Skip to content

DRAFT Test aggmetrics3 - #25237

Draft
kosiew wants to merge 35 commits into
apache:mainfrom
kosiew:test-aggmetrics3
Draft

DRAFT Test aggmetrics3#25237
kosiew wants to merge 35 commits into
apache:mainfrom
kosiew:test-aggmetrics3

Conversation

@kosiew

@kosiew kosiew commented Sep 12, 2026

Copy link
Copy Markdown
Contributor

Test aggmetrics
benchmark runs for #25051

- Added harness implementation:
- `benchmarks/sql_benchmarks/array_agg_distinct/array_agg_distinct.suite` – defines the benchmark suite, test parameters, and execution configuration for the data‑free SQL harness.
- `benchmarks/sql_benchmarks/array_agg_distinct/benchmarks/q01.benchmark` – contains the specific query benchmark (`q01`) that exercises the `array_agg(DISTINCT …)` workload without requiring any input data.

- Workload characteristics:
- Simulates **2 M range rows** → **1 M groups**.
- Each group contains **2 rows** with **2 distinct values**, providing a realistic yet data‑free test scenario for aggregation performance.
- add bench.sh wrapper for array_agg_distinct
Reduce the data-free grouped array_agg(DISTINCT) workload while preserving its two-rows-per-group and two-distinct-values-per-group shape.
@github-actions github-actions Bot added documentation Improvements or additions to documentation logical-expr Logical plan and expressions physical-expr Changes to the physical-expr crates functions Changes to functions implementation physical-plan Changes to the physical-plan crate labels Sep 12, 2026
@kosiew

kosiew commented Sep 12, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark sql

env:
  CARGO_BUILD_JOBS: 1
  BENCH_NAME: array_agg_distinct
  BENCH_QUERY: 1
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

@kosiew

kosiew commented Sep 12, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark groups_accumulator_adapter nth_value percentile_cont

env:
  CARGO_BUILD_JOBS: 1
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5646609656-2324-7rh25 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark sql
env:
  BENCH_NAME: "array_agg_distinct"
  BENCH_QUERY: "1"
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

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-c5646624113-2327-mb5nk 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark percentile_cont
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

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-c5646624113-2326-bhp7q 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark nth_value
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

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-c5646624113-2325-dcmr6 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark groups_accumulator_adapter
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark groups_accumulator_adapter
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"
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

group                                                  HEAD                                   test-aggmetrics3
-----                                                  ----                                   ----------------
groups_accumulator_adapter/update_batch/8192_groups    1.00   1374.0±3.47µs        ? ?/sec    1.02   1398.7±3.32µs        ? ?/sec

Resource Usage

groups_accumulator_adapter — base (merge-base)

Metric Value
Wall time 210.0s
Peak memory 7.8 MiB
Avg memory 515.6 KiB
CPU user 12.1s
CPU sys 0.0s
Peak spill 0 B

groups_accumulator_adapter — branch

Metric Value
Wall time 160.0s
Peak memory 7.9 MiB
Avg memory 676.9 KiB
CPU user 12.7s
CPU sys 0.0s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark percentile_cont
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"
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

group                                                              HEAD                                   test-aggmetrics3
-----                                                              ----                                   ----------------
approx_percentile_cont/groups_accumulator_adapter/8192_groups      1.00      3.8±0.05ms        ? ?/sec    1.01      3.8±0.02ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=16384      1.00      2.0±0.00ms        ? ?/sec    1.00      2.0±0.00ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=256        1.00    165.9±0.37µs        ? ?/sec    1.00    166.6±0.36µs        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=4096       1.00    618.9±0.58µs        ? ?/sec    1.00    620.3±1.43µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=16384    1.00   1415.1±1.56µs        ? ?/sec    1.00   1415.9±1.21µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=256      1.00    165.9±0.26µs        ? ?/sec    1.01    166.9±0.26µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=4096     1.00    480.3±0.78µs        ? ?/sec    1.00    479.9±0.43µs        ? ?/sec

Resource Usage

percentile_cont — base (merge-base)

Metric Value
Wall time 350.1s
Peak memory 43.2 MiB
Avg memory 6.1 MiB
CPU user 88.7s
CPU sys 0.2s
Peak spill 0 B

percentile_cont — branch

Metric Value
Wall time 435.1s
Peak memory 41.3 MiB
Avg memory 4.6 MiB
CPU user 88.7s
CPU sys 0.2s
Peak spill 0 B

File an issue against this benchmark runner

@codecov-commenter

codecov-commenter commented Sep 12, 2026

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 88.15642% with 106 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.92%. Comparing base (c2cf289) to head (d30fe36).
⚠️ Report is 5 commits behind head on main.

