Skip to content

Defer task spawning in CoalescePartitionsExec to first poll#21326

Closed
Dandandan wants to merge 1 commit intoapache:mainfrom
Dandandan:lazy-coalesce-partitions
Closed

Defer task spawning in CoalescePartitionsExec to first poll#21326
Dandandan wants to merge 1 commit intoapache:mainfrom
Dandandan:lazy-coalesce-partitions

Conversation

@Dandandan
Copy link
Copy Markdown
Contributor

@Dandandan Dandandan commented Apr 2, 2026 •
edited
Loading

Which issue does this PR close?

N/A

Rationale for this change

CoalescePartitionsExec::execute() eagerly spawns tasks for all input partitions immediately, even before any batch is polled from the output stream. This wastes resources when streams are created but never consumed (e.g., query cancelled before first poll), and increases peak memory usage.

I am benchmarking it to see the current perf impact.

What changes are included in this PR?

Introduces a CoalescePartitionsStream that defers spawning input partition tasks until the first poll_next() call, rather than in execute(). The single-partition fast path is unchanged.

How are these changes tested?

All existing tests pass (8/8), covering:

  • Multi-partition merge
  • Single/multi partition with fetch limits
  • Drop cancellation
  • Panic propagation

🤖 Generated with Claude Code

@Dandandan @claude
Defer task spawning in CoalescePartitionsExec to first poll
e2339fb 
Previously, CoalescePartitionsExec eagerly spawned tasks for all input
partitions in execute(). This meant that even if the output stream was
never polled (e.g., query cancelled), all tasks would be spawned.

This changes the multi-partition path to use a lazy stream that defers
spawning until first poll_next(), reducing unnecessary work when streams
are created but not consumed.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@Dandandan
Copy link
Copy Markdown
Contributor Author

Dandandan commented Apr 2, 2026

run benchmarks

@github-actions github-actions bot added the physical-plan Changes to the physical-plan crate label Apr 2, 2026
@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4180001312-667-8p9qw 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 lazy-coalesce-partitions (e2339fb) to 4084a18 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4180001312-668-7n4cg 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 lazy-coalesce-partitions (e2339fb) to 4084a18 (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4180001312-666-txkzj 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 lazy-coalesce-partitions (e2339fb) to 4084a18 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

@Dandandan Dandandan mentioned this pull request Apr 2, 2026
Open
@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 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 lazy-coalesce-partitions
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃       lazy-coalesce-partitions ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 45.29 / 46.03 ±0.76 / 47.32 ms │ 45.36 / 46.22 ±0.93 / 47.49 ms │    no change │
│ QQuery 2  │ 20.84 / 21.39 ±0.50 / 22.20 ms │ 24.70 / 24.89 ±0.15 / 25.15 ms │ 1.16x slower │
│ QQuery 3  │ 31.80 / 32.02 ±0.21 / 32.41 ms │ 32.18 / 33.34 ±1.29 / 35.21 ms │    no change │
│ QQuery 4  │ 20.56 / 21.34 ±0.82 / 22.63 ms │ 20.19 / 21.33 ±0.79 / 22.36 ms │    no change │
│ QQuery 5  │ 47.96 / 50.02 ±1.61 / 51.73 ms │ 49.80 / 51.08 ±0.98 / 52.22 ms │    no change │
│ QQuery 6  │ 17.11 / 17.35 ±0.18 / 17.56 ms │ 17.16 / 17.76 ±0.94 / 19.62 ms │    no change │
│ QQuery 7  │ 56.28 / 57.26 ±0.87 / 58.35 ms │ 55.37 / 56.42 ±1.53 / 59.43 ms │    no change │
│ QQuery 8  │ 47.64 / 48.54 ±0.73 / 49.82 ms │ 52.01 / 52.26 ±0.23 / 52.63 ms │ 1.08x slower │
│ QQuery 9  │ 54.41 / 55.36 ±1.00 / 57.21 ms │ 65.51 / 67.19 ±1.21 / 69.03 ms │ 1.21x slower │
│ QQuery 10 │ 70.54 / 72.48 ±1.42 / 74.78 ms │ 71.71 / 74.30 ±2.11 / 77.56 ms │    no change │
│ QQuery 11 │ 13.95 / 14.25 ±0.31 / 14.80 ms │ 14.69 / 14.88 ±0.23 / 15.32 ms │    no change │
│ QQuery 12 │ 28.02 / 28.36 ±0.37 / 29.04 ms │ 27.50 / 27.88 ±0.21 / 28.09 ms │    no change │
│ QQuery 13 │ 39.00 / 40.46 ±1.71 / 43.78 ms │ 38.68 / 40.01 ±1.03 / 41.62 ms │    no change │
│ QQuery 14 │ 28.41 / 28.78 ±0.44 / 29.63 ms │ 28.43 / 28.60 ±0.12 / 28.76 ms │    no change │
│ QQuery 15 │ 33.56 / 34.26 ±0.61 / 35.31 ms │ 33.63 / 35.27 ±0.90 / 36.25 ms │    no change │
│ QQuery 16 │ 15.93 / 16.25 ±0.19 / 16.46 ms │ 17.93 / 18.36 ±0.36 / 18.97 ms │ 1.13x slower │
│ QQuery 17 │ 72.74 / 74.09 ±1.02 / 75.19 ms │ 71.94 / 73.83 ±1.49 / 76.43 ms │    no change │
│ QQuery 18 │ 76.29 / 79.28 ±1.83 / 81.32 ms │ 76.89 / 79.02 ±1.27 / 80.40 ms │    no change │
│ QQuery 19 │ 37.39 / 37.75 ±0.37 / 38.38 ms │ 37.09 / 37.52 ±0.48 / 38.39 ms │    no change │
│ QQuery 20 │ 39.63 / 40.48 ±0.61 / 41.24 ms │ 41.86 / 42.89 ±0.88 / 44.19 ms │ 1.06x slower │
│ QQuery 21 │ 63.92 / 65.25 ±1.04 / 66.85 ms │ 62.84 / 64.52 ±1.24 / 66.02 ms │    no change │
│ QQuery 22 │ 17.94 / 18.17 ±0.15 / 18.35 ms │ 25.42 / 26.01 ±0.38 / 26.59 ms │ 1.43x slower │
└───────────┴────────────────────────────────┴────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                       ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 899.18ms │
│ Total Time (lazy-coalesce-partitions)   │ 933.59ms │
│ Average Time (HEAD)                     │  40.87ms │
│ Average Time (lazy-coalesce-partitions) │  42.44ms │
│ Queries Faster                          │        0 │
│ Queries Slower                          │        6 │
│ Queries with No Change                  │       16 │
│ Queries with Failure                    │        0 │
└─────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric Value
Wall time 4.8s
Peak memory 4.1 GiB
Avg memory 3.6 GiB
CPU user 33.4s
CPU sys 3.0s
Disk read 0 B
Disk write 140.0 KiB

