532.24 ±29.29 / 1579.91 ms │  1.06x slower │
│ QQuery 35 │     371.67 / 376.10 ±3.93 / 381.24 ms │     375.92 / 385.01 ±6.61 / 395.37 ms │     no change │
│ QQuery 36 │     115.08 / 118.55 ±2.92 / 123.85 ms │     111.25 / 119.71 ±4.73 / 124.52 ms │     no change │
│ QQuery 37 │        45.94 / 48.23 ±1.18 / 49.14 ms │        47.02 / 48.69 ±1.35 / 50.36 ms │     no change │
│ QQuery 38 │        74.25 / 75.72 ±1.46 / 78.38 ms │        74.70 / 75.74 ±0.66 / 76.54 ms │     no change │
│ QQuery 39 │     205.45 / 212.04 ±4.16 / 218.51 ms │     218.89 / 219.55 ±0.54 / 220.43 ms │     no change │
│ QQuery 40 │        22.39 / 24.93 ±2.09 / 28.51 ms │        21.66 / 24.24 ±1.57 / 26.59 ms │     no change │
│ QQuery 41 │        19.74 / 20.12 ±0.30 / 20.62 ms │        20.47 / 21.70 ±1.14 / 23.18 ms │  1.08x slower │
│ QQuery 42 │        18.05 / 20.08 ±2.92 / 25.83 ms │        18.84 / 19.63 ±0.55 / 20.47 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 27001.26ms │
│ Total Time (fix-join-cardinality-estimation)   │ 27281.22ms │
│ Average Time (HEAD)                            │   627.94ms │
│ Average Time (fix-join-cardinality-estimation) │   634.45ms │
│ Queries Faster                                 │          1 │
│ Queries Slower                                 │          5 │
│ Queries with No Change                         │         37 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	136.0s
Peak memory	37.4 GiB
Avg memory	27.0 GiB
CPU user	1276.7s
CPU sys	99.2s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	137.5s
Peak memory	36.7 GiB
Avg memory	29.8 GiB
CPU user	1283.9s
CPU sys	105.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:50:03Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.27 / 4.64 ±6.55 / 17.75 ms │          1.29 / 4.64 ±6.50 / 17.64 ms │     no change │
│ QQuery 1  │        14.52 / 15.11 ±0.33 / 15.46 ms │        15.03 / 15.77 ±0.38 / 16.10 ms │     no change │
│ QQuery 2  │        44.54 / 45.06 ±0.57 / 46.11 ms │        44.54 / 45.21 ±0.48 / 45.81 ms │     no change │
│ QQuery 3  │        41.67 / 45.77 ±3.29 / 49.85 ms │        43.38 / 46.05 ±2.28 / 49.04 ms │     no change │
│ QQuery 4  │     313.76 / 320.16 ±4.15 / 326.49 ms │    286.42 / 316.08 ±19.70 / 340.33 ms │     no change │
│ QQuery 5  │     362.11 / 376.29 ±8.62 / 384.24 ms │    337.24 / 354.03 ±17.01 / 378.52 ms │ +1.06x faster │
│ QQuery 6  │          5.62 / 7.24 ±1.58 / 10.22 ms │           5.44 / 7.10 ±1.50 / 9.01 ms │     no change │
│ QQuery 7  │        17.31 / 17.55 ±0.17 / 17.74 ms │        17.52 / 18.01 ±0.47 / 18.91 ms │     no change │
│ QQuery 8  │     446.43 / 453.09 ±5.14 / 460.55 ms │     447.02 / 456.57 ±6.52 / 467.54 ms │     no change │
│ QQuery 9  │    636.40 / 685.25 ±33.50 / 718.76 ms │    639.14 / 710.62 ±36.78 / 739.89 ms │     no change │
│ QQuery 10 │        90.46 / 92.16 ±1.78 / 94.76 ms │       92.48 / 95.51 ±4.40 / 104.27 ms │     no change │
│ QQuery 11 │     103.43 / 104.05 ±0.68 / 105.06 ms │     105.26 / 105.71 ±0.42 / 106.45 ms │     no change │
│ QQuery 12 │     340.10 / 347.90 ±4.64 / 352.76 ms │     338.07 / 345.64 ±7.65 / 359.44 ms │     no change │
│ QQuery 13 │     464.76 / 474.82 ±6.63 / 484.47 ms │    499.44 / 518.10 ±13.12 / 539.47 ms │  1.09x slower │
│ QQuery 14 │     369.86 / 382.42 ±6.53 / 387.42 ms │     381.13 / 387.70 ±4.78 / 394.77 ms │     no change │
│ QQuery 15 │    400.04 / 409.91 ±12.42 / 434.42 ms │    373.01 / 436.85 ±43.80 / 502.88 ms │  1.07x slower │
│ QQuery 16 │    715.19 / 777.01 ±31.57 / 798.89 ms │    725.17 / 740.90 ±19.93 / 779.37 ms │     no change │
│ QQuery 17 │    699.92 / 750.73 ±25.78 / 770.49 ms │     710.87 / 715.21 ±4.29 / 722.02 ms │     no change │
│ QQuery 18 │ 1486.78 / 1535.28 ±48.49 / 1596.07 ms │ 1436.78 / 1466.48 ±19.99 / 1487.01 ms │     no change │
│ QQuery 19 │        37.46 / 41.86 ±5.65 / 52.85 ms │        36.06 / 37.28 ±1.57 / 40.25 ms │ +1.12x faster │
│ QQuery 20 │    744.12 / 757.09 ±14.51 / 781.54 ms │     727.90 / 734.71 ±7.73 / 747.56 ms │     no change │
│ QQuery 21 │    766.59 / 790.11 ±12.09 / 798.95 ms │     786.34 / 789.23 ±2.77 / 793.32 ms │     no change │
│ QQuery 22 │  1136.37 / 1143.00 ±4.98 / 1149.90 ms │ 1156.05 / 1173.54 ±13.78 / 1191.00 ms │     no change │
│ QQuery 23 │ 3121.52 / 3205.37 ±66.29 / 3292.75 ms │ 3137.04 / 3241.13 ±74.67 / 3349.53 ms │     no change │
│ QQuery 24 │     105.60 / 107.72 ±1.29 / 109.40 ms │     103.15 / 106.45 ±2.15 / 109.64 ms │     no change │
│ QQuery 25 │     143.70 / 146.15 ±1.59 / 148.27 ms │     140.77 / 142.31 ±1.38 / 144.75 ms │     no change │
│ QQuery 26 │     104.77 / 106.23 ±1.75 / 109.24 ms │     105.35 / 106.58 ±0.85 / 107.95 ms │     no change │
│ QQuery 27 │     853.97 / 863.33 ±6.98 / 871.21 ms │     857.88 / 875.28 ±9.77 / 886.27 ms │     no change │
│ QQuery 28 │ 7778.14 / 7820.47 ±43.49 / 7873.87 ms │ 7767.27 / 7806.43 ±33.47 / 7848.16 ms │     no change │
│ QQuery 29 │        50.86 / 55.28 ±4.58 / 62.06 ms │        51.04 / 56.45 ±4.83 / 64.53 ms │     no change │
│ QQuery 30 │     357.40 / 367.49 ±6.28 / 374.14 ms │     354.24 / 365.96 ±7.13 / 375.09 ms │     no change │
│ QQuery 31 │    349.77 / 367.67 ±12.38 / 385.81 ms │    384.55 / 399.23 ±12.60 / 419.65 ms │  1.09x slower │
│ QQuery 32 │ 1085.00 / 1184.37 ±77.66 / 1320.06 ms │ 1314.63 / 1443.67 ±77.04 / 1556.41 ms │  1.22x slower │
│ QQuery 33 │ 1657.46 / 1705.84 ±37.62 / 1763.25 ms │ 1550.36 / 1602.99 ±35.97 / 1643.07 ms │ +1.06x faster │
│ QQuery 34 │ 1465.59 / 1547.18 ±72.00 / 1632.07 ms │ 1453.44 / 1541.14 ±73.32 / 1623.00 ms │     no change │
│ QQuery 35 │    382.04 / 410.49 ±26.26 / 457.06 ms │     467.20 / 468.02 ±0.90 / 469.74 ms │  1.14x slower │
│ QQuery 36 │     118.44 / 127.95 ±5.70 / 134.39 ms │     112.15 / 121.24 ±5.41 / 127.69 ms │ +1.06x faster │
│ QQuery 37 │        52.16 / 53.75 ±1.25 / 54.87 ms │        47.16 / 49.20 ±2.27 / 53.55 ms │ +1.09x faster │
│ QQuery 38 │        76.84 / 79.14 ±1.56 / 81.63 ms │        73.61 / 75.02 ±0.98 / 76.34 ms │ +1.05x faster │
│ QQuery 39 │     234.94 / 243.72 ±6.71 / 251.97 ms │     206.67 / 213.14 ±7.21 / 226.75 ms │ +1.14x faster │
│ QQuery 40 │        26.08 / 27.93 ±1.84 / 30.28 ms │        21.47 / 25.97 ±3.13 / 29.91 ms │ +1.08x faster │
│ QQuery 41 │        21.87 / 23.88 ±1.57 / 25.95 ms │        18.71 / 19.97 ±0.86 / 21.26 ms │ +1.20x faster │
│ QQuery 42 │        21.27 / 21.48 ±0.17 / 21.78 ms │        19.61 / 20.32 ±0.74 / 21.75 ms │ +1.06x faster │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 28041.94ms │
│ Total Time (fix-join-cardinality-estimation)   │ 28201.46ms │
│ Average Time (HEAD)                            │   652.14ms │
│ Average Time (fix-join-cardinality-estimation) │   655.85ms │
│ Queries Faster                                 │         10 │
│ Queries Slower                                 │          5 │
│ Queries with No Change                         │         28 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	141.5s
Peak memory	39.6 GiB
Avg memory	28.2 GiB
CPU user	1327.0s
CPU sys	104.7s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	142.1s
Peak memory	39.2 GiB
Avg memory	27.5 GiB
CPU user	1330.7s
CPU sys	106.6s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T10:50:12Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃       fix-join-cardinality-estimation ┃          Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.70 / 7.15 ±0.77 / 8.69 ms │        10.53 / 10.91 ±0.69 / 12.30 ms │    1.53x slower │
│ QQuery 2  │        146.00 / 146.50 ±0.37 / 146.93 ms │     145.12 / 146.06 ±0.75 / 147.30 ms │       no change │
│ QQuery 3  │        113.43 / 113.98 ±0.39 / 114.47 ms │     114.04 / 114.66 ±0.55 / 115.50 ms │       no change │
│ QQuery 4  │    1290.15 / 1316.28 ±23.17 / 1351.29 ms │ 1501.24 / 1525.50 ±17.50 / 1546.76 ms │    1.16x slower │
│ QQuery 5  │        172.15 / 175.31 ±1.95 / 177.67 ms │     172.14 / 173.32 ±1.07 / 174.97 ms │       no change │
│ QQuery 6  │       842.91 / 858.26 ±15.58 / 882.22 ms │     188.92 / 190.30 ±0.86 / 191.27 ms │   +4.51x faster │
│ QQuery 7  │        343.43 / 345.87 ±1.66 / 347.59 ms │     339.26 / 342.47 ±2.61 / 345.93 ms │       no change │
│ QQuery 8  │        116.28 / 117.22 ±0.87 / 118.62 ms │     114.00 / 115.33 ±0.80 / 116.29 ms │       no change │
│ QQuery 9  │        101.44 / 109.57 ±9.80 / 128.29 ms │     100.66 / 105.06 ±2.23 / 106.54 ms │       no change │
│ QQuery 10 │        104.89 / 106.12 ±0.93 / 107.40 ms │     131.88 / 133.28 ±0.82 / 134.44 ms │    1.26x slower │
│ QQuery 11 │        897.77 / 906.94 ±6.45 / 917.31 ms │    968.41 / 982.48 ±10.29 / 991.20 ms │    1.08x slower │
│ QQuery 12 │           45.24 / 46.75 ±0.96 / 48.07 ms │        45.08 / 45.77 ±0.59 / 46.86 ms │       no change │
│ QQuery 13 │        405.61 / 409.30 ±3.70 / 414.66 ms │     396.33 / 401.47 ±3.89 / 407.78 ms │       no change │
│ QQuery 14 │      998.30 / 1010.08 ±7.64 / 1019.62 ms │  1595.04 / 1598.85 ±3.02 / 1604.22 ms │    1.58x slower │
│ QQuery 15 │           15.76 / 17.02 ±1.23 / 19.05 ms │        15.10 / 16.15 ±1.48 / 19.05 ms │   +1.05x faster │
│ QQuery 16 │              7.24 / 7.37 ±0.16 / 7.68 ms │        46.52 / 47.24 ±0.56 / 48.12 ms │    6.41x slower │
│ QQuery 17 │        228.44 / 230.10 ±1.12 / 231.30 ms │     228.32 / 229.54 ±0.91 / 231.02 ms │       no change │
│ QQuery 18 │        127.62 / 129.02 ±1.28 / 131.15 ms │     121.71 / 123.15 ±1.05 / 124.99 ms │       no change │
│ QQuery 19 │        155.50 / 156.97 ±1.70 / 159.54 ms │     153.05 / 154.12 ±0.95 / 155.39 ms │       no change │
│ QQuery 20 │           13.73 / 14.43 ±0.61 / 15.47 ms │        13.24 / 13.65 ±0.31 / 14.01 ms │   +1.06x faster │
│ QQuery 21 │           19.53 / 19.74 ±0.15 / 19.98 ms │        18.66 / 19.52 ±0.54 / 20.00 ms │       no change │