Files with missing lines Patch % Lines
...gregate-common/src/aggregate/groups_accumulator.rs 80.82% 43 Missing and 13 partials ⚠️
...n/physical-plan/src/aggregates/aggregate_stream.rs 88.15% 6 Missing and 12 partials ⚠️
...hysical-plan/src/aggregates/grouped_hash_stream.rs 86.95% 7 Missing and 11 partials ⚠️
datafusion/functions-aggregate/src/array_agg.rs 91.91% 0 Missing and 8 partials ⚠️
...plan/src/aggregates/aggregate_hash_table/common.rs 80.00% 0 Missing and 4 partials ⚠️
datafusion/physical-expr/src/aggregate.rs 93.75% 0 Missing and 1 partial ⚠️
.../aggregates/aggregate_hash_table/common_ordered.rs 88.88% 0 Missing and 1 partial ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main   #25237      +/-   ##
==========================================
+ Coverage   81.90%   81.92%   +0.01%     
==========================================
  Files        1134     1134              
  Lines      425261   426481    +1220     
  Branches   425261   426481    +1220     
==========================================
+ Hits       348325   349398    +1073     
- Misses      56295    56366      +71     
- Partials    20641    20717      +76     

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

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

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark nth_value
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"
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

group                                                         HEAD                                    test-aggmetrics3
-----                                                         ----                                    ----------------
nth_value/ordered/n=-1                                        1.00  1344.7±217.43ns        ? ?/sec    1.00  1349.1±179.97ns        ? ?/sec
nth_value/ordered/n=-10                                       1.00     10.6±0.08µs        ? ?/sec     1.01     10.6±0.18µs        ? ?/sec
nth_value/ordered/n=-100                                      1.00    102.5±0.47µs        ? ?/sec     1.01    103.6±0.36µs        ? ?/sec
nth_value/ordered/n=1                                         1.01  421.3±598.06ns        ? ?/sec     1.00  417.5±620.92ns        ? ?/sec
nth_value/ordered/n=10                                        1.00  1590.5±167.48ns        ? ?/sec    1.01  1612.9±472.10ns        ? ?/sec
nth_value/ordered/n=100                                       1.00     12.7±0.12µs        ? ?/sec     1.00     12.6±0.05µs        ? ?/sec
nth_value/ordered_groups_accumulator_adapter/8192_groups      1.00      8.0±0.05ms        ? ?/sec     1.02      8.2±0.04ms        ? ?/sec
nth_value/trivial/n=-1                                        1.00    664.2±5.55ns        ? ?/sec     1.00    666.0±5.53ns        ? ?/sec
nth_value/trivial/n=-10                                       1.00      5.2±0.02µs        ? ?/sec     1.00      5.2±0.02µs        ? ?/sec
nth_value/trivial/n=-100                                      1.00     48.9±0.13µs        ? ?/sec     1.01     49.6±0.16µs        ? ?/sec
nth_value/trivial/n=1                                         1.00    123.1±0.75ns        ? ?/sec     1.01    124.8±0.86ns        ? ?/sec
nth_value/trivial/n=10                                        1.00    658.5±2.63ns        ? ?/sec     1.01    665.2±3.80ns        ? ?/sec
nth_value/trivial/n=100                                       1.00      5.3±0.01µs        ? ?/sec     1.01      5.3±0.02µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/0%_nulls         1.01    683.4±6.81µs        ? ?/sec     1.00    676.9±8.65µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/30%_nulls        1.01   1051.4±5.98µs        ? ?/sec     1.00   1044.5±5.25µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/50%_nulls        1.02   1054.8±6.32µs        ? ?/sec     1.00   1036.9±5.17µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/80%_nulls        1.05  1094.8±31.24µs        ? ?/sec     1.00   1043.6±4.92µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/0%_nulls       1.01   681.4±15.20µs        ? ?/sec     1.00   675.5±13.67µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/30%_nulls      1.00   1016.4±8.95µs        ? ?/sec     1.00   1011.7±4.10µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/50%_nulls      1.01   1015.2±7.94µs        ? ?/sec     1.00   1000.8±4.95µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/80%_nulls      1.05  1062.1±32.51µs        ? ?/sec     1.00   1008.5±4.09µs        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/0%_nulls          1.02    687.1±6.14µs        ? ?/sec     1.00    675.6±8.70µs        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/30%_nulls         1.00     16.8±0.03ms        ? ?/sec     1.00     16.7±0.01ms        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/50%_nulls         1.00     11.5±0.02ms        ? ?/sec     1.00     11.5±0.01ms        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/80%_nulls         1.01      4.5±0.02ms        ? ?/sec     1.00      4.5±0.00ms        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/0%_nulls        1.01   680.6±14.99µs        ? ?/sec     1.00   674.6±13.69µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/30%_nulls       1.00   1412.7±8.16µs        ? ?/sec     1.00   1406.7±3.92µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/50%_nulls       1.01   1304.5±7.32µs        ? ?/sec     1.00   1291.2±4.07µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/80%_nulls       1.04  1163.6±34.17µs        ? ?/sec     1.00   1119.1±3.93µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/0%_nulls        1.01    684.1±6.12µs        ? ?/sec     1.00    677.8±8.38µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/30%_nulls       1.01   1076.7±5.25µs        ? ?/sec     1.00   1068.5±4.04µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/50%_nulls       1.01   1079.1±4.76µs        ? ?/sec     1.00   1064.0±4.14µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/80%_nulls       1.05  1120.5±32.18µs        ? ?/sec     1.00   1069.4±4.33µs        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/0%_nulls     1.01    683.7±6.18µs        ? ?/sec     1.00    677.7±8.56µs        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/30%_nulls    1.00     13.8±0.07ms        ? ?/sec     1.03     14.3±0.10ms        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/50%_nulls    1.00      9.3±0.14ms        ? ?/sec     1.00      9.4±0.12ms        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/80%_nulls    1.01      4.1±0.04ms        ? ?/sec     1.00      4.0±0.01ms        ? ?/sec
nth_value_nulls_comparison/first_value/ignore_nulls           1.00   1044.2±6.15µs        ? ?/sec     1.00   1041.3±4.86µs        ? ?/sec
nth_value_nulls_comparison/first_value/respect_nulls          1.05   124.0±12.65µs        ? ?/sec     1.00    118.3±5.70µs        ? ?/sec
nth_value_nulls_comparison/nth_value_10/ignore_nulls          1.01   1068.4±5.19µs        ? ?/sec     1.00   1062.8±3.79µs        ? ?/sec
nth_value_nulls_comparison/nth_value_10/respect_nulls         1.06   121.6±13.44µs        ? ?/sec     1.00    114.9±1.19µs        ? ?/sec

Resource Usage

nth_value — base (merge-base)

Metric Value
Wall time 765.2s
Peak memory 290.2 MiB
Avg memory 19.7 MiB
CPU user 526.2s
CPU sys 1.0s
Peak spill 0 B

nth_value — branch

Metric Value
Wall time 685.2s
Peak memory 293.6 MiB
Avg memory 22.2 MiB
CPU user 520.9s
CPU sys 1.0s
Peak spill 0 B

File an issue against this benchmark runner

@kosiew

kosiew commented Sep 12, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark groups_accumulator_adapter

env:
  CARGO_BUILD_JOBS: 1
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5646783600-2328-lrd9x 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark groups_accumulator_adapter
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark groups_accumulator_adapter
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"
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

group                                                  HEAD                                   test-aggmetrics3
-----                                                  ----                                   ----------------
groups_accumulator_adapter/update_batch/8192_groups    1.00   1371.6±3.96µs        ? ?/sec    1.02   1395.5±3.16µs        ? ?/sec

Resource Usage

groups_accumulator_adapter — base (merge-base)

Metric Value
Wall time 155.0s
Peak memory 7.8 MiB
Avg memory 697.6 KiB
CPU user 12.5s
CPU sys 0.0s
Peak spill 0 B

groups_accumulator_adapter — branch

Metric Value
Wall time 145.0s
Peak memory 23.9 MiB
Avg memory 859.6 KiB
CPU user 15.2s
CPU sys 0.0s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark sql
env:
  BENCH_NAME: "array_agg_distinct"
  BENCH_QUERY: "1"
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"
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

group                     HEAD                                   test-aggmetrics3
-----                     ----                                   ----------------
array_agg_distinct/Q01    1.00     27.8±0.44ms        ? ?/sec    1.02     28.2±1.00ms        ? ?/sec

Resource Usage

sql — base (merge-base)

Metric Value
Wall time 3015.7s
Peak memory 586.6 MiB
Avg memory 1.7 MiB
CPU user 72.9s
CPU sys 2.5s
Peak spill 0 B

sql — branch

Metric Value
Wall time 3460.8s
Peak memory 630.0 MiB
Avg memory 1.6 MiB
CPU user 79.5s
CPU sys 2.7s
Peak spill 0 B

File an issue against this benchmark runner

@kosiew

kosiew commented Sep 13, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark sql

env:
  BENCH_NAME: "array_agg_distinct"
  BENCH_QUERY: "1"
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5652940396-2329-jj68d 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 e4b81f9 (e4b81f9) to 2bc3a3f diff

Run configuration
run benchmark sql
env:
  BENCH_NAME: "array_agg_distinct"
  BENCH_QUERY: "1"
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "e4b81f9"

Results will be posted here when complete


File an issue against this benchmark runner

@kosiew

kosiew commented Sep 13, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark groups_accumulator_adapter nth_value percentile_cont

env:
  CARGO_BUILD_JOBS: 1
baseline:
  ref: "2bc3a3f"
changed:
  ref: "9b4e95b"

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5653353544-2332-tjcbl 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 9b4e95b (9b4e95b) to 2bc3a3f diff

Run configuration
run benchmark percentile_cont
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "9b4e95b"

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-c5653353544-2331-ms9zk 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 9b4e95b (9b4e95b) to 2bc3a3f diff

Run configuration
run benchmark nth_value
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "9b4e95b"

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-c5653353544-2330-gfl8x 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 9b4e95b (9b4e95b) to 2bc3a3f diff

Run configuration
run benchmark groups_accumulator_adapter
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "9b4e95b"

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing 9b4e95b (9b4e95b) to 2bc3a3f diff

Run configuration
run benchmark groups_accumulator_adapter
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "9b4e95b"
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

group                                                  HEAD                                   test-aggmetrics3
-----                                                  ----                                   ----------------
groups_accumulator_adapter/update_batch/8192_groups    1.00   1381.5±5.82µs        ? ?/sec    1.10   1516.1±4.51µs        ? ?/sec

Resource Usage

groups_accumulator_adapter — base (merge-base)

Metric Value
Wall time 190.0s
Peak memory 23.8 MiB
Avg memory 681.3 KiB
CPU user 14.6s
CPU sys 0.0s
Peak spill 0 B

groups_accumulator_adapter — branch

Metric Value
Wall time 195.0s
Peak memory 16.5 MiB
Avg memory 522.2 KiB
CPU user 11.1s
CPU sys 0.0s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing 9b4e95b (9b4e95b) to 2bc3a3f diff

Run configuration
run benchmark percentile_cont
env:
  CARGO_BUILD_JOBS: "1"
baseline:
  ref: "2bc3a3f"
changed:
  ref: "9b4e95b"
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

group                                                              HEAD                                   test-aggmetrics3
-----                                                              ----                                   ----------------
approx_percentile_cont/groups_accumulator_adapter/8192_groups      1.00      3.8±0.02ms        ? ?/sec    1.13      4.3±0.01ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=16384      1.00      2.0±0.00ms        ? ?/sec    1.00      2.0±0.00ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=256        1.00    165.8±0.40µs        ? ?/sec    1.00    166.1±0.30µs        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=4096       1.00    619.0±0.80µs        ? ?/sec    1.00    618.0±1.37µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=16384    1.00   1414.8±1.13µs        ? ?/sec    1.00   1414.9±1.25µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=256      1.00    166.5±0.47µs        ? ?/sec    1.00    166.6±0.47µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=4096     1.00    479.9±0.33µs        ? ?/sec    1.00    479.9±0.55µs        ? ?/sec

Resource Usage

percentile_cont — base (merge-base)

Metric Value
Wall time 280.1s
Peak memory 40.7 MiB
Avg memory 7.2 MiB
CPU user 90.5s
CPU sys 0.2s
Peak spill 0 B

percentile_cont — branch

Metric Value
Wall time 405.1s
Peak memory 39.6 MiB
Avg memory 4.9 MiB
CPU user 89.1s
CPU sys 0.2s
Peak spill 0 B

File an issue against this benchmark runner

…etric-count regressions

- array_agg(DISTINCT) merge: one metric timer/state batch.
- convert_to_state: untimed row updates + one conversion timer.
- Added metric-count regressions.
…ng, bound prep chunks, time only accumulator calls, and emit one metric update per logical batch

- Preps slice/filter outside submetric timing.
- Times only accumulator calls.
- Bounded 64‑group prep chunks; avoids unbounded retained arrays.
- One metric update per logical batch.
…pped batch updates and adds regression test

**Details:**
- Bounded 64‑row prep chunks for `convert_to_state`.
- Timer only wraps `update_batch_grouped`.
- Excludes factory, slice/filter, `state/evaluate`, and result materialization steps.
- Single metric update / logical conversion batch per operation.
- Added RED→GREEN filtered multi‑row state‑materialization regression test.
…s, import AggregateMetrics

- **Clear conditional metric lookup** – Refactored the conditional logic that retrieves metrics, removing redundant checks and making the code path easier to follow.
- **Merged adjacent metrics impl blocks** – Consolidated neighboring implementation blocks that handled similar metric types, reducing duplication and improving maintainability.
- **Imported AggregateMetrics; removed qualified repeats** – Added the `AggregateMetrics` import and eliminated unnecessary fully‑qualified references throughout the codebase.
… propagate to accumulators

- Updated `common.rs` to introduce a helper that constructs an `AggregateMetrics` owned by the execution context.
- The helper returns an `Arc<dyn AggregateMetrics>` which is now passed to both the initial and replacement accumulator instances.
- Ensures consistent metric aggregation ownership across accumulator updates, resolving the blocker.
- Internal submetrics may overlap enclosing phase timers.
- They are supplementary diagnostics.
- Do not add them as phase‑time breakdowns.
…handling in GroupsAccumulatorAdapter

The `sizes_pre` accumulation is now performed before the timer starts,
preventing adapter work from being measured as part of the aggregate
duration. The inner accumulation loop now breaks early on error and uses
borrowed references, improving error propagation and reducing unnecessary copies.
- Added `groups_accumulator.rs` regression test file.
- Introduced a sleeping `SlowSizeAccumulator` that delays `size()` by 50 ms.
- Asserts that the grouped metric excludes the 50 ms `size()` delay, confirming correct grouping behavior.
… add retry regression

- Clear successful groups before propagating later-group error to prevent contaminating subsequent error handling.
- Added retry regression test to catch duplicate update scenarios.
- GroupsAccumulatorAdapter: uninstrumented aggregates skip chunk/Vec path.
- Removed dead `time_grouped_update`.
- Timed submetric path unchanged.
- `convert_to_state` unchanged.
@kosiew

kosiew commented Sep 14, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark groups_accumulator_adapter nth_value percentile_cont

baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5665428883-2352-pkk8c 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark groups_accumulator_adapter
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

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-c5665428883-2353-rmclb 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark nth_value
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

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-c5665428883-2354-rlr6j 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark percentile_cont
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark groups_accumulator_adapter
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"
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

group                                                  HEAD                                   test-aggmetrics3
-----                                                  ----                                   ----------------
groups_accumulator_adapter/update_batch/8192_groups    1.00   1379.1±3.65µs        ? ?/sec    1.03   1420.9±6.31µs        ? ?/sec

Resource Usage

groups_accumulator_adapter — base (merge-base)

Metric Value
Wall time 115.0s
Peak memory 7.7 MiB
Avg memory 930.6 KiB
CPU user 12.6s
CPU sys 0.0s
Peak spill 0 B

groups_accumulator_adapter — branch

Metric Value
Wall time 120.0s
Peak memory 24.5 MiB
Avg memory 1.0 MiB
CPU user 14.6s
CPU sys 0.0s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark percentile_cont
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"
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

group                                                              HEAD                                   test-aggmetrics3
-----                                                              ----                                   ----------------
approx_percentile_cont/groups_accumulator_adapter/8192_groups      1.00      3.9±0.01ms        ? ?/sec    1.01      4.0±0.02ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=16384      1.00      2.0±0.00ms        ? ?/sec    1.00      2.0±0.00ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=256        1.00    166.2±0.43µs        ? ?/sec    1.00    166.5±0.40µs        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=4096       1.00    620.5±1.28µs        ? ?/sec    1.00    620.6±0.34µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=16384    1.00   1415.0±0.96µs        ? ?/sec    1.00   1416.1±1.30µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=256      1.00    165.9±0.22µs        ? ?/sec    1.00    166.5±0.32µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=4096     1.00    481.0±0.51µs        ? ?/sec    1.00    480.8±0.64µs        ? ?/sec

Resource Usage

percentile_cont — base (merge-base)

Metric Value
Wall time 190.0s
Peak memory 51.2 MiB
Avg memory 11.3 MiB
CPU user 91.4s
CPU sys 0.2s
Peak spill 0 B

percentile_cont — branch

Metric Value
Wall time 305.1s
Peak memory 44.1 MiB
Avg memory 7.2 MiB
CPU user 89.0s
CPU sys 0.2s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark nth_value
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"
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

group                                                         HEAD                                    test-aggmetrics3
-----                                                         ----                                    ----------------
nth_value/ordered/n=-1                                        1.00  1337.7±226.64ns        ? ?/sec    1.00  1341.5±177.47ns        ? ?/sec
nth_value/ordered/n=-10                                       1.00     10.6±0.08µs        ? ?/sec     1.01     10.7±0.09µs        ? ?/sec
nth_value/ordered/n=-100                                      1.00    103.3±0.46µs        ? ?/sec     1.00    103.6±0.51µs        ? ?/sec
nth_value/ordered/n=1                                         1.00  422.4±672.82ns        ? ?/sec     1.02  430.0±669.59ns        ? ?/sec
nth_value/ordered/n=10                                        1.00  1541.2±229.42ns        ? ?/sec    1.01  1551.7±188.61ns        ? ?/sec
nth_value/ordered/n=100                                       1.00     12.7±0.05µs        ? ?/sec     1.01     12.8±0.13µs        ? ?/sec
nth_value/ordered_groups_accumulator_adapter/8192_groups      1.00      7.9±0.20ms        ? ?/sec     1.01      8.0±0.03ms        ? ?/sec
nth_value/trivial/n=-1                                        1.00    659.6±3.72ns        ? ?/sec     1.01    666.4±4.73ns        ? ?/sec
nth_value/trivial/n=-10                                       1.00      5.2±0.02µs        ? ?/sec     1.01      5.2±0.02µs        ? ?/sec
nth_value/trivial/n=-100                                      1.00     49.1±0.18µs        ? ?/sec     1.01     49.6±0.18µs        ? ?/sec
nth_value/trivial/n=1                                         1.01    125.0±0.63ns        ? ?/sec     1.00    123.7±0.86ns        ? ?/sec
nth_value/trivial/n=10                                        1.01    672.8±2.53ns        ? ?/sec     1.00    666.1±4.04ns        ? ?/sec
nth_value/trivial/n=100                                       1.00      5.3±0.02µs        ? ?/sec     1.01      5.3±0.02µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/0%_nulls         1.01    681.8±6.70µs        ? ?/sec     1.00    676.1±9.09µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/30%_nulls        1.04  1083.4±32.75µs        ? ?/sec     1.00   1038.7±5.24µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/50%_nulls        1.00   1046.8±5.98µs        ? ?/sec     1.00   1047.5±5.32µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/80%_nulls        1.01   1047.2±6.25µs        ? ?/sec     1.00   1040.3±5.98µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/0%_nulls       1.01   679.7±14.95µs        ? ?/sec     1.00   673.7±12.40µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/30%_nulls      1.04  1039.2±29.95µs        ? ?/sec     1.00    999.5±5.64µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/50%_nulls      1.00   1012.4±8.73µs        ? ?/sec     1.00   1009.8±5.64µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/80%_nulls      1.02   1017.6±8.58µs        ? ?/sec     1.00    998.6±6.04µs        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/0%_nulls          1.01    682.5±5.71µs        ? ?/sec     1.00    675.4±8.67µs        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/30%_nulls         1.00     17.1±0.03ms        ? ?/sec     1.00     17.0±0.04ms        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/50%_nulls         1.00     11.5±0.02ms        ? ?/sec     1.01     11.6±0.03ms        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/80%_nulls         1.00      4.5±0.01ms        ? ?/sec     1.01      4.5±0.01ms        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/0%_nulls        1.01   678.6±14.98µs        ? ?/sec     1.00   672.4±12.28µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/30%_nulls       1.02  1427.0±24.56µs        ? ?/sec     1.00   1398.2±4.51µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/50%_nulls       1.00   1298.5±6.82µs        ? ?/sec     1.00   1303.0±5.78µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/80%_nulls       1.01   1116.4±6.89µs        ? ?/sec     1.00   1107.3±8.38µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/0%_nulls        1.01    682.1±5.96µs        ? ?/sec     1.00    674.8±7.62µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/30%_nulls       1.04  1106.9±31.67µs        ? ?/sec     1.00   1066.3±4.26µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/50%_nulls       1.00   1071.0±4.90µs        ? ?/sec     1.00   1072.3±4.26µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/80%_nulls       1.00   1070.7±4.98µs        ? ?/sec     1.00   1065.9±5.10µs        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/0%_nulls     1.01    681.6±6.03µs        ? ?/sec     1.00    674.4±7.77µs        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/30%_nulls    1.00     13.8±0.09ms        ? ?/sec     1.00     13.8±0.05ms        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/50%_nulls    1.00      9.2±0.09ms        ? ?/sec     1.00      9.2±0.11ms        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/80%_nulls    1.00      4.0±0.01ms        ? ?/sec     1.01      4.1±0.01ms        ? ?/sec
nth_value_nulls_comparison/first_value/ignore_nulls           1.12  1175.6±14.44µs        ? ?/sec     1.00   1047.8±4.97µs        ? ?/sec
nth_value_nulls_comparison/first_value/respect_nulls          1.07   123.9±12.38µs        ? ?/sec     1.00    115.4±3.19µs        ? ?/sec
nth_value_nulls_comparison/nth_value_10/ignore_nulls          1.12  1204.7±14.40µs        ? ?/sec     1.00   1073.4±3.90µs        ? ?/sec
nth_value_nulls_comparison/nth_value_10/respect_nulls         1.06   122.0±12.24µs        ? ?/sec     1.00    114.8±0.87µs        ? ?/sec

Resource Usage

nth_value — base (merge-base)

Metric Value
Wall time 535.1s
Peak memory 291.1 MiB
Avg memory 27.2 MiB
CPU user 525.6s
CPU sys 1.1s
Peak spill 0 B

nth_value — branch

Metric Value
Wall time 530.1s
Peak memory 294.1 MiB
Avg memory 29.7 MiB
CPU user 520.0s
CPU sys 1.1s
Peak spill 0 B

File an issue against this benchmark runner

@kosiew

kosiew commented Sep 14, 2026

Copy link
Copy Markdown
Contributor Author

run benchmark groups_accumulator_adapter nth_value percentile_cont

baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5666008098-2355-4vjdf 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark groups_accumulator_adapter
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

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-c5666008098-2356-ndttq 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark nth_value
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

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-c5666008098-2357-jvmvs 6.12.94+ #1 SMP Tue Aug 4 08:44:15 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 d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark percentile_cont
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"

Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark groups_accumulator_adapter
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"
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

group                                                  HEAD                                   test-aggmetrics3
-----                                                  ----                                   ----------------
groups_accumulator_adapter/update_batch/8192_groups    1.00   1377.7±5.41µs        ? ?/sec    1.02   1411.4±8.35µs        ? ?/sec

Resource Usage

groups_accumulator_adapter — base (merge-base)

Metric Value
Wall time 80.0s
Peak memory 12.5 MiB
Avg memory 1.4 MiB
CPU user 12.9s
CPU sys 0.0s
Peak spill 0 B

groups_accumulator_adapter — branch

Metric Value
Wall time 70.0s
Peak memory 23.8 MiB
Avg memory 1.7 MiB
CPU user 13.8s
CPU sys 0.0s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark percentile_cont
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"
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

group                                                              HEAD                                   test-aggmetrics3
-----                                                              ----                                   ----------------
approx_percentile_cont/groups_accumulator_adapter/8192_groups      1.02      3.8±0.03ms        ? ?/sec    1.00      3.8±0.01ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=16384      1.00      2.0±0.00ms        ? ?/sec    1.00      2.0±0.00ms        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=256        1.00    166.8±2.33µs        ? ?/sec    1.00    166.4±0.39µs        ? ?/sec
percentile_cont sliding_window f64 no_nulls window_size=4096       1.00    619.9±0.72µs        ? ?/sec    1.00    620.4±0.86µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=16384    1.00   1416.4±1.87µs        ? ?/sec    1.00   1417.1±1.14µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=256      1.00    166.8±0.39µs        ? ?/sec    1.00    166.9±0.31µs        ? ?/sec
percentile_cont sliding_window f64 with_nulls window_size=4096     1.00    481.3±0.39µs        ? ?/sec    1.00    482.5±0.76µs        ? ?/sec

Resource Usage

percentile_cont — base (merge-base)

Metric Value
Wall time 185.0s
Peak memory 39.1 MiB
Avg memory 10.1 MiB
CPU user 85.8s
CPU sys 0.2s
Peak spill 0 B

percentile_cont — branch

Metric Value
Wall time 180.0s
Peak memory 41.2 MiB
Avg memory 10.9 MiB
CPU user 86.3s
CPU sys 0.2s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot

Copy link
Copy Markdown

🤖 Benchmark completed (GKE) | trigger

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

Comparing d30fe36 (d30fe36) to ea83270 diff

Run configuration
run benchmark nth_value
baseline:
  ref: "ea83270d70"
changed:
  ref: "d30fe36039"
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

group                                                         HEAD                                    test-aggmetrics3
-----                                                         ----                                    ----------------
nth_value/ordered/n=-1                                        1.00  1350.5±308.09ns        ? ?/sec    1.00  1348.5±227.20ns        ? ?/sec
nth_value/ordered/n=-10                                       1.00     10.6±0.10µs        ? ?/sec     1.01     10.7±0.22µs        ? ?/sec
nth_value/ordered/n=-100                                      1.00    103.3±0.53µs        ? ?/sec     1.00    103.7±0.84µs        ? ?/sec
nth_value/ordered/n=1                                         1.00  424.8±651.59ns        ? ?/sec     1.04  440.9±783.18ns        ? ?/sec
nth_value/ordered/n=10                                        1.00  1555.0±304.99ns        ? ?/sec    1.00  1549.2±210.17ns        ? ?/sec
nth_value/ordered/n=100                                       1.00     12.6±0.05µs        ? ?/sec     1.01     12.7±0.09µs        ? ?/sec
nth_value/ordered_groups_accumulator_adapter/8192_groups      1.00      8.6±0.11ms        ? ?/sec     1.00      8.6±0.13ms        ? ?/sec
nth_value/trivial/n=-1                                        1.00    659.8±4.09ns        ? ?/sec     1.01    663.7±3.47ns        ? ?/sec
nth_value/trivial/n=-10                                       1.00      5.2±0.02µs        ? ?/sec     1.01      5.2±0.02µs        ? ?/sec
nth_value/trivial/n=-100                                      1.00     49.1±0.20µs        ? ?/sec     1.01     49.5±0.15µs        ? ?/sec
nth_value/trivial/n=1                                         1.01    124.7±0.79ns        ? ?/sec     1.00    123.5±0.80ns        ? ?/sec
nth_value/trivial/n=10                                        1.00    672.6±2.34ns        ? ?/sec     1.00    673.6±2.79ns        ? ?/sec
nth_value/trivial/n=100                                       1.00      5.3±0.02µs        ? ?/sec     1.01      5.3±0.02µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/0%_nulls         1.00    675.9±6.90µs        ? ?/sec     1.00    676.0±7.05µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/30%_nulls        1.00   1043.2±5.74µs        ? ?/sec     1.01   1052.7±7.95µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/50%_nulls        1.00   1036.3±5.98µs        ? ?/sec     1.01   1041.6±6.70µs        ? ?/sec
nth_value_ignore_nulls/first_value_expanding/80%_nulls        1.00   1043.7±5.99µs        ? ?/sec     1.00   1041.5±7.02µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/0%_nulls       1.00   674.8±13.52µs        ? ?/sec     1.01    678.9±9.35µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/30%_nulls      1.00   1006.9±5.31µs        ? ?/sec     1.01  1018.4±10.04µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/50%_nulls      1.00    998.0±5.36µs        ? ?/sec     1.01  1007.3±12.81µs        ? ?/sec
nth_value_ignore_nulls/first_value_sliding_100/80%_nulls      1.00   1006.6±5.32µs        ? ?/sec     1.01  1015.7±13.22µs        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/0%_nulls          1.00    672.3±6.29µs        ? ?/sec     1.01    676.5±6.65µs        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/30%_nulls         1.00     17.1±0.03ms        ? ?/sec     1.00     17.0±0.05ms        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/50%_nulls         1.00     11.5±0.02ms        ? ?/sec     1.01     11.6±0.05ms        ? ?/sec
nth_value_ignore_nulls/last_value_expanding/80%_nulls         1.00      4.5±0.01ms        ? ?/sec     1.01      4.5±0.01ms        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/0%_nulls        1.00   673.6±13.30µs        ? ?/sec     1.01    677.6±9.45µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/30%_nulls       1.00   1401.9±4.45µs        ? ?/sec     1.01   1411.4±9.04µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/50%_nulls       1.00   1283.7±4.17µs        ? ?/sec     1.01  1298.1±10.30µs        ? ?/sec
nth_value_ignore_nulls/last_value_sliding_100/80%_nulls       1.00   1117.2±5.29µs        ? ?/sec     1.00  1122.7±11.70µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/0%_nulls        1.00    675.3±6.49µs        ? ?/sec     1.00    677.8±6.25µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/30%_nulls       1.00   1070.6±4.56µs        ? ?/sec     1.01   1076.7±6.55µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/50%_nulls       1.00   1063.2±4.37µs        ? ?/sec     1.01   1068.6±5.64µs        ? ?/sec
nth_value_ignore_nulls/nth_value_10_expanding/80%_nulls       1.00   1070.7±4.88µs        ? ?/sec     1.00   1073.9±5.71µs        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/0%_nulls     1.00    675.1±6.60µs        ? ?/sec     1.00    677.3±6.44µs        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/30%_nulls    1.00     13.8±0.06ms        ? ?/sec     1.00     13.7±0.07ms        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/50%_nulls    1.00      9.2±0.11ms        ? ?/sec     1.01      9.3±0.12ms        ? ?/sec
nth_value_ignore_nulls/nth_value_neg10_expanding/80%_nulls    1.00      4.0±0.01ms        ? ?/sec     1.01      4.0±0.01ms        ? ?/sec
nth_value_nulls_comparison/first_value/ignore_nulls           1.02   1068.9±6.10µs        ? ?/sec     1.00   1043.4±6.91µs        ? ?/sec
nth_value_nulls_comparison/first_value/respect_nulls          1.00    118.5±5.35µs        ? ?/sec     1.00    118.0±4.59µs        ? ?/sec
nth_value_nulls_comparison/nth_value_10/ignore_nulls          1.02   1093.8±3.97µs        ? ?/sec     1.00   1068.0±5.38µs        ? ?/sec
nth_value_nulls_comparison/nth_value_10/respect_nulls         1.00    115.3±0.94µs        ? ?/sec     1.00    114.8±1.25µs        ? ?/sec

Resource Usage

nth_value — base (merge-base)

Metric Value
Wall time 540.1s
Peak memory 289.8 MiB
Avg memory 29.4 MiB
CPU user 521.7s
CPU sys 1.1s
Peak spill 0 B

nth_value — branch

Metric Value
Wall time 540.1s
Peak memory 291.1 MiB
Avg memory 29.3 MiB
CPU user 523.9s
CPU sys 1.1s
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

documentation Improvements or additions to documentation functions Changes to functions implementation logical-expr Logical plan and expressions physical-expr Changes to the physical-expr crates physical-plan Changes to the physical-plan crate

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants