Skip to content

refactor(hash-aggr): Migrate single mode hash aggregation#23408

Open
2010YOUY01 wants to merge 2 commits into
apache:mainfrom
2010YOUY01:split-aggr-single
Open

refactor(hash-aggr): Migrate single mode hash aggregation#23408
2010YOUY01 wants to merge 2 commits into
apache:mainfrom
2010YOUY01:split-aggr-single

Conversation

@2010YOUY01

@2010YOUY01 2010YOUY01 commented Jul 9, 2026

Copy link
Copy Markdown
Contributor

Which issue does this PR close?

Rationale for this change

This PRs implements the single aggregation (fused partial and final aggregation into a single physical plan operator). This mode can be planed if input is already hash/range partitioned.

See comments at datafusion/physical-plan/src/aggregates/single_stream.rs for details.

What changes are included in this PR?

  • Adds an AggregateHashTable<SingleMarker> variant to handle single aggregation.
  • Adds SingleHashAggregateStream that uses AggregateHashTable<SingleMarker> to implement the single aggregation state machine.

Are these changes tested?

Existing tests for functionalities + UT for planning

Are there any user-facing changes?

No

@github-actions github-actions Bot added the physical-plan Changes to the physical-plan crate label Jul 9, 2026
@2010YOUY01 2010YOUY01 requested a review from alamb July 9, 2026 07:01

@alamb alamb left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I hope to review this tomorrow

@alamb

alamb commented Jul 10, 2026

Copy link
Copy Markdown
Contributor

run benchmarks

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4938227863-960-kfnxb 6.12.85+ #1 SMP Mon May 11 08:17:35 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 split-aggr-single (ebdd966) to b790763 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4938227863-962-f9v2l 6.12.85+ #1 SMP Mon May 11 08:17:35 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 split-aggr-single (ebdd966) to b790763 (merge-base) diff using: tpch
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4938227863-961-p7rxj 6.12.85+ #1 SMP Mon May 11 08:17:35 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 split-aggr-single (ebdd966) to b790763 (merge-base) diff using: tpcds
Results will be posted here when complete


File an issue against this benchmark runner

@alamb alamb left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you @2010YOUY01 -- there are a few ignored results that I think are worth looking at, but otherwise this looks ready to go

cc @Rachelint

let baseline_metrics = BaselineMetrics::new(&agg.metrics, partition);

// Preserve the existing aggregate metric surface for this plan node.
let _spill_metrics = SpillMetrics::new(&agg.metrics, partition);

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This variable is ignored -- why bother to make it? Or is this a TODO when we implement spilling? If so, perhaps we can leave the TODO / link to more context

hash_table: &mut AggregateHashTable<SingleMarker>,
) -> Result<()> {
let input_schema = self.input.schema();
self.input = Box::pin(EmptyRecordBatchStream::new(input_schema));

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe worth a comment that this eagerly releases the input stream


match result {
Ok(Some(batch)) => {
let _ = self

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we should return the error if the reservation fails (rather than ignoring it), right (as is done above in

            if let Err(e) = self
                    .reservation
                    .try_resize(original_state.hash_table().memory_size())
                {
                    return ControlFlow::Break((
                        Poll::Ready(Some(Err(e))),
                        original_state,
                    ));
                }