│ QQuery 22 │        489.31 / 492.20 ±3.55 / 499.03 ms │     487.44 / 491.42 ±2.51 / 493.80 ms │       no change │
│ QQuery 23 │        862.62 / 869.65 ±7.23 / 881.81 ms │  1095.31 / 1099.97 ±5.72 / 1110.59 ms │    1.26x slower │
│ QQuery 24 │        380.07 / 382.85 ±1.88 / 385.41 ms │     760.53 / 765.41 ±2.96 / 769.71 ms │    2.00x slower │
│ QQuery 25 │        343.02 / 344.48 ±1.58 / 347.39 ms │     272.24 / 274.78 ±1.62 / 277.35 ms │   +1.25x faster │
│ QQuery 26 │           81.95 / 82.92 ±0.78 / 84.01 ms │        81.94 / 83.66 ±1.17 / 85.57 ms │       no change │
│ QQuery 27 │              7.01 / 7.21 ±0.19 / 7.50 ms │           7.62 / 7.74 ±0.10 / 7.91 ms │    1.07x slower │
│ QQuery 28 │        150.34 / 152.18 ±1.36 / 154.28 ms │     148.98 / 151.01 ±1.34 / 153.05 ms │       no change │
│ QQuery 29 │        281.02 / 284.06 ±2.35 / 287.26 ms │     257.39 / 261.03 ±2.19 / 263.05 ms │   +1.09x faster │
│ QQuery 30 │           45.14 / 46.34 ±0.96 / 47.83 ms │        38.43 / 40.20 ±1.03 / 41.25 ms │   +1.15x faster │
│ QQuery 31 │        169.93 / 173.23 ±2.35 / 176.59 ms │     417.55 / 420.68 ±2.22 / 423.26 ms │    2.43x slower │
│ QQuery 32 │           56.96 / 57.91 ±0.65 / 58.53 ms │        13.52 / 15.13 ±1.10 / 16.76 ms │   +3.83x faster │
│ QQuery 33 │        141.18 / 142.99 ±1.30 / 144.71 ms │     139.25 / 141.10 ±1.95 / 144.83 ms │       no change │
│ QQuery 34 │              7.09 / 7.34 ±0.23 / 7.75 ms │        14.37 / 15.54 ±1.21 / 17.75 ms │    2.12x slower │
│ QQuery 35 │        104.96 / 106.69 ±1.42 / 108.62 ms │     125.05 / 126.72 ±1.17 / 127.94 ms │    1.19x slower │
│ QQuery 36 │              6.65 / 6.84 ±0.13 / 7.05 ms │           6.86 / 7.15 ±0.35 / 7.63 ms │       no change │
│ QQuery 37 │              8.23 / 8.71 ±0.34 / 9.02 ms │           8.70 / 8.80 ±0.09 / 8.94 ms │       no change │
│ QQuery 38 │           81.49 / 87.67 ±5.25 / 97.40 ms │     138.82 / 140.11 ±1.22 / 142.41 ms │    1.60x slower │
│ QQuery 39 │        122.22 / 127.29 ±4.33 / 134.70 ms │     124.93 / 127.84 ±1.59 / 129.72 ms │       no change │
│ QQuery 40 │        108.32 / 114.06 ±7.50 / 128.84 ms │        20.67 / 21.41 ±0.96 / 23.19 ms │   +5.33x faster │
│ QQuery 41 │           14.28 / 14.87 ±0.49 / 15.55 ms │        14.17 / 14.73 ±0.39 / 15.27 ms │       no change │
│ QQuery 42 │        107.36 / 108.17 ±0.92 / 109.58 ms │     108.62 / 109.30 ±0.55 / 110.15 ms │       no change │
│ QQuery 43 │              5.90 / 6.09 ±0.24 / 6.56 ms │           5.89 / 6.14 ±0.20 / 6.51 ms │       no change │
│ QQuery 44 │           11.75 / 12.61 ±0.97 / 14.49 ms │        11.94 / 12.16 ±0.16 / 12.38 ms │       no change │
│ QQuery 45 │           50.00 / 51.36 ±0.92 / 52.35 ms │        47.62 / 48.24 ±0.48 / 48.89 ms │   +1.06x faster │
│ QQuery 46 │              8.37 / 8.89 ±0.39 / 9.55 ms │        15.85 / 16.97 ±0.75 / 18.06 ms │    1.91x slower │
│ QQuery 47 │        695.34 / 700.36 ±2.63 / 702.37 ms │     687.30 / 697.79 ±7.87 / 710.94 ms │       no change │
│ QQuery 48 │        289.50 / 294.25 ±4.25 / 300.56 ms │     280.29 / 282.10 ±1.37 / 284.08 ms │       no change │
│ QQuery 49 │        251.20 / 252.96 ±1.21 / 254.15 ms │     253.01 / 254.28 ±1.31 / 256.58 ms │       no change │
│ QQuery 50 │        217.49 / 226.82 ±5.32 / 234.00 ms │     186.11 / 187.67 ±1.34 / 189.91 ms │   +1.21x faster │
│ QQuery 51 │        179.36 / 184.39 ±3.35 / 188.84 ms │     177.58 / 180.84 ±1.71 / 182.31 ms │       no change │
│ QQuery 52 │        107.35 / 108.70 ±1.41 / 111.20 ms │     107.94 / 109.48 ±0.98 / 110.81 ms │       no change │
│ QQuery 53 │        101.97 / 103.49 ±1.25 / 105.62 ms │     102.94 / 104.13 ±1.57 / 107.22 ms │       no change │
│ QQuery 54 │        146.16 / 147.49 ±1.10 / 149.13 ms │     186.24 / 189.29 ±2.37 / 192.64 ms │    1.28x slower │
│ QQuery 55 │        107.10 / 108.25 ±1.18 / 110.02 ms │     106.20 / 107.82 ±1.44 / 109.93 ms │       no change │
│ QQuery 56 │        142.70 / 143.73 ±0.89 / 144.89 ms │     138.88 / 140.45 ±0.99 / 141.90 ms │       no change │
│ QQuery 57 │        171.71 / 173.94 ±1.27 / 175.16 ms │     172.57 / 175.71 ±2.21 / 178.76 ms │       no change │
│ QQuery 58 │       291.83 / 309.22 ±10.00 / 319.69 ms │     315.40 / 319.28 ±2.14 / 321.32 ms │       no change │
│ QQuery 59 │        199.77 / 202.19 ±1.39 / 203.96 ms │     370.12 / 375.21 ±3.28 / 380.45 ms │    1.86x slower │
│ QQuery 60 │        143.66 / 145.11 ±1.15 / 146.88 ms │     139.74 / 142.28 ±1.35 / 143.31 ms │       no change │
│ QQuery 61 │           13.16 / 13.30 ±0.17 / 13.62 ms │        18.23 / 18.67 ±0.33 / 19.15 ms │    1.40x slower │
│ QQuery 62 │       883.20 / 929.75 ±31.03 / 964.83 ms │        62.89 / 64.61 ±1.09 / 66.01 ms │  +14.39x faster │
│ QQuery 63 │        102.53 / 104.76 ±1.33 / 106.70 ms │     102.37 / 104.10 ±1.33 / 106.21 ms │       no change │
│ QQuery 64 │        678.42 / 684.88 ±6.76 / 696.80 ms │  1258.28 / 1270.11 ±7.32 / 1280.68 ms │    1.85x slower │
│ QQuery 65 │        248.68 / 252.32 ±2.84 / 256.99 ms │     260.69 / 264.25 ±2.09 / 266.45 ms │       no change │
│ QQuery 66 │       229.76 / 250.76 ±14.38 / 269.74 ms │     104.53 / 107.31 ±1.84 / 110.23 ms │   +2.34x faster │
│ QQuery 67 │        310.32 / 321.01 ±9.41 / 335.32 ms │    622.25 / 635.02 ±10.83 / 647.42 ms │    1.98x slower │
│ QQuery 68 │           10.25 / 11.01 ±0.76 / 12.42 ms │        14.91 / 15.90 ±0.91 / 17.36 ms │    1.44x slower │
│ QQuery 69 │        102.17 / 103.25 ±0.90 / 104.67 ms │     126.80 / 130.01 ±2.24 / 132.77 ms │    1.26x slower │
│ QQuery 70 │       332.82 / 345.49 ±11.89 / 367.83 ms │     321.06 / 325.87 ±3.79 / 331.92 ms │   +1.06x faster │
│ QQuery 71 │        134.96 / 137.21 ±2.59 / 142.01 ms │     133.54 / 134.51 ±0.70 / 135.66 ms │       no change │
│ QQuery 72 │       603.40 / 623.23 ±10.95 / 636.28 ms │     605.77 / 610.61 ±3.57 / 615.22 ms │       no change │
│ QQuery 73 │             7.13 / 8.82 ±1.02 / 10.03 ms │        11.73 / 13.10 ±1.09 / 14.61 ms │    1.48x slower │
│ QQuery 74 │        549.25 / 559.30 ±5.56 / 564.85 ms │     598.53 / 606.42 ±5.31 / 613.15 ms │    1.08x slower │
│ QQuery 75 │        275.67 / 278.06 ±1.50 / 279.34 ms │     436.14 / 439.78 ±2.46 / 443.34 ms │    1.58x slower │
│ QQuery 76 │        132.29 / 133.19 ±0.93 / 134.55 ms │     130.13 / 132.88 ±1.63 / 135.18 ms │       no change │
│ QQuery 77 │        188.20 / 191.59 ±2.86 / 195.42 ms │     189.44 / 191.77 ±1.96 / 194.33 ms │       no change │
│ QQuery 78 │        341.92 / 344.48 ±2.08 / 347.83 ms │     367.35 / 372.64 ±4.27 / 380.43 ms │    1.08x slower │
│ QQuery 79 │        233.80 / 236.21 ±1.54 / 238.47 ms │     231.16 / 234.13 ±3.05 / 238.62 ms │       no change │
│ QQuery 80 │        321.75 / 327.45 ±4.08 / 333.79 ms │     322.21 / 323.44 ±1.09 / 325.48 ms │       no change │
│ QQuery 81 │           25.69 / 26.97 ±0.82 / 28.28 ms │        29.99 / 31.67 ±1.19 / 33.62 ms │    1.17x slower │
│ QQuery 82 │        197.95 / 200.19 ±1.85 / 203.39 ms │        40.39 / 41.76 ±1.52 / 43.92 ms │   +4.79x faster │
│ QQuery 83 │           37.77 / 39.72 ±1.64 / 41.98 ms │        46.77 / 48.35 ±1.47 / 50.44 ms │    1.22x slower │
│ QQuery 84 │           47.88 / 49.24 ±1.00 / 50.78 ms │        41.09 / 41.40 ±0.23 / 41.71 ms │   +1.19x faster │
│ QQuery 85 │        147.72 / 149.64 ±1.15 / 151.21 ms │     150.00 / 151.38 ±0.96 / 152.94 ms │       no change │
│ QQuery 86 │           38.86 / 40.70 ±1.58 / 43.35 ms │        40.50 / 41.25 ±0.83 / 42.83 ms │       no change │
│ QQuery 87 │           85.81 / 88.00 ±2.03 / 91.24 ms │     137.19 / 141.11 ±3.01 / 145.03 ms │    1.60x slower │
│ QQuery 88 │         99.64 / 104.09 ±7.92 / 119.92 ms │      99.68 / 101.13 ±1.02 / 102.43 ms │       no change │
│ QQuery 89 │        119.96 / 121.53 ±1.17 / 123.54 ms │     117.63 / 118.86 ±1.28 / 121.19 ms │       no change │
│ QQuery 90 │           23.84 / 24.55 ±0.88 / 26.23 ms │        22.00 / 22.73 ±0.37 / 23.01 ms │   +1.08x faster │
│ QQuery 91 │           61.94 / 64.54 ±1.75 / 66.50 ms │        54.49 / 55.46 ±0.68 / 56.54 ms │   +1.16x faster │
│ QQuery 92 │           57.76 / 58.78 ±0.80 / 60.01 ms │        56.75 / 57.31 ±0.70 / 58.67 ms │       no change │
│ QQuery 93 │        190.17 / 191.30 ±0.72 / 192.38 ms │     184.68 / 188.11 ±2.06 / 190.10 ms │       no change │
│ QQuery 94 │           61.72 / 62.53 ±0.64 / 63.39 ms │        66.56 / 68.17 ±1.11 / 70.02 ms │    1.09x slower │
│ QQuery 95 │        128.73 / 129.93 ±1.53 / 132.90 ms │     125.90 / 127.24 ±1.06 / 128.99 ms │       no change │
│ QQuery 96 │           74.06 / 75.58 ±1.47 / 78.35 ms │        67.72 / 68.49 ±0.49 / 69.25 ms │   +1.10x faster │
│ QQuery 97 │        127.89 / 129.72 ±1.46 / 132.17 ms │     113.86 / 115.32 ±1.19 / 117.32 ms │   +1.12x faster │
│ QQuery 98 │        154.65 / 157.85 ±1.78 / 160.10 ms │     152.74 / 154.52 ±1.62 / 157.44 ms │       no change │
│ QQuery 99 │ 10692.48 / 10745.04 ±30.28 / 10770.89 ms │        61.16 / 61.87 ±0.44 / 62.36 ms │ +173.68x faster │
└───────────┴──────────────────────────────────────────┴───────────────────────────────────────┴─────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 31357.47ms │
│ Total Time (fix-join-cardinality-estimation)   │ 21793.68ms │
│ Average Time (HEAD)                            │   316.74ms │
│ Average Time (fix-join-cardinality-estimation) │   220.14ms │
│ Queries Faster                                 │         20 │
│ Queries Slower                                 │         29 │
│ Queries with No Change                         │         50 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	157.2s
Peak memory	5.6 GiB
Avg memory	4.7 GiB
CPU user	259.2s
CPU sys	17.2s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	109.2s
Peak memory	4.7 GiB
Avg memory	4.1 GiB
CPU user	198.2s
CPU sys	14.6s
Peak spill	0 B