tpch — branch

Metric Value
Wall time 4.9s
Peak memory 4.1 GiB
Avg memory 3.6 GiB
CPU user 33.3s
CPU sys 3.0s
Disk read 0 B
Disk write 80.0 KiB

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 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 lazy-coalesce-partitions
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃              lazy-coalesce-partitions ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.30 / 4.48 ±6.29 / 17.05 ms │          1.30 / 4.49 ±6.30 / 17.08 ms │     no change │
│ QQuery 1  │        14.07 / 14.47 ±0.25 / 14.78 ms │        14.05 / 14.48 ±0.26 / 14.86 ms │     no change │
│ QQuery 2  │        44.02 / 44.50 ±0.42 / 45.16 ms │        43.96 / 44.31 ±0.25 / 44.61 ms │     no change │
│ QQuery 3  │        43.34 / 44.81 ±1.28 / 46.97 ms │        41.83 / 45.22 ±2.44 / 48.48 ms │     no change │
│ QQuery 4  │     286.77 / 294.40 ±4.07 / 298.04 ms │     292.58 / 303.72 ±7.01 / 312.89 ms │     no change │
│ QQuery 5  │     344.12 / 348.24 ±3.08 / 352.55 ms │     342.68 / 345.99 ±4.48 / 354.84 ms │     no change │
│ QQuery 6  │           5.28 / 6.14 ±0.55 / 7.01 ms │           5.21 / 5.96 ±0.51 / 6.51 ms │     no change │
│ QQuery 7  │        16.45 / 17.07 ±0.84 / 18.74 ms │        16.29 / 16.91 ±0.80 / 18.46 ms │     no change │
│ QQuery 8  │     416.65 / 425.74 ±7.51 / 435.13 ms │     424.12 / 430.53 ±4.73 / 438.67 ms │     no change │
│ QQuery 9  │    626.14 / 650.39 ±46.04 / 742.47 ms │    625.23 / 639.55 ±10.06 / 651.58 ms │     no change │
│ QQuery 10 │        91.82 / 94.87 ±2.21 / 98.04 ms │        92.24 / 95.14 ±2.00 / 98.12 ms │     no change │
│ QQuery 11 │     103.65 / 105.56 ±1.10 / 106.64 ms │     102.58 / 103.43 ±0.73 / 104.60 ms │     no change │
│ QQuery 12 │     339.60 / 344.79 ±3.21 / 348.87 ms │     347.61 / 351.58 ±2.95 / 355.38 ms │     no change │
│ QQuery 13 │    459.53 / 477.69 ±14.85 / 495.69 ms │    442.96 / 472.88 ±16.69 / 492.94 ms │     no change │
│ QQuery 14 │     350.36 / 356.42 ±4.60 / 363.76 ms │     350.69 / 357.19 ±5.49 / 366.32 ms │     no change │
│ QQuery 15 │    358.79 / 381.35 ±24.32 / 428.52 ms │    367.85 / 380.39 ±11.28 / 400.61 ms │     no change │
│ QQuery 16 │  738.12 / 892.19 ±269.77 / 1430.95 ms │    731.52 / 755.38 ±25.42 / 801.76 ms │ +1.18x faster │
│ QQuery 17 │     711.91 / 721.03 ±7.24 / 732.87 ms │     728.02 / 734.35 ±4.18 / 739.54 ms │     no change │
│ QQuery 18 │ 1439.32 / 1510.03 ±37.82 / 1551.57 ms │ 1445.45 / 1472.06 ±26.29 / 1518.53 ms │     no change │
│ QQuery 19 │        35.72 / 37.28 ±1.26 / 38.97 ms │      35.87 / 64.04 ±35.76 / 127.11 ms │  1.72x slower │
│ QQuery 20 │    713.27 / 727.75 ±14.20 / 749.14 ms │    714.19 / 735.34 ±24.50 / 780.16 ms │     no change │
│ QQuery 21 │    758.01 / 773.45 ±21.01 / 814.30 ms │     756.78 / 763.36 ±4.96 / 772.17 ms │     no change │
│ QQuery 22 │  1124.82 / 1130.75 ±5.73 / 1140.69 ms │  1126.16 / 1129.07 ±2.56 / 1133.75 ms │     no change │
│ QQuery 23 │ 3053.39 / 3072.91 ±19.07 / 3099.67 ms │ 3042.73 / 3066.87 ±15.36 / 3084.04 ms │     no change │
│ QQuery 24 │      97.14 / 101.91 ±2.83 / 105.35 ms │      99.76 / 102.95 ±3.14 / 106.81 ms │     no change │
│ QQuery 25 │     136.61 / 139.36 ±1.95 / 142.45 ms │     137.78 / 140.03 ±1.23 / 141.06 ms │     no change │
│ QQuery 26 │      97.05 / 100.46 ±2.24 / 104.02 ms │       95.49 / 99.81 ±2.47 / 102.49 ms │     no change │
│ QQuery 27 │     847.44 / 852.61 ±5.15 / 861.78 ms │     846.17 / 853.55 ±8.86 / 868.46 ms │     no change │
│ QQuery 28 │ 7745.51 / 7823.45 ±50.59 / 7890.84 ms │ 7678.20 / 7731.42 ±30.23 / 7766.80 ms │     no change │
│ QQuery 29 │        50.64 / 55.82 ±5.09 / 64.62 ms │        49.98 / 53.92 ±4.40 / 59.46 ms │     no change │
│ QQuery 30 │     362.11 / 367.82 ±6.19 / 378.74 ms │     364.09 / 369.39 ±6.95 / 382.59 ms │     no change │
│ QQuery 31 │     364.99 / 381.51 ±8.96 / 390.87 ms │    368.02 / 380.47 ±11.97 / 399.87 ms │     no change │
│ QQuery 32 │ 1039.12 / 1051.88 ±16.98 / 1084.97 ms │ 1170.50 / 1222.72 ±53.04 / 1299.37 ms │  1.16x slower │
│ QQuery 33 │ 1449.21 / 1470.76 ±11.36 / 1480.17 ms │ 1438.55 / 1458.49 ±16.79 / 1483.50 ms │     no change │
│ QQuery 34 │  1450.04 / 1457.91 ±8.55 / 1473.82 ms │ 1426.98 / 1446.02 ±19.37 / 1483.35 ms │     no change │
│ QQuery 35 │     383.70 / 391.35 ±7.03 / 400.84 ms │     385.45 / 391.07 ±3.71 / 394.80 ms │     no change │
│ QQuery 36 │     118.47 / 122.91 ±2.48 / 125.17 ms │     122.38 / 123.87 ±1.64 / 126.39 ms │     no change │
│ QQuery 37 │        47.49 / 48.94 ±1.68 / 52.04 ms │        48.63 / 50.08 ±1.23 / 52.25 ms │     no change │
│ QQuery 38 │        75.02 / 76.46 ±1.74 / 79.76 ms │        74.78 / 76.22 ±0.91 / 77.55 ms │     no change │
│ QQuery 39 │     219.93 / 222.85 ±3.56 / 229.82 ms │     208.06 / 213.70 ±4.45 / 220.04 ms │     no change │
│ QQuery 40 │        23.58 / 25.70 ±1.75 / 28.38 ms │        22.97 / 25.63 ±1.54 / 27.26 ms │     no change │
│ QQuery 41 │        18.74 / 20.57 ±1.18 / 22.08 ms │        20.59 / 20.80 ±0.25 / 21.28 ms │     no change │
│ QQuery 42 │        18.92 / 19.38 ±0.32 / 19.78 ms │        18.51 / 19.49 ±0.61 / 20.30 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 27207.98ms │
│ Total Time (lazy-coalesce-partitions)   │ 27111.86ms │
│ Average Time (HEAD)                     │   632.74ms │
│ Average Time (lazy-coalesce-partitions) │   630.51ms │
│ Queries Faster                          │          1 │
│ Queries Slower                          │          2 │
│ Queries with No Change                  │         40 │
│ Queries with Failure                    │          0 │
└─────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 137.2s
Peak memory 44.5 GiB
Avg memory 32.5 GiB
CPU user 1289.7s
CPU sys 91.9s
Disk read 0 B
Disk write 4.4 GiB

clickbench_partitioned — branch

Metric Value
Wall time 136.6s
Peak memory 40.3 GiB
Avg memory 30.1 GiB
CPU user 1277.2s
CPU sys 96.9s
Disk read 0 B
Disk write 748.0 KiB

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 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 lazy-coalesce-partitions
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃                 lazy-coalesce-partitions ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           43.65 / 44.31 ±0.88 / 45.99 ms │           44.56 / 45.57 ±0.70 / 46.72 ms │     no change │
│ QQuery 2  │        145.06 / 146.47 ±0.74 / 147.00 ms │        150.14 / 151.73 ±1.03 / 153.13 ms │     no change │
│ QQuery 3  │        115.23 / 116.21 ±0.99 / 117.93 ms │        116.95 / 118.25 ±0.96 / 119.15 ms │     no change │
│ QQuery 4  │    1386.84 / 1422.69 ±23.95 / 1454.01 ms │    1718.64 / 1763.56 ±27.71 / 1797.66 ms │  1.24x slower │
│ QQuery 5  │        176.05 / 178.21 ±1.66 / 181.08 ms │        175.49 / 177.08 ±1.20 / 178.68 ms │     no change │
│ QQuery 6  │    1047.37 / 1093.65 ±30.09 / 1128.21 ms │    1032.39 / 1064.29 ±35.51 / 1110.85 ms │     no change │
│ QQuery 7  │        361.65 / 366.25 ±3.32 / 371.53 ms │        355.82 / 361.09 ±5.47 / 371.63 ms │     no change │
│ QQuery 8  │        118.88 / 119.45 ±0.43 / 120.19 ms │        120.38 / 122.10 ±1.04 / 123.37 ms │     no change │
│ QQuery 9  │        102.37 / 106.58 ±3.23 / 110.43 ms │        812.67 / 816.71 ±3.08 / 821.84 ms │  7.66x slower │
│ QQuery 10 │        108.48 / 110.88 ±2.05 / 113.50 ms │        130.80 / 132.55 ±1.10 / 134.11 ms │  1.20x slower │
│ QQuery 11 │    1009.78 / 1023.99 ±15.63 / 1053.16 ms │     1150.50 / 1160.09 ±7.30 / 1172.72 ms │  1.13x slower │
│ QQuery 12 │           45.78 / 47.77 ±1.17 / 48.98 ms │           47.51 / 48.95 ±1.60 / 51.97 ms │     no change │
│ QQuery 13 │        407.40 / 409.72 ±1.63 / 411.84 ms │        406.38 / 409.82 ±2.80 / 414.79 ms │     no change │
│ QQuery 14 │    1025.40 / 1036.56 ±12.20 / 1059.35 ms │     1741.56 / 1752.12 ±8.84 / 1765.37 ms │  1.69x slower │
│ QQuery 15 │           16.26 / 17.23 ±0.77 / 18.06 ms │           16.51 / 17.77 ±1.15 / 19.25 ms │     no change │
│ QQuery 16 │           41.41 / 42.52 ±0.75 / 43.63 ms │           42.23 / 43.44 ±0.70 / 44.31 ms │     no change │
│ QQuery 17 │        244.04 / 245.74 ±1.11 / 247.16 ms │        242.55 / 245.71 ±3.11 / 251.68 ms │     no change │
│ QQuery 18 │        130.71 / 132.46 ±1.51 / 135.12 ms │        132.30 / 134.25 ±1.56 / 136.23 ms │     no change │
│ QQuery 19 │        158.08 / 159.66 ±1.09 / 161.38 ms │        156.37 / 157.99 ±1.28 / 159.67 ms │     no change │
│ QQuery 20 │           14.36 / 15.04 ±0.77 / 16.24 ms │           15.90 / 16.64 ±0.44 / 17.06 ms │  1.11x slower │
│ QQuery 21 │           20.05 / 20.71 ±0.61 / 21.80 ms │           21.10 / 21.66 ±0.65 / 22.91 ms │     no change │
│ QQuery 22 │        491.13 / 493.92 ±1.52 / 495.70 ms │        496.03 / 498.69 ±2.70 / 501.99 ms │     no change │
│ QQuery 23 │        919.44 / 932.73 ±8.45 / 942.20 ms │    1230.52 / 1260.30 ±20.77 / 1293.27 ms │  1.35x slower │
│ QQuery 24 │        422.32 / 424.25 ±2.58 / 429.29 ms │        816.31 / 820.61 ±4.35 / 828.97 ms │  1.93x slower │
│ QQuery 25 │        360.72 / 363.27 ±1.97 / 365.63 ms │        359.14 / 362.76 ±2.62 / 366.35 ms │     no change │
│ QQuery 26 │           82.06 / 84.46 ±1.39 / 85.86 ms │           82.95 / 84.53 ±1.40 / 86.81 ms │     no change │
│ QQuery 27 │        353.82 / 357.94 ±3.32 / 362.41 ms │        351.04 / 352.95 ±2.04 / 356.68 ms │     no change │
│ QQuery 28 │        151.54 / 152.79 ±0.98 / 154.23 ms │        844.73 / 847.12 ±2.16 / 849.90 ms │  5.54x slower │
│ QQuery 29 │        302.99 / 307.74 ±3.15 / 312.32 ms │        303.55 / 304.99 ±1.26 / 306.96 ms │     no change │
│ QQuery 30 │           42.60 / 44.73 ±1.75 / 47.88 ms │           51.65 / 52.80 ±0.90 / 53.90 ms │  1.18x slower │
│ QQuery 31 │        172.08 / 174.81 ±1.94 / 178.02 ms │        426.34 / 428.69 ±2.41 / 432.98 ms │  2.45x slower │
│ QQuery 32 │           57.73 / 59.83 ±1.35 / 61.51 ms │           58.28 / 59.29 ±0.77 / 60.56 ms │     no change │
│ QQuery 33 │        143.46 / 145.63 ±1.28 / 147.13 ms │        143.70 / 145.81 ±1.15 / 146.96 ms │     no change │
