aggregate blocked impl by rluvaton · Pull Request #24928 · apache/datafusion

rluvaton · 2026-09-03T16:55:44Z

Huge blocked impl that is backwards compatible
And just see the performance cost

Currently it only contain blocked impl for single group by and some aggregate expression

[EPIC] Use blocked / chunked memory management in hash aggregation #24704

Which issue does this PR close?

Closes #.

Rationale for this change

What changes are included in this PR?

What is the testing strategy for this PR?

Are there any user-facing changes?

…so all usages will be migrated

…warded

…ockedGroupColumn`

…not `usize`

…e grouped_hash_stream code

# Conflicts: # datafusion/expr-common/src/groups_accumulator.rs

adriangbot · 2026-09-09T18:47:36Z

Benchmark for this request failed before finishing (Kubernetes reason: BackoffLimitExceeded).

Benchmarks requested: h2o_medium

Runner log (last 40 lines)

2026-09-09T18:45:22.676291Z  INFO runner starting benchmark runner bench_type=Datafusion, pr_url=https://github.com/apache/datafusion/pull/24928, benchmarks=h2o_medium
2026-09-09T18:45:22.771467Z  INFO benchmark_controller::runner::bench_datafusion === Cloning PR branch ===
2026-09-09T18:45:22.771578Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["clone", "--depth=200", "https://github.com/apache/datafusion.git", "/workspace/datafusion-branch"], cwd="/"
2026-09-09T18:45:27.780079Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["fetch", "origin", "refs/pull/24928/head:add-blocks-impl-from-scratch", "main"], cwd="/workspace/datafusion-branch"
2026-09-09T18:45:32.782703Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["checkout", "add-blocks-impl-from-scratch"], cwd="/workspace/datafusion-branch"
2026-09-09T18:45:37.783989Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["merge-base", "HEAD", "origin/main"], cwd="/workspace/datafusion-branch"
2026-09-09T18:45:42.786842Z  INFO benchmark_controller::runner::bench_datafusion === Cloning merge-base ===
2026-09-09T18:45:42.786858Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["clone", "--depth=200", "https://github.com/apache/datafusion.git", "/workspace/datafusion-base"], cwd="/"
2026-09-09T18:45:47.787929Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["-c", "advice.detachedHead=false", "checkout", "40488988ad596c9b093ad60e1453430d803ce33c"], cwd="/workspace/datafusion-base"
2026-09-09T18:45:52.789146Z  INFO benchmark_controller::runner::shell running command cmd=rustc, args=["--version"], cwd="/"
2026-09-09T18:45:57.791473Z  INFO benchmark_controller::runner::shell running command cmd=cargo, args=["metadata", "--no-deps", "--format-version", "1"], cwd="/workspace/datafusion-branch/benchmarks"
2026-09-09T18:46:02.795549Z  INFO benchmark_controller::runner::bench_datafusion === Compiling dfbench for PR branch and merge-base in parallel ===
2026-09-09T18:46:02.835389Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["rev-parse", "HEAD"], cwd="/workspace/datafusion-branch"
2026-09-09T18:46:07.855026Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["rev-parse", "HEAD"], cwd="/workspace/datafusion-base"
2026-09-09T18:46:12.858039Z  INFO benchmark_controller::runner::controller_client post_comment _repo=apache/datafusion, _pr_number=24928, job_id=2276
2026-09-09T18:47:28.101519Z ERROR runner benchmark failed error=send request: error sending request for url (http://benchmark-controller.benchmarking.svc.cluster.local:8080/jobs/2276/comment): client error (Connect): tcp connect error: Connection refused (os error 111)

Kubernetes message

Job has reached the specified backoff limit

File an issue against this benchmark runner

adriangbot · 2026-09-09T18:47:38Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5606968809-2278-4xgxl 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 add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-09T18:47:43Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5606967120-2275-ll72d 6.12.94+ #1 SMP Fri Jul 17 09:42:57 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 add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark external_aggr

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-09T18:47:49Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5606968809-2279-vs6tz 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 add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark tpcds

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-09T18:47:49Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5606968809-2280-jlpz6 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 add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark tpch

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-09T18:47:52Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c5606967120-2277-pnd78 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 add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark tpch10

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-09-09T19:02:57Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark tpch

CPU Details (lscpu)

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

Details

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ add-blocks-impl-from-scratch ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 39.08 ms │                     42.01 ms │ 1.07x slower │
│ QQuery 2  │ 19.13 ms │                     19.83 ms │    no change │
│ QQuery 3  │ 29.75 ms │                     29.62 ms │    no change │
│ QQuery 4  │ 18.21 ms │                     18.50 ms │    no change │
│ QQuery 5  │ 36.93 ms │                     37.58 ms │    no change │
│ QQuery 6  │ 16.44 ms │                     16.85 ms │    no change │
│ QQuery 7  │ 42.20 ms │                     42.36 ms │    no change │
│ QQuery 8  │ 41.78 ms │                     42.15 ms │    no change │
│ QQuery 9  │ 51.30 ms │                     50.12 ms │    no change │
│ QQuery 10 │ 43.57 ms │                     43.80 ms │    no change │
│ QQuery 11 │ 14.24 ms │                     14.14 ms │    no change │
│ QQuery 12 │ 24.56 ms │                     24.44 ms │    no change │
│ QQuery 13 │ 42.09 ms │                     46.43 ms │ 1.10x slower │
│ QQuery 14 │ 25.26 ms │                     25.50 ms │    no change │
│ QQuery 15 │ 31.78 ms │                     32.72 ms │    no change │
│ QQuery 16 │ 14.17 ms │                     14.68 ms │    no change │
│ QQuery 17 │ 75.07 ms │                     89.77 ms │ 1.20x slower │
│ QQuery 18 │ 63.00 ms │                     63.42 ms │    no change │
│ QQuery 19 │ 33.63 ms │                     33.75 ms │    no change │
│ QQuery 20 │ 33.14 ms │                     33.07 ms │    no change │
│ QQuery 21 │ 58.70 ms │                     57.88 ms │    no change │
│ QQuery 22 │ 14.71 ms │                     14.98 ms │    no change │
└───────────┴──────────┴──────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                           ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 768.75ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 793.59ms │
│ Average Time (HEAD)                         │  34.94ms │
│ Average Time (add-blocks-impl-from-scratch) │  36.07ms │
│ Queries Faster                              │        0 │
│ Queries Slower                              │        3 │
│ Queries with No Change                      │       19 │
│ Queries with Failure                        │        0 │
└─────────────────────────────────────────────┴──────────┘

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

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃   add-blocks-impl-from-scratch ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 39.08 / 39.82 ±0.94 / 41.65 ms │ 42.01 / 43.34 ±1.28 / 44.93 ms │ 1.09x slower │
│ QQuery 2  │ 19.13 / 19.93 ±0.78 / 21.41 ms │ 19.83 / 19.89 ±0.07 / 20.02 ms │    no change │
│ QQuery 3  │ 29.75 / 30.07 ±0.19 / 30.32 ms │ 29.62 / 29.94 ±0.21 / 30.15 ms │    no change │
│ QQuery 4  │ 18.21 / 19.80 ±1.92 / 23.36 ms │ 18.50 / 19.08 ±0.94 / 20.96 ms │    no change │
│ QQuery 5  │ 36.93 / 37.99 ±1.11 / 40.00 ms │ 37.58 / 37.72 ±0.17 / 38.06 ms │    no change │
│ QQuery 6  │ 16.44 / 16.58 ±0.19 / 16.95 ms │ 16.85 / 16.98 ±0.08 / 17.10 ms │    no change │
│ QQuery 7  │ 42.20 / 44.20 ±3.11 / 50.40 ms │ 42.36 / 44.00 ±1.20 / 45.46 ms │    no change │
│ QQuery 8  │ 41.78 / 42.01 ±0.24 / 42.47 ms │ 42.15 / 43.26 ±1.45 / 46.06 ms │    no change │
│ QQuery 9  │ 51.30 / 52.82 ±1.38 / 55.19 ms │ 50.12 / 51.52 ±1.21 / 53.30 ms │    no change │
│ QQuery 10 │ 43.57 / 44.82 ±1.27 / 46.38 ms │ 43.80 / 44.03 ±0.17 / 44.29 ms │    no change │
│ QQuery 11 │ 14.24 / 14.31 ±0.09 / 14.49 ms │ 14.14 / 14.31 ±0.16 / 14.53 ms │    no change │
│ QQuery 12 │ 24.56 / 25.67 ±1.05 / 27.66 ms │ 24.44 / 25.98 ±1.83 / 29.47 ms │    no change │
│ QQuery 13 │ 42.09 / 42.49 ±0.47 / 43.20 ms │ 46.43 / 47.39 ±1.01 / 49.06 ms │ 1.12x slower │
│ QQuery 14 │ 25.26 / 25.42 ±0.15 / 25.65 ms │ 25.50 / 25.69 ±0.14 / 25.86 ms │    no change │
│ QQuery 15 │ 31.78 / 32.40 ±0.53 / 33.34 ms │ 32.72 / 32.85 ±0.17 / 33.17 ms │    no change │
│ QQuery 16 │ 14.17 / 14.36 ±0.12 / 14.55 ms │ 14.68 / 14.91 ±0.18 / 15.13 ms │    no change │
│ QQuery 17 │ 75.07 / 76.95 ±2.39 / 81.51 ms │ 89.77 / 91.51 ±1.35 / 93.61 ms │ 1.19x slower │
│ QQuery 18 │ 63.00 / 63.84 ±0.62 / 64.59 ms │ 63.42 / 65.87 ±1.76 / 68.89 ms │    no change │
│ QQuery 19 │ 33.63 / 33.99 ±0.27 / 34.29 ms │ 33.75 / 34.34 ±0.77 / 35.83 ms │    no change │
│ QQuery 20 │ 33.14 / 33.29 ±0.12 / 33.42 ms │ 33.07 / 33.80 ±0.42 / 34.30 ms │    no change │
│ QQuery 21 │ 58.70 / 59.61 ±1.10 / 61.77 ms │ 57.88 / 58.35 ±0.32 / 58.82 ms │    no change │
│ QQuery 22 │ 14.71 / 15.31 ±0.50 / 16.06 ms │ 14.98 / 16.60 ±1.70 / 19.48 ms │ 1.08x slower │
└───────────┴────────────────────────────────┴────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                           ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 785.67ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 811.36ms │
│ Average Time (HEAD)                         │  35.71ms │
│ Average Time (add-blocks-impl-from-scratch) │  36.88ms │
│ Queries Faster                              │        0 │
│ Queries Slower                              │        4 │
│ Queries with No Change                      │       18 │
│ Queries with Failure                        │        0 │
└─────────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.2 GiB
Avg memory	502.4 MiB
CPU user	22.2s
CPU sys	1.8s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.3 GiB
Avg memory	687.7 MiB
CPU user	23.2s
CPU sys	1.8s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-09T19:03:40Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark tpch10

CPU Details (lscpu)

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

Details

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃      HEAD ┃ add-blocks-impl-from-scratch ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 309.57 ms │                    335.93 ms │ 1.09x slower │
│ QQuery 2  │  90.48 ms │                     92.49 ms │    no change │
│ QQuery 3  │ 216.41 ms │                    218.59 ms │    no change │
│ QQuery 4  │ 113.14 ms │                    113.91 ms │    no change │
│ QQuery 5  │ 350.59 ms │                    351.57 ms │    no change │
│ QQuery 6  │ 123.14 ms │                    125.04 ms │    no change │
│ QQuery 7  │ 448.85 ms │                    454.60 ms │    no change │
│ QQuery 8  │ 360.08 ms │                    356.79 ms │    no change │
│ QQuery 9  │ 540.98 ms │                    519.62 ms │    no change │
│ QQuery 10 │ 309.89 ms │                    302.95 ms │    no change │
│ QQuery 11 │  64.55 ms │                     64.08 ms │    no change │
│ QQuery 12 │ 182.51 ms │                    180.21 ms │    no change │
│ QQuery 13 │ 307.48 ms │                    322.92 ms │ 1.05x slower │
│ QQuery 14 │ 172.16 ms │                    170.09 ms │    no change │
│ QQuery 15 │ 301.70 ms │                    310.07 ms │    no change │
│ QQuery 16 │  64.35 ms │                     66.23 ms │    no change │
│ QQuery 17 │ 569.74 ms │                    684.02 ms │ 1.20x slower │
│ QQuery 18 │ 703.89 ms │                    747.30 ms │ 1.06x slower │
│ QQuery 19 │ 244.96 ms │                    250.50 ms │    no change │
│ QQuery 20 │ 256.49 ms │                    285.20 ms │ 1.11x slower │
│ QQuery 21 │ 661.01 ms │                    691.69 ms │    no change │
│ QQuery 22 │  61.30 ms │                     64.71 ms │ 1.06x slower │
└───────────┴───────────┴──────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 6453.26ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 6708.50ms │
│ Average Time (HEAD)                         │  293.33ms │
│ Average Time (add-blocks-impl-from-scratch) │  304.93ms │
│ Queries Faster                              │         0 │
│ Queries Slower                              │         6 │
│ Queries with No Change                      │        16 │
│ Queries with Failure                        │         0 │
└─────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                               HEAD ┃       add-blocks-impl-from-scratch ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │  309.57 / 311.89 ±1.97 / 315.04 ms │  335.93 / 340.59 ±6.08 / 352.55 ms │ 1.09x slower │
│ QQuery 2  │     90.48 / 91.97 ±1.46 / 94.37 ms │     92.49 / 94.10 ±1.24 / 96.22 ms │    no change │
│ QQuery 3  │  216.41 / 219.46 ±3.49 / 226.25 ms │  218.59 / 222.47 ±3.26 / 227.10 ms │    no change │
│ QQuery 4  │  113.14 / 115.12 ±1.76 / 118.24 ms │  113.91 / 115.85 ±1.40 / 117.59 ms │    no change │
│ QQuery 5  │  350.59 / 355.83 ±4.42 / 363.79 ms │  351.57 / 362.45 ±5.82 / 366.73 ms │    no change │
│ QQuery 6  │  123.14 / 124.73 ±1.77 / 128.05 ms │  125.04 / 128.56 ±3.41 / 134.38 ms │    no change │
│ QQuery 7  │  448.85 / 456.21 ±5.73 / 466.47 ms │ 454.60 / 469.62 ±10.91 / 485.24 ms │    no change │
│ QQuery 8  │  360.08 / 363.44 ±2.85 / 367.84 ms │  356.79 / 362.37 ±4.05 / 367.67 ms │    no change │
│ QQuery 9  │ 540.98 / 554.64 ±14.34 / 582.03 ms │  519.62 / 527.81 ±4.66 / 532.59 ms │    no change │
│ QQuery 10 │ 309.89 / 321.08 ±10.02 / 338.90 ms │  302.95 / 306.18 ±3.50 / 312.41 ms │    no change │
│ QQuery 11 │     64.55 / 66.07 ±1.89 / 69.65 ms │     64.08 / 67.76 ±6.44 / 80.63 ms │    no change │
│ QQuery 12 │  182.51 / 193.21 ±8.69 / 204.83 ms │  180.21 / 184.56 ±3.97 / 189.50 ms │    no change │
│ QQuery 13 │  307.48 / 319.78 ±9.45 / 335.27 ms │  322.92 / 328.55 ±4.43 / 333.62 ms │    no change │
│ QQuery 14 │  172.16 / 178.32 ±9.02 / 196.07 ms │  170.09 / 176.75 ±6.76 / 187.07 ms │    no change │
│ QQuery 15 │  301.70 / 309.82 ±5.14 / 317.58 ms │  310.07 / 314.12 ±4.36 / 320.61 ms │    no change │
│ QQuery 16 │     64.35 / 68.59 ±3.73 / 73.40 ms │     66.23 / 67.48 ±1.53 / 70.46 ms │    no change │
│ QQuery 17 │ 569.74 / 590.36 ±14.44 / 613.42 ms │ 684.02 / 697.57 ±11.73 / 717.30 ms │ 1.18x slower │
│ QQuery 18 │  703.89 / 713.60 ±8.91 / 730.16 ms │ 747.30 / 766.87 ±15.68 / 785.53 ms │ 1.07x slower │
│ QQuery 19 │  244.96 / 256.77 ±7.29 / 267.14 ms │ 250.50 / 265.74 ±16.83 / 293.46 ms │    no change │
│ QQuery 20 │ 256.49 / 273.02 ±10.31 / 288.92 ms │  285.20 / 291.77 ±4.57 / 297.34 ms │ 1.07x slower │
│ QQuery 21 │ 661.01 / 676.76 ±10.14 / 686.15 ms │ 691.69 / 710.59 ±15.82 / 735.94 ms │    no change │
│ QQuery 22 │     61.30 / 65.39 ±3.84 / 70.84 ms │     64.71 / 69.24 ±4.50 / 76.63 ms │ 1.06x slower │
└───────────┴────────────────────────────────────┴────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 6626.02ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 6870.98ms │
│ Average Time (HEAD)                         │  301.18ms │
│ Average Time (add-blocks-impl-from-scratch) │  312.32ms │
│ Queries Faster                              │         0 │
│ Queries Slower                              │         5 │
│ Queries with No Change                      │        17 │
│ Queries with Failure                        │         0 │
└─────────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — base (merge-base)

Metric	Value
Wall time	35.0s
Peak memory	5.3 GiB
Avg memory	1.7 GiB
CPU user	337.5s
CPU sys	19.4s
Peak spill	0 B

tpch10 — branch

Metric	Value
Wall time	35.0s
Peak memory	5.3 GiB
Avg memory	1.6 GiB
CPU user	347.9s
CPU sys	19.1s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-09T19:03:55Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark tpcds

CPU Details (lscpu)

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

Details

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ add-blocks-impl-from-scratch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.52 ms │                      6.34 ms │  1.15x slower │
│ QQuery 2  │   79.20 ms │                     85.30 ms │  1.08x slower │
│ QQuery 3  │   28.48 ms │                     29.77 ms │     no change │
│ QQuery 4  │  465.00 ms │                    491.42 ms │  1.06x slower │
│ QQuery 5  │   51.61 ms │                     51.88 ms │     no change │
│ QQuery 6  │   35.09 ms │                     35.28 ms │     no change │
│ QQuery 7  │   74.88 ms │                     74.10 ms │     no change │
│ QQuery 8  │   35.51 ms │                     36.20 ms │     no change │
│ QQuery 9  │   53.89 ms │                     53.33 ms │     no change │
│ QQuery 10 │   62.49 ms │                     62.17 ms │     no change │
│ QQuery 11 │  287.07 ms │                    347.73 ms │  1.21x slower │
│ QQuery 12 │   28.14 ms │                     29.79 ms │  1.06x slower │
│ QQuery 13 │  117.36 ms │                    121.85 ms │     no change │
│ QQuery 14 │  417.43 ms │                    432.34 ms │     no change │
│ QQuery 15 │   56.49 ms │                     62.66 ms │  1.11x slower │
│ QQuery 16 │    6.65 ms │                      7.53 ms │  1.13x slower │
│ QQuery 17 │   79.28 ms │                     82.27 ms │     no change │
│ QQuery 18 │  103.05 ms │                    110.87 ms │  1.08x slower │
│ QQuery 19 │   40.83 ms │                     43.48 ms │  1.06x slower │
│ QQuery 20 │   34.97 ms │                     38.69 ms │  1.11x slower │
│ QQuery 21 │   16.81 ms │                     17.54 ms │     no change │
│ QQuery 22 │   62.47 ms │                     70.53 ms │  1.13x slower │
│ QQuery 23 │  303.52 ms │                    359.86 ms │  1.19x slower │
│ QQuery 24 │  190.20 ms │                    214.66 ms │  1.13x slower │
│ QQuery 25 │  108.41 ms │                    115.48 ms │  1.07x slower │
│ QQuery 26 │   48.60 ms │                     50.86 ms │     no change │
│ QQuery 27 │    6.02 ms │                      6.84 ms │  1.14x slower │
│ QQuery 28 │   60.40 ms │                     58.83 ms │     no change │
│ QQuery 29 │   95.58 ms │                    101.72 ms │  1.06x slower │
│ QQuery 30 │   31.89 ms │                     34.62 ms │  1.09x slower │
│ QQuery 31 │  109.27 ms │                    118.05 ms │  1.08x slower │
│ QQuery 32 │   19.88 ms │                     22.22 ms │  1.12x slower │
│ QQuery 33 │   37.07 ms │                     39.87 ms │  1.08x slower │
│ QQuery 34 │    9.72 ms │                     10.79 ms │  1.11x slower │
│ QQuery 35 │   71.92 ms │                     78.66 ms │  1.09x slower │
│ QQuery 36 │    5.68 ms │                      6.16 ms │  1.08x slower │
│ QQuery 37 │    6.67 ms │                      7.05 ms │  1.06x slower │
│ QQuery 38 │   60.91 ms │                     69.11 ms │  1.13x slower │
│ QQuery 39 │   88.54 ms │                    103.08 ms │  1.16x slower │
│ QQuery 40 │   23.17 ms │                     26.02 ms │  1.12x slower │
│ QQuery 41 │   11.24 ms │                     12.23 ms │  1.09x slower │
│ QQuery 42 │   23.33 ms │                     24.85 ms │  1.07x slower │
│ QQuery 43 │    5.19 ms │                      5.55 ms │  1.07x slower │
│ QQuery 44 │    9.49 ms │                      9.96 ms │  1.05x slower │
│ QQuery 45 │   38.03 ms │                     42.72 ms │  1.12x slower │
│ QQuery 46 │   11.77 ms │                     12.36 ms │     no change │
│ QQuery 47 │  219.66 ms │                    259.75 ms │  1.18x slower │
│ QQuery 48 │   94.70 ms │                     98.57 ms │     no change │
│ QQuery 49 │   69.91 ms │                     74.15 ms │  1.06x slower │
│ QQuery 50 │   60.18 ms │                     61.19 ms │     no change │
│ QQuery 51 │   92.98 ms │                     93.76 ms │     no change │
│ QQuery 52 │   24.71 ms │                     25.13 ms │     no change │
│ QQuery 53 │   30.00 ms │                     29.65 ms │     no change │
│ QQuery 54 │   55.87 ms │                     57.47 ms │     no change │
│ QQuery 55 │   24.17 ms │                     24.44 ms │     no change │
│ QQuery 56 │   40.09 ms │                     40.98 ms │     no change │
│ QQuery 57 │  181.28 ms │                    190.49 ms │  1.05x slower │
│ QQuery 58 │  113.44 ms │                    115.67 ms │     no change │
│ QQuery 59 │  117.23 ms │                    127.48 ms │  1.09x slower │
│ QQuery 60 │   40.37 ms │                     40.82 ms │     no change │
│ QQuery 61 │   12.84 ms │                     13.52 ms │  1.05x slower │
│ QQuery 62 │   47.77 ms │                     49.08 ms │     no change │
│ QQuery 63 │   30.11 ms │                     30.20 ms │     no change │
│ QQuery 64 │  385.31 ms │                    386.87 ms │     no change │
│ QQuery 65 │  124.91 ms │                    134.80 ms │  1.08x slower │
│ QQuery 66 │   81.03 ms │                     84.80 ms │     no change │
│ QQuery 67 │  264.96 ms │                    275.52 ms │     no change │
│ QQuery 68 │   12.56 ms │                     13.13 ms │     no change │
│ QQuery 69 │   58.62 ms │                     59.02 ms │     no change │
│ QQuery 70 │  109.38 ms │                    112.43 ms │     no change │
│ QQuery 71 │   36.01 ms │                     36.68 ms │     no change │
│ QQuery 72 │ 1907.51 ms │                   1887.64 ms │     no change │
│ QQuery 73 │   10.54 ms │                      9.82 ms │ +1.07x faster │
│ QQuery 74 │  196.12 ms │                    175.46 ms │ +1.12x faster │
│ QQuery 75 │  150.88 ms │                    148.59 ms │     no change │
│ QQuery 76 │   36.86 ms │                     35.33 ms │     no change │
│ QQuery 77 │   62.67 ms │                     61.63 ms │     no change │
│ QQuery 78 │  242.10 ms │                    207.49 ms │ +1.17x faster │
│ QQuery 79 │   69.15 ms │                     66.20 ms │     no change │
│ QQuery 80 │  102.75 ms │                     99.36 ms │     no change │
│ QQuery 81 │   27.45 ms │                     25.90 ms │ +1.06x faster │
│ QQuery 82 │   17.04 ms │                     16.26 ms │     no change │
│ QQuery 83 │   34.97 ms │                     34.09 ms │     no change │
│ QQuery 84 │   30.42 ms │                     29.42 ms │     no change │
│ QQuery 85 │  105.07 ms │                    102.13 ms │     no change │
│ QQuery 86 │   27.37 ms │                     24.91 ms │ +1.10x faster │
│ QQuery 87 │   67.66 ms │                     63.83 ms │ +1.06x faster │
│ QQuery 88 │   64.65 ms │                     63.34 ms │     no change │
│ QQuery 89 │   36.83 ms │                     35.40 ms │     no change │
│ QQuery 90 │   17.91 ms │                     17.14 ms │     no change │
│ QQuery 91 │   46.09 ms │                     44.50 ms │     no change │
│ QQuery 92 │   30.14 ms │                     28.90 ms │     no change │
│ QQuery 93 │   50.71 ms │                     49.85 ms │     no change │
│ QQuery 94 │   38.68 ms │                     37.78 ms │     no change │
│ QQuery 95 │   83.94 ms │                     80.79 ms │     no change │
│ QQuery 96 │   24.37 ms │                     23.76 ms │     no change │
│ QQuery 97 │   53.74 ms │                     54.93 ms │     no change │
│ QQuery 98 │   43.04 ms │                     42.53 ms │     no change │
│ QQuery 99 │   70.59 ms │                     70.99 ms │     no change │
└───────────┴────────────┴──────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 9496.11ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 9796.18ms │
│ Average Time (HEAD)                         │   95.92ms │
│ Average Time (add-blocks-impl-from-scratch) │   98.95ms │
│ Queries Faster                              │         6 │
│ Queries Slower                              │        38 │
│ Queries with No Change                      │        55 │
│ Queries with Failure                        │         0 │
└─────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃          add-blocks-impl-from-scratch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.52 / 6.08 ±0.95 / 7.97 ms │           6.34 / 6.84 ±0.90 / 8.65 ms │  1.12x slower │
│ QQuery 2  │        79.20 / 80.19 ±1.03 / 82.15 ms │        85.30 / 85.94 ±0.42 / 86.46 ms │  1.07x slower │
│ QQuery 3  │        28.48 / 28.89 ±0.26 / 29.23 ms │        29.77 / 30.15 ±0.24 / 30.50 ms │     no change │
│ QQuery 4  │     465.00 / 474.60 ±9.54 / 492.85 ms │    491.42 / 535.23 ±22.78 / 554.14 ms │  1.13x slower │
│ QQuery 5  │        51.61 / 54.83 ±4.36 / 63.34 ms │        51.88 / 52.28 ±0.32 / 52.72 ms │     no change │
│ QQuery 6  │        35.09 / 35.96 ±0.57 / 36.86 ms │        35.28 / 35.93 ±0.46 / 36.63 ms │     no change │
│ QQuery 7  │        74.88 / 75.31 ±0.31 / 75.73 ms │        74.10 / 75.59 ±1.75 / 78.85 ms │     no change │
│ QQuery 8  │        35.51 / 35.92 ±0.30 / 36.39 ms │        36.20 / 36.67 ±0.50 / 37.62 ms │     no change │
│ QQuery 9  │        53.89 / 56.55 ±3.31 / 62.97 ms │        53.33 / 54.04 ±0.71 / 55.37 ms │     no change │
│ QQuery 10 │        62.49 / 62.70 ±0.27 / 63.24 ms │        62.17 / 63.74 ±1.22 / 65.13 ms │     no change │
│ QQuery 11 │     287.07 / 289.34 ±1.88 / 292.45 ms │     347.73 / 353.22 ±3.83 / 358.46 ms │  1.22x slower │
│ QQuery 12 │        28.14 / 30.42 ±2.22 / 34.61 ms │        29.79 / 30.52 ±0.61 / 31.38 ms │     no change │
│ QQuery 13 │     117.36 / 118.20 ±0.95 / 119.92 ms │     121.85 / 123.75 ±2.06 / 127.56 ms │     no change │
│ QQuery 14 │     417.43 / 422.01 ±5.93 / 433.59 ms │     432.34 / 437.48 ±4.29 / 444.95 ms │     no change │
│ QQuery 15 │        56.49 / 57.38 ±1.53 / 60.42 ms │        62.66 / 65.16 ±1.63 / 67.42 ms │  1.14x slower │
│ QQuery 16 │           6.65 / 6.75 ±0.08 / 6.84 ms │          7.53 / 8.98 ±2.61 / 14.17 ms │  1.33x slower │
│ QQuery 17 │        79.28 / 80.02 ±0.70 / 81.22 ms │        82.27 / 84.62 ±2.10 / 88.52 ms │  1.06x slower │
│ QQuery 18 │     103.05 / 104.21 ±0.88 / 105.23 ms │     110.87 / 111.60 ±0.83 / 113.22 ms │  1.07x slower │
│ QQuery 19 │        40.83 / 41.07 ±0.27 / 41.55 ms │        43.48 / 45.22 ±1.53 / 47.68 ms │  1.10x slower │
│ QQuery 20 │        34.97 / 35.49 ±0.30 / 35.80 ms │        38.69 / 39.58 ±0.71 / 40.78 ms │  1.12x slower │
│ QQuery 21 │        16.81 / 17.01 ±0.12 / 17.14 ms │        17.54 / 18.27 ±0.45 / 18.83 ms │  1.07x slower │
│ QQuery 22 │        62.47 / 62.96 ±0.55 / 63.96 ms │        70.53 / 72.07 ±0.98 / 73.63 ms │  1.14x slower │
│ QQuery 23 │     303.52 / 308.68 ±3.32 / 313.90 ms │     359.86 / 370.55 ±6.14 / 378.92 ms │  1.20x slower │
│ QQuery 24 │     190.20 / 193.26 ±3.65 / 200.20 ms │     214.66 / 219.31 ±5.32 / 228.85 ms │  1.13x slower │
│ QQuery 25 │     108.41 / 110.27 ±1.72 / 113.44 ms │     115.48 / 117.82 ±2.20 / 121.91 ms │  1.07x slower │
│ QQuery 26 │        48.60 / 49.15 ±0.58 / 50.25 ms │        50.86 / 51.14 ±0.24 / 51.51 ms │     no change │
│ QQuery 27 │           6.02 / 6.14 ±0.12 / 6.36 ms │           6.84 / 7.06 ±0.16 / 7.31 ms │  1.15x slower │
│ QQuery 28 │        60.40 / 60.66 ±0.23 / 61.03 ms │        58.83 / 61.47 ±1.88 / 63.28 ms │     no change │
│ QQuery 29 │       95.58 / 98.55 ±2.74 / 103.30 ms │     101.72 / 106.62 ±7.23 / 120.91 ms │  1.08x slower │
│ QQuery 30 │        31.89 / 32.65 ±0.59 / 33.38 ms │        34.62 / 35.24 ±0.50 / 36.12 ms │  1.08x slower │
│ QQuery 31 │     109.27 / 110.40 ±1.51 / 113.37 ms │     118.05 / 121.44 ±3.44 / 127.53 ms │  1.10x slower │
│ QQuery 32 │        19.88 / 20.26 ±0.43 / 21.09 ms │        22.22 / 22.65 ±0.49 / 23.30 ms │  1.12x slower │
│ QQuery 33 │        37.07 / 37.77 ±0.41 / 38.20 ms │        39.87 / 40.13 ±0.35 / 40.82 ms │  1.06x slower │
│ QQuery 34 │         9.72 / 10.07 ±0.33 / 10.62 ms │        10.79 / 11.15 ±0.38 / 11.74 ms │  1.11x slower │
│ QQuery 35 │        71.92 / 72.44 ±0.35 / 73.00 ms │        78.66 / 81.46 ±4.52 / 90.47 ms │  1.12x slower │
│ QQuery 36 │           5.68 / 5.81 ±0.14 / 6.09 ms │           6.16 / 6.50 ±0.19 / 6.72 ms │  1.12x slower │
│ QQuery 37 │           6.67 / 6.76 ±0.08 / 6.89 ms │           7.05 / 7.41 ±0.23 / 7.79 ms │  1.10x slower │
│ QQuery 38 │        60.91 / 62.91 ±2.11 / 66.48 ms │        69.11 / 70.70 ±1.47 / 73.42 ms │  1.12x slower │
│ QQuery 39 │        88.54 / 89.77 ±1.27 / 91.94 ms │     103.08 / 105.52 ±1.80 / 107.82 ms │  1.18x slower │
│ QQuery 40 │        23.17 / 23.49 ±0.21 / 23.77 ms │        26.02 / 29.02 ±3.44 / 35.73 ms │  1.24x slower │
│ QQuery 41 │        11.24 / 11.38 ±0.11 / 11.50 ms │        12.23 / 12.36 ±0.07 / 12.42 ms │  1.09x slower │
│ QQuery 42 │        23.33 / 23.65 ±0.34 / 24.31 ms │        24.85 / 25.67 ±0.65 / 26.42 ms │  1.09x slower │
│ QQuery 43 │           5.19 / 5.29 ±0.12 / 5.53 ms │           5.55 / 5.71 ±0.15 / 5.98 ms │  1.08x slower │
│ QQuery 44 │           9.49 / 9.54 ±0.05 / 9.64 ms │         9.96 / 10.23 ±0.19 / 10.50 ms │  1.07x slower │
│ QQuery 45 │        38.03 / 39.23 ±1.35 / 41.81 ms │        42.72 / 43.59 ±0.80 / 44.60 ms │  1.11x slower │
│ QQuery 46 │        11.77 / 12.39 ±0.38 / 12.90 ms │        12.36 / 12.70 ±0.25 / 13.12 ms │     no change │
│ QQuery 47 │     219.66 / 223.17 ±2.52 / 227.38 ms │     259.75 / 266.01 ±5.69 / 273.35 ms │  1.19x slower │
│ QQuery 48 │        94.70 / 95.30 ±0.63 / 96.22 ms │      98.57 / 101.66 ±4.41 / 110.38 ms │  1.07x slower │
│ QQuery 49 │        69.91 / 72.82 ±2.94 / 78.21 ms │        74.15 / 75.19 ±0.74 / 76.07 ms │     no change │
│ QQuery 50 │        60.18 / 60.48 ±0.30 / 61.01 ms │        61.19 / 64.22 ±3.04 / 69.76 ms │  1.06x slower │
│ QQuery 51 │        92.98 / 94.51 ±0.91 / 95.34 ms │       93.76 / 97.42 ±2.15 / 100.53 ms │     no change │
│ QQuery 52 │        24.71 / 24.94 ±0.18 / 25.22 ms │        25.13 / 25.51 ±0.34 / 26.06 ms │     no change │
│ QQuery 53 │        30.00 / 32.54 ±4.20 / 40.90 ms │        29.65 / 30.08 ±0.24 / 30.27 ms │ +1.08x faster │
│ QQuery 54 │        55.87 / 56.77 ±0.74 / 58.10 ms │        57.47 / 60.42 ±5.29 / 70.99 ms │  1.06x slower │
│ QQuery 55 │        24.17 / 24.75 ±0.72 / 26.16 ms │        24.44 / 25.02 ±0.54 / 25.97 ms │     no change │
│ QQuery 56 │        40.09 / 40.59 ±0.52 / 41.52 ms │        40.98 / 41.72 ±0.55 / 42.47 ms │     no change │
│ QQuery 57 │     181.28 / 183.66 ±2.83 / 189.18 ms │     190.49 / 193.86 ±2.55 / 198.21 ms │  1.06x slower │
│ QQuery 58 │     113.44 / 115.00 ±1.23 / 117.19 ms │     115.67 / 118.17 ±3.01 / 123.63 ms │     no change │
│ QQuery 59 │     117.23 / 118.93 ±1.13 / 120.72 ms │     127.48 / 128.74 ±1.07 / 130.68 ms │  1.08x slower │
│ QQuery 60 │        40.37 / 42.24 ±3.17 / 48.56 ms │        40.82 / 41.59 ±0.43 / 42.15 ms │     no change │
│ QQuery 61 │        12.84 / 13.07 ±0.12 / 13.18 ms │        13.52 / 15.22 ±2.74 / 20.64 ms │  1.17x slower │
│ QQuery 62 │        47.77 / 47.90 ±0.11 / 48.04 ms │        49.08 / 49.60 ±0.65 / 50.87 ms │     no change │
│ QQuery 63 │        30.11 / 30.33 ±0.23 / 30.72 ms │        30.20 / 30.67 ±0.56 / 31.37 ms │     no change │
│ QQuery 64 │     385.31 / 388.72 ±2.94 / 393.66 ms │     386.87 / 395.79 ±6.74 / 407.06 ms │     no change │
│ QQuery 65 │     124.91 / 129.67 ±3.72 / 135.16 ms │     134.80 / 138.59 ±3.21 / 142.69 ms │  1.07x slower │
│ QQuery 66 │        81.03 / 84.51 ±1.76 / 85.82 ms │        84.80 / 85.65 ±0.59 / 86.28 ms │     no change │
│ QQuery 67 │     264.96 / 271.30 ±6.08 / 282.10 ms │     275.52 / 280.56 ±4.24 / 288.09 ms │     no change │
│ QQuery 68 │        12.56 / 12.69 ±0.18 / 13.04 ms │        13.13 / 13.75 ±0.84 / 15.40 ms │  1.08x slower │
│ QQuery 69 │        58.62 / 60.49 ±2.04 / 64.45 ms │        59.02 / 63.30 ±6.58 / 76.35 ms │     no change │
│ QQuery 70 │     109.38 / 111.13 ±1.55 / 112.88 ms │     112.43 / 114.75 ±3.26 / 121.12 ms │     no change │
│ QQuery 71 │        36.01 / 36.26 ±0.23 / 36.64 ms │        36.68 / 38.11 ±1.33 / 40.25 ms │  1.05x slower │
│ QQuery 72 │ 1907.51 / 1941.37 ±48.71 / 2037.27 ms │ 1887.64 / 2008.16 ±89.14 / 2127.37 ms │     no change │
│ QQuery 73 │        10.54 / 11.32 ±0.67 / 12.29 ms │         9.82 / 12.90 ±4.04 / 20.85 ms │  1.14x slower │
│ QQuery 74 │     196.12 / 199.34 ±2.99 / 204.15 ms │     175.46 / 177.00 ±1.62 / 180.01 ms │ +1.13x faster │
│ QQuery 75 │     150.88 / 155.36 ±4.85 / 164.60 ms │     148.59 / 150.99 ±3.52 / 157.86 ms │     no change │
│ QQuery 76 │        36.86 / 37.51 ±0.53 / 38.27 ms │        35.33 / 36.15 ±0.99 / 38.08 ms │     no change │
│ QQuery 77 │        62.67 / 63.46 ±0.75 / 64.45 ms │        61.63 / 62.00 ±0.32 / 62.40 ms │     no change │
│ QQuery 78 │    242.10 / 255.03 ±14.16 / 279.15 ms │    207.49 / 213.91 ±10.33 / 234.43 ms │ +1.19x faster │
│ QQuery 79 │        69.15 / 71.02 ±0.98 / 72.07 ms │        66.20 / 66.70 ±0.40 / 67.31 ms │ +1.06x faster │
│ QQuery 80 │     102.75 / 108.85 ±8.54 / 125.79 ms │      99.36 / 101.80 ±3.36 / 108.33 ms │ +1.07x faster │
│ QQuery 81 │        27.45 / 28.53 ±1.75 / 32.03 ms │        25.90 / 26.42 ±0.54 / 27.14 ms │ +1.08x faster │
│ QQuery 82 │        17.04 / 17.42 ±0.38 / 18.08 ms │        16.26 / 16.66 ±0.34 / 17.23 ms │     no change │
│ QQuery 83 │        34.97 / 35.41 ±0.31 / 35.84 ms │        34.09 / 34.29 ±0.14 / 34.47 ms │     no change │
│ QQuery 84 │        30.42 / 30.90 ±0.39 / 31.51 ms │        29.42 / 29.56 ±0.19 / 29.91 ms │     no change │
│ QQuery 85 │     105.07 / 107.92 ±3.74 / 115.24 ms │     102.13 / 105.37 ±2.93 / 110.58 ms │     no change │
│ QQuery 86 │        27.37 / 28.69 ±1.27 / 31.06 ms │        24.91 / 25.42 ±0.34 / 25.83 ms │ +1.13x faster │
│ QQuery 87 │        67.66 / 69.18 ±1.52 / 71.77 ms │        63.83 / 64.72 ±0.47 / 65.21 ms │ +1.07x faster │
│ QQuery 88 │        64.65 / 65.28 ±0.75 / 66.70 ms │        63.34 / 64.52 ±1.65 / 67.77 ms │     no change │
│ QQuery 89 │        36.83 / 38.80 ±3.15 / 45.05 ms │        35.40 / 38.61 ±4.84 / 48.25 ms │     no change │
│ QQuery 90 │        17.91 / 19.02 ±1.14 / 21.16 ms │        17.14 / 17.46 ±0.20 / 17.71 ms │ +1.09x faster │
│ QQuery 91 │        46.09 / 46.47 ±0.26 / 46.77 ms │        44.50 / 45.39 ±0.84 / 46.55 ms │     no change │
│ QQuery 92 │        30.14 / 30.95 ±0.58 / 31.86 ms │        28.90 / 29.74 ±1.14 / 31.95 ms │     no change │
│ QQuery 93 │        50.71 / 51.93 ±0.78 / 53.18 ms │        49.85 / 50.39 ±0.33 / 50.83 ms │     no change │
│ QQuery 94 │        38.68 / 40.91 ±1.58 / 43.29 ms │        37.78 / 39.34 ±1.52 / 41.56 ms │     no change │
│ QQuery 95 │        83.94 / 84.90 ±0.88 / 86.22 ms │        80.79 / 83.13 ±3.22 / 89.52 ms │     no change │
│ QQuery 96 │        24.37 / 24.65 ±0.22 / 24.96 ms │        23.76 / 24.13 ±0.29 / 24.58 ms │     no change │
│ QQuery 97 │        53.74 / 55.14 ±1.10 / 56.64 ms │        54.93 / 56.13 ±0.91 / 57.59 ms │     no change │
│ QQuery 98 │        43.04 / 45.53 ±1.31 / 46.78 ms │        42.53 / 42.93 ±0.30 / 43.21 ms │ +1.06x faster │
│ QQuery 99 │        70.59 / 71.76 ±1.50 / 74.64 ms │        70.99 / 73.06 ±3.90 / 80.85 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                           │  9684.78ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 10134.84ms │
│ Average Time (HEAD)                         │    97.83ms │
│ Average Time (add-blocks-impl-from-scratch) │   102.37ms │
│ Queries Faster                              │         10 │
│ Queries Slower                              │         44 │
│ Queries with No Change                      │         45 │
│ Queries with Failure                        │          0 │
└─────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	50.0s
Peak memory	2.2 GiB
Avg memory	1.5 GiB
CPU user	213.7s
CPU sys	5.6s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	2.0 GiB
Avg memory	1.3 GiB
CPU user	221.6s
CPU sys	6.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-09T19:08:03Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark clickbench_partitioned

CPU Details (lscpu)

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

Details

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ add-blocks-impl-from-scratch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.24 ms │                      1.30 ms │     no change │
│ QQuery 1  │   11.80 ms │                     11.03 ms │ +1.07x faster │
│ QQuery 2  │   36.81 ms │                     35.57 ms │     no change │
│ QQuery 3  │   31.68 ms │                     30.84 ms │     no change │
│ QQuery 4  │  238.76 ms │                    312.66 ms │  1.31x slower │
│ QQuery 5  │  288.97 ms │                    314.38 ms │  1.09x slower │
│ QQuery 6  │    1.28 ms │                      1.35 ms │     no change │
│ QQuery 7  │   13.88 ms │                     12.10 ms │ +1.15x faster │
│ QQuery 8  │  355.21 ms │                    362.16 ms │     no change │
│ QQuery 9  │  479.60 ms │                    544.55 ms │  1.14x slower │
│ QQuery 10 │   71.68 ms │                     69.60 ms │     no change │
│ QQuery 11 │   83.23 ms │                     80.59 ms │     no change │
│ QQuery 12 │  283.99 ms │                    303.25 ms │  1.07x slower │
│ QQuery 13 │  368.85 ms │                    411.87 ms │  1.12x slower │
│ QQuery 14 │  288.40 ms │                    325.15 ms │  1.13x slower │
│ QQuery 15 │  273.47 ms │                    381.47 ms │  1.39x slower │
│ QQuery 16 │  622.77 ms │                    686.75 ms │  1.10x slower │
│ QQuery 17 │  617.21 ms │                    700.73 ms │  1.14x slower │
│ QQuery 18 │ 1266.00 ms │                   1356.13 ms │  1.07x slower │
│ QQuery 19 │   27.46 ms │                     26.35 ms │     no change │
│ QQuery 20 │  521.67 ms │                    523.41 ms │     no change │
│ QQuery 21 │  526.48 ms │                    519.98 ms │     no change │
│ QQuery 22 │ 1000.65 ms │                    993.77 ms │     no change │
│ QQuery 23 │ 3136.79 ms │                   3137.60 ms │     no change │
│ QQuery 24 │   41.52 ms │                     40.10 ms │     no change │
│ QQuery 25 │  114.02 ms │                    109.35 ms │     no change │
│ QQuery 26 │   41.67 ms │                     40.59 ms │     no change │
│ QQuery 27 │  518.60 ms │                    533.47 ms │     no change │
│ QQuery 28 │ 2989.39 ms │                   3009.32 ms │     no change │
│ QQuery 29 │   42.54 ms │                     41.94 ms │     no change │
│ QQuery 30 │  311.57 ms │                    336.47 ms │  1.08x slower │
│ QQuery 31 │  285.21 ms │                    286.45 ms │     no change │
│ QQuery 32 │  956.20 ms │                    981.79 ms │     no change │
│ QQuery 33 │ 1471.69 ms │                   1682.47 ms │  1.14x slower │
│ QQuery 34 │ 1593.22 ms │                   1703.10 ms │  1.07x slower │
│ QQuery 35 │  291.11 ms │                    382.71 ms │  1.31x slower │
│ QQuery 36 │   67.23 ms │                     71.78 ms │  1.07x slower │
│ QQuery 37 │   38.32 ms │                     37.07 ms │     no change │
│ QQuery 38 │   44.02 ms │                     40.78 ms │ +1.08x faster │
│ QQuery 39 │  142.99 ms │                    159.40 ms │  1.11x slower │
│ QQuery 40 │   14.54 ms │                     13.13 ms │ +1.11x faster │
│ QQuery 41 │   14.18 ms │                     12.42 ms │ +1.14x faster │
│ QQuery 42 │   13.83 ms │                     12.92 ms │ +1.07x faster │
└───────────┴────────────┴──────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 19539.75ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 20637.80ms │
│ Average Time (HEAD)                         │   454.41ms │
│ Average Time (add-blocks-impl-from-scratch) │   479.95ms │
│ Queries Faster                              │          6 │
│ Queries Slower                              │         16 │
│ Queries with No Change                      │         21 │
│ Queries with Failure                        │          0 │
└─────────────────────────────────────────────┴────────────┘

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

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃          add-blocks-impl-from-scratch ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.24 / 4.14 ±5.62 / 15.38 ms │          1.30 / 4.25 ±5.70 / 15.64 ms │     no change │
│ QQuery 1  │        11.80 / 12.37 ±0.33 / 12.65 ms │        11.03 / 11.15 ±0.08 / 11.26 ms │ +1.11x faster │
│ QQuery 2  │        36.81 / 37.16 ±0.31 / 37.71 ms │        35.57 / 35.86 ±0.30 / 36.40 ms │     no change │
│ QQuery 3  │        31.68 / 32.60 ±0.50 / 33.15 ms │        30.84 / 31.88 ±1.10 / 33.48 ms │     no change │
│ QQuery 4  │     238.76 / 242.91 ±3.60 / 247.72 ms │     312.66 / 317.95 ±3.84 / 322.57 ms │  1.31x slower │
│ QQuery 5  │     288.97 / 291.94 ±2.29 / 295.34 ms │     314.38 / 320.41 ±5.25 / 326.99 ms │  1.10x slower │
│ QQuery 6  │           1.28 / 1.42 ±0.22 / 1.85 ms │           1.35 / 1.49 ±0.22 / 1.93 ms │     no change │
│ QQuery 7  │        13.88 / 14.92 ±1.68 / 18.28 ms │        12.10 / 13.31 ±2.00 / 17.29 ms │ +1.12x faster │
│ QQuery 8  │     355.21 / 356.92 ±2.35 / 361.50 ms │     362.16 / 369.29 ±5.41 / 377.20 ms │     no change │
│ QQuery 9  │     479.60 / 484.96 ±6.14 / 494.34 ms │     544.55 / 559.75 ±8.46 / 567.49 ms │  1.15x slower │
│ QQuery 10 │        71.68 / 76.04 ±6.51 / 88.95 ms │        69.60 / 75.13 ±5.92 / 82.84 ms │     no change │
│ QQuery 11 │        83.23 / 83.98 ±0.61 / 85.04 ms │        80.59 / 81.14 ±0.41 / 81.62 ms │     no change │
│ QQuery 12 │    283.99 / 295.02 ±12.41 / 318.37 ms │     303.25 / 312.02 ±7.89 / 321.81 ms │  1.06x slower │
│ QQuery 13 │     368.85 / 381.56 ±7.67 / 391.32 ms │    411.87 / 424.48 ±12.96 / 447.13 ms │  1.11x slower │
│ QQuery 14 │     288.40 / 294.52 ±4.31 / 300.87 ms │     325.15 / 331.46 ±5.19 / 339.09 ms │  1.13x slower │
│ QQuery 15 │     273.47 / 280.73 ±8.99 / 297.96 ms │    381.47 / 405.07 ±19.83 / 428.84 ms │  1.44x slower │
│ QQuery 16 │     622.77 / 629.55 ±3.74 / 633.60 ms │     686.75 / 691.43 ±3.13 / 695.79 ms │  1.10x slower │
│ QQuery 17 │     617.21 / 630.52 ±9.48 / 646.05 ms │    700.73 / 709.56 ±11.90 / 732.07 ms │  1.13x slower │
│ QQuery 18 │ 1266.00 / 1287.85 ±14.16 / 1305.53 ms │ 1356.13 / 1460.18 ±52.50 / 1498.91 ms │  1.13x slower │
│ QQuery 19 │        27.46 / 27.93 ±0.37 / 28.45 ms │        26.35 / 33.86 ±9.01 / 47.78 ms │  1.21x slower │
│ QQuery 20 │    521.67 / 531.05 ±11.99 / 552.73 ms │    523.41 / 541.68 ±11.96 / 556.34 ms │     no change │
│ QQuery 21 │     526.48 / 536.73 ±9.62 / 553.01 ms │    519.98 / 529.36 ±10.57 / 549.28 ms │     no change │
│ QQuery 22 │ 1000.65 / 1020.79 ±13.57 / 1040.98 ms │  993.77 / 1012.67 ±18.06 / 1046.45 ms │     no change │
│ QQuery 23 │ 3136.79 / 3194.10 ±37.65 / 3245.70 ms │ 3137.60 / 3167.68 ±21.99 / 3199.78 ms │     no change │
│ QQuery 24 │        41.52 / 42.58 ±1.38 / 45.19 ms │        40.10 / 42.84 ±3.46 / 49.61 ms │     no change │
│ QQuery 25 │     114.02 / 118.66 ±4.02 / 125.95 ms │     109.35 / 111.15 ±1.66 / 113.70 ms │ +1.07x faster │
│ QQuery 26 │        41.67 / 42.95 ±0.79 / 44.01 ms │        40.59 / 41.86 ±1.13 / 43.34 ms │     no change │
│ QQuery 27 │     518.60 / 529.34 ±6.53 / 538.10 ms │     533.47 / 545.42 ±7.44 / 555.30 ms │     no change │
│ QQuery 28 │  2989.39 / 2998.00 ±8.93 / 3013.56 ms │ 3009.32 / 3037.05 ±23.91 / 3071.72 ms │     no change │
│ QQuery 29 │       42.54 / 63.03 ±18.73 / 91.47 ms │       41.94 / 58.99 ±21.93 / 95.91 ms │ +1.07x faster │
│ QQuery 30 │     311.57 / 318.82 ±8.16 / 333.79 ms │     336.47 / 345.26 ±6.95 / 356.63 ms │  1.08x slower │
│ QQuery 31 │     285.21 / 292.98 ±6.45 / 304.77 ms │     286.45 / 293.95 ±5.59 / 300.51 ms │     no change │
│ QQuery 32 │  956.20 / 1000.27 ±36.96 / 1067.24 ms │  981.79 / 1007.80 ±28.93 / 1061.44 ms │     no change │
│ QQuery 33 │ 1471.69 / 1517.71 ±46.25 / 1604.72 ms │ 1682.47 / 1719.67 ±39.58 / 1795.40 ms │  1.13x slower │
│ QQuery 34 │ 1593.22 / 1608.44 ±11.93 / 1628.97 ms │ 1703.10 / 1800.87 ±60.69 / 1876.95 ms │  1.12x slower │
│ QQuery 35 │    291.11 / 323.85 ±30.02 / 380.45 ms │    382.71 / 446.93 ±59.31 / 536.92 ms │  1.38x slower │
│ QQuery 36 │        67.23 / 78.35 ±6.58 / 85.16 ms │        71.78 / 81.13 ±8.01 / 92.46 ms │     no change │
│ QQuery 37 │        38.32 / 40.38 ±2.57 / 45.42 ms │        37.07 / 42.02 ±3.90 / 48.02 ms │     no change │
│ QQuery 38 │        44.02 / 45.83 ±1.81 / 49.32 ms │        40.78 / 47.30 ±7.14 / 60.92 ms │     no change │
│ QQuery 39 │    142.99 / 157.71 ±10.98 / 170.96 ms │     159.40 / 164.43 ±4.62 / 172.80 ms │     no change │
│ QQuery 40 │        14.54 / 14.85 ±0.37 / 15.56 ms │        13.13 / 13.46 ±0.38 / 14.20 ms │ +1.10x faster │
│ QQuery 41 │        14.18 / 15.76 ±2.62 / 20.95 ms │        12.42 / 12.80 ±0.36 / 13.24 ms │ +1.23x faster │
│ QQuery 42 │        13.83 / 14.07 ±0.17 / 14.22 ms │        12.92 / 14.58 ±2.99 / 20.55 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 19973.29ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 21268.55ms │
│ Average Time (HEAD)                         │   464.50ms │
│ Average Time (add-blocks-impl-from-scratch) │   494.62ms │
│ Queries Faster                              │          6 │
│ Queries Slower                              │         15 │
│ Queries with No Change                      │         22 │
│ Queries with Failure                        │          0 │
└─────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	105.0s
Peak memory	11.7 GiB
Avg memory	4.5 GiB
CPU user	1022.1s
CPU sys	74.1s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	110.0s
Peak memory	11.4 GiB
Avg memory	4.6 GiB
CPU user	1094.0s
CPU sys	77.9s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-09-09T19:20:13Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing add-blocks-impl-from-scratch (78834e5) to 4048898 (merge-base) diff

Run configuration

run benchmark external_aggr

CPU Details (lscpu)

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

Details

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark external_aggr.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query        ┃      HEAD ┃ add-blocks-impl-from-scratch ┃       Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ Q1(64.0 MB)  │  51.57 ms │                     64.53 ms │ 1.25x slower │
│ Q1(32.0 MB)  │  51.31 ms │                     55.53 ms │ 1.08x slower │
│ Q1(16.0 MB)  │  47.90 ms │                     52.53 ms │ 1.10x slower │
│ Q2(512.0 MB) │ 277.39 ms │                    291.35 ms │ 1.05x slower │
│ Q2(256.0 MB) │ 254.93 ms │                    266.64 ms │    no change │
│ Q2(128.0 MB) │ 239.40 ms │                    264.40 ms │ 1.10x slower │
│ Q2(64.0 MB)  │ 237.10 ms │                    260.94 ms │ 1.10x slower │
│ Q2(32.0 MB)  │ 293.12 ms │                    319.73 ms │ 1.09x slower │
└──────────────┴───────────┴──────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 1452.72ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 1575.64ms │
│ Average Time (HEAD)                         │  181.59ms │
│ Average Time (add-blocks-impl-from-scratch) │  196.96ms │
│ Queries Faster                              │         0 │
│ Queries Slower                              │         7 │
│ Queries with No Change                      │         1 │
│ Queries with Failure                        │         0 │
└─────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and add-blocks-impl-from-scratch
--------------------
Benchmark external_aggr.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query        ┃                               HEAD ┃       add-blocks-impl-from-scratch ┃       Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ Q1(64.0 MB)  │     51.57 / 54.29 ±2.57 / 59.17 ms │     64.53 / 68.85 ±5.10 / 78.69 ms │ 1.27x slower │
│ Q1(32.0 MB)  │     51.31 / 52.10 ±0.45 / 52.53 ms │     55.53 / 56.45 ±1.35 / 59.13 ms │ 1.08x slower │
│ Q1(16.0 MB)  │     47.90 / 50.39 ±1.39 / 51.90 ms │     52.53 / 54.26 ±2.06 / 58.19 ms │ 1.08x slower │
│ Q2(512.0 MB) │ 277.39 / 288.72 ±10.82 / 305.67 ms │ 291.35 / 303.78 ±12.84 / 324.98 ms │ 1.05x slower │
│ Q2(256.0 MB) │ 254.93 / 285.34 ±23.12 / 315.67 ms │  266.64 / 271.33 ±2.70 / 274.60 ms │    no change │
│ Q2(128.0 MB) │  239.40 / 245.30 ±7.30 / 257.94 ms │  264.40 / 268.77 ±5.06 / 278.33 ms │ 1.10x slower │
│ Q2(64.0 MB)  │  237.10 / 238.44 ±1.11 / 239.79 ms │  260.94 / 264.56 ±4.00 / 272.16 ms │ 1.11x slower │
│ Q2(32.0 MB)  │  293.12 / 298.12 ±3.79 / 302.35 ms │  319.73 / 323.91 ±3.99 / 329.31 ms │ 1.09x slower │
└──────────────┴────────────────────────────────────┴────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                           ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                           │ 1512.69ms │
│ Total Time (add-blocks-impl-from-scratch)   │ 1611.90ms │
│ Average Time (HEAD)                         │  189.09ms │
│ Average Time (add-blocks-impl-from-scratch) │  201.49ms │
│ Queries Faster                              │         0 │
│ Queries Slower                              │         7 │
│ Queries with No Change                      │         1 │
│ Queries with Failure                        │         0 │
└─────────────────────────────────────────────┴───────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only when the benchmark runs with DATAFUSION_RUNTIME_MEMORY_LIMIT set.

Base: 4048898 (merge-base) | Changed: add-blocks-impl-from-scratch

external_aggr

Query	Base	Changed	Change
1(64.0 MB)	35.0 MiB	29.2 MiB	-16.5%
1(32.0 MB)	18.6 MiB	18.6 MiB	-0.1%
1(16.0 MB)	11.4 MiB	13.4 MiB	+17.5%
2(512.0 MB)	136.2 MiB	186.3 MiB	+36.8%
2(256.0 MB)	97.9 MiB	100.1 MiB	+2.2%
2(128.0 MB)	49.4 MiB	52.7 MiB	+6.8%
2(64.0 MB)	30.5 MiB	29.0 MiB	-5.1%
2(32.0 MB)	30.0 MiB	30.0 MiB	+0.0%

Pool accounting vs. process RSS

Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
external_aggr	base (`4048898 (merge-base)`)	136.2 MiB	435.7 MiB	299.5 MiB	3.2×
external_aggr	changed (`add-blocks-impl-from-scratch`)	186.3 MiB	531.6 MiB	345.3 MiB	2.9×

Resource Usage

external_aggr — base (merge-base)

Metric	Value
Wall time	525.1s
Peak memory	435.7 MiB
Avg memory	9.2 MiB
CPU user	25.7s
CPU sys	3.7s
Peak spill	0 B

external_aggr — branch

Metric	Value
Wall time	560.1s
Peak memory	531.6 MiB
Avg memory	10.1 MiB
CPU user	27.1s
CPU sys	3.2s
Peak spill	0 B

File an issue against this benchmark runner

# Conflicts: # datafusion/functions-aggregate-common/src/aggregate/groups_accumulator/accumulate.rs # datafusion/functions-aggregate/src/average.rs

…size like their non blocked counterparts

rluvaton added 30 commits August 31, 2026 10:54

copy paste aggregates as is to aggregates_blocked

5cdf7d5

update imports and make it compile

cc0e79a

add steps

5979c90

use existing structs from aggregate to avoid full duplication

16e3848

rename AggregateExec in blocked to BlockedAggregateExec

1a2e531

update AggregateExec::try_new to return the BlockedAggregateExec …

f0f7b8d

…so all usages will be migrated

add fallback to BlockedAggregateExec so unsupported logic will be for…

672ebb5

…warded

Rename GroupValues to BlockedGroupValues and GroupColumn to `Bl…

41b926c

…ockedGroupColumn`

Add BlockedIndex and BlockedEmitTo

358d618

Change in group indices to be stored internally as BlocksIndex and …

464f8fe

…not `usize`

Remove all BlockedGroupValues impl and replace with adapter

e5d3895

Add batch_size to BlockedGroupValues for later emit change

1e9c6d1

fallback to non blocked in grouped_hash_stream and removed duplicat…

2ae5d2c

…e grouped_hash_stream code

add blocked emit to

6970f8c

changed intern to get BlocksIndex and not usize

a605809

create blocked groups accumulator

e6cdab2

remove print

d6a5479

fix compile for tests

50796d5

add group indices to ordering

f76ba2a

Merge branch 'main' into add-blocks-impl-from-scratch

22e0f32

# Conflicts: # datafusion/expr-common/src/groups_accumulator.rs

fix emit too much

9082071

add blocked group by boolean

0c67d68

add state_preserving to blocked group accumulator adapter

923ee62

add evaluate_preserving to blocked group accumulator adapter

0aeb359

fix compile for BlockedGroupsSelection

7fb5b7a

impl blocked group accumulator for sum

f511677

fix group by boolean

11216d2

add support for blocked min max

ee5f265

add more blocked helpers and tests

98e7c1a

re-export and fix clippy

e34750f

rluvaton added 2 commits September 10, 2026 16:19

add iter and iter mut, undo index mut for heap based

e280a96

Merge branch 'main' into add-blocks-impl-from-scratch

e831736

# Conflicts: # datafusion/functions-aggregate-common/src/aggregate/groups_accumulator/accumulate.rs # datafusion/functions-aggregate/src/average.rs

rluvaton mentioned this pull request Sep 14, 2026

feat: blocked hash aggregation state via an independent address-resolve pass, with block-wise output release #25246

Draft

rluvaton added 4 commits September 15, 2026 13:50

fix blocked rows builder to have working take_n

a5b888e

impl row fallback group by

033be42

rename and update comments

5d6018b

add list and row backed blocked group values in multi group by

db6903c

rluvaton force-pushed the add-blocks-impl-from-scratch branch from f935f74 to db6903c Compare September 16, 2026 17:03

rluvaton added 9 commits September 16, 2026 22:15

fixed take blocked to be as expected by the user

476cc63

fix bugs in used block or not

10cfdcb

rename and remove - 1

5304918

fix size calculation

f298eac

changes from 2 usize to one u32

f8f0bd0

remove block size from index in block and block index

9be21d4

remove locate

e9cbf80

change blocked binary map and binary view map to hash map allocation …

cbf0485

…size like their non blocked counterparts

fix metrics sync and dynamic filter

13d8603

rluvaton mentioned this pull request Sep 27, 2026

feat(agg): add initial Blocked aggregate api #25707

Draft

jayzhan211 mentioned this pull request Sep 30, 2026

feat: blocked agg with dynamic flat and blocked #25877

Open

Navigation Menu

Repository navigation

aggregate blocked impl - #24928

aggregate blocked impl#24928
rluvaton wants to merge 95 commits into
apache:mainfrom
rluvaton:add-blocks-impl-from-scratch

rluvaton commented Sep 3, 2026 •

edited

Loading

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

rluvaton commented Sep 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

What is the testing strategy for this PR?

Are there any user-facing changes?

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

adriangbot commented Sep 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

rluvaton commented Sep 3, 2026 •

edited

Loading