File an issue against this benchmark runner

Dandandan · 2026-04-09T11:00:27Z

run benchmarks

adriangbot · 2026-04-09T11:00:54Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213621310-1002-tnhvd 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (322e5f6) to 5ba06ac (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:02:29Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213621310-1003-gpvjw 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (322e5f6) to 5ba06ac (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:03:10Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213621310-1004-mdrz7 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (322e5f6) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

Dandandan · 2026-04-09T11:13:20Z

run benchmarks

adriangbot · 2026-04-09T11:14:06Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213689280-1005-qdwhn 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (03037cd) to 5ba06ac (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:15:18Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213689280-1006-9vhdl 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (03037cd) to 5ba06ac (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:16:03Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213689280-1007-bptg9 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (03037cd) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:17:21Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃          fix-join-cardinality-estimation ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.60 / 7.05 ±0.78 / 8.60 ms │              6.61 / 7.10 ±0.85 / 8.80 ms │    no change │
│ QQuery 2  │        144.05 / 145.20 ±0.75 / 146.40 ms │        144.79 / 145.53 ±0.63 / 146.55 ms │    no change │
│ QQuery 3  │        113.70 / 114.73 ±0.88 / 116.14 ms │        114.20 / 114.72 ±0.26 / 114.92 ms │    no change │
│ QQuery 4  │    1331.50 / 1358.15 ±16.29 / 1373.85 ms │     1349.00 / 1362.37 ±8.09 / 1373.25 ms │    no change │
│ QQuery 5  │        171.59 / 173.66 ±1.74 / 176.28 ms │        171.80 / 174.73 ±2.25 / 177.58 ms │    no change │
│ QQuery 6  │       848.60 / 871.52 ±12.47 / 883.74 ms │       820.62 / 843.50 ±23.66 / 879.95 ms │    no change │
│ QQuery 7  │        340.52 / 343.57 ±3.01 / 348.81 ms │        341.66 / 344.44 ±1.63 / 346.53 ms │    no change │
│ QQuery 8  │        114.39 / 116.49 ±1.17 / 117.87 ms │        116.01 / 117.38 ±1.08 / 119.26 ms │    no change │
│ QQuery 9  │       100.70 / 108.92 ±10.51 / 129.25 ms │        106.50 / 111.81 ±8.25 / 127.91 ms │    no change │
│ QQuery 10 │        104.76 / 106.25 ±1.13 / 107.61 ms │        106.66 / 107.42 ±0.65 / 108.41 ms │    no change │
│ QQuery 11 │        946.15 / 952.17 ±5.64 / 960.43 ms │       914.55 / 943.61 ±15.87 / 958.24 ms │    no change │
│ QQuery 12 │           45.43 / 46.49 ±0.92 / 47.95 ms │           44.66 / 46.02 ±1.35 / 48.40 ms │    no change │
│ QQuery 13 │        396.99 / 403.54 ±4.36 / 409.01 ms │        397.58 / 400.47 ±1.88 / 402.42 ms │    no change │
│ QQuery 14 │      995.85 / 1005.94 ±8.74 / 1021.05 ms │      998.10 / 1005.09 ±5.99 / 1014.49 ms │    no change │
│ QQuery 15 │           15.49 / 16.70 ±0.98 / 17.95 ms │           15.45 / 16.14 ±0.59 / 16.88 ms │    no change │
│ QQuery 16 │              7.10 / 7.74 ±0.78 / 9.26 ms │              7.35 / 8.19 ±0.79 / 9.68 ms │ 1.06x slower │
│ QQuery 17 │        224.47 / 227.12 ±2.08 / 229.12 ms │        227.72 / 228.44 ±0.53 / 229.15 ms │    no change │
│ QQuery 18 │        126.02 / 126.93 ±0.72 / 128.10 ms │        126.06 / 127.10 ±0.74 / 128.28 ms │    no change │
│ QQuery 19 │        154.73 / 156.68 ±1.01 / 157.65 ms │        157.26 / 158.42 ±0.69 / 159.37 ms │    no change │
│ QQuery 20 │           13.62 / 13.96 ±0.23 / 14.35 ms │           12.99 / 13.98 ±0.64 / 14.77 ms │    no change │
│ QQuery 21 │           19.35 / 19.84 ±0.59 / 20.84 ms │           19.30 / 19.95 ±0.52 / 20.57 ms │    no change │
│ QQuery 22 │        486.48 / 490.79 ±4.07 / 498.31 ms │        485.43 / 494.58 ±6.54 / 504.35 ms │    no change │
│ QQuery 23 │        863.65 / 871.36 ±4.87 / 878.89 ms │        875.58 / 886.14 ±6.22 / 892.33 ms │    no change │
│ QQuery 24 │        378.58 / 383.51 ±2.83 / 386.14 ms │        381.83 / 385.44 ±2.75 / 390.28 ms │    no change │
│ QQuery 25 │        337.68 / 339.66 ±1.86 / 343.12 ms │        340.24 / 342.82 ±2.03 / 345.63 ms │    no change │
│ QQuery 26 │           81.34 / 82.20 ±0.66 / 83.30 ms │           80.58 / 82.31 ±1.49 / 84.95 ms │    no change │
│ QQuery 27 │              6.94 / 7.11 ±0.15 / 7.38 ms │              6.93 / 7.43 ±0.69 / 8.78 ms │    no change │
│ QQuery 28 │        148.97 / 150.42 ±1.16 / 152.40 ms │        148.31 / 149.22 ±0.55 / 149.85 ms │    no change │
│ QQuery 29 │        277.47 / 280.39 ±2.27 / 283.65 ms │        279.74 / 281.62 ±2.37 / 286.11 ms │    no change │
│ QQuery 30 │           43.29 / 46.18 ±2.11 / 49.79 ms │           43.49 / 45.08 ±1.31 / 47.09 ms │    no change │
│ QQuery 31 │        169.27 / 172.50 ±1.87 / 175.00 ms │        172.49 / 173.60 ±1.43 / 176.21 ms │    no change │
│ QQuery 32 │           56.95 / 57.67 ±0.48 / 58.19 ms │           57.44 / 58.26 ±0.51 / 58.96 ms │    no change │
│ QQuery 33 │        139.59 / 140.82 ±0.97 / 141.98 ms │        140.66 / 142.28 ±1.53 / 144.71 ms │    no change │
│ QQuery 34 │              7.03 / 7.19 ±0.21 / 7.60 ms │              6.95 / 7.15 ±0.28 / 7.69 ms │    no change │
│ QQuery 35 │        106.64 / 107.33 ±1.13 / 109.58 ms │        108.17 / 109.86 ±1.32 / 111.79 ms │    no change │
│ QQuery 36 │              6.53 / 6.65 ±0.14 / 6.91 ms │              6.47 / 6.65 ±0.18 / 6.99 ms │    no change │
│ QQuery 37 │              8.42 / 8.83 ±0.31 / 9.33 ms │             8.44 / 9.23 ±0.75 / 10.47 ms │    no change │
│ QQuery 38 │           82.56 / 86.80 ±3.07 / 91.34 ms │           84.79 / 88.09 ±4.64 / 97.28 ms │    no change │
│ QQuery 39 │        125.78 / 127.40 ±1.28 / 129.63 ms │        122.68 / 125.37 ±1.86 / 127.36 ms │    no change │
│ QQuery 40 │        109.95 / 113.83 ±4.89 / 123.46 ms │        108.16 / 114.81 ±7.32 / 128.88 ms │    no change │
│ QQuery 41 │           14.61 / 15.29 ±0.55 / 15.79 ms │           14.11 / 15.50 ±1.16 / 17.17 ms │    no change │
│ QQuery 42 │        108.05 / 109.63 ±1.36 / 111.53 ms │        110.76 / 111.57 ±0.74 / 112.54 ms │    no change │
│ QQuery 43 │              6.13 / 6.21 ±0.15 / 6.51 ms │              6.05 / 6.47 ±0.60 / 7.63 ms │    no change │
│ QQuery 44 │           11.85 / 12.15 ±0.28 / 12.67 ms │           11.62 / 11.88 ±0.18 / 12.11 ms │    no change │
│ QQuery 45 │           50.89 / 51.54 ±0.64 / 52.69 ms │           50.38 / 52.34 ±1.25 / 54.25 ms │    no change │
│ QQuery 46 │              8.36 / 8.78 ±0.24 / 9.07 ms │              8.65 / 8.92 ±0.29 / 9.44 ms │    no change │
│ QQuery 47 │        700.52 / 707.91 ±7.04 / 718.57 ms │        712.22 / 716.78 ±5.30 / 726.67 ms │    no change │
│ QQuery 48 │        288.35 / 290.85 ±2.44 / 295.43 ms │        292.89 / 297.55 ±3.45 / 303.59 ms │    no change │
│ QQuery 49 │        252.27 / 255.29 ±2.46 / 257.55 ms │        255.61 / 257.68 ±2.15 / 261.66 ms │    no change │
│ QQuery 50 │        223.40 / 225.82 ±1.64 / 228.01 ms │        222.38 / 225.75 ±2.76 / 230.49 ms │    no change │
│ QQuery 51 │        178.62 / 181.30 ±2.85 / 186.83 ms │        182.31 / 185.49 ±2.58 / 190.02 ms │    no change │
│ QQuery 52 │        106.48 / 107.49 ±0.58 / 108.22 ms │        109.20 / 111.32 ±2.11 / 115.37 ms │    no change │
│ QQuery 53 │        102.03 / 102.95 ±0.84 / 104.49 ms │        102.33 / 104.03 ±1.08 / 105.13 ms │    no change │
│ QQuery 54 │        144.81 / 146.42 ±1.47 / 149.16 ms │        145.37 / 147.54 ±1.77 / 150.29 ms │    no change │
│ QQuery 55 │        106.74 / 107.38 ±0.72 / 108.72 ms │        107.22 / 109.09 ±1.83 / 111.38 ms │    no change │
│ QQuery 56 │        140.35 / 142.04 ±2.56 / 147.13 ms │        142.18 / 143.37 ±1.23 / 145.03 ms │    no change │
│ QQuery 57 │        172.89 / 174.52 ±0.85 / 175.26 ms │        172.34 / 173.52 ±1.23 / 175.21 ms │    no change │
│ QQuery 58 │        293.52 / 301.67 ±4.12 / 304.39 ms │        287.49 / 298.27 ±8.13 / 309.99 ms │    no change │
│ QQuery 59 │        195.93 / 198.75 ±2.12 / 201.12 ms │        196.28 / 197.33 ±0.65 / 198.04 ms │    no change │
│ QQuery 60 │        143.31 / 144.24 ±0.82 / 145.62 ms │        143.37 / 146.08 ±1.95 / 149.09 ms │    no change │
│ QQuery 61 │           13.04 / 13.22 ±0.22 / 13.64 ms │           13.13 / 13.54 ±0.33 / 14.06 ms │    no change │
│ QQuery 62 │      886.37 / 937.71 ±42.77 / 1008.52 ms │       889.90 / 925.67 ±28.11 / 952.60 ms │    no change │
│ QQuery 63 │        102.76 / 105.14 ±2.12 / 108.93 ms │        102.24 / 104.49 ±1.41 / 105.95 ms │    no change │
│ QQuery 64 │        681.67 / 682.91 ±1.60 / 686.02 ms │        682.42 / 687.11 ±4.08 / 692.77 ms │    no change │
│ QQuery 65 │        252.81 / 255.12 ±3.46 / 261.95 ms │        248.13 / 251.55 ±2.92 / 256.56 ms │    no change │
│ QQuery 66 │        244.91 / 256.15 ±9.36 / 269.82 ms │        240.75 / 252.72 ±8.79 / 263.39 ms │    no change │
│ QQuery 67 │        308.79 / 312.65 ±5.86 / 324.15 ms │       303.96 / 316.07 ±12.04 / 337.92 ms │    no change │
│ QQuery 68 │             8.40 / 9.97 ±0.98 / 11.03 ms │            9.43 / 11.20 ±1.56 / 13.96 ms │ 1.12x slower │
│ QQuery 69 │        101.03 / 102.65 ±1.02 / 103.87 ms │        102.67 / 104.55 ±3.32 / 111.18 ms │    no change │
│ QQuery 70 │        332.99 / 341.35 ±6.51 / 353.01 ms │       334.34 / 347.91 ±10.33 / 365.78 ms │    no change │