│ QQuery 34 │        108.72 / 110.20 ±1.21 / 112.10 ms │        109.44 / 110.65 ±0.99 / 112.28 ms │     no change │
│ QQuery 35 │        111.13 / 112.14 ±1.31 / 114.63 ms │        136.96 / 138.14 ±1.03 / 139.99 ms │  1.23x slower │
│ QQuery 36 │        219.48 / 223.16 ±2.48 / 226.22 ms │        214.68 / 220.36 ±3.69 / 226.19 ms │     no change │
│ QQuery 37 │        181.11 / 183.16 ±1.13 / 184.20 ms │        178.89 / 181.96 ±1.99 / 184.79 ms │     no change │
│ QQuery 38 │           86.84 / 90.04 ±3.05 / 93.83 ms │        118.76 / 121.43 ±1.68 / 123.86 ms │  1.35x slower │
│ QQuery 39 │        127.04 / 129.82 ±2.49 / 134.33 ms │        127.98 / 134.06 ±3.87 / 139.63 ms │     no change │
│ QQuery 40 │        111.89 / 118.91 ±6.43 / 130.60 ms │        110.28 / 119.63 ±9.39 / 137.25 ms │     no change │
│ QQuery 41 │           14.67 / 15.72 ±0.99 / 17.33 ms │           15.56 / 16.66 ±0.91 / 18.15 ms │  1.06x slower │
│ QQuery 42 │        108.66 / 110.24 ±1.15 / 111.39 ms │        109.60 / 111.22 ±1.98 / 114.47 ms │     no change │
│ QQuery 43 │           84.34 / 85.84 ±1.41 / 87.63 ms │           85.40 / 86.02 ±1.06 / 88.13 ms │     no change │
│ QQuery 44 │           12.06 / 12.46 ±0.42 / 13.25 ms │           11.38 / 11.60 ±0.12 / 11.71 ms │ +1.07x faster │
│ QQuery 45 │           53.46 / 54.34 ±0.98 / 55.76 ms │           53.57 / 54.53 ±0.77 / 55.48 ms │     no change │
│ QQuery 46 │        235.65 / 238.81 ±2.71 / 242.76 ms │        240.34 / 242.92 ±2.20 / 246.67 ms │     no change │
│ QQuery 47 │        734.05 / 740.68 ±3.91 / 745.77 ms │        754.30 / 759.29 ±5.91 / 770.78 ms │     no change │
│ QQuery 48 │        294.73 / 298.10 ±2.56 / 301.14 ms │        281.89 / 292.82 ±6.04 / 299.64 ms │     no change │
│ QQuery 49 │        256.75 / 259.44 ±2.08 / 262.57 ms │        256.47 / 258.57 ±1.68 / 261.30 ms │     no change │
│ QQuery 50 │        237.07 / 240.53 ±3.20 / 245.52 ms │        234.42 / 238.39 ±3.67 / 244.69 ms │     no change │
│ QQuery 51 │        184.91 / 186.44 ±1.07 / 188.11 ms │        186.15 / 188.97 ±1.69 / 191.19 ms │     no change │
│ QQuery 52 │        109.19 / 110.03 ±0.82 / 111.05 ms │        109.61 / 110.77 ±1.47 / 113.62 ms │     no change │
│ QQuery 53 │        104.35 / 105.37 ±0.94 / 106.73 ms │        106.05 / 106.70 ±0.71 / 107.79 ms │     no change │
│ QQuery 54 │        150.47 / 151.62 ±0.91 / 153.08 ms │        151.38 / 154.20 ±1.57 / 155.59 ms │     no change │
│ QQuery 55 │        108.53 / 109.16 ±0.54 / 109.90 ms │        109.46 / 110.50 ±1.50 / 113.45 ms │     no change │
│ QQuery 56 │        142.34 / 143.70 ±1.56 / 146.62 ms │        142.64 / 145.56 ±1.74 / 147.82 ms │     no change │
│ QQuery 57 │        174.79 / 175.62 ±0.42 / 175.89 ms │        176.96 / 180.63 ±2.96 / 185.66 ms │     no change │
│ QQuery 58 │        299.64 / 310.90 ±7.98 / 320.51 ms │        387.61 / 395.90 ±8.53 / 410.80 ms │  1.27x slower │
│ QQuery 59 │        200.04 / 204.63 ±5.70 / 215.63 ms │        381.08 / 385.06 ±3.04 / 389.47 ms │  1.88x slower │
│ QQuery 60 │        145.13 / 146.29 ±1.01 / 147.92 ms │        147.54 / 149.30 ±1.35 / 151.71 ms │     no change │
│ QQuery 61 │        173.41 / 174.84 ±1.24 / 176.60 ms │        322.70 / 326.26 ±2.30 / 328.61 ms │  1.87x slower │
│ QQuery 62 │      925.94 / 982.11 ±45.10 / 1031.40 ms │       946.29 / 970.27 ±18.32 / 989.77 ms │     no change │
│ QQuery 63 │        106.22 / 107.45 ±1.43 / 110.18 ms │        107.73 / 110.83 ±2.73 / 114.17 ms │     no change │
│ QQuery 64 │        710.74 / 714.84 ±3.70 / 720.59 ms │    1378.78 / 1390.37 ±12.78 / 1413.96 ms │  1.95x slower │
│ QQuery 65 │        254.10 / 258.95 ±2.82 / 261.65 ms │        259.35 / 261.52 ±2.53 / 266.27 ms │     no change │
│ QQuery 66 │       232.55 / 254.65 ±15.58 / 278.45 ms │       248.78 / 269.87 ±14.56 / 290.52 ms │  1.06x slower │
│ QQuery 67 │        320.84 / 324.89 ±2.92 / 329.78 ms │        323.44 / 331.52 ±5.17 / 339.08 ms │     no change │
│ QQuery 68 │        284.99 / 289.20 ±3.07 / 293.12 ms │        289.91 / 291.88 ±2.00 / 295.52 ms │     no change │
│ QQuery 69 │        102.47 / 104.27 ±1.17 / 105.95 ms │        124.42 / 126.22 ±1.47 / 128.49 ms │  1.21x slower │
│ QQuery 70 │        336.64 / 348.92 ±6.85 / 356.50 ms │       351.19 / 363.98 ±13.68 / 385.64 ms │     no change │
│ QQuery 71 │        135.12 / 136.76 ±1.10 / 138.06 ms │        137.03 / 140.51 ±3.35 / 145.99 ms │     no change │
│ QQuery 72 │        731.73 / 734.55 ±2.75 / 738.76 ms │       739.00 / 747.82 ±10.02 / 761.12 ms │     no change │
│ QQuery 73 │        104.84 / 106.77 ±2.01 / 110.11 ms │        105.10 / 106.39 ±1.42 / 109.17 ms │     no change │
│ QQuery 74 │        603.27 / 618.14 ±9.09 / 626.11 ms │        703.05 / 719.18 ±8.45 / 727.77 ms │  1.16x slower │
│ QQuery 75 │        278.50 / 279.56 ±0.83 / 280.71 ms │        459.07 / 463.84 ±2.96 / 468.28 ms │  1.66x slower │
│ QQuery 76 │        136.48 / 137.87 ±1.50 / 140.59 ms │        134.83 / 136.95 ±1.69 / 139.25 ms │     no change │
│ QQuery 77 │        189.12 / 191.74 ±1.60 / 193.95 ms │        191.08 / 194.78 ±2.12 / 197.24 ms │     no change │
│ QQuery 78 │        358.00 / 361.08 ±2.37 / 364.07 ms │        403.52 / 410.05 ±4.93 / 416.53 ms │  1.14x slower │
│ QQuery 79 │        237.12 / 240.41 ±2.28 / 244.05 ms │        247.05 / 249.20 ±1.94 / 252.61 ms │     no change │
│ QQuery 80 │        335.73 / 338.63 ±2.43 / 341.64 ms │        334.68 / 338.93 ±3.58 / 342.79 ms │     no change │
│ QQuery 81 │           27.06 / 28.14 ±0.70 / 29.08 ms │           36.32 / 39.11 ±3.94 / 46.86 ms │  1.39x slower │
│ QQuery 82 │        204.22 / 208.16 ±2.43 / 211.60 ms │        202.19 / 204.13 ±2.40 / 208.58 ms │     no change │
│ QQuery 83 │           41.12 / 42.41 ±1.15 / 44.18 ms │           53.15 / 54.19 ±1.07 / 56.04 ms │  1.28x slower │
│ QQuery 84 │           50.42 / 50.93 ±0.42 / 51.68 ms │           50.34 / 51.44 ±0.84 / 52.77 ms │     no change │
│ QQuery 85 │        152.69 / 154.76 ±2.47 / 159.32 ms │        150.77 / 152.92 ±1.52 / 154.90 ms │     no change │
│ QQuery 86 │           40.10 / 41.19 ±0.66 / 42.10 ms │           39.91 / 41.07 ±0.71 / 42.08 ms │     no change │
│ QQuery 87 │           90.50 / 93.73 ±2.65 / 98.42 ms │        121.66 / 124.38 ±2.49 / 128.31 ms │  1.33x slower │
│ QQuery 88 │        101.54 / 103.01 ±1.18 / 105.08 ms │        591.65 / 594.48 ±2.25 / 597.31 ms │  5.77x slower │
│ QQuery 89 │        121.79 / 123.14 ±0.84 / 124.43 ms │        121.51 / 123.08 ±1.41 / 125.57 ms │     no change │
│ QQuery 90 │           24.53 / 25.91 ±1.76 / 29.37 ms │           39.46 / 41.40 ±1.13 / 42.51 ms │  1.60x slower │
│ QQuery 91 │           65.45 / 66.39 ±1.02 / 68.34 ms │           64.32 / 64.99 ±0.41 / 65.50 ms │     no change │
│ QQuery 92 │           60.06 / 61.32 ±1.47 / 64.18 ms │           59.09 / 60.33 ±0.91 / 61.92 ms │     no change │
│ QQuery 93 │        198.58 / 201.10 ±2.63 / 205.11 ms │        193.58 / 195.80 ±1.72 / 197.78 ms │     no change │
│ QQuery 94 │           62.83 / 64.21 ±0.76 / 65.11 ms │           63.29 / 63.85 ±0.53 / 64.85 ms │     no change │
│ QQuery 95 │        137.78 / 138.86 ±1.26 / 141.22 ms │        135.15 / 136.98 ±1.54 / 138.83 ms │     no change │
│ QQuery 96 │           75.14 / 76.17 ±0.63 / 77.12 ms │           71.72 / 74.60 ±1.58 / 76.56 ms │     no change │
│ QQuery 97 │        132.27 / 137.45 ±3.03 / 141.32 ms │        129.46 / 133.09 ±1.99 / 135.04 ms │     no change │
│ QQuery 98 │        156.44 / 161.32 ±2.95 / 165.35 ms │        155.30 / 158.06 ±2.39 / 161.30 ms │     no change │
│ QQuery 99 │ 10847.83 / 10901.62 ±39.37 / 10946.22 ms │ 10795.79 / 10850.41 ±37.54 / 10902.77 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 34450.98ms │
│ Total Time (lazy-coalesce-partitions)   │ 40120.45ms │
│ Average Time (HEAD)                     │   347.99ms │
│ Average Time (lazy-coalesce-partitions) │   405.26ms │
│ Queries Faster                          │          1 │
│ Queries Slower                          │         28 │
│ Queries with No Change                  │         70 │
│ Queries with Failure                    │          0 │
└─────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 172.6s
Peak memory 5.8 GiB
Avg memory 4.7 GiB
CPU user 276.2s
CPU sys 19.5s
Disk read 0 B
Disk write 707.5 MiB

tpcds — branch

Metric Value
Wall time 200.9s
Peak memory 5.7 GiB
Avg memory 4.4 GiB
CPU user 276.1s
CPU sys 20.9s
Disk read 0 B
Disk write 796.0 KiB

File an issue against this benchmark runner

@Dandandan
Copy link
Copy Markdown
Contributor Author

Dandandan commented Apr 2, 2026 •
edited
Loading

run benchmark tpcds10 tpch10

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4180178576-676-srgtw 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 lazy-coalesce-partitions (e2339fb) to 4084a18 (merge-base) diff using: tpch10
Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 Criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4180178576-675-csc82 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 lazy-coalesce-partitions (e2339fb) to 4084a18 (merge-base) diff
BENCH_NAME=tpcd10
BENCH_COMMAND=cargo bench --features=parquet --bench tpcd10
BENCH_FILTER=
Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

Benchmark for this request failed.

Last 20 lines of output:

Click to expand 
    stats_merge
    strpos
    struct_query_sql
    substr
    substr_index
    substring
    sum
    to_char
    to_hex
    to_timestamp
    topk_aggregate
    topk_repartition
    translate
    trim
    trunc
    unhex
    upper
    uuid
    window_query_sql
    with_hashes

File an issue against this benchmark runner

@Dandandan
Copy link
Copy Markdown
Contributor Author

Dandandan commented Apr 2, 2026

run benchmark tpcds10

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 Criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4180220240-677-hlmvg 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 lazy-coalesce-partitions (e2339fb) to 4084a18 (merge-base) diff
BENCH_NAME=tpcds10
BENCH_COMMAND=cargo bench --features=parquet --bench tpcds10
BENCH_FILTER=
Results will be posted here when complete

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

Benchmark for this request failed.

Last 20 lines of output:

Click to expand 
    stats_merge
    strpos
    struct_query_sql
    substr
    substr_index
    substring
    sum
    to_char
    to_hex
    to_timestamp
    topk_aggregate
    topk_repartition
    translate
    trim
    trunc
    unhex
    upper
    uuid
    window_query_sql
    with_hashes

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

adriangbot commented Apr 2, 2026

🤖 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 lazy-coalesce-partitions
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                               HEAD ┃           lazy-coalesce-partitions ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │  375.54 / 380.19 ±4.02 / 385.25 ms │  374.01 / 381.51 ±5.07 / 386.71 ms │    no change │
│ QQuery 2  │  139.11 / 143.43 ±2.45 / 146.08 ms │  142.51 / 146.69 ±2.60 / 150.67 ms │    no change │
│ QQuery 3  │  291.44 / 301.66 ±8.41 / 310.63 ms │  307.38 / 311.28 ±2.93 / 315.87 ms │    no change │
│ QQuery 4  │  155.68 / 157.66 ±2.09 / 161.64 ms │  158.59 / 163.28 ±2.89 / 166.34 ms │    no change │
│ QQuery 5  │ 433.43 / 457.81 ±12.53 / 468.98 ms │  476.18 / 482.11 ±8.52 / 498.95 ms │ 1.05x slower │
│ QQuery 6  │  133.20 / 134.93 ±1.00 / 136.12 ms │  140.06 / 140.99 ±0.82 / 142.51 ms │    no change │
│ QQuery 7  │  573.24 / 587.07 ±9.34 / 600.60 ms │ 590.60 / 612.64 ±18.48 / 640.42 ms │    no change │
│ QQuery 8  │ 475.53 / 498.44 ±17.22 / 522.40 ms │ 481.04 / 499.65 ±15.94 / 523.67 ms │    no change │
│ QQuery 9  │ 699.10 / 711.96 ±13.30 / 734.30 ms │ 691.48 / 707.61 ±11.99 / 726.40 ms │    no change │
│ QQuery 10 │ 342.21 / 357.16 ±12.60 / 377.59 ms │  337.50 / 349.70 ±8.68 / 364.39 ms │    no change │
│ QQuery 11 │  107.09 / 114.32 ±4.56 / 120.49 ms │  109.98 / 115.89 ±4.77 / 122.74 ms │    no change │
│ QQuery 12 │  211.31 / 215.33 ±4.36 / 222.89 ms │  208.04 / 212.86 ±3.48 / 218.13 ms │    no change │
│ QQuery 13 │ 306.02 / 327.74 ±13.99 / 346.79 ms │ 316.70 / 334.96 ±14.77 / 361.51 ms │    no change │
│ QQuery 14 │  184.76 / 189.95 ±3.70 / 195.79 ms │  186.58 / 193.78 ±4.26 / 199.06 ms │    no change │
│ QQuery 15 │  344.70 / 351.45 ±5.63 / 361.16 ms │  340.53 / 347.38 ±4.09 / 352.38 ms │    no change │
│ QQuery 16 │     84.03 / 86.69 ±2.29 / 89.46 ms │     83.18 / 90.16 ±4.26 / 95.32 ms │    no change │
│ QQuery 17 │ 768.73 / 793.22 ±18.43 / 813.46 ms │ 773.48 / 799.10 ±15.98 / 823.40 ms │    no change │
│ QQuery 18 │ 877.35 / 898.91 ±11.82 / 911.79 ms │ 881.69 / 902.54 ±10.92 / 912.13 ms │    no change │
│ QQuery 19 │  278.22 / 283.62 ±5.73 / 294.65 ms │ 271.59 / 282.25 ±12.53 / 305.74 ms │    no change │
│ QQuery 20 │  327.55 / 335.36 ±6.79 / 346.40 ms │  328.31 / 341.28 ±8.93 / 351.43 ms │    no change │
│ QQuery 21 │ 837.46 / 873.32 ±18.71 / 890.70 ms │ 856.88 / 886.85 ±22.27 / 918.29 ms │    no change │
│ QQuery 22 │     82.27 / 83.99 ±1.93 / 87.02 ms │     81.51 / 86.84 ±4.19 / 90.62 ms │    no change │
└───────────┴────────────────────────────────────┴────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                       ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                       │ 8284.21ms │
│ Total Time (lazy-coalesce-partitions)   │ 8389.33ms │
│ Average Time (HEAD)                     │  376.56ms │
│ Average Time (lazy-coalesce-partitions) │  381.33ms │
│ Queries Faster                          │         0 │
│ Queries Slower                          │         1 │
│ Queries with No Change                  │        21 │
│ Queries with Failure                    │         0 │
└─────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — base (merge-base)

Metric Value
Wall time 41.8s
Peak memory 10.0 GiB
Avg memory 7.7 GiB
CPU user 434.8s
CPU sys 31.6s
Disk read 0 B
Disk write 2.9 GiB

tpch10 — branch

Metric Value
Wall time 42.3s
Peak memory 10.0 GiB
Avg memory 7.7 GiB
CPU user 440.1s
CPU sys 32.5s
Disk read 0 B
Disk write 204.0 KiB

File an issue against this benchmark runner

@Dandandan Dandandan closed this Apr 3, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

physical-plan Changes to the physical-plan crate

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@Dandandan @adriangbot