@adriangbot

Copy link
Copy Markdown

🤖 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 split-aggr-single
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃              split-aggr-single ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 38.45 / 39.46 ±0.82 / 40.96 ms │ 38.02 / 39.39 ±1.43 / 41.56 ms │ no change │
│ QQuery 2  │ 19.26 / 19.67 ±0.46 / 20.52 ms │ 19.28 / 19.45 ±0.12 / 19.58 ms │ no change │
│ QQuery 3  │ 33.27 / 34.13 ±1.27 / 36.65 ms │ 31.46 / 33.06 ±0.88 / 33.98 ms │ no change │
│ QQuery 4  │ 17.54 / 17.84 ±0.44 / 18.72 ms │ 17.51 / 17.80 ±0.47 / 18.72 ms │ no change │
│ QQuery 5  │ 38.15 / 40.42 ±1.16 / 41.36 ms │ 37.97 / 40.24 ±1.15 / 41.00 ms │ no change │
│ QQuery 6  │ 16.10 / 16.37 ±0.19 / 16.68 ms │ 16.12 / 16.20 ±0.06 / 16.27 ms │ no change │
│ QQuery 7  │ 43.49 / 45.69 ±1.66 / 47.52 ms │ 44.93 / 45.98 ±0.86 / 47.43 ms │ no change │
│ QQuery 8  │ 43.17 / 43.29 ±0.11 / 43.49 ms │ 43.03 / 43.89 ±0.75 / 44.90 ms │ no change │
│ QQuery 9  │ 48.83 / 50.00 ±0.77 / 51.20 ms │ 48.98 / 50.52 ±0.89 / 51.51 ms │ no change │
│ QQuery 10 │ 41.98 / 42.23 ±0.19 / 42.52 ms │ 42.18 / 42.35 ±0.15 / 42.60 ms │ no change │
│ QQuery 11 │ 13.41 / 13.78 ±0.54 / 14.84 ms │ 13.38 / 13.74 ±0.46 / 14.62 ms │ no change │
│ QQuery 12 │ 23.90 / 24.66 ±0.98 / 26.56 ms │ 24.07 / 24.24 ±0.10 / 24.37 ms │ no change │
│ QQuery 13 │ 32.65 / 34.34 ±1.77 / 37.77 ms │ 32.45 / 34.40 ±1.85 / 37.65 ms │ no change │
│ QQuery 14 │ 23.78 / 24.20 ±0.57 / 25.33 ms │ 24.01 / 24.20 ±0.19 / 24.55 ms │ no change │
│ QQuery 15 │ 30.97 / 31.12 ±0.12 / 31.26 ms │ 30.92 / 31.16 ±0.26 / 31.60 ms │ no change │
│ QQuery 16 │ 13.97 / 14.19 ±0.13 / 14.36 ms │ 13.91 / 14.01 ±0.09 / 14.14 ms │ no change │
│ QQuery 17 │ 73.69 / 75.95 ±1.52 / 77.70 ms │ 74.09 / 75.87 ±1.67 / 78.09 ms │ no change │
│ QQuery 18 │ 60.22 / 61.25 ±0.59 / 62.01 ms │ 59.93 / 60.47 ±0.62 / 61.65 ms │ no change │
│ QQuery 19 │ 33.41 / 33.63 ±0.30 / 34.22 ms │ 33.06 / 33.37 ±0.23 / 33.75 ms │ no change │
│ QQuery 20 │ 32.09 / 32.55 ±0.60 / 33.71 ms │ 32.04 / 32.16 ±0.08 / 32.27 ms │ no change │
│ QQuery 21 │ 54.71 / 56.28 ±1.38 / 58.76 ms │ 54.81 / 56.05 ±0.90 / 57.29 ms │ no change │
│ QQuery 22 │ 14.01 / 14.33 ±0.51 / 15.35 ms │ 13.85 / 14.24 ±0.41 / 15.02 ms │ no change │
└───────────┴────────────────────────────────┴────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                │ 765.40ms │
│ Total Time (split-aggr-single)   │ 762.80ms │
│ Average Time (HEAD)              │  34.79ms │
│ Average Time (split-aggr-single) │  34.67ms │
│ Queries Faster                   │        0 │
│ Queries Slower                   │        0 │
│ Queries with No Change           │       22 │
│ Queries with Failure             │        0 │
└──────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric Value
Wall time 5.0s
Peak memory 1.2 GiB
Avg memory 520.8 MiB
CPU user 21.9s
CPU sys 1.8s
Peak spill 0 B

tpch — branch

Metric Value
Wall time 5.0s
Peak memory 1.2 GiB
Avg memory 506.8 MiB
CPU user 22.1s
CPU sys 1.6s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