│ QQuery 71 │        133.83 / 135.73 ±1.86 / 138.26 ms │        135.82 / 138.02 ±2.37 / 141.76 ms │    no change │
│ QQuery 72 │        606.76 / 617.79 ±7.40 / 625.46 ms │        599.63 / 610.61 ±6.17 / 617.88 ms │    no change │
│ QQuery 73 │              7.29 / 8.47 ±0.79 / 9.28 ms │              7.65 / 8.47 ±0.77 / 9.77 ms │    no change │
│ QQuery 74 │        578.23 / 590.07 ±6.21 / 596.52 ms │       571.96 / 583.05 ±11.44 / 604.47 ms │    no change │
│ QQuery 75 │        273.85 / 277.74 ±2.06 / 280.00 ms │        275.45 / 277.74 ±1.36 / 279.61 ms │    no change │
│ QQuery 76 │        131.57 / 132.71 ±0.91 / 134.32 ms │        131.26 / 132.15 ±0.89 / 133.85 ms │    no change │
│ QQuery 77 │        188.02 / 189.38 ±0.89 / 190.69 ms │        188.75 / 190.94 ±1.90 / 193.98 ms │    no change │
│ QQuery 78 │        334.84 / 341.51 ±4.08 / 346.39 ms │        339.89 / 341.93 ±1.93 / 345.53 ms │    no change │
│ QQuery 79 │        230.23 / 232.38 ±1.69 / 234.94 ms │        233.89 / 236.41 ±1.84 / 239.43 ms │    no change │
│ QQuery 80 │        322.91 / 323.87 ±1.27 / 326.34 ms │        322.42 / 325.64 ±2.16 / 328.86 ms │    no change │
│ QQuery 81 │           26.58 / 27.71 ±1.99 / 31.67 ms │           26.35 / 27.34 ±0.68 / 28.29 ms │    no change │
│ QQuery 82 │        194.36 / 197.89 ±2.70 / 202.30 ms │        194.11 / 196.72 ±1.43 / 197.86 ms │    no change │
│ QQuery 83 │           38.20 / 38.79 ±0.43 / 39.54 ms │           38.42 / 39.59 ±0.95 / 41.13 ms │    no change │
│ QQuery 84 │           48.07 / 48.63 ±0.51 / 49.28 ms │           48.25 / 48.99 ±0.46 / 49.59 ms │    no change │
│ QQuery 85 │        145.56 / 148.87 ±2.34 / 152.19 ms │        147.51 / 148.94 ±1.15 / 150.42 ms │    no change │
│ QQuery 86 │           38.67 / 40.23 ±1.01 / 41.77 ms │           39.26 / 40.08 ±0.53 / 40.76 ms │    no change │
│ QQuery 87 │           84.27 / 87.47 ±3.67 / 94.60 ms │           84.76 / 87.96 ±2.90 / 93.37 ms │    no change │
│ QQuery 88 │         99.33 / 100.19 ±0.61 / 101.00 ms │        100.04 / 101.82 ±0.93 / 102.71 ms │    no change │
│ QQuery 89 │        117.32 / 118.65 ±0.94 / 120.08 ms │        116.67 / 118.55 ±1.32 / 120.38 ms │    no change │
│ QQuery 90 │           24.12 / 24.40 ±0.20 / 24.72 ms │           23.10 / 23.83 ±0.57 / 24.80 ms │    no change │
│ QQuery 91 │           62.47 / 63.40 ±0.71 / 64.32 ms │           62.03 / 62.94 ±0.94 / 64.48 ms │    no change │
│ QQuery 92 │           57.14 / 57.66 ±0.42 / 58.30 ms │           57.78 / 58.26 ±0.52 / 59.26 ms │    no change │
│ QQuery 93 │        186.01 / 187.76 ±1.26 / 189.13 ms │        184.91 / 186.57 ±1.46 / 188.93 ms │    no change │
│ QQuery 94 │           60.73 / 61.87 ±0.98 / 63.17 ms │           61.30 / 61.79 ±0.40 / 62.41 ms │    no change │
│ QQuery 95 │        126.68 / 128.09 ±0.82 / 128.96 ms │        126.35 / 128.25 ±1.49 / 130.10 ms │    no change │
│ QQuery 96 │           70.13 / 72.27 ±1.56 / 73.83 ms │           72.03 / 74.42 ±1.20 / 75.25 ms │    no change │
│ QQuery 97 │        126.31 / 126.66 ±0.32 / 127.16 ms │        124.00 / 126.47 ±1.49 / 128.22 ms │    no change │
│ QQuery 98 │        152.44 / 153.78 ±0.98 / 155.19 ms │        151.64 / 155.18 ±2.93 / 158.70 ms │    no change │
│ QQuery 99 │ 10713.93 / 10755.69 ±38.55 / 10826.41 ms │ 10716.65 / 10759.95 ±27.98 / 10795.11 ms │    no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 31400.03ms │
│ Total Time (fix-join-cardinality-estimation)   │ 31434.32ms │
│ Average Time (HEAD)                            │   317.17ms │
│ Average Time (fix-join-cardinality-estimation) │   317.52ms │
│ Queries Faster                                 │          0 │
│ Queries Slower                                 │          2 │
│ Queries with No Change                         │         97 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	157.3s
Peak memory	5.8 GiB
Avg memory	4.8 GiB
CPU user	259.7s
CPU sys	16.3s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	157.4s
Peak memory	5.5 GiB
Avg memory	4.5 GiB
CPU user	260.0s
CPU sys	16.6s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:17:44Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.19 / 4.36 ±6.28 / 16.92 ms │          1.22 / 4.55 ±6.49 / 17.52 ms │     no change │
│ QQuery 1  │        14.18 / 14.37 ±0.10 / 14.46 ms │        14.24 / 14.51 ±0.14 / 14.61 ms │     no change │
│ QQuery 2  │        43.53 / 43.93 ±0.30 / 44.41 ms │        43.56 / 43.88 ±0.28 / 44.38 ms │     no change │
│ QQuery 3  │        42.90 / 44.52 ±1.32 / 46.42 ms │        42.30 / 43.62 ±1.22 / 45.19 ms │     no change │
│ QQuery 4  │     279.55 / 282.89 ±4.16 / 290.94 ms │     287.82 / 293.67 ±5.13 / 299.88 ms │     no change │
│ QQuery 5  │     334.14 / 339.17 ±4.14 / 345.76 ms │     343.02 / 345.30 ±2.49 / 349.78 ms │     no change │
│ QQuery 6  │          5.46 / 8.13 ±2.54 / 12.60 ms │           5.54 / 6.49 ±0.81 / 7.65 ms │ +1.25x faster │
│ QQuery 7  │        16.52 / 16.81 ±0.21 / 17.18 ms │        16.52 / 17.45 ±0.94 / 19.25 ms │     no change │
│ QQuery 8  │     400.60 / 413.31 ±8.98 / 426.14 ms │    410.52 / 426.98 ±16.51 / 458.23 ms │     no change │
│ QQuery 9  │     631.59 / 632.97 ±1.86 / 636.57 ms │     635.19 / 642.52 ±6.30 / 650.73 ms │     no change │
│ QQuery 10 │        89.67 / 92.57 ±3.64 / 98.82 ms │        89.86 / 93.37 ±2.77 / 97.34 ms │     no change │
│ QQuery 11 │     102.92 / 105.67 ±1.77 / 108.43 ms │     102.03 / 103.09 ±1.00 / 104.78 ms │     no change │
│ QQuery 12 │     341.89 / 345.59 ±4.22 / 352.43 ms │     334.20 / 341.20 ±3.83 / 344.82 ms │     no change │
│ QQuery 13 │     436.31 / 453.57 ±9.14 / 460.91 ms │    443.18 / 463.96 ±21.33 / 490.68 ms │     no change │
│ QQuery 14 │     335.64 / 338.13 ±2.45 / 342.44 ms │     338.33 / 343.16 ±4.46 / 349.40 ms │     no change │
│ QQuery 15 │    340.93 / 353.41 ±16.26 / 385.18 ms │    348.17 / 370.87 ±23.93 / 417.32 ms │     no change │
│ QQuery 16 │    707.31 / 730.50 ±17.56 / 748.57 ms │    700.01 / 715.62 ±16.84 / 745.89 ms │     no change │
│ QQuery 17 │     685.89 / 694.68 ±6.63 / 701.34 ms │     698.09 / 706.24 ±8.03 / 719.86 ms │     no change │
│ QQuery 18 │ 1360.38 / 1434.44 ±51.50 / 1503.65 ms │ 1369.76 / 1453.97 ±43.98 / 1490.49 ms │     no change │
│ QQuery 19 │       38.68 / 67.78 ±16.02 / 79.77 ms │        34.80 / 37.29 ±2.68 / 41.89 ms │ +1.82x faster │
│ QQuery 20 │    725.56 / 762.94 ±33.18 / 803.86 ms │    712.11 / 729.23 ±23.14 / 775.04 ms │     no change │
│ QQuery 21 │     764.28 / 769.27 ±3.42 / 772.70 ms │     760.88 / 765.29 ±4.47 / 772.64 ms │     no change │
│ QQuery 22 │  1137.16 / 1146.26 ±5.61 / 1152.32 ms │  1137.25 / 1138.18 ±0.73 / 1139.15 ms │     no change │
│ QQuery 23 │ 3070.82 / 3101.50 ±18.38 / 3127.26 ms │ 3062.27 / 3084.12 ±18.87 / 3113.20 ms │     no change │
│ QQuery 24 │     100.62 / 102.65 ±2.09 / 105.20 ms │     101.37 / 103.39 ±1.69 / 105.99 ms │     no change │
│ QQuery 25 │     139.60 / 140.08 ±0.41 / 140.83 ms │     137.16 / 139.03 ±1.36 / 141.26 ms │     no change │
│ QQuery 26 │      97.46 / 100.63 ±1.64 / 102.06 ms │      96.33 / 101.59 ±2.87 / 105.10 ms │     no change │
│ QQuery 27 │     851.85 / 854.95 ±4.74 / 864.32 ms │     846.27 / 854.76 ±7.34 / 863.61 ms │     no change │
│ QQuery 28 │ 7698.73 / 7756.81 ±31.82 / 7787.53 ms │ 7721.12 / 7752.20 ±25.78 / 7780.32 ms │     no change │
│ QQuery 29 │       50.20 / 61.56 ±13.82 / 87.15 ms │        49.53 / 54.50 ±3.55 / 60.10 ms │ +1.13x faster │
│ QQuery 30 │     357.83 / 361.82 ±4.38 / 370.26 ms │     358.92 / 368.81 ±5.64 / 374.77 ms │     no change │
│ QQuery 31 │    358.53 / 382.46 ±16.65 / 407.94 ms │     369.15 / 375.42 ±6.71 / 386.12 ms │     no change │
│ QQuery 32 │ 1228.32 / 1291.64 ±43.29 / 1351.76 ms │ 1028.78 / 1052.43 ±25.50 / 1098.37 ms │ +1.23x faster │
│ QQuery 33 │ 1487.90 / 1546.58 ±38.76 / 1590.50 ms │  1452.45 / 1461.80 ±9.75 / 1479.92 ms │ +1.06x faster │
│ QQuery 34 │ 1526.16 / 1536.11 ±13.20 / 1562.00 ms │ 1446.74 / 1479.21 ±24.12 / 1504.18 ms │     no change │
│ QQuery 35 │    429.44 / 464.77 ±46.04 / 553.52 ms │     375.72 / 384.40 ±6.02 / 392.59 ms │ +1.21x faster │
│ QQuery 36 │     121.72 / 123.88 ±1.67 / 125.86 ms │     114.73 / 119.69 ±2.91 / 123.48 ms │     no change │
│ QQuery 37 │        50.33 / 51.69 ±1.11 / 53.60 ms │        48.16 / 49.79 ±1.26 / 52.04 ms │     no change │
│ QQuery 38 │        80.00 / 83.30 ±2.68 / 87.89 ms │        75.07 / 75.88 ±0.77 / 77.03 ms │ +1.10x faster │
│ QQuery 39 │    218.61 / 240.20 ±11.43 / 251.92 ms │     198.56 / 212.14 ±7.77 / 221.30 ms │ +1.13x faster │
│ QQuery 40 │        22.64 / 27.55 ±2.92 / 30.58 ms │        22.94 / 24.93 ±1.04 / 25.80 ms │ +1.10x faster │
│ QQuery 41 │       20.45 / 29.89 ±15.77 / 61.27 ms │        18.89 / 20.99 ±1.61 / 23.48 ms │ +1.42x faster │
│ QQuery 42 │        20.29 / 20.63 ±0.31 / 21.10 ms │        18.74 / 19.33 ±0.37 / 19.76 ms │ +1.07x faster │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 27373.97ms │
│ Total Time (fix-join-cardinality-estimation)   │ 26834.85ms │
│ Average Time (HEAD)                            │   636.60ms │
│ Average Time (fix-join-cardinality-estimation) │   624.07ms │
│ Queries Faster                                 │         11 │
│ Queries Slower                                 │          0 │
│ Queries with No Change                         │         32 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	137.4s
Peak memory	38.3 GiB
Avg memory	28.5 GiB
CPU user	1281.1s
CPU sys	104.9s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	135.2s
Peak memory	39.2 GiB
Avg memory	30.2 GiB
CPU user	1281.0s
CPU sys	88.7s
Peak spill	0 B

File an issue against this benchmark runner

When join keys involve CAST expressions (e.g. CAST(id AS Float64)), the column statistics were lost in ProjectionExec because it only propagated stats for plain Column and Literal expressions. Propagate distinct_count and null_count through numeric CAST expressions in ProjectionExec::project_statistics. The distinct count is made Inexact since casting can reduce (but never increase) the number of distinct values. min/max/sum are cleared since they may not be valid after type conversion. This fixes join cardinality estimation for queries with type mismatches in join keys (common in TPC-DS where fact table FKs are Float64 and dimension PKs are Int32). TPC-DS Q99: 10.4s → ~60ms. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Dandandan · 2026-04-09T11:29:37Z

run benchmarks

adriangbot · 2026-04-09T11:30:49Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃          fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.67 / 7.16 ±0.78 / 8.71 ms │              6.60 / 7.08 ±0.78 / 8.63 ms │     no change │
│ QQuery 2  │        145.04 / 145.99 ±0.56 / 146.74 ms │        146.07 / 146.76 ±0.49 / 147.53 ms │     no change │
│ QQuery 3  │        112.76 / 113.56 ±0.56 / 114.28 ms │        113.27 / 114.06 ±0.63 / 115.15 ms │     no change │
│ QQuery 4  │    1245.05 / 1282.17 ±21.34 / 1303.64 ms │    1269.94 / 1292.47 ±18.81 / 1319.86 ms │     no change │
│ QQuery 5  │        171.40 / 172.00 ±0.52 / 172.64 ms │        172.89 / 174.62 ±1.44 / 177.16 ms │     no change │
│ QQuery 6  │       841.92 / 875.78 ±24.95 / 908.86 ms │       809.82 / 862.74 ±30.84 / 902.32 ms │     no change │
│ QQuery 7  │        337.34 / 341.67 ±3.28 / 346.37 ms │        342.26 / 345.24 ±2.22 / 347.86 ms │     no change │
│ QQuery 8  │        114.59 / 116.05 ±1.01 / 117.62 ms │        117.25 / 117.90 ±0.64 / 118.85 ms │     no change │
│ QQuery 9  │        101.40 / 103.93 ±2.69 / 107.22 ms │        100.54 / 103.39 ±3.02 / 107.74 ms │     no change │
│ QQuery 10 │        106.22 / 107.40 ±0.94 / 108.68 ms │        107.27 / 108.08 ±0.79 / 109.40 ms │     no change │
│ QQuery 11 │        872.44 / 881.61 ±4.97 / 886.07 ms │       862.78 / 875.88 ±11.52 / 894.30 ms │     no change │
│ QQuery 12 │           44.26 / 45.29 ±0.82 / 46.22 ms │           44.49 / 47.26 ±2.06 / 49.74 ms │     no change │
│ QQuery 13 │        398.71 / 400.96 ±2.75 / 406.24 ms │        397.01 / 400.83 ±3.40 / 405.77 ms │     no change │
│ QQuery 14 │      997.15 / 1006.79 ±6.51 / 1014.64 ms │     1009.44 / 1016.54 ±5.65 / 1026.00 ms │     no change │
│ QQuery 15 │           15.65 / 16.58 ±1.04 / 18.41 ms │           15.22 / 16.04 ±0.71 / 17.16 ms │     no change │
│ QQuery 16 │              7.17 / 7.65 ±0.78 / 9.19 ms │              7.04 / 7.79 ±0.99 / 9.73 ms │     no change │
│ QQuery 17 │        227.27 / 229.15 ±1.82 / 232.06 ms │        227.54 / 229.18 ±1.66 / 232.15 ms │     no change │
│ QQuery 18 │        126.68 / 128.83 ±1.35 / 130.48 ms │        126.80 / 127.88 ±0.61 / 128.47 ms │     no change │
│ QQuery 19 │        155.23 / 156.12 ±1.16 / 158.31 ms │        155.30 / 156.66 ±0.98 / 157.84 ms │     no change │
│ QQuery 20 │           13.46 / 14.38 ±0.71 / 15.54 ms │           13.53 / 13.83 ±0.20 / 14.10 ms │     no change │
│ QQuery 21 │           19.11 / 19.92 ±0.86 / 21.34 ms │           19.25 / 20.05 ±0.59 / 21.08 ms │     no change │
│ QQuery 22 │        488.58 / 492.66 ±4.25 / 499.05 ms │        489.25 / 492.12 ±3.30 / 498.34 ms │     no change │
│ QQuery 23 │        869.26 / 876.19 ±5.12 / 883.54 ms │        863.05 / 868.40 ±3.98 / 873.48 ms │     no change │
│ QQuery 24 │        381.85 / 384.39 ±2.01 / 387.29 ms │        380.57 / 385.17 ±2.69 / 388.49 ms │     no change │
│ QQuery 25 │        339.05 / 340.05 ±0.74 / 340.88 ms │        338.21 / 341.60 ±1.96 / 344.13 ms │     no change │
│ QQuery 26 │           81.14 / 82.33 ±1.77 / 85.86 ms │           79.58 / 81.85 ±1.49 / 84.05 ms │     no change │
│ QQuery 27 │              7.08 / 7.21 ±0.14 / 7.46 ms │              6.93 / 7.36 ±0.47 / 8.26 ms │     no change │
│ QQuery 28 │        149.03 / 150.40 ±0.91 / 151.50 ms │        149.62 / 150.68 ±0.78 / 151.67 ms │     no change │
│ QQuery 29 │        281.77 / 283.41 ±1.17 / 284.51 ms │        281.42 / 283.07 ±1.58 / 285.58 ms │     no change │
│ QQuery 30 │           43.18 / 45.21 ±2.54 / 49.84 ms │           43.32 / 44.64 ±1.12 / 46.14 ms │     no change │
│ QQuery 31 │        168.57 / 171.19 ±1.33 / 172.21 ms │        168.45 / 171.89 ±2.10 / 175.07 ms │     no change │
│ QQuery 32 │           56.59 / 57.10 ±0.35 / 57.68 ms │           56.65 / 57.62 ±0.89 / 59.19 ms │     no change │
│ QQuery 33 │        137.66 / 140.65 ±1.77 / 142.83 ms │        139.63 / 141.29 ±1.22 / 143.13 ms │     no change │
│ QQuery 34 │              6.88 / 7.20 ±0.19 / 7.49 ms │              7.10 / 7.31 ±0.27 / 7.84 ms │     no change │
│ QQuery 35 │        106.25 / 107.69 ±1.13 / 109.57 ms │        106.70 / 108.07 ±1.23 / 110.00 ms │     no change │
│ QQuery 36 │              6.50 / 6.74 ±0.24 / 7.06 ms │              6.54 / 6.93 ±0.25 / 7.31 ms │     no change │
│ QQuery 37 │              8.02 / 8.37 ±0.32 / 8.84 ms │              8.41 / 8.89 ±0.46 / 9.73 ms │  1.06x slower │
│ QQuery 38 │           84.06 / 86.77 ±2.77 / 91.85 ms │           85.76 / 88.56 ±2.12 / 92.30 ms │     no change │
│ QQuery 39 │        121.32 / 126.13 ±3.55 / 131.21 ms │        123.45 / 128.17 ±2.53 / 130.51 ms │     no change │
│ QQuery 40 │        115.34 / 117.84 ±4.11 / 126.02 ms │        110.03 / 115.31 ±5.30 / 124.56 ms │     no change │
│ QQuery 41 │           14.67 / 16.36 ±1.81 / 19.64 ms │           15.16 / 15.81 ±0.62 / 16.96 ms │     no change │
│ QQuery 42 │        106.87 / 107.97 ±0.83 / 109.22 ms │        106.45 / 109.03 ±2.37 / 113.26 ms │     no change │
│ QQuery 43 │              5.99 / 6.16 ±0.12 / 6.34 ms │              6.08 / 6.30 ±0.22 / 6.72 ms │     no change │
│ QQuery 44 │           11.75 / 12.32 ±0.77 / 13.83 ms │           11.52 / 11.98 ±0.23 / 12.16 ms │     no change │
│ QQuery 45 │           49.92 / 51.48 ±1.31 / 53.75 ms │           51.19 / 51.42 ±0.16 / 51.66 ms │     no change │
│ QQuery 46 │              8.48 / 8.65 ±0.15 / 8.86 ms │              8.41 / 8.63 ±0.18 / 8.92 ms │     no change │
│ QQuery 47 │        683.03 / 691.62 ±6.90 / 699.11 ms │        685.82 / 699.13 ±7.64 / 709.35 ms │     no change │
│ QQuery 48 │        286.72 / 294.14 ±5.26 / 299.91 ms │        283.75 / 290.61 ±4.44 / 296.98 ms │     no change │
│ QQuery 49 │        251.81 / 253.78 ±1.72 / 255.81 ms │        253.13 / 254.30 ±1.09 / 256.37 ms │     no change │
│ QQuery 50 │        219.96 / 223.89 ±4.14 / 229.97 ms │        217.55 / 223.83 ±3.42 / 227.74 ms │     no change │
│ QQuery 51 │        178.51 / 181.13 ±2.53 / 184.33 ms │        180.59 / 183.23 ±1.38 / 184.50 ms │     no change │
│ QQuery 52 │        105.58 / 106.67 ±0.71 / 107.81 ms │        107.16 / 108.68 ±1.03 / 109.74 ms │     no change │
│ QQuery 53 │        100.69 / 102.01 ±0.85 / 103.02 ms │        101.38 / 103.14 ±1.58 / 106.02 ms │     no change │
│ QQuery 54 │        144.44 / 145.13 ±0.67 / 146.28 ms │        146.06 / 147.54 ±1.46 / 150.00 ms │     no change │
│ QQuery 55 │        106.64 / 108.08 ±0.90 / 109.33 ms │        107.35 / 109.31 ±1.06 / 110.56 ms │     no change │
│ QQuery 56 │        139.09 / 139.81 ±0.67 / 141.06 ms │        138.51 / 141.55 ±1.66 / 143.08 ms │     no change │
│ QQuery 57 │        172.99 / 176.01 ±2.07 / 178.68 ms │        172.57 / 174.75 ±1.87 / 177.65 ms │     no change │
│ QQuery 58 │       286.65 / 305.68 ±14.34 / 329.43 ms │        286.98 / 296.59 ±6.96 / 307.76 ms │     no change │
│ QQuery 59 │        197.85 / 200.00 ±1.73 / 202.56 ms │        197.38 / 199.77 ±1.59 / 202.08 ms │     no change │
│ QQuery 60 │        141.55 / 142.79 ±1.19 / 144.78 ms │        140.05 / 145.18 ±2.92 / 149.15 ms │     no change │
│ QQuery 61 │           13.23 / 13.39 ±0.19 / 13.75 ms │           12.99 / 13.35 ±0.24 / 13.68 ms │     no change │
│ QQuery 62 │      885.34 / 931.50 ±46.68 / 1004.40 ms │       870.06 / 910.79 ±42.30 / 973.22 ms │     no change │
│ QQuery 63 │        101.26 / 103.37 ±2.52 / 108.18 ms │        101.90 / 104.63 ±2.70 / 108.21 ms │     no change │
│ QQuery 64 │        681.10 / 685.44 ±3.59 / 689.61 ms │        680.42 / 682.91 ±1.57 / 685.17 ms │     no change │
│ QQuery 65 │        248.78 / 252.03 ±2.52 / 256.44 ms │        247.98 / 252.37 ±3.07 / 255.93 ms │     no change │
│ QQuery 66 │       245.06 / 266.25 ±16.18 / 293.27 ms │        250.08 / 254.08 ±2.88 / 257.55 ms │     no change │
│ QQuery 67 │        301.90 / 311.31 ±8.70 / 322.94 ms │        306.65 / 313.03 ±4.98 / 320.14 ms │     no change │
│ QQuery 68 │            8.58 / 10.27 ±0.95 / 11.12 ms │             8.97 / 9.56 ±0.58 / 10.67 ms │ +1.07x faster │
│ QQuery 69 │        101.18 / 102.92 ±1.18 / 104.22 ms │        101.21 / 103.40 ±1.50 / 105.50 ms │     no change │
│ QQuery 70 │        337.78 / 351.57 ±9.15 / 363.90 ms │        343.85 / 350.56 ±5.43 / 357.46 ms │     no change │
│ QQuery 71 │        132.22 / 135.33 ±2.44 / 139.57 ms │        135.52 / 137.67 ±1.59 / 140.02 ms │     no change │
│ QQuery 72 │        611.17 / 623.18 ±8.71 / 636.75 ms │        613.19 / 617.96 ±4.07 / 624.27 ms │     no change │
│ QQuery 73 │              6.79 / 7.62 ±0.71 / 8.90 ms │              6.47 / 6.75 ±0.31 / 7.25 ms │ +1.13x faster │
│ QQuery 74 │        538.21 / 543.20 ±3.56 / 547.27 ms │        540.10 / 546.88 ±6.31 / 557.43 ms │     no change │
│ QQuery 75 │        276.27 / 277.44 ±1.39 / 280.10 ms │        270.45 / 276.61 ±3.34 / 279.89 ms │     no change │
│ QQuery 76 │        131.54 / 133.55 ±1.45 / 135.81 ms │        131.23 / 132.42 ±0.99 / 133.70 ms │     no change │
│ QQuery 77 │        186.95 / 187.47 ±0.39 / 188.07 ms │        186.12 / 188.48 ±2.02 / 190.66 ms │     no change │
│ QQuery 78 │        335.15 / 342.33 ±3.99 / 346.43 ms │        339.65 / 343.33 ±2.68 / 347.75 ms │     no change │
│ QQuery 79 │        231.85 / 233.63 ±2.18 / 237.51 ms │        234.84 / 236.73 ±1.53 / 239.08 ms │     no change │
│ QQuery 80 │        323.56 / 324.71 ±1.66 / 328.01 ms │        320.99 / 324.08 ±1.92 / 326.43 ms │     no change │
│ QQuery 81 │           25.61 / 26.77 ±1.29 / 29.27 ms │           25.82 / 26.95 ±1.08 / 28.86 ms │     no change │
│ QQuery 82 │        195.11 / 198.27 ±2.65 / 202.87 ms │        196.88 / 199.60 ±2.16 / 203.42 ms │     no change │
│ QQuery 83 │           38.06 / 39.73 ±1.02 / 40.97 ms │           37.98 / 38.67 ±0.77 / 40.09 ms │     no change │
│ QQuery 84 │           47.99 / 49.58 ±1.15 / 51.35 ms │           48.24 / 48.81 ±0.55 / 49.58 ms │     no change │
│ QQuery 85 │        145.75 / 148.34 ±1.37 / 149.66 ms │        147.10 / 148.42 ±1.20 / 150.32 ms │     no change │
│ QQuery 86 │           39.29 / 41.30 ±1.40 / 43.02 ms │           38.34 / 39.42 ±0.68 / 40.39 ms │     no change │
│ QQuery 87 │           83.53 / 87.73 ±4.16 / 95.39 ms │           85.27 / 88.74 ±3.23 / 94.63 ms │     no change │
│ QQuery 88 │         98.63 / 100.26 ±1.22 / 101.87 ms │        100.36 / 104.83 ±6.71 / 118.12 ms │     no change │
│ QQuery 89 │        116.21 / 118.06 ±1.05 / 119.25 ms │        117.57 / 118.64 ±1.07 / 120.23 ms │     no change │
│ QQuery 90 │           23.19 / 23.77 ±0.40 / 24.29 ms │           23.20 / 23.78 ±0.38 / 24.32 ms │     no change │
│ QQuery 91 │           63.01 / 64.36 ±1.35 / 66.80 ms │           61.60 / 63.49 ±1.32 / 65.56 ms │     no change │
│ QQuery 92 │           56.68 / 57.34 ±0.50 / 58.10 ms │           56.60 / 58.02 ±1.25 / 60.25 ms │     no change │
│ QQuery 93 │        186.07 / 188.16 ±1.91 / 190.62 ms │        185.52 / 186.81 ±1.35 / 189.18 ms │     no change │
│ QQuery 94 │           60.59 / 61.41 ±0.58 / 62.13 ms │           60.71 / 62.14 ±0.72 / 62.70 ms │     no change │
│ QQuery 95 │        127.32 / 128.82 ±1.10 / 130.71 ms │        127.47 / 127.92 ±0.52 / 128.82 ms │     no change │
│ QQuery 96 │           71.62 / 72.93 ±1.34 / 75.36 ms │           73.23 / 74.15 ±0.56 / 74.80 ms │     no change │
│ QQuery 97 │        123.47 / 125.61 ±1.77 / 128.54 ms │        124.12 / 125.67 ±1.21 / 127.45 ms │     no change │
│ QQuery 98 │        152.81 / 155.06 ±1.60 / 157.19 ms │        149.58 / 154.23 ±2.91 / 157.66 ms │     no change │
│ QQuery 99 │ 10722.91 / 10769.15 ±46.22 / 10846.96 ms │ 10714.65 / 10825.30 ±69.45 / 10917.91 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 31231.30ms │