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 split-aggr-single
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                     split-aggr-single ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.54 / 6.02 ±0.80 / 7.60 ms │           5.34 / 5.80 ±0.84 / 7.48 ms │     no change │
│ QQuery 2  │        81.98 / 82.20 ±0.18 / 82.49 ms │        80.70 / 81.04 ±0.20 / 81.26 ms │     no change │
│ QQuery 3  │        29.43 / 29.79 ±0.28 / 30.12 ms │        29.02 / 29.40 ±0.27 / 29.70 ms │     no change │
│ QQuery 4  │     495.62 / 499.16 ±2.80 / 503.72 ms │     494.36 / 497.07 ±2.07 / 499.96 ms │     no change │
│ QQuery 5  │        51.59 / 52.05 ±0.51 / 53.01 ms │        51.32 / 51.74 ±0.22 / 51.97 ms │     no change │
│ QQuery 6  │        36.60 / 37.16 ±0.32 / 37.46 ms │        36.57 / 36.96 ±0.36 / 37.57 ms │     no change │
│ QQuery 7  │       93.78 / 96.56 ±3.32 / 103.06 ms │        94.21 / 94.99 ±0.55 / 95.90 ms │     no change │
│ QQuery 8  │        36.90 / 37.86 ±0.70 / 38.86 ms │        36.56 / 38.35 ±1.83 / 41.73 ms │     no change │
│ QQuery 9  │        53.33 / 54.85 ±1.30 / 56.41 ms │        51.32 / 53.40 ±1.59 / 55.33 ms │     no change │
│ QQuery 10 │        63.76 / 64.24 ±0.47 / 65.06 ms │        63.94 / 64.36 ±0.41 / 65.13 ms │     no change │
│ QQuery 11 │     307.92 / 310.34 ±1.94 / 313.15 ms │     308.68 / 316.01 ±4.95 / 322.33 ms │     no change │
│ QQuery 12 │        28.83 / 29.23 ±0.36 / 29.76 ms │        28.99 / 29.27 ±0.16 / 29.43 ms │     no change │
│ QQuery 13 │     118.24 / 119.63 ±1.21 / 121.29 ms │     117.97 / 119.05 ±0.64 / 119.95 ms │     no change │
│ QQuery 14 │     410.54 / 412.84 ±1.24 / 414.27 ms │     411.09 / 414.57 ±2.79 / 417.64 ms │     no change │
│ QQuery 15 │        58.37 / 58.92 ±0.40 / 59.44 ms │        58.53 / 59.44 ±0.55 / 60.15 ms │     no change │
│ QQuery 16 │           6.58 / 6.74 ±0.22 / 7.18 ms │           6.47 / 6.62 ±0.20 / 7.02 ms │     no change │
│ QQuery 17 │        80.08 / 80.82 ±0.85 / 82.46 ms │        79.33 / 80.44 ±1.29 / 82.84 ms │     no change │
│ QQuery 18 │     125.06 / 125.52 ±0.45 / 126.36 ms │     123.84 / 125.67 ±2.47 / 130.39 ms │     no change │
│ QQuery 19 │        41.30 / 41.80 ±0.33 / 42.32 ms │        41.19 / 42.11 ±0.66 / 43.23 ms │     no change │
│ QQuery 20 │        35.76 / 36.58 ±0.78 / 37.96 ms │        35.57 / 35.96 ±0.49 / 36.89 ms │     no change │
│ QQuery 21 │        17.52 / 17.69 ±0.12 / 17.82 ms │        17.75 / 17.89 ±0.24 / 18.37 ms │     no change │
│ QQuery 22 │        62.14 / 62.87 ±1.04 / 64.91 ms │        62.88 / 63.47 ±0.73 / 64.73 ms │     no change │
│ QQuery 23 │     341.48 / 346.33 ±2.86 / 350.43 ms │     340.09 / 346.47 ±4.25 / 352.27 ms │     no change │
│ QQuery 24 │     223.18 / 226.50 ±3.10 / 232.13 ms │     223.97 / 228.00 ±3.22 / 232.44 ms │     no change │
│ QQuery 25 │     109.14 / 111.55 ±1.89 / 114.51 ms │     109.98 / 111.58 ±1.12 / 113.44 ms │     no change │
│ QQuery 26 │        57.53 / 58.33 ±0.67 / 59.37 ms │        57.91 / 58.97 ±1.40 / 61.74 ms │     no change │
│ QQuery 27 │           6.08 / 6.21 ±0.15 / 6.51 ms │          6.11 / 7.24 ±1.97 / 11.17 ms │  1.17x slower │
│ QQuery 28 │        56.95 / 60.61 ±1.84 / 61.85 ms │        56.21 / 60.37 ±2.18 / 62.52 ms │     no change │
│ QQuery 29 │       96.99 / 98.43 ±1.11 / 100.39 ms │       96.25 / 98.01 ±1.48 / 100.65 ms │     no change │
│ QQuery 30 │        33.27 / 34.89 ±2.57 / 40.00 ms │        32.20 / 33.52 ±1.56 / 36.54 ms │     no change │
│ QQuery 31 │     112.55 / 112.94 ±0.58 / 114.07 ms │     112.37 / 113.13 ±0.72 / 114.37 ms │     no change │
│ QQuery 32 │        20.42 / 20.66 ±0.25 / 21.06 ms │        20.32 / 20.79 ±0.37 / 21.35 ms │     no change │
│ QQuery 33 │        37.82 / 37.94 ±0.14 / 38.16 ms │        37.64 / 37.97 ±0.22 / 38.27 ms │     no change │