│ Total Time (fix-join-cardinality-estimation)   │ 31275.18ms │
│ Average Time (HEAD)                            │   315.47ms │
│ Average Time (fix-join-cardinality-estimation) │   315.91ms │
│ Queries Faster                                 │          2 │
│ Queries Slower                                 │          1 │
│ Queries with No Change                         │         96 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	156.5s
Peak memory	5.0 GiB
Avg memory	4.4 GiB
CPU user	258.0s
CPU sys	17.3s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	156.6s
Peak memory	5.5 GiB
Avg memory	4.5 GiB
CPU user	259.2s
CPU sys	16.6s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:31:03Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213800450-1009-qdz9v 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (c8f2fa6) to 5ba06ac (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:31:15Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213800450-1010-n5fv6 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (c8f2fa6) to 5ba06ac (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:31:30Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃        fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.20 / 4.51 ±6.45 / 17.41 ms │           1.21 / 4.49 ±6.41 / 17.30 ms │     no change │
│ QQuery 1  │        14.61 / 15.01 ±0.20 / 15.18 ms │         14.27 / 14.75 ±0.28 / 15.09 ms │     no change │
│ QQuery 2  │        44.16 / 44.77 ±0.90 / 46.56 ms │         43.62 / 43.88 ±0.22 / 44.20 ms │     no change │
│ QQuery 3  │        44.23 / 45.74 ±1.22 / 47.25 ms │         40.77 / 41.24 ±0.39 / 41.85 ms │ +1.11x faster │
│ QQuery 4  │     282.95 / 291.65 ±5.79 / 300.83 ms │      285.77 / 290.75 ±3.49 / 296.68 ms │     no change │
│ QQuery 5  │     341.50 / 344.88 ±2.75 / 348.03 ms │      341.08 / 347.30 ±7.34 / 360.41 ms │     no change │
│ QQuery 6  │           4.73 / 5.85 ±0.95 / 7.50 ms │            5.34 / 7.21 ±1.02 / 8.22 ms │  1.23x slower │
│ QQuery 7  │        16.84 / 17.36 ±0.68 / 18.69 ms │         16.74 / 16.85 ±0.09 / 17.00 ms │     no change │
│ QQuery 8  │    407.78 / 426.15 ±10.60 / 435.98 ms │      411.33 / 419.58 ±6.25 / 427.38 ms │     no change │
│ QQuery 9  │     643.80 / 653.38 ±6.19 / 659.68 ms │      647.20 / 651.82 ±4.17 / 658.32 ms │     no change │
│ QQuery 10 │        91.61 / 95.94 ±2.49 / 98.43 ms │        91.04 / 93.69 ±4.36 / 102.37 ms │     no change │
│ QQuery 11 │     104.33 / 105.72 ±0.80 / 106.74 ms │      101.89 / 104.29 ±2.29 / 108.49 ms │     no change │
│ QQuery 12 │     342.07 / 347.35 ±3.24 / 351.67 ms │      339.84 / 349.51 ±9.55 / 367.52 ms │     no change │
│ QQuery 13 │     472.77 / 477.84 ±3.65 / 483.82 ms │      455.23 / 467.88 ±9.19 / 477.85 ms │     no change │
│ QQuery 14 │     350.92 / 355.31 ±3.25 / 360.50 ms │      341.78 / 348.82 ±8.56 / 365.68 ms │     no change │
│ QQuery 15 │     358.84 / 372.40 ±8.89 / 382.36 ms │     352.52 / 368.91 ±17.88 / 400.33 ms │     no change │
│ QQuery 16 │     716.27 / 725.94 ±6.23 / 735.58 ms │      716.17 / 722.77 ±3.83 / 727.72 ms │     no change │
│ QQuery 17 │     704.20 / 711.32 ±8.81 / 727.14 ms │      708.04 / 714.02 ±4.41 / 720.86 ms │     no change │
│ QQuery 18 │ 1427.04 / 1474.52 ±29.23 / 1505.06 ms │  1402.06 / 1500.22 ±56.40 / 1556.48 ms │     no change │
│ QQuery 19 │        34.90 / 36.48 ±1.36 / 38.66 ms │         37.31 / 37.79 ±0.40 / 38.51 ms │     no change │
│ QQuery 20 │    716.61 / 733.26 ±18.35 / 758.00 ms │     721.18 / 732.41 ±15.22 / 761.65 ms │     no change │
│ QQuery 21 │     763.31 / 767.86 ±2.46 / 770.58 ms │      761.33 / 767.67 ±5.53 / 775.82 ms │     no change │
│ QQuery 22 │  1134.18 / 1137.77 ±2.82 / 1142.24 ms │   1132.57 / 1138.04 ±5.48 / 1146.18 ms │     no change │
│ QQuery 23 │ 3070.24 / 3095.36 ±17.95 / 3125.45 ms │  3075.02 / 3088.36 ±11.89 / 3104.86 ms │     no change │
│ QQuery 24 │      99.43 / 103.90 ±2.63 / 107.52 ms │       97.70 / 101.70 ±2.18 / 103.81 ms │     no change │
│ QQuery 25 │     138.57 / 139.21 ±0.67 / 140.39 ms │      137.62 / 139.28 ±1.53 / 141.18 ms │     no change │
│ QQuery 26 │      99.02 / 100.43 ±0.94 / 101.36 ms │      100.70 / 103.46 ±1.84 / 105.98 ms │     no change │
│ QQuery 27 │     851.07 / 859.24 ±6.53 / 870.01 ms │      852.02 / 857.41 ±7.48 / 872.18 ms │     no change │
│ QQuery 28 │ 7716.26 / 7802.65 ±59.17 / 7885.19 ms │  7692.09 / 7743.28 ±34.83 / 7783.22 ms │     no change │
│ QQuery 29 │        52.17 / 57.47 ±5.10 / 65.93 ms │         49.46 / 54.45 ±6.62 / 66.97 ms │ +1.06x faster │
│ QQuery 30 │     365.54 / 373.12 ±5.16 / 380.00 ms │      363.80 / 367.08 ±2.75 / 372.08 ms │     no change │
│ QQuery 31 │    377.44 / 387.97 ±14.17 / 415.22 ms │     370.37 / 385.43 ±13.08 / 404.29 ms │     no change │
│ QQuery 32 │ 1059.35 / 1074.85 ±17.88 / 1103.02 ms │  1246.56 / 1266.79 ±23.52 / 1312.00 ms │  1.18x slower │
│ QQuery 33 │ 1488.61 / 1500.00 ±11.19 / 1519.07 ms │ 1489.51 / 1661.21 ±114.90 / 1769.72 ms │  1.11x slower │
│ QQuery 34 │  1467.63 / 1481.82 ±7.39 / 1487.94 ms │  1507.35 / 1538.49 ±18.69 / 1564.25 ms │     no change │
│ QQuery 35 │     399.22 / 402.74 ±2.90 / 407.24 ms │      384.70 / 392.18 ±6.82 / 402.78 ms │     no change │
│ QQuery 36 │     121.57 / 125.80 ±5.80 / 137.14 ms │      114.51 / 120.82 ±4.05 / 127.23 ms │     no change │
│ QQuery 37 │        50.87 / 52.41 ±1.64 / 55.39 ms │         47.02 / 50.33 ±2.23 / 53.25 ms │     no change │
│ QQuery 38 │        77.56 / 78.53 ±0.64 / 79.32 ms │         77.80 / 80.01 ±2.65 / 85.07 ms │     no change │
│ QQuery 39 │     219.63 / 221.06 ±1.23 / 222.94 ms │      220.17 / 225.32 ±4.17 / 232.42 ms │     no change │
│ QQuery 40 │        22.03 / 24.66 ±1.44 / 26.29 ms │         21.99 / 26.15 ±2.64 / 30.32 ms │  1.06x slower │
│ QQuery 41 │        20.24 / 21.34 ±0.88 / 22.47 ms │         20.29 / 21.25 ±0.89 / 22.89 ms │     no change │
│ QQuery 42 │        19.94 / 20.98 ±0.67 / 21.80 ms │         19.67 / 20.53 ±0.75 / 21.82 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 27114.54ms │
│ Total Time (fix-join-cardinality-estimation)   │ 27427.41ms │
│ Average Time (HEAD)                            │   630.57ms │
│ Average Time (fix-join-cardinality-estimation) │   637.85ms │
│ Queries Faster                                 │          2 │
│ Queries Slower                                 │          4 │
│ Queries with No Change                         │         37 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	136.8s
Peak memory	44.3 GiB
Avg memory	29.5 GiB
CPU user	1299.6s
CPU sys	84.5s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	138.1s
Peak memory	35.0 GiB
Avg memory	27.9 GiB
CPU user	1297.7s
CPU sys	102.9s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:31:37Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4213800450-1008-mvtgn 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing fix-join-cardinality-estimation (c8f2fa6) to 5ba06ac (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:48:43Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃          fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.59 / 7.07 ±0.87 / 8.81 ms │              6.64 / 7.14 ±0.87 / 8.88 ms │     no change │
│ QQuery 2  │        144.49 / 146.49 ±1.50 / 148.64 ms │        146.32 / 146.80 ±0.46 / 147.67 ms │     no change │
│ QQuery 3  │        114.10 / 115.04 ±0.81 / 116.46 ms │        114.20 / 114.83 ±0.57 / 115.66 ms │     no change │
│ QQuery 4  │    1363.11 / 1389.88 ±24.50 / 1420.83 ms │    1375.54 / 1404.14 ±21.89 / 1432.55 ms │     no change │
│ QQuery 5  │        171.29 / 172.37 ±0.73 / 173.19 ms │        170.15 / 171.77 ±1.15 / 173.14 ms │     no change │
│ QQuery 6  │       821.28 / 853.44 ±21.65 / 889.24 ms │       856.43 / 880.36 ±19.01 / 906.54 ms │     no change │
│ QQuery 7  │        340.15 / 343.91 ±2.40 / 347.34 ms │        345.28 / 347.12 ±1.59 / 349.57 ms │     no change │
│ QQuery 8  │        117.15 / 117.98 ±0.43 / 118.35 ms │        114.28 / 116.63 ±1.24 / 117.61 ms │     no change │
│ QQuery 9  │        102.51 / 106.12 ±1.91 / 108.11 ms │        107.04 / 107.41 ±0.27 / 107.73 ms │     no change │
│ QQuery 10 │        105.47 / 107.45 ±1.40 / 108.91 ms │        107.05 / 108.39 ±1.20 / 110.30 ms │     no change │
│ QQuery 11 │       982.32 / 991.29 ±7.49 / 1000.36 ms │       988.91 / 998.11 ±4.99 / 1002.44 ms │     no change │
│ QQuery 12 │           44.40 / 46.28 ±1.06 / 47.48 ms │           44.20 / 47.17 ±1.56 / 48.48 ms │     no change │
│ QQuery 13 │        399.38 / 400.92 ±1.14 / 402.41 ms │        401.46 / 404.68 ±2.44 / 407.38 ms │     no change │
│ QQuery 14 │      998.99 / 1010.32 ±7.96 / 1021.81 ms │      997.04 / 1010.83 ±9.46 / 1024.54 ms │     no change │
│ QQuery 15 │           15.70 / 16.90 ±0.81 / 17.91 ms │           16.11 / 17.78 ±1.82 / 21.11 ms │  1.05x slower │
│ QQuery 16 │              7.33 / 8.02 ±0.75 / 9.49 ms │              7.69 / 8.19 ±0.78 / 9.74 ms │     no change │
│ QQuery 17 │        231.01 / 231.98 ±1.35 / 234.62 ms │        233.14 / 234.17 ±1.64 / 237.43 ms │     no change │
│ QQuery 18 │        127.42 / 128.57 ±0.84 / 129.97 ms │        129.64 / 130.86 ±1.19 / 132.83 ms │     no change │
│ QQuery 19 │        155.16 / 156.59 ±1.94 / 160.32 ms │        155.96 / 157.83 ±1.09 / 159.21 ms │     no change │
│ QQuery 20 │           14.02 / 14.60 ±0.36 / 14.91 ms │           14.13 / 14.69 ±0.47 / 15.49 ms │     no change │
│ QQuery 21 │           19.52 / 20.31 ±0.60 / 21.20 ms │           19.85 / 20.10 ±0.27 / 20.58 ms │     no change │
│ QQuery 22 │        494.01 / 497.92 ±3.17 / 502.99 ms │        488.64 / 493.55 ±4.93 / 502.10 ms │     no change │
│ QQuery 23 │       873.25 / 889.44 ±10.03 / 904.54 ms │       883.06 / 901.22 ±12.23 / 913.98 ms │     no change │
│ QQuery 24 │        382.19 / 385.27 ±2.45 / 389.14 ms │        389.18 / 394.51 ±4.63 / 402.21 ms │     no change │
│ QQuery 25 │        341.76 / 343.51 ±1.84 / 345.83 ms │        346.00 / 346.83 ±0.86 / 348.08 ms │     no change │
│ QQuery 26 │           80.86 / 83.82 ±1.90 / 86.45 ms │           84.02 / 84.55 ±0.45 / 85.25 ms │     no change │
│ QQuery 27 │              6.85 / 7.38 ±0.88 / 9.13 ms │              7.08 / 7.68 ±0.60 / 8.59 ms │     no change │
│ QQuery 28 │        150.37 / 151.18 ±0.51 / 151.74 ms │        150.35 / 152.10 ±1.20 / 153.16 ms │     no change │
│ QQuery 29 │        283.42 / 285.60 ±1.73 / 288.49 ms │        282.67 / 286.41 ±3.10 / 292.03 ms │     no change │
│ QQuery 30 │           43.31 / 45.44 ±1.56 / 47.32 ms │           43.12 / 46.38 ±1.97 / 49.28 ms │     no change │
│ QQuery 31 │        171.48 / 174.02 ±1.64 / 175.85 ms │        170.60 / 172.89 ±1.38 / 174.56 ms │     no change │
│ QQuery 32 │           57.11 / 58.09 ±0.67 / 59.20 ms │           57.05 / 58.17 ±1.24 / 60.47 ms │     no change │
│ QQuery 33 │        141.08 / 143.24 ±1.67 / 145.22 ms │        140.06 / 141.58 ±1.40 / 143.38 ms │     no change │
│ QQuery 34 │              7.33 / 7.90 ±0.77 / 9.37 ms │              6.92 / 7.17 ±0.29 / 7.72 ms │ +1.10x faster │
│ QQuery 35 │        105.59 / 108.21 ±1.64 / 109.90 ms │        104.47 / 107.57 ±1.90 / 110.11 ms │     no change │
│ QQuery 36 │              6.53 / 6.73 ±0.16 / 6.93 ms │              6.64 / 6.77 ±0.11 / 6.94 ms │     no change │
│ QQuery 37 │              8.38 / 8.99 ±0.35 / 9.34 ms │              8.62 / 8.88 ±0.18 / 9.18 ms │     no change │
│ QQuery 38 │           86.91 / 90.01 ±2.50 / 94.25 ms │           86.86 / 89.63 ±3.52 / 96.53 ms │     no change │
│ QQuery 39 │        127.63 / 130.58 ±1.78 / 132.82 ms │        122.79 / 127.46 ±3.07 / 131.89 ms │     no change │
│ QQuery 40 │        113.26 / 117.09 ±4.81 / 126.21 ms │        112.26 / 116.89 ±5.79 / 128.20 ms │     no change │
│ QQuery 41 │           14.95 / 16.30 ±1.15 / 17.64 ms │           14.71 / 15.89 ±0.87 / 17.32 ms │     no change │
│ QQuery 42 │        109.18 / 110.39 ±1.54 / 113.38 ms │        106.33 / 110.47 ±2.10 / 111.92 ms │     no change │
│ QQuery 43 │              6.15 / 6.44 ±0.36 / 7.06 ms │              5.90 / 6.06 ±0.14 / 6.33 ms │ +1.06x faster │
│ QQuery 44 │           12.17 / 12.43 ±0.18 / 12.64 ms │           11.49 / 11.96 ±0.32 / 12.42 ms │     no change │
│ QQuery 45 │           50.51 / 52.34 ±1.63 / 54.45 ms │           50.57 / 51.65 ±0.92 / 53.01 ms │     no change │
│ QQuery 46 │              8.55 / 8.80 ±0.26 / 9.27 ms │              8.46 / 8.79 ±0.26 / 9.07 ms │     no change │
│ QQuery 47 │        718.66 / 726.86 ±7.31 / 740.33 ms │        728.61 / 737.08 ±5.80 / 743.78 ms │     no change │
│ QQuery 48 │        288.16 / 292.95 ±4.40 / 298.39 ms │        291.87 / 293.25 ±1.54 / 295.76 ms │     no change │
│ QQuery 49 │        254.61 / 255.14 ±0.61 / 256.32 ms │        252.39 / 254.38 ±1.95 / 257.33 ms │     no change │
│ QQuery 50 │        220.63 / 228.46 ±4.26 / 232.29 ms │        218.94 / 225.79 ±4.36 / 231.33 ms │     no change │
│ QQuery 51 │        179.50 / 183.32 ±2.48 / 186.07 ms │        180.23 / 183.24 ±2.70 / 187.86 ms │     no change │
│ QQuery 52 │        107.34 / 108.40 ±0.55 / 108.85 ms │        107.97 / 110.50 ±3.00 / 116.27 ms │     no change │
│ QQuery 53 │        102.90 / 103.56 ±0.47 / 104.16 ms │        101.73 / 103.41 ±1.08 / 105.04 ms │     no change │
│ QQuery 54 │        144.71 / 147.49 ±1.68 / 149.96 ms │        145.11 / 148.10 ±1.83 / 150.88 ms │     no change │
│ QQuery 55 │        106.67 / 108.53 ±1.43 / 111.02 ms │        106.81 / 108.13 ±1.74 / 111.53 ms │     no change │
│ QQuery 56 │        139.85 / 142.21 ±2.00 / 145.88 ms │        140.65 / 141.39 ±0.69 / 142.59 ms │     no change │
│ QQuery 57 │        173.80 / 177.06 ±1.98 / 179.19 ms │        174.30 / 176.08 ±0.91 / 176.90 ms │     no change │
│ QQuery 58 │        289.32 / 298.65 ±5.88 / 305.46 ms │        285.49 / 292.08 ±5.01 / 298.43 ms │     no change │
│ QQuery 59 │        197.75 / 199.94 ±1.47 / 201.84 ms │        198.21 / 200.18 ±1.56 / 202.87 ms │     no change │
│ QQuery 60 │        142.33 / 143.83 ±1.50 / 146.61 ms │        142.70 / 144.32 ±1.47 / 146.12 ms │     no change │
│ QQuery 61 │           13.12 / 13.53 ±0.29 / 13.98 ms │           13.08 / 13.40 ±0.25 / 13.80 ms │     no change │
│ QQuery 62 │      921.78 / 960.31 ±55.16 / 1069.13 ms │      890.35 / 965.26 ±63.04 / 1039.05 ms │     no change │
│ QQuery 63 │        104.81 / 107.34 ±1.87 / 109.61 ms │        103.44 / 105.14 ±1.52 / 107.90 ms │     no change │
│ QQuery 64 │        696.47 / 702.01 ±3.77 / 706.65 ms │        690.68 / 697.05 ±4.42 / 702.38 ms │     no change │
│ QQuery 65 │        267.66 / 271.80 ±3.23 / 277.07 ms │        252.06 / 257.78 ±3.36 / 261.08 ms │ +1.05x faster │
│ QQuery 66 │        243.06 / 254.80 ±7.46 / 262.12 ms │       245.54 / 265.10 ±13.90 / 281.96 ms │     no change │
│ QQuery 67 │        311.05 / 321.97 ±7.24 / 332.54 ms │        312.76 / 321.01 ±9.12 / 337.13 ms │     no change │
│ QQuery 68 │            8.72 / 10.35 ±1.11 / 11.89 ms │            9.44 / 10.17 ±0.51 / 10.79 ms │     no change │
│ QQuery 69 │        102.11 / 105.40 ±3.37 / 110.39 ms │        103.25 / 104.12 ±0.89 / 105.62 ms │     no change │
│ QQuery 70 │        338.35 / 347.06 ±6.00 / 355.38 ms │        347.69 / 355.46 ±8.33 / 370.50 ms │     no change │
│ QQuery 71 │        134.39 / 137.17 ±2.19 / 140.63 ms │        133.33 / 136.91 ±2.62 / 140.32 ms │     no change │
│ QQuery 72 │        621.65 / 634.09 ±9.13 / 648.43 ms │        619.30 / 625.14 ±7.62 / 639.46 ms │     no change │
│ QQuery 73 │              6.98 / 7.39 ±0.41 / 8.09 ms │             7.17 / 8.86 ±1.06 / 10.06 ms │  1.20x slower │
│ QQuery 74 │        611.57 / 617.66 ±3.16 / 619.99 ms │        603.19 / 608.45 ±4.16 / 613.60 ms │     no change │
│ QQuery 75 │        279.35 / 280.85 ±1.34 / 282.42 ms │        274.80 / 277.43 ±1.60 / 279.40 ms │     no change │
│ QQuery 76 │        131.93 / 134.80 ±1.84 / 136.85 ms │        130.75 / 133.47 ±2.07 / 136.29 ms │     no change │
│ QQuery 77 │        187.47 / 189.24 ±1.52 / 191.31 ms │        185.28 / 188.12 ±2.35 / 192.26 ms │     no change │
│ QQuery 78 │        343.78 / 348.14 ±2.54 / 351.08 ms │        337.81 / 345.02 ±3.97 / 349.74 ms │     no change │
│ QQuery 79 │        234.92 / 237.70 ±2.33 / 241.39 ms │        232.49 / 236.42 ±2.06 / 238.46 ms │     no change │
│ QQuery 80 │        321.12 / 324.12 ±2.38 / 328.42 ms │        320.87 / 323.00 ±2.46 / 326.91 ms │     no change │
│ QQuery 81 │           26.18 / 27.67 ±1.52 / 30.37 ms │           26.38 / 27.60 ±1.05 / 28.97 ms │     no change │
│ QQuery 82 │        198.59 / 200.48 ±2.58 / 205.55 ms │        195.70 / 200.13 ±2.87 / 203.90 ms │     no change │
│ QQuery 83 │           38.54 / 40.34 ±1.05 / 41.38 ms │           38.72 / 40.00 ±0.98 / 41.53 ms │     no change │
│ QQuery 84 │           48.36 / 49.52 ±0.75 / 50.58 ms │           48.61 / 49.63 ±0.72 / 50.84 ms │     no change │
│ QQuery 85 │        147.68 / 150.25 ±1.64 / 152.69 ms │        149.68 / 151.56 ±1.46 / 153.94 ms │     no change │
│ QQuery 86 │           39.59 / 39.91 ±0.22 / 40.18 ms │           40.01 / 40.45 ±0.36 / 40.90 ms │     no change │
│ QQuery 87 │           86.86 / 89.03 ±1.88 / 91.98 ms │           86.43 / 90.34 ±2.58 / 93.29 ms │     no change │
│ QQuery 88 │        100.35 / 101.28 ±1.17 / 103.54 ms │        100.45 / 101.76 ±0.69 / 102.45 ms │     no change │
│ QQuery 89 │        119.00 / 120.46 ±1.24 / 122.25 ms │        119.16 / 120.04 ±0.80 / 121.03 ms │     no change │
│ QQuery 90 │           23.48 / 23.92 ±0.37 / 24.47 ms │           23.50 / 23.95 ±0.36 / 24.50 ms │     no change │
│ QQuery 91 │           63.28 / 65.63 ±1.71 / 67.74 ms │           63.09 / 65.16 ±2.03 / 68.60 ms │     no change │
│ QQuery 92 │           58.01 / 58.62 ±0.55 / 59.43 ms │           57.72 / 58.08 ±0.24 / 58.47 ms │     no change │
│ QQuery 93 │        186.38 / 189.50 ±1.97 / 192.21 ms │        187.87 / 189.68 ±1.30 / 191.51 ms │     no change │
│ QQuery 94 │           61.00 / 61.84 ±0.46 / 62.41 ms │           60.12 / 61.25 ±0.86 / 62.25 ms │     no change │
│ QQuery 95 │        125.18 / 128.33 ±1.75 / 130.48 ms │        125.90 / 127.86 ±1.03 / 128.70 ms │     no change │
│ QQuery 96 │           74.77 / 75.59 ±0.81 / 77.13 ms │           70.00 / 73.86 ±1.99 / 75.64 ms │     no change │
│ QQuery 97 │        124.46 / 126.27 ±1.60 / 129.07 ms │        127.20 / 128.96 ±1.86 / 132.09 ms │     no change │
│ QQuery 98 │        152.47 / 156.43 ±2.75 / 160.69 ms │        152.56 / 155.35 ±2.16 / 158.14 ms │     no change │
│ QQuery 99 │ 10808.81 / 10852.95 ±38.38 / 10907.68 ms │ 10810.05 / 10875.08 ±51.49 / 10936.91 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 31806.80ms │
│ Total Time (fix-join-cardinality-estimation)   │ 31880.01ms │
│ Average Time (HEAD)                            │   321.28ms │
│ Average Time (fix-join-cardinality-estimation) │   322.02ms │
│ Queries Faster                                 │          3 │
│ Queries Slower                                 │          2 │
│ Queries with No Change                         │         94 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	159.4s
Peak memory	5.4 GiB
Avg memory	4.4 GiB
CPU user	262.9s
CPU sys	17.4s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	159.7s
Peak memory	5.6 GiB
Avg memory	4.5 GiB
CPU user	263.7s
CPU sys	17.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-09T11:49:58Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and fix-join-cardinality-estimation
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       fix-join-cardinality-estimation ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.26 / 4.62 ±6.53 / 17.66 ms │          1.24 / 4.54 ±6.45 / 17.43 ms │     no change │
│ QQuery 1  │        14.87 / 15.13 ±0.14 / 15.26 ms │        13.97 / 14.47 ±0.28 / 14.78 ms │     no change │
│ QQuery 2  │        45.13 / 45.54 ±0.40 / 46.25 ms │        42.80 / 43.05 ±0.29 / 43.42 ms │ +1.06x faster │
│ QQuery 3  │        44.34 / 46.88 ±1.38 / 48.42 ms │        40.80 / 44.85 ±3.30 / 49.62 ms │     no change │
│ QQuery 4  │     310.22 / 316.69 ±5.69 / 326.03 ms │    292.94 / 316.38 ±20.02 / 343.19 ms │     no change │
│ QQuery 5  │     369.52 / 371.80 ±2.27 / 375.38 ms │    350.86 / 365.37 ±10.45 / 378.24 ms │     no change │
│ QQuery 6  │           5.57 / 6.91 ±0.70 / 7.61 ms │           6.24 / 7.42 ±1.15 / 9.47 ms │  1.07x slower │
│ QQuery 7  │        17.35 / 17.56 ±0.12 / 17.68 ms │        16.91 / 18.14 ±1.03 / 19.79 ms │     no change │
│ QQuery 8  │    441.97 / 455.38 ±13.06 / 480.02 ms │    435.03 / 458.85 ±16.13 / 479.39 ms │     no change │
│ QQuery 9  │    717.74 / 735.64 ±10.21 / 746.90 ms │    651.59 / 684.52 ±23.99 / 710.51 ms │ +1.07x faster │
│ QQuery 10 │        89.36 / 92.43 ±3.51 / 99.21 ms │       91.81 / 96.65 ±3.41 / 100.89 ms │     no change │
│ QQuery 11 │     104.15 / 107.71 ±2.31 / 110.44 ms │     104.09 / 106.16 ±2.25 / 109.48 ms │     no change │
│ QQuery 12 │     379.20 / 385.25 ±5.84 / 394.34 ms │    334.73 / 358.20 ±12.10 / 367.78 ms │ +1.08x faster │
│ QQuery 13 │    452.81 / 481.95 ±22.25 / 520.62 ms │    464.14 / 484.59 ±15.03 / 499.58 ms │     no change │
│ QQuery 14 │    338.40 / 357.95 ±15.54 / 379.62 ms │    347.75 / 364.55 ±11.05 / 379.45 ms │     no change │
│ QQuery 15 │    347.18 / 384.57 ±23.01 / 418.33 ms │    356.55 / 380.34 ±22.41 / 418.52 ms │     no change │
│ QQuery 16 │    748.65 / 762.47 ±14.79 / 783.61 ms │    731.46 / 752.67 ±15.59 / 776.72 ms │     no change │
│ QQuery 17 │    730.93 / 742.76 ±13.29 / 767.47 ms │     733.02 / 744.16 ±7.72 / 752.86 ms │     no change │
│ QQuery 18 │ 1425.46 / 1531.87 ±65.55 / 1632.12 ms │ 1461.99 / 1512.71 ±41.62 / 1560.42 ms │     no change │
│ QQuery 19 │        36.94 / 38.80 ±2.15 / 42.77 ms │       36.09 / 44.02 ±13.45 / 70.88 ms │  1.13x slower │
│ QQuery 20 │    721.07 / 743.17 ±18.43 / 775.03 ms │    714.31 / 729.00 ±17.89 / 763.51 ms │     no change │
│ QQuery 21 │     761.86 / 772.35 ±7.47 / 783.87 ms │     772.01 / 778.35 ±5.58 / 787.65 ms │     no change │
│ QQuery 22 │  1150.04 / 1156.45 ±4.60 / 1161.02 ms │  1166.30 / 1174.57 ±7.37 / 1187.71 ms │     no change │
│ QQuery 23 │ 3117.15 / 3156.61 ±25.25 / 3186.26 ms │ 3203.62 / 3231.98 ±21.08 / 3267.81 ms │     no change │
│ QQuery 24 │      98.26 / 104.05 ±4.36 / 111.61 ms │     101.80 / 107.03 ±3.71 / 110.92 ms │     no change │
│ QQuery 25 │     142.92 / 144.64 ±1.34 / 146.48 ms │     140.80 / 142.24 ±1.52 / 145.16 ms │     no change │
│ QQuery 26 │     104.71 / 107.02 ±1.69 / 109.37 ms │     102.61 / 107.06 ±2.69 / 109.92 ms │     no change │
│ QQuery 27 │     867.77 / 875.83 ±5.05 / 882.22 ms │     852.28 / 860.29 ±4.39 / 864.24 ms │     no change │
│ QQuery 28 │ 7779.60 / 7812.38 ±19.50 / 7834.48 ms │ 7779.51 / 7828.33 ±35.60 / 7888.27 ms │     no change │
│ QQuery 29 │        52.42 / 55.96 ±3.44 / 61.60 ms │        51.39 / 54.52 ±3.63 / 60.84 ms │     no change │
│ QQuery 30 │     370.55 / 381.93 ±8.87 / 393.80 ms │    362.06 / 380.85 ±15.88 / 404.88 ms │     no change │
│ QQuery 31 │    377.97 / 390.31 ±12.51 / 408.19 ms │     383.66 / 392.39 ±7.89 / 404.60 ms │     no change │
│ QQuery 32 │ 1018.06 / 1062.36 ±23.42 / 1086.61 ms │ 1096.89 / 1122.17 ±33.17 / 1187.10 ms │  1.06x slower │
│ QQuery 33 │ 1519.63 / 1563.19 ±27.08 / 1602.54 ms │ 1511.82 / 1546.43 ±39.25 / 1622.47 ms │     no change │
│ QQuery 34 │ 1550.62 / 1612.92 ±34.92 / 1656.02 ms │ 1530.66 / 1558.12 ±20.37 / 1586.53 ms │     no change │
│ QQuery 35 │     435.85 / 445.65 ±5.72 / 452.41 ms │    394.64 / 427.90 ±28.04 / 462.27 ms │     no change │
│ QQuery 36 │     119.62 / 123.04 ±2.68 / 127.38 ms │     119.90 / 125.36 ±3.39 / 129.68 ms │     no change │
│ QQuery 37 │        50.34 / 52.72 ±1.53 / 54.44 ms │        50.82 / 53.90 ±2.18 / 56.92 ms │     no change │
│ QQuery 38 │        74.55 / 78.05 ±3.12 / 83.05 ms │        75.94 / 77.17 ±1.12 / 79.03 ms │     no change │
│ QQuery 39 │    204.56 / 224.27 ±15.70 / 247.13 ms │     221.80 / 230.31 ±7.08 / 239.85 ms │     no change │
│ QQuery 40 │        23.21 / 25.80 ±1.40 / 27.12 ms │        22.18 / 27.08 ±3.41 / 32.11 ms │     no change │
│ QQuery 41 │        21.86 / 22.96 ±0.93 / 24.10 ms │        21.02 / 23.19 ±1.40 / 25.17 ms │     no change │
│ QQuery 42 │        20.86 / 21.64 ±0.42 / 22.05 ms │        20.99 / 21.88 ±0.57 / 22.74 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 27835.17ms │
│ Total Time (fix-join-cardinality-estimation)   │ 27801.72ms │
│ Average Time (HEAD)                            │   647.33ms │
│ Average Time (fix-join-cardinality-estimation) │   646.55ms │
│ Queries Faster                                 │          3 │
│ Queries Slower                                 │          3 │
│ Queries with No Change                         │         37 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	140.6s
Peak memory	41.3 GiB
Avg memory	30.7 GiB
CPU user	1331.1s
CPU sys	91.3s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	140.1s
Peak memory	38.0 GiB
Avg memory	27.9 GiB
CPU user	1324.7s
CPU sys	95.1s
Peak spill	0 B

File an issue against this benchmark runner

asolimando · 2026-04-09T16:57:41Z

Hey @Dandandan, I noticed this PR and it resonated a lot with what I am working on lately.

Specifically, a combination of #21483 (introducing StatisticRegistry to provide customizable support for statistic at the operator level and in physical rules) and #21122 (introducing ExpressionAnalyzer, adding customizable support for statistics at the expression level, including casts, which is totally missing at the moment, as you noticed) would allow to support similar improvements easily and in a way that downstream consumers could override fully (including custom statistics, but also override the way built-in operator deal with statistics, if one needs different statistics propagation or cardinality estimation).

In local testing I had a similar speed up for Q99 (~93x speedup) and others, in local testing, for the same exact reasons you described, improving on statistics allowed to better select the join operand parameters and join sides.

At the moment there is still a limitation due to the fully additive and non-breaking stance we took for introducing the framework, but once #20184 gets in we will be able to have full support for all operators, and improve current CBO decisions like the JoinSelection you mentioned, and show similar improvements in TPC-* benchmarks.

Additionally, this could lead to moving beyond some per-query flags like prefer_hash_join, and use a CBO approach for setting the join type at individual operator level.

Apart from sharing a potential connection to what I am working on (in case you are interested), I also wanted to ask you why the PR closed, are you exploring other angles or you hit a roadblock?

github-actions bot added the physical-plan Changes to the physical-plan crate label Apr 9, 2026

Dandandan force-pushed the fix-join-cardinality-estimation branch from 991084b to ad81053 Compare April 9, 2026 09:57

Dandandan force-pushed the fix-join-cardinality-estimation branch from ad81053 to 775906d Compare April 9, 2026 10:02

Dandandan changed the title ~~fix: improve inner join cardinality estimation for FK joins~~ Improve inner join cardinality estimation for FK joins Apr 9, 2026

Dandandan marked this pull request as ready for review April 9, 2026 10:06

Dandandan force-pushed the fix-join-cardinality-estimation branch from 775906d to 3e0ab51 Compare April 9, 2026 10:19

Dandandan changed the title ~~Improve inner join cardinality estimation for FK joins~~ Improve inner join cardinality estimation for FK joins without distinct count statistics Apr 9, 2026

Dandandan force-pushed the fix-join-cardinality-estimation branch from 3e0ab51 to d4ac04a Compare April 9, 2026 10:25

Dandandan force-pushed the fix-join-cardinality-estimation branch from d4ac04a to 322e5f6 Compare April 9, 2026 10:57

Dandandan changed the title ~~Improve inner join cardinality estimation for FK joins without distinct count statistics~~ fix: propagate column statistics through CAST in join key expressions Apr 9, 2026

github-actions bot removed the core Core DataFusion crate label Apr 9, 2026

Dandandan force-pushed the fix-join-cardinality-estimation branch from 322e5f6 to 03037cd Compare April 9, 2026 11:06

Dandandan force-pushed the fix-join-cardinality-estimation branch from 03037cd to c8f2fa6 Compare April 9, 2026 11:23

Dandandan changed the title ~~fix: propagate column statistics through CAST in join key expressions~~ fix: propagate column statistics through CAST in projections Apr 9, 2026

Dandandan closed this Apr 9, 2026

Conversation

Dandandan commented Apr 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

adriangbot commented Apr 9, 2026

Uh oh!

Dandandan commented Apr 9, 2026

Dandandan commented Apr 9, 2026 •

edited

Loading