│ QQuery 34 │         9.61 / 11.85 ±3.08 / 17.91 ms │        10.30 / 11.53 ±1.81 / 15.08 ms │     no change │
│ QQuery 35 │        73.24 / 74.58 ±1.56 / 77.60 ms │        72.79 / 74.10 ±1.60 / 77.20 ms │     no change │
│ QQuery 36 │           5.72 / 5.83 ±0.17 / 6.17 ms │           5.70 / 5.82 ±0.18 / 6.17 ms │     no change │
│ QQuery 37 │           6.99 / 7.04 ±0.05 / 7.12 ms │           6.88 / 6.96 ±0.07 / 7.09 ms │     no change │
│ QQuery 38 │        62.27 / 62.96 ±0.57 / 63.87 ms │        62.37 / 63.26 ±0.74 / 64.18 ms │     no change │
│ QQuery 39 │        89.87 / 92.55 ±2.50 / 96.94 ms │        90.19 / 91.93 ±1.35 / 94.34 ms │     no change │
│ QQuery 40 │        23.45 / 23.86 ±0.28 / 24.27 ms │        24.18 / 24.25 ±0.05 / 24.32 ms │     no change │
│ QQuery 41 │        11.60 / 11.76 ±0.13 / 11.93 ms │        11.71 / 11.90 ±0.16 / 12.16 ms │     no change │
│ QQuery 42 │        23.94 / 24.48 ±0.80 / 26.06 ms │        24.01 / 24.55 ±0.68 / 25.78 ms │     no change │
│ QQuery 43 │           4.85 / 4.99 ±0.14 / 5.26 ms │           4.82 / 4.98 ±0.25 / 5.48 ms │     no change │
│ QQuery 44 │           9.16 / 9.36 ±0.21 / 9.69 ms │           9.13 / 9.21 ±0.06 / 9.30 ms │     no change │
│ QQuery 45 │        38.52 / 38.85 ±0.44 / 39.67 ms │        37.87 / 38.37 ±0.42 / 39.12 ms │     no change │
│ QQuery 46 │        11.56 / 12.04 ±0.26 / 12.37 ms │        11.60 / 11.90 ±0.25 / 12.35 ms │     no change │
│ QQuery 47 │     230.43 / 233.26 ±2.54 / 237.62 ms │     228.58 / 229.98 ±1.16 / 231.22 ms │     no change │
│ QQuery 48 │        96.36 / 97.37 ±0.73 / 98.61 ms │       95.81 / 97.79 ±1.69 / 100.46 ms │     no change │
│ QQuery 49 │        76.10 / 77.08 ±0.73 / 78.26 ms │        75.91 / 76.19 ±0.35 / 76.89 ms │     no change │
│ QQuery 50 │        58.81 / 62.23 ±4.28 / 70.57 ms │        58.79 / 59.19 ±0.31 / 59.64 ms │     no change │
│ QQuery 51 │        91.75 / 93.46 ±1.21 / 95.03 ms │        91.28 / 93.38 ±1.38 / 95.02 ms │     no change │
│ QQuery 52 │        23.92 / 24.34 ±0.22 / 24.56 ms │        24.07 / 24.35 ±0.15 / 24.47 ms │     no change │
│ QQuery 53 │        29.31 / 29.42 ±0.15 / 29.70 ms │        29.28 / 29.51 ±0.16 / 29.78 ms │     no change │
│ QQuery 54 │        56.03 / 57.67 ±2.43 / 62.51 ms │        55.35 / 58.03 ±2.24 / 60.87 ms │     no change │
│ QQuery 55 │        23.73 / 24.25 ±0.41 / 24.74 ms │        23.49 / 24.06 ±0.79 / 25.60 ms │     no change │
│ QQuery 56 │        39.33 / 39.81 ±0.34 / 40.26 ms │        39.04 / 39.41 ±0.35 / 39.86 ms │     no change │
│ QQuery 57 │     176.11 / 178.53 ±3.06 / 184.39 ms │     175.75 / 176.81 ±1.42 / 179.54 ms │     no change │
│ QQuery 58 │     114.54 / 115.48 ±0.75 / 116.58 ms │     114.20 / 115.38 ±0.94 / 116.90 ms │     no change │
│ QQuery 59 │     118.00 / 118.75 ±0.69 / 120.06 ms │     117.65 / 118.72 ±0.94 / 120.47 ms │     no change │
│ QQuery 60 │        39.78 / 40.14 ±0.40 / 40.86 ms │        39.11 / 40.05 ±0.80 / 41.34 ms │     no change │
│ QQuery 61 │        12.21 / 12.38 ±0.19 / 12.71 ms │        11.99 / 12.24 ±0.20 / 12.52 ms │     no change │
│ QQuery 62 │        46.63 / 48.48 ±2.50 / 53.38 ms │        46.61 / 47.74 ±1.75 / 51.20 ms │     no change │
│ QQuery 63 │        29.61 / 29.99 ±0.39 / 30.62 ms │        29.40 / 30.52 ±1.47 / 33.42 ms │     no change │
│ QQuery 64 │     408.34 / 415.70 ±4.66 / 422.96 ms │     409.09 / 414.50 ±4.75 / 421.82 ms │     no change │
│ QQuery 65 │     144.50 / 148.72 ±3.88 / 155.48 ms │     140.37 / 143.17 ±2.08 / 145.93 ms │     no change │
│ QQuery 66 │        80.82 / 81.88 ±1.09 / 83.82 ms │        80.67 / 82.36 ±2.43 / 87.09 ms │     no change │
│ QQuery 67 │     242.67 / 250.16 ±4.13 / 254.92 ms │     242.47 / 247.07 ±3.51 / 251.82 ms │     no change │
│ QQuery 68 │        11.87 / 12.07 ±0.19 / 12.43 ms │        11.68 / 11.90 ±0.20 / 12.24 ms │     no change │
│ QQuery 69 │        57.94 / 60.21 ±3.18 / 66.50 ms │        58.16 / 58.42 ±0.16 / 58.63 ms │     no change │
│ QQuery 70 │     106.48 / 108.30 ±1.54 / 110.94 ms │     105.84 / 111.54 ±8.12 / 127.50 ms │     no change │
│ QQuery 71 │        35.54 / 36.31 ±1.46 / 39.22 ms │        35.31 / 35.84 ±0.58 / 36.84 ms │     no change │
│ QQuery 72 │ 2053.94 / 2131.76 ±61.00 / 2242.16 ms │ 2112.27 / 2161.11 ±48.52 / 2233.87 ms │     no change │
│ QQuery 73 │           9.52 / 9.72 ±0.15 / 9.98 ms │          9.65 / 9.89 ±0.19 / 10.13 ms │     no change │
│ QQuery 74 │     173.04 / 176.63 ±3.30 / 182.13 ms │     172.03 / 175.09 ±4.20 / 183.36 ms │     no change │
│ QQuery 75 │     148.39 / 151.39 ±3.84 / 158.77 ms │     147.50 / 152.09 ±6.28 / 164.37 ms │     no change │
│ QQuery 76 │        35.40 / 35.80 ±0.44 / 36.66 ms │        35.47 / 35.71 ±0.15 / 35.92 ms │     no change │
│ QQuery 77 │        61.22 / 61.87 ±0.92 / 63.66 ms │        61.19 / 61.65 ±0.28 / 61.98 ms │     no change │
│ QQuery 78 │     195.96 / 202.23 ±9.04 / 220.12 ms │     198.03 / 201.87 ±6.25 / 214.33 ms │     no change │
│ QQuery 79 │        66.80 / 68.08 ±1.78 / 71.55 ms │        66.91 / 67.46 ±0.38 / 67.94 ms │     no change │
│ QQuery 80 │      99.80 / 101.26 ±1.48 / 103.27 ms │      99.15 / 103.05 ±3.66 / 107.90 ms │     no change │
│ QQuery 81 │        26.01 / 26.17 ±0.15 / 26.40 ms │        25.87 / 26.16 ±0.27 / 26.64 ms │     no change │
│ QQuery 82 │        16.44 / 16.56 ±0.10 / 16.69 ms │        16.49 / 16.65 ±0.12 / 16.84 ms │     no change │
│ QQuery 83 │        39.87 / 40.44 ±0.73 / 41.75 ms │        39.65 / 39.94 ±0.17 / 40.13 ms │     no change │
│ QQuery 84 │        30.37 / 32.55 ±3.84 / 40.23 ms │        30.45 / 30.50 ±0.06 / 30.62 ms │ +1.07x faster │
│ QQuery 85 │     107.04 / 108.90 ±1.15 / 110.43 ms │     107.96 / 111.93 ±7.52 / 126.96 ms │     no change │
│ QQuery 86 │        25.28 / 25.43 ±0.15 / 25.70 ms │        24.94 / 25.35 ±0.26 / 25.68 ms │     no change │
│ QQuery 87 │        62.39 / 63.96 ±2.50 / 68.91 ms │        61.67 / 62.55 ±0.62 / 63.40 ms │     no change │
│ QQuery 88 │        63.66 / 64.19 ±0.47 / 64.83 ms │        63.50 / 65.27 ±2.43 / 70.08 ms │     no change │
│ QQuery 89 │        35.88 / 36.57 ±0.37 / 36.97 ms │        35.77 / 36.36 ±0.50 / 37.15 ms │     no change │
│ QQuery 90 │        17.19 / 17.58 ±0.33 / 17.97 ms │        17.17 / 17.26 ±0.09 / 17.43 ms │     no change │
│ QQuery 91 │        46.12 / 46.53 ±0.38 / 47.22 ms │        46.75 / 47.18 ±0.27 / 47.46 ms │     no change │
│ QQuery 92 │        29.44 / 29.56 ±0.13 / 29.80 ms │        29.70 / 30.70 ±0.68 / 31.58 ms │     no change │
│ QQuery 93 │        49.54 / 50.69 ±1.14 / 52.14 ms │        49.81 / 52.06 ±1.97 / 55.44 ms │     no change │
│ QQuery 94 │        38.56 / 38.97 ±0.22 / 39.15 ms │        38.77 / 39.25 ±0.31 / 39.53 ms │     no change │
│ QQuery 95 │        79.95 / 80.72 ±0.54 / 81.50 ms │        80.45 / 81.30 ±0.56 / 81.95 ms │     no change │
│ QQuery 96 │        24.38 / 24.59 ±0.25 / 25.07 ms │        24.09 / 24.35 ±0.22 / 24.70 ms │     no change │
│ QQuery 97 │        46.96 / 48.64 ±2.87 / 54.37 ms │        46.55 / 48.71 ±3.14 / 54.90 ms │     no change │
│ QQuery 98 │        42.50 / 43.30 ±0.41 / 43.58 ms │        42.41 / 43.20 ±0.50 / 43.99 ms │     no change │
│ QQuery 99 │        70.72 / 71.03 ±0.32 / 71.64 ms │        70.34 / 70.71 ±0.22 / 71.04 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                │ 9958.55ms │
│ Total Time (split-aggr-single)   │ 9973.96ms │
│ Average Time (HEAD)              │  100.59ms │
│ Average Time (split-aggr-single) │  100.75ms │
│ Queries Faster                   │         1 │
│ Queries Slower                   │         1 │
│ Queries with No Change           │        97 │
│ Queries with Failure             │         0 │
└──────────────────────────────────┴───────────┘

Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 55.0s
Peak memory 2.1 GiB
Avg memory 1.4 GiB
CPU user 223.1s
CPU sys 5.7s
Peak spill 0 B

tpcds — branch

Metric Value
Wall time 55.0s
Peak memory 1.9 GiB
Avg memory 1.3 GiB
CPU user 228.0s
CPU sys 5.7s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

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 split-aggr-single
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃                      split-aggr-single ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.57 / 4.55 ±5.70 / 15.95 ms │           1.26 / 4.16 ±5.56 / 15.26 ms │ +1.09x faster │
│ QQuery 1  │         13.61 / 14.09 ±0.32 / 14.45 ms │         13.44 / 14.00 ±0.32 / 14.34 ms │     no change │
│ QQuery 2  │         37.15 / 37.63 ±0.31 / 38.00 ms │         36.57 / 37.08 ±0.33 / 37.45 ms │     no change │
│ QQuery 3  │         35.02 / 35.59 ±0.67 / 36.84 ms │         33.01 / 33.69 ±0.52 / 34.57 ms │ +1.06x faster │
│ QQuery 4  │      340.39 / 349.84 ±9.28 / 365.18 ms │      245.03 / 247.35 ±2.07 / 250.51 ms │ +1.41x faster │
│ QQuery 5  │     362.05 / 377.47 ±15.73 / 407.49 ms │      275.67 / 284.06 ±5.86 / 291.97 ms │ +1.33x faster │
│ QQuery 6  │            2.28 / 2.36 ±0.11 / 2.57 ms │            1.30 / 1.45 ±0.23 / 1.90 ms │ +1.62x faster │
│ QQuery 7  │         16.50 / 18.64 ±2.86 / 24.30 ms │         14.59 / 17.96 ±6.24 / 30.44 ms │     no change │
│ QQuery 8  │     476.67 / 502.89 ±17.23 / 525.24 ms │     355.34 / 394.63 ±23.20 / 427.10 ms │ +1.27x faster │
│ QQuery 9  │     669.90 / 698.88 ±16.17 / 714.18 ms │     519.99 / 547.06 ±16.41 / 568.07 ms │ +1.28x faster │
│ QQuery 10 │         84.12 / 87.49 ±1.70 / 88.73 ms │         78.32 / 78.86 ±0.39 / 79.52 ms │ +1.11x faster │
│ QQuery 11 │        87.57 / 95.17 ±5.83 / 105.52 ms │         88.25 / 93.86 ±3.82 / 99.39 ms │     no change │
│ QQuery 12 │     297.95 / 335.74 ±22.48 / 357.55 ms │      303.59 / 311.87 ±6.56 / 322.99 ms │ +1.08x faster │
│ QQuery 13 │     390.84 / 462.02 ±42.00 / 516.17 ms │     409.14 / 450.56 ±27.52 / 495.64 ms │     no change │
│ QQuery 14 │     305.99 / 323.00 ±14.04 / 345.30 ms │     320.99 / 351.55 ±39.48 / 418.76 ms │  1.09x slower │
│ QQuery 15 │     290.56 / 305.30 ±10.93 / 322.20 ms │     304.06 / 372.15 ±56.22 / 449.18 ms │  1.22x slower │
│ QQuery 16 │     652.08 / 678.41 ±18.11 / 703.56 ms │     771.54 / 810.54 ±31.93 / 852.89 ms │  1.19x slower │
│ QQuery 17 │     681.55 / 781.98 ±68.73 / 867.50 ms │     762.24 / 815.33 ±33.41 / 858.55 ms │     no change │
│ QQuery 18 │  1628.09 / 1749.42 ±75.84 / 1844.12 ms │ 1511.37 / 1745.27 ±130.18 / 1876.48 ms │     no change │
│ QQuery 19 │         33.02 / 33.67 ±0.55 / 34.50 ms │         31.85 / 35.23 ±5.05 / 45.23 ms │     no change │
│ QQuery 20 │     525.45 / 544.40 ±15.90 / 566.24 ms │     549.00 / 581.55 ±21.46 / 606.41 ms │  1.07x slower │
│ QQuery 21 │     525.85 / 547.37 ±23.99 / 584.59 ms │     564.54 / 590.44 ±20.09 / 617.26 ms │  1.08x slower │
│ QQuery 22 │  1056.31 / 1083.93 ±21.49 / 1119.59 ms │  1026.13 / 1068.55 ±43.05 / 1132.24 ms │     no change │
│ QQuery 23 │ 3151.60 / 3369.39 ±178.48 / 3573.96 ms │ 3236.53 / 3411.92 ±124.23 / 3598.93 ms │     no change │
│ QQuery 24 │         45.50 / 46.09 ±0.68 / 47.33 ms │        45.24 / 56.61 ±15.84 / 86.81 ms │  1.23x slower │
│ QQuery 25 │      119.05 / 121.63 ±3.05 / 125.99 ms │      120.56 / 127.36 ±4.08 / 132.57 ms │     no change │
│ QQuery 26 │         43.90 / 44.59 ±0.67 / 45.82 ms │         46.06 / 48.66 ±2.38 / 52.44 ms │  1.09x slower │
│ QQuery 27 │      683.29 / 698.09 ±7.70 / 705.65 ms │     705.92 / 732.80 ±15.16 / 748.44 ms │     no change │
│ QQuery 28 │  3148.77 / 3228.59 ±56.63 / 3321.55 ms │  3163.50 / 3279.70 ±73.65 / 3392.51 ms │     no change │
│ QQuery 29 │         41.68 / 48.19 ±8.44 / 63.31 ms │        43.07 / 56.03 ±14.95 / 80.25 ms │  1.16x slower │
│ QQuery 30 │      313.80 / 320.22 ±5.10 / 326.13 ms │     342.25 / 371.61 ±15.30 / 386.42 ms │  1.16x slower │
│ QQuery 31 │     298.25 / 311.05 ±10.72 / 327.56 ms │     306.98 / 334.31 ±22.23 / 374.13 ms │  1.07x slower │
│ QQuery 32 │  1077.36 / 1177.95 ±85.90 / 1330.74 ms │ 1040.16 / 1152.90 ±120.72 / 1382.67 ms │     no change │
│ QQuery 33 │  1583.64 / 1636.45 ±46.79 / 1712.33 ms │ 1793.76 / 2126.32 ±204.56 / 2349.46 ms │  1.30x slower │
│ QQuery 34 │  1587.18 / 1705.10 ±92.73 / 1841.90 ms │ 1717.09 / 1826.25 ±111.01 / 2036.59 ms │  1.07x slower │
│ QQuery 35 │     330.05 / 381.04 ±45.45 / 440.40 ms │     316.86 / 385.14 ±76.33 / 528.87 ms │     no change │
│ QQuery 36 │         76.04 / 90.49 ±7.52 / 96.87 ms │         71.72 / 79.45 ±7.39 / 89.48 ms │ +1.14x faster │
│ QQuery 37 │         37.14 / 41.15 ±3.26 / 46.66 ms │         37.71 / 39.02 ±0.85 / 40.07 ms │ +1.05x faster │
│ QQuery 38 │         45.87 / 48.71 ±2.89 / 53.25 ms │         42.57 / 48.21 ±4.77 / 55.79 ms │     no change │
│ QQuery 39 │     153.58 / 166.99 ±10.41 / 179.82 ms │      161.57 / 172.39 ±7.33 / 182.51 ms │     no change │
│ QQuery 40 │         14.74 / 15.52 ±0.54 / 16.32 ms │         15.00 / 20.38 ±6.88 / 32.25 ms │  1.31x slower │
│ QQuery 41 │         14.44 / 14.53 ±0.07 / 14.66 ms │         14.54 / 15.74 ±0.72 / 16.37 ms │  1.08x slower │
│ QQuery 42 │         13.69 / 15.95 ±3.82 / 23.57 ms │         14.71 / 16.85 ±3.86 / 24.56 ms │  1.06x slower │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                │ 22551.54ms │
│ Total Time (split-aggr-single)   │ 23188.85ms │
│ Average Time (HEAD)              │   524.45ms │
│ Average Time (split-aggr-single) │   539.28ms │
│ Queries Faster                   │         11 │
│ Queries Slower                   │         15 │
│ Queries with No Change           │         17 │
│ Queries with Failure             │          0 │
└──────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 115.0s
Peak memory 11.3 GiB
Avg memory 4.6 GiB
CPU user 1149.0s
CPU sys 89.8s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 120.0s
Peak memory 11.4 GiB
Avg memory 4.2 GiB
CPU user 1173.1s
CPU sys 99.7s
Peak spill 0 B

File an issue against this benchmark runner

@alamb

alamb commented Jul 10, 2026

Copy link
Copy Markdown
Contributor

run benchmark clickbench_partitioned

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4938865473-963-bcf7l 6.12.85+ #1 SMP Mon May 11 08:17:35 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 split-aggr-single (ebdd966) to b790763 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

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 split-aggr-single
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                      split-aggr-single ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.21 / 3.90 ±5.29 / 14.49 ms │           1.22 / 4.05 ±5.52 / 15.09 ms │     no change │
│ QQuery 1  │        12.83 / 13.17 ±0.20 / 13.45 ms │         12.45 / 12.79 ±0.19 / 12.98 ms │     no change │
│ QQuery 2  │        36.90 / 37.27 ±0.28 / 37.70 ms │         36.50 / 36.71 ±0.29 / 37.26 ms │     no change │
│ QQuery 3  │        30.91 / 31.85 ±1.12 / 33.96 ms │         31.13 / 33.77 ±1.96 / 36.39 ms │  1.06x slower │
│ QQuery 4  │     220.74 / 225.15 ±3.39 / 230.85 ms │      213.72 / 220.55 ±4.37 / 226.66 ms │     no change │
│ QQuery 5  │    271.17 / 297.32 ±13.72 / 309.32 ms │     273.27 / 299.84 ±13.34 / 307.88 ms │     no change │
│ QQuery 6  │           1.54 / 1.70 ±0.19 / 2.07 ms │            1.51 / 2.14 ±0.80 / 3.68 ms │  1.26x slower │
│ QQuery 7  │        15.19 / 15.30 ±0.11 / 15.52 ms │         14.93 / 15.15 ±0.15 / 15.38 ms │     no change │
│ QQuery 8  │     371.60 / 376.16 ±4.14 / 381.40 ms │     317.84 / 347.21 ±30.93 / 387.31 ms │ +1.08x faster │
│ QQuery 9  │     525.37 / 530.59 ±2.83 / 533.88 ms │      446.89 / 455.13 ±7.35 / 465.88 ms │ +1.17x faster │
│ QQuery 10 │        75.56 / 76.02 ±0.42 / 76.75 ms │         69.87 / 70.72 ±0.52 / 71.47 ms │ +1.08x faster │
│ QQuery 11 │        88.52 / 90.32 ±1.51 / 92.44 ms │         82.23 / 85.73 ±6.10 / 97.89 ms │ +1.05x faster │
│ QQuery 12 │    265.53 / 287.34 ±20.12 / 318.49 ms │      262.84 / 269.05 ±8.32 / 285.27 ms │ +1.07x faster │
│ QQuery 13 │    361.73 / 376.71 ±16.04 / 406.58 ms │     357.33 / 375.00 ±12.60 / 389.84 ms │     no change │
│ QQuery 14 │     278.94 / 283.65 ±4.71 / 292.65 ms │      281.32 / 286.38 ±4.34 / 292.49 ms │     no change │
│ QQuery 15 │     269.81 / 278.64 ±8.67 / 295.25 ms │     266.52 / 277.21 ±12.17 / 300.24 ms │     no change │
│ QQuery 16 │    606.85 / 621.71 ±16.56 / 650.81 ms │     609.71 / 632.32 ±23.84 / 672.56 ms │     no change │
│ QQuery 17 │    617.09 / 666.33 ±42.71 / 719.24 ms │      706.26 / 711.10 ±4.95 / 720.12 ms │  1.07x slower │
│ QQuery 18 │ 1250.84 / 1267.00 ±10.11 / 1277.98 ms │  1280.59 / 1362.40 ±57.15 / 1444.37 ms │  1.08x slower │
│ QQuery 19 │        28.02 / 28.29 ±0.17 / 28.53 ms │        27.93 / 38.07 ±14.74 / 67.09 ms │  1.35x slower │
│ QQuery 20 │    516.92 / 526.30 ±11.05 / 547.49 ms │     512.74 / 523.19 ±15.00 / 552.89 ms │     no change │
│ QQuery 21 │     513.73 / 517.86 ±5.26 / 527.52 ms │     517.64 / 528.94 ±10.46 / 543.18 ms │     no change │
│ QQuery 22 │  983.97 / 1014.99 ±31.01 / 1065.61 ms │    986.51 / 996.36 ±13.61 / 1022.34 ms │     no change │
│ QQuery 23 │ 3069.28 / 3125.19 ±68.88 / 3252.82 ms │  3078.84 / 3127.54 ±54.04 / 3218.58 ms │     no change │
│ QQuery 24 │        41.05 / 41.26 ±0.15 / 41.49 ms │         42.33 / 43.66 ±2.16 / 47.96 ms │  1.06x slower │
│ QQuery 25 │     111.10 / 113.42 ±2.94 / 119.08 ms │      111.20 / 116.37 ±7.48 / 131.09 ms │     no change │
│ QQuery 26 │        41.32 / 42.33 ±0.78 / 43.35 ms │         41.47 / 42.51 ±1.30 / 44.94 ms │     no change │
│ QQuery 27 │    670.57 / 680.42 ±10.84 / 694.98 ms │      672.08 / 680.23 ±5.31 / 685.32 ms │     no change │
│ QQuery 28 │ 3071.59 / 3138.93 ±65.17 / 3246.84 ms │  3049.43 / 3083.16 ±35.14 / 3150.57 ms │     no change │
│ QQuery 29 │       41.16 / 50.71 ±11.83 / 71.58 ms │        41.15 / 51.15 ±16.29 / 83.21 ms │     no change │
│ QQuery 30 │     300.83 / 306.86 ±4.98 / 314.01 ms │      295.79 / 303.09 ±5.94 / 312.21 ms │     no change │
│ QQuery 31 │     288.95 / 295.52 ±3.82 / 300.02 ms │     283.62 / 297.91 ±13.66 / 323.56 ms │     no change │
│ QQuery 32 │  948.17 / 1018.08 ±45.62 / 1084.93 ms │   986.16 / 1077.76 ±69.71 / 1172.82 ms │  1.06x slower │
│ QQuery 33 │ 1450.93 / 1497.93 ±45.89 / 1575.42 ms │ 1467.79 / 1622.52 ±107.77 / 1735.12 ms │  1.08x slower │
│ QQuery 34 │ 1483.84 / 1554.83 ±55.02 / 1645.78 ms │  1509.48 / 1572.09 ±82.76 / 1735.19 ms │     no change │
│ QQuery 35 │    279.09 / 294.88 ±26.01 / 346.55 ms │     278.14 / 303.38 ±22.60 / 338.04 ms │     no change │
│ QQuery 36 │        68.95 / 74.39 ±4.36 / 80.00 ms │         66.81 / 74.08 ±5.80 / 82.61 ms │     no change │
│ QQuery 37 │        35.37 / 36.05 ±0.49 / 36.68 ms │         35.47 / 38.51 ±2.75 / 42.06 ms │  1.07x slower │
│ QQuery 38 │        39.88 / 44.27 ±3.75 / 49.11 ms │         41.92 / 42.92 ±0.76 / 43.79 ms │     no change │
│ QQuery 39 │     141.01 / 151.25 ±6.43 / 159.67 ms │      150.78 / 157.82 ±5.30 / 165.90 ms │     no change │
│ QQuery 40 │        14.01 / 14.28 ±0.27 / 14.68 ms │         13.76 / 14.58 ±0.73 / 15.56 ms │     no change │
│ QQuery 41 │        13.75 / 15.00 ±1.92 / 18.80 ms │         13.54 / 13.76 ±0.15 / 13.96 ms │ +1.09x faster │
│ QQuery 42 │        12.95 / 14.24 ±2.03 / 18.28 ms │         13.01 / 14.23 ±2.14 / 18.51 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                │ 20077.40ms │
│ Total Time (split-aggr-single)   │ 20261.08ms │
│ Average Time (HEAD)              │   466.92ms │
│ Average Time (split-aggr-single) │   471.19ms │
│ Queries Faster                   │          6 │
│ Queries Slower                   │          9 │
│ Queries with No Change           │         28 │
│ Queries with Failure             │          0 │
└──────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 105.0s
Peak memory 12.1 GiB
Avg memory 4.1 GiB
CPU user 1029.1s
CPU sys 72.6s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 105.0s
Peak memory 10.3 GiB
Avg memory 4.1 GiB
CPU user 1034.5s
CPU sys 75.9s
Peak spill 0 B

File an issue against this benchmark runner

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

Labels

physical-plan Changes to the physical-plan crate

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants