Optimize existence-only right semi and anti joins by Dandandan · Pull Request #22828 · apache/datafusion

Dandandan · 2026-06-08T19:15:30Z

Which issue does this PR close?

None.

Rationale for this change

Right semi, right anti, and right mark joins only need to know whether each probe-side row has at least one build-side match. The existing hash join probe path can materialize every build-side match before the semi/anti/mark output projection, which is unnecessarily expensive for high-fanout build keys.

What changes are included in this PR?

Add existence-only lookup helpers for JoinHashMap and ArrayMap that return at most one build-side match per probe row.
Use those helpers for unfiltered RightSemi, RightAnti, and RightMark hash joins.
Keep unique-hash JoinHashMap probes on the existing path to avoid comparator overhead when there are no chains to skip.
Keep filtered joins on the existing path because join filters must evaluate candidate pairs.
Add focused tests for hash collisions and ArrayMap duplicate-key behavior.
Add a RightMark high-fanout benchmark case.

Are these changes tested?

cargo fmt --all
cargo test -p datafusion-physical-plan --lib join_right -- --nocapture
cargo test -p datafusion-physical-plan --lib matching_indices -- --nocapture
cargo clippy --all-targets --all-features -- -D warnings
cargo bench -p datafusion-physical-plan --features test_utils --bench hash_join_semi_anti -- --baseline main-inplace --sample-size 10 --warm-up-time 1 --measurement-time 2
cargo bench -p datafusion-physical-plan --features test_utils --bench hash_join_semi_anti -- right_mark_fanout100_h1 --sample-size 10 --warm-up-time 1 --measurement-time 2

Selected Criterion results:

Benchmark	main median	PR median	change
right_semi_d100_h10	1.5872 ms	1.5070 ms	-3.70%
right_semi_d50_h10	1.6041 ms	1.4900 ms	-7.03%
right_semi_fanout100_h1	3.9918 ms	820.04 us	-79.47%
right_mark_fanout100_h1	10.940 ms	5.6948 ms	-47.94%
right_anti_d100_h100	2.6888 ms	2.4691 ms	-8.49%
right_anti_d50_h100	2.7308 ms	2.4674 ms	-9.86%
right_anti_d50_h10	6.4212 ms	6.3363 ms	-2.32%
sparse d10 cases	unchanged	unchanged	no change detected

Are there any user-facing changes?

No.

github-actions · 2026-06-08T19:19:06Z

Thank you for opening this pull request!

Reviewer note: cargo-semver-checks reported the current version number is not SemVer-compatible with the changes in this pull request (compared against the base branch).

Details

     Cloning apache/main
    Building datafusion-physical-plan v53.1.0 (current)
       Built [  36.406s] (current)
     Parsing datafusion-physical-plan v53.1.0 (current)
      Parsed [   0.127s] (current)
    Building datafusion-physical-plan v53.1.0 (baseline)
       Built [  34.817s] (baseline)
     Parsing datafusion-physical-plan v53.1.0 (baseline)
      Parsed [   0.126s] (baseline)
    Checking datafusion-physical-plan v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   0.585s] 223 checks: 222 pass, 1 fail, 0 warn, 30 skip

--- failure trait_method_added: pub trait method added ---

Description:
A non-sealed public trait added a new method without a default implementation, which breaks downstream implementations of the trait
        ref: https://doc.rust-lang.org/cargo/reference/semver.html#trait-new-item-no-default
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.48.0/src/lints/trait_method_added.ron

Failed in:
  trait method datafusion_physical_plan::joins::utils::JoinHashMapType::get_matching_indices_with_limit_offset in file /home/runner/work/datafusion/datafusion/datafusion/physical-plan/src/joins/join_hash_map.rs:134
  trait method datafusion_physical_plan::joins::utils::JoinHashMapType::may_contain_hash_chains in file /home/runner/work/datafusion/datafusion/datafusion/physical-plan/src/joins/join_hash_map.rs:153
  trait method datafusion_physical_plan::joins::join_hash_map::JoinHashMapType::get_matching_indices_with_limit_offset in file /home/runner/work/datafusion/datafusion/datafusion/physical-plan/src/joins/join_hash_map.rs:134
  trait method datafusion_physical_plan::joins::join_hash_map::JoinHashMapType::may_contain_hash_chains in file /home/runner/work/datafusion/datafusion/datafusion/physical-plan/src/joins/join_hash_map.rs:153

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [  73.871s] datafusion-physical-plan

Dandandan · 2026-06-08T19:40:25Z

run benchmark hash_join_semi_anti tpch10 tpcds

adriangbot · 2026-06-08T19:42:18Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4652769700-488-vcsdl 6.12.68+ #1 SMP Sat May 2 07:49:07 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 codex/existence-semi-join-kernel (72363e8) to 883c38e (merge-base) diff using: tpch10
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-08T19:42:56Z

🤖 Criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4652769700-487-hb5d8 6.12.68+ #1 SMP Sat May 2 07:49:07 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 codex/existence-semi-join-kernel (72363e8) to 883c38e (merge-base) diff
BENCH_NAME=hash_join_semi_anti
BENCH_COMMAND=cargo bench --features=parquet --bench hash_join_semi_anti
BENCH_FILTER=
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-08T19:43:01Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

Cloning into '/workspace/datafusion-branch'...
From https://github.com/apache/datafusion
 * [new ref]         refs/pull/22828/head -> codex/existence-semi-join-kernel
 * branch            main                 -> FETCH_HEAD
Switched to branch 'codex/existence-semi-join-kernel'
883c38ee022ccd3e77cd0d4a4647e6a304762faf
Cloning into '/workspace/datafusion-base'...
HEAD is now at 883c38e docs: add Boston DataFusion meetup (#22722)
rustc 1.95.0 (59807616e 2026-04-14)
72363e89e4f38ebcab3b9c415766d4abc7f19681
883c38ee022ccd3e77cd0d4a4647e6a304762faf
    Blocking waiting for file lock on package cache
    Blocking waiting for file lock on package cache
    Blocking waiting for file lock on package cache
error: target `hash_join_semi_anti` in package `datafusion-physical-plan` requires the features: `test_utils`
Consider enabling them by passing, e.g., `--features="test_utils"`

File an issue against this benchmark runner

adriangbot · 2026-06-08T19:43:10Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4652769700-489-jkp5r 6.12.68+ #1 SMP Sat May 2 07:49:07 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 codex/existence-semi-join-kernel (72363e8) to 883c38e (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

mbutrovich · 2026-06-08T19:46:10Z

Rationale for this change

Right semi and right anti joins only need to know whether each probe-side row has at least one build-side match. The existing hash join probe path can materialize every build-side match before the semi/anti output projection, which is unnecessarily expensive for high-fanout build keys.

Seems similar to what I did for SMJ in #20806. There was a ton of wasted materialization work before. I later used mark to the same stream (#21184), can this optimization benefit mark as well?

adriangbot · 2026-06-08T19:57:48Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and codex_existence-semi-join-kernel
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                               HEAD ┃   codex_existence-semi-join-kernel ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │  311.93 / 314.06 ±1.09 / 314.94 ms │  314.90 / 316.89 ±1.32 / 318.79 ms │    no change │
│ QQuery 2  │  100.07 / 104.99 ±3.30 / 108.67 ms │  101.10 / 105.45 ±2.62 / 108.97 ms │    no change │
│ QQuery 3  │  239.74 / 242.46 ±2.84 / 247.43 ms │  236.05 / 238.81 ±2.08 / 241.56 ms │    no change │
│ QQuery 4  │  116.38 / 117.19 ±0.66 / 118.40 ms │  118.50 / 118.99 ±0.48 / 119.77 ms │    no change │
│ QQuery 5  │  342.08 / 351.30 ±6.74 / 362.47 ms │  344.48 / 355.47 ±5.88 / 361.34 ms │    no change │
│ QQuery 6  │  124.66 / 127.94 ±3.60 / 134.83 ms │  127.36 / 130.87 ±3.04 / 135.75 ms │    no change │
│ QQuery 7  │  457.18 / 464.12 ±6.08 / 475.48 ms │ 461.02 / 480.55 ±11.08 / 492.43 ms │    no change │
│ QQuery 8  │  381.08 / 393.08 ±7.96 / 400.92 ms │  377.42 / 385.03 ±4.27 / 390.23 ms │    no change │
│ QQuery 9  │  547.16 / 553.85 ±5.54 / 562.42 ms │  540.95 / 549.77 ±6.24 / 557.17 ms │    no change │
│ QQuery 10 │ 303.21 / 322.81 ±14.12 / 338.07 ms │  308.89 / 320.34 ±9.41 / 337.63 ms │    no change │
│ QQuery 11 │   82.69 / 94.10 ±11.33 / 113.03 ms │    84.14 / 90.15 ±6.53 / 102.60 ms │    no change │
│ QQuery 12 │  177.38 / 182.65 ±7.28 / 196.93 ms │  176.49 / 183.01 ±6.36 / 194.57 ms │    no change │
│ QQuery 13 │  281.03 / 292.58 ±6.89 / 301.47 ms │ 284.97 / 302.79 ±10.14 / 315.53 ms │    no change │
│ QQuery 14 │  171.49 / 179.38 ±6.97 / 189.19 ms │  173.34 / 178.83 ±4.93 / 185.55 ms │    no change │
│ QQuery 15 │  311.19 / 314.64 ±4.04 / 321.47 ms │  308.77 / 313.65 ±4.15 / 319.91 ms │    no change │
│ QQuery 16 │     64.22 / 67.18 ±1.85 / 69.45 ms │     66.50 / 68.95 ±1.71 / 71.52 ms │    no change │
│ QQuery 17 │  629.72 / 643.51 ±9.20 / 657.71 ms │  625.25 / 632.08 ±9.15 / 649.81 ms │    no change │
│ QQuery 18 │ 688.11 / 702.29 ±11.38 / 715.20 ms │ 671.62 / 693.18 ±17.57 / 721.49 ms │    no change │
│ QQuery 19 │ 248.71 / 273.78 ±28.37 / 317.98 ms │ 248.81 / 269.82 ±24.45 / 308.86 ms │    no change │
│ QQuery 20 │  289.89 / 298.86 ±5.60 / 305.56 ms │ 278.37 / 289.11 ±11.50 / 309.71 ms │    no change │
│ QQuery 21 │ 769.23 / 807.08 ±29.94 / 846.01 ms │ 751.87 / 778.13 ±24.67 / 814.87 ms │    no change │
│ QQuery 22 │   62.03 / 71.24 ±15.32 / 101.83 ms │   59.34 / 84.83 ±29.64 / 130.25 ms │ 1.19x slower │
└───────────┴────────────────────────────────────┴────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                               ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                               │ 6919.12ms │
│ Total Time (codex_existence-semi-join-kernel)   │ 6886.69ms │
│ Average Time (HEAD)                             │  314.51ms │
│ Average Time (codex_existence-semi-join-kernel) │  313.03ms │
│ Queries Faster                                  │         0 │
│ Queries Slower                                  │         1 │
│ Queries with No Change                          │        21 │
│ Queries with Failure                            │         0 │
└─────────────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — base (merge-base)

Metric	Value
Wall time	35.0s
Peak memory	12.4 GiB
Avg memory	8.9 GiB
CPU user	357.2s
CPU sys	22.6s
Peak spill	0 B

tpch10 — branch

Metric	Value
Wall time	35.0s
Peak memory	12.4 GiB
Avg memory	8.9 GiB
CPU user	357.1s
CPU sys	22.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-06-08T19:58:51Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and codex_existence-semi-join-kernel
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃      codex_existence-semi-join-kernel ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           6.15 / 6.61 ±0.88 / 8.37 ms │           5.84 / 6.29 ±0.83 / 7.94 ms │     no change │
│ QQuery 2  │        82.54 / 82.84 ±0.36 / 83.54 ms │        80.33 / 80.78 ±0.26 / 81.13 ms │     no change │
│ QQuery 3  │        29.28 / 30.02 ±0.48 / 30.78 ms │        29.06 / 29.27 ±0.16 / 29.54 ms │     no change │
│ QQuery 4  │     563.40 / 569.58 ±4.38 / 574.26 ms │     497.72 / 505.67 ±5.20 / 513.55 ms │ +1.13x faster │
│ QQuery 5  │        54.36 / 55.13 ±0.91 / 56.84 ms │        52.12 / 52.84 ±0.60 / 53.93 ms │     no change │
│ QQuery 6  │        39.67 / 39.90 ±0.12 / 40.00 ms │        36.72 / 37.11 ±0.23 / 37.37 ms │ +1.08x faster │
│ QQuery 7  │      99.53 / 101.01 ±0.85 / 101.80 ms │       95.37 / 97.52 ±3.23 / 103.85 ms │     no change │
│ QQuery 8  │        38.82 / 39.30 ±0.46 / 40.15 ms │        37.17 / 37.66 ±0.44 / 38.48 ms │     no change │
│ QQuery 9  │        55.38 / 58.36 ±1.98 / 61.39 ms │        52.80 / 54.49 ±0.99 / 55.76 ms │ +1.07x faster │
│ QQuery 10 │        65.64 / 68.33 ±3.97 / 76.20 ms │        63.17 / 63.51 ±0.25 / 63.92 ms │ +1.08x faster │
│ QQuery 11 │     362.31 / 369.61 ±7.21 / 382.77 ms │     354.29 / 360.51 ±3.27 / 363.99 ms │     no change │
│ QQuery 12 │        31.82 / 32.09 ±0.20 / 32.30 ms │        30.83 / 31.24 ±0.43 / 31.94 ms │     no change │
│ QQuery 13 │     126.34 / 127.36 ±0.88 / 128.52 ms │     125.84 / 126.64 ±0.90 / 128.37 ms │     no change │
│ QQuery 14 │     430.32 / 433.10 ±2.51 / 437.33 ms │     418.51 / 423.83 ±4.95 / 430.32 ms │     no change │
│ QQuery 15 │        59.29 / 59.92 ±0.61 / 60.97 ms │        68.99 / 70.99 ±1.35 / 72.72 ms │  1.18x slower │
│ QQuery 16 │           6.89 / 7.15 ±0.26 / 7.64 ms │           7.47 / 7.62 ±0.18 / 7.98 ms │  1.07x slower │
│ QQuery 17 │        81.67 / 82.32 ±0.77 / 83.82 ms │        85.09 / 86.34 ±1.81 / 89.94 ms │     no change │
│ QQuery 18 │     126.31 / 133.81 ±3.82 / 136.97 ms │     132.91 / 134.51 ±2.24 / 138.84 ms │     no change │
│ QQuery 19 │        44.60 / 44.80 ±0.13 / 44.96 ms │        44.22 / 45.24 ±1.51 / 48.23 ms │     no change │
│ QQuery 20 │        39.20 / 39.60 ±0.25 / 39.99 ms │        38.45 / 39.28 ±0.73 / 40.53 ms │     no change │
│ QQuery 21 │        18.25 / 18.37 ±0.11 / 18.55 ms │        17.91 / 18.54 ±0.89 / 20.30 ms │     no change │
│ QQuery 22 │        71.11 / 72.06 ±0.51 / 72.60 ms │        70.46 / 70.96 ±0.36 / 71.41 ms │     no change │
│ QQuery 23 │    330.50 / 355.46 ±16.47 / 370.92 ms │     352.49 / 361.57 ±6.03 / 367.44 ms │     no change │
│ QQuery 24 │     228.60 / 232.66 ±5.17 / 241.48 ms │     226.58 / 229.63 ±2.95 / 233.29 ms │     no change │
│ QQuery 25 │     110.87 / 113.12 ±1.81 / 116.27 ms │     111.90 / 113.73 ±1.56 / 115.96 ms │     no change │
│ QQuery 26 │        58.83 / 60.86 ±3.12 / 67.07 ms │        58.93 / 59.31 ±0.38 / 59.94 ms │     no change │
│ QQuery 27 │           6.73 / 6.89 ±0.13 / 7.12 ms │           6.82 / 6.94 ±0.17 / 7.28 ms │     no change │
│ QQuery 28 │        57.28 / 60.67 ±1.71 / 61.74 ms │        61.86 / 63.44 ±2.15 / 67.57 ms │     no change │
│ QQuery 29 │        98.42 / 98.98 ±0.46 / 99.81 ms │     104.44 / 105.34 ±1.27 / 107.78 ms │  1.06x slower │
│ QQuery 30 │        32.87 / 35.03 ±3.29 / 41.58 ms │        35.93 / 36.41 ±0.45 / 37.09 ms │     no change │
│ QQuery 31 │     113.26 / 114.78 ±1.54 / 117.22 ms │     120.26 / 122.54 ±1.96 / 125.51 ms │  1.07x slower │
│ QQuery 32 │        20.87 / 21.04 ±0.15 / 21.25 ms │        22.45 / 22.62 ±0.19 / 22.98 ms │  1.07x slower │
│ QQuery 33 │        38.81 / 39.59 ±0.54 / 40.43 ms │        41.13 / 41.80 ±0.67 / 43.09 ms │  1.06x slower │
│ QQuery 34 │         9.79 / 12.47 ±4.15 / 20.73 ms │        10.61 / 11.12 ±0.64 / 12.31 ms │ +1.12x faster │
│ QQuery 35 │        72.96 / 73.48 ±0.60 / 74.61 ms │        80.56 / 82.01 ±1.86 / 85.67 ms │  1.12x slower │
│ QQuery 36 │           6.00 / 6.14 ±0.15 / 6.43 ms │           6.57 / 6.68 ±0.09 / 6.81 ms │  1.09x slower │
│ QQuery 37 │           7.08 / 7.14 ±0.05 / 7.20 ms │           7.65 / 7.81 ±0.16 / 8.03 ms │  1.09x slower │
│ QQuery 38 │        64.56 / 65.24 ±0.37 / 65.60 ms │        70.80 / 71.32 ±0.41 / 71.89 ms │  1.09x slower │
│ QQuery 39 │     461.86 / 468.39 ±6.77 / 479.51 ms │    458.85 / 479.13 ±17.74 / 502.02 ms │     no change │
│ QQuery 40 │        23.63 / 23.96 ±0.34 / 24.59 ms │        23.09 / 24.29 ±1.63 / 27.52 ms │     no change │
│ QQuery 41 │        11.53 / 11.73 ±0.19 / 12.07 ms │        11.17 / 11.40 ±0.17 / 11.69 ms │     no change │
│ QQuery 42 │        24.36 / 25.63 ±1.75 / 29.03 ms │        24.58 / 24.91 ±0.38 / 25.61 ms │     no change │
│ QQuery 43 │           4.92 / 5.00 ±0.12 / 5.24 ms │           4.87 / 4.96 ±0.13 / 5.22 ms │     no change │
│ QQuery 44 │           9.32 / 9.44 ±0.09 / 9.56 ms │           9.13 / 9.22 ±0.07 / 9.34 ms │     no change │
│ QQuery 45 │        39.82 / 41.19 ±1.04 / 43.00 ms │        38.35 / 39.41 ±0.81 / 40.45 ms │     no change │
│ QQuery 46 │        12.16 / 12.38 ±0.21 / 12.76 ms │        11.76 / 11.88 ±0.20 / 12.29 ms │     no change │
│ QQuery 47 │     236.60 / 238.77 ±2.78 / 244.24 ms │     233.12 / 239.97 ±5.26 / 249.30 ms │     no change │
│ QQuery 48 │      98.18 / 100.42 ±2.11 / 103.85 ms │        97.22 / 97.77 ±0.50 / 98.45 ms │     no change │
│ QQuery 49 │        78.28 / 78.88 ±0.50 / 79.79 ms │        77.26 / 78.47 ±0.97 / 80.23 ms │     no change │
│ QQuery 50 │        59.85 / 63.21 ±4.74 / 72.60 ms │        59.72 / 60.22 ±0.47 / 61.03 ms │     no change │
│ QQuery 51 │        92.06 / 94.77 ±1.64 / 97.18 ms │       93.31 / 95.56 ±2.78 / 100.97 ms │     no change │
│ QQuery 52 │        24.20 / 24.80 ±0.36 / 25.32 ms │        24.18 / 24.52 ±0.26 / 24.98 ms │     no change │
│ QQuery 53 │        30.10 / 30.34 ±0.22 / 30.68 ms │        29.93 / 30.09 ±0.11 / 30.25 ms │     no change │
│ QQuery 54 │        56.08 / 59.01 ±4.37 / 67.69 ms │        55.97 / 58.15 ±3.58 / 65.29 ms │     no change │
│ QQuery 55 │        24.36 / 24.65 ±0.31 / 25.15 ms │        24.46 / 24.87 ±0.35 / 25.31 ms │     no change │
│ QQuery 56 │        39.45 / 39.99 ±0.55 / 40.91 ms │        39.44 / 39.99 ±0.59 / 41.07 ms │     no change │
│ QQuery 57 │     183.32 / 190.18 ±3.72 / 193.32 ms │     177.66 / 180.49 ±3.33 / 186.91 ms │ +1.05x faster │
│ QQuery 58 │     122.00 / 123.91 ±1.98 / 127.65 ms │     116.00 / 116.42 ±0.27 / 116.83 ms │ +1.06x faster │
│ QQuery 59 │     122.66 / 123.21 ±0.40 / 123.64 ms │     117.85 / 120.51 ±2.69 / 125.63 ms │     no change │
│ QQuery 60 │        40.22 / 41.23 ±0.87 / 42.56 ms │        39.63 / 40.55 ±0.62 / 41.54 ms │     no change │
│ QQuery 61 │        13.49 / 15.14 ±3.08 / 21.29 ms │        13.29 / 13.40 ±0.14 / 13.67 ms │ +1.13x faster │
│ QQuery 62 │        47.00 / 47.18 ±0.23 / 47.62 ms │        46.40 / 46.79 ±0.34 / 47.31 ms │     no change │
│ QQuery 63 │        30.31 / 30.90 ±0.48 / 31.32 ms │        29.97 / 30.53 ±0.53 / 31.51 ms │     no change │
│ QQuery 64 │     406.18 / 418.92 ±8.34 / 432.06 ms │     396.86 / 405.01 ±7.65 / 418.62 ms │     no change │
│ QQuery 65 │     147.68 / 152.83 ±5.37 / 162.10 ms │     149.11 / 151.94 ±2.57 / 156.06 ms │     no change │
│ QQuery 66 │        81.00 / 83.71 ±3.06 / 89.67 ms │        85.62 / 90.04 ±4.82 / 98.27 ms │  1.08x slower │
│ QQuery 67 │     254.75 / 263.63 ±6.60 / 273.74 ms │    265.36 / 290.31 ±13.47 / 302.48 ms │  1.10x slower │
│ QQuery 68 │        12.31 / 15.01 ±3.61 / 22.02 ms │        11.97 / 12.14 ±0.20 / 12.50 ms │ +1.24x faster │
│ QQuery 69 │        58.16 / 58.44 ±0.35 / 59.07 ms │        57.65 / 60.77 ±3.54 / 65.32 ms │     no change │
│ QQuery 70 │     104.89 / 109.25 ±3.62 / 115.85 ms │     105.94 / 111.36 ±3.31 / 116.38 ms │     no change │
│ QQuery 71 │        36.13 / 36.74 ±0.41 / 37.20 ms │        36.39 / 37.01 ±0.57 / 37.93 ms │     no change │
│ QQuery 72 │ 2168.86 / 2248.43 ±57.30 / 2312.82 ms │ 2192.43 / 2241.87 ±48.89 / 2332.38 ms │     no change │
│ QQuery 73 │         9.82 / 10.12 ±0.25 / 10.56 ms │         9.87 / 12.46 ±4.48 / 21.40 ms │  1.23x slower │
│ QQuery 74 │     181.40 / 183.93 ±2.22 / 187.53 ms │    183.44 / 207.45 ±17.08 / 233.68 ms │  1.13x slower │
│ QQuery 75 │     150.19 / 154.56 ±3.91 / 160.62 ms │     160.66 / 161.63 ±0.61 / 162.39 ms │     no change │
│ QQuery 76 │        36.07 / 36.86 ±0.78 / 38.08 ms │        38.22 / 41.17 ±3.80 / 48.62 ms │  1.12x slower │
│ QQuery 77 │        61.13 / 62.35 ±0.96 / 63.36 ms │        62.96 / 65.62 ±2.06 / 68.67 ms │  1.05x slower │
│ QQuery 78 │     187.67 / 191.82 ±3.69 / 198.41 ms │     188.18 / 193.95 ±7.81 / 209.29 ms │     no change │
│ QQuery 79 │        67.31 / 69.56 ±3.15 / 75.73 ms │        66.82 / 67.41 ±0.39 / 67.85 ms │     no change │
│ QQuery 80 │     102.54 / 103.47 ±0.86 / 104.71 ms │      99.73 / 104.11 ±4.32 / 112.08 ms │     no change │
│ QQuery 81 │        25.72 / 26.48 ±0.43 / 27.01 ms │        25.96 / 26.39 ±0.29 / 26.73 ms │     no change │
│ QQuery 82 │        16.56 / 16.76 ±0.14 / 16.98 ms │        16.72 / 16.94 ±0.17 / 17.21 ms │     no change │
│ QQuery 83 │        40.56 / 43.33 ±4.70 / 52.70 ms │        40.06 / 40.37 ±0.23 / 40.76 ms │ +1.07x faster │
│ QQuery 84 │        35.06 / 35.38 ±0.23 / 35.72 ms │        34.97 / 35.12 ±0.13 / 35.30 ms │     no change │
│ QQuery 85 │     109.30 / 112.23 ±3.60 / 119.26 ms │     107.52 / 113.34 ±7.34 / 127.68 ms │     no change │
│ QQuery 86 │        25.58 / 27.19 ±2.33 / 31.80 ms │        25.22 / 25.73 ±0.28 / 25.97 ms │ +1.06x faster │
│ QQuery 87 │        64.64 / 65.67 ±0.58 / 66.26 ms │        64.79 / 65.05 ±0.26 / 65.43 ms │     no change │
│ QQuery 88 │        63.39 / 63.84 ±0.54 / 64.84 ms │        62.44 / 65.90 ±3.89 / 73.32 ms │     no change │
│ QQuery 89 │        36.30 / 36.74 ±0.29 / 37.17 ms │        36.15 / 36.94 ±0.84 / 38.34 ms │     no change │
│ QQuery 90 │        17.26 / 17.39 ±0.24 / 17.88 ms │        16.90 / 17.15 ±0.25 / 17.63 ms │     no change │
│ QQuery 91 │        45.58 / 47.15 ±2.06 / 51.21 ms │        45.22 / 45.45 ±0.20 / 45.77 ms │     no change │
│ QQuery 92 │        30.32 / 30.67 ±0.27 / 31.03 ms │        29.53 / 30.29 ±0.52 / 31.02 ms │     no change │
│ QQuery 93 │        50.73 / 53.43 ±2.66 / 58.34 ms │        51.32 / 53.02 ±1.44 / 55.00 ms │     no change │
│ QQuery 94 │        39.04 / 39.96 ±0.73 / 41.23 ms │        39.50 / 39.95 ±0.38 / 40.44 ms │     no change │
│ QQuery 95 │        84.50 / 86.10 ±1.08 / 87.43 ms │        84.47 / 85.10 ±0.55 / 86.12 ms │     no change │
│ QQuery 96 │        24.45 / 24.67 ±0.19 / 24.94 ms │        24.28 / 24.69 ±0.24 / 24.91 ms │     no change │
│ QQuery 97 │        46.40 / 47.86 ±2.12 / 52.07 ms │        46.55 / 48.87 ±2.38 / 52.99 ms │     no change │
│ QQuery 98 │        43.18 / 43.82 ±0.54 / 44.73 ms │        42.77 / 43.53 ±0.59 / 44.32 ms │     no change │
│ QQuery 99 │        69.72 / 71.66 ±2.86 / 77.36 ms │        69.64 / 70.34 ±0.57 / 71.35 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                               │ 10789.07ms │
│ Total Time (codex_existence-semi-join-kernel)   │ 10765.02ms │
│ Average Time (HEAD)                             │   108.98ms │
│ Average Time (codex_existence-semi-join-kernel) │   108.74ms │
│ Queries Faster                                  │         11 │
│ Queries Slower                                  │         16 │
│ Queries with No Change                          │         72 │
│ Queries with Failure                            │          0 │
└─────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	6.8 GiB
Avg memory	6.2 GiB
CPU user	247.7s
CPU sys	6.9s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	6.8 GiB
Avg memory	6.2 GiB
CPU user	247.1s
CPU sys	6.9s
Peak spill	0 B

File an issue against this benchmark runner

Dandandan · 2026-06-08T20:04:46Z

run benchmark tpcds tpch10

adriangbot · 2026-06-08T20:07:12Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4652990766-492-q4l62 6.12.68+ #1 SMP Sat May 2 07:49:07 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 codex/existence-semi-join-kernel (53bec28) to 883c38e (merge-base) diff using: tpch10
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-08T20:26:03Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and codex_existence-semi-join-kernel
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                               HEAD ┃   codex_existence-semi-join-kernel ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │  329.14 / 330.71 ±0.86 / 331.70 ms │  324.11 / 325.13 ±0.81 / 326.51 ms │    no change │
│ QQuery 2  │  106.20 / 113.96 ±7.38 / 127.18 ms │  106.18 / 109.56 ±1.99 / 111.61 ms │    no change │
│ QQuery 3  │  262.69 / 264.97 ±2.12 / 268.51 ms │  250.52 / 257.26 ±5.65 / 265.05 ms │    no change │
│ QQuery 4  │  123.37 / 129.22 ±3.98 / 134.94 ms │  122.96 / 128.16 ±4.10 / 135.37 ms │    no change │
│ QQuery 5  │  381.97 / 392.50 ±8.02 / 402.40 ms │  384.55 / 395.90 ±7.02 / 405.49 ms │    no change │
│ QQuery 6  │  135.15 / 136.82 ±1.38 / 138.76 ms │  131.83 / 133.99 ±1.77 / 136.94 ms │    no change │
│ QQuery 7  │  513.72 / 526.09 ±6.89 / 533.04 ms │ 514.90 / 532.78 ±12.60 / 550.03 ms │    no change │
│ QQuery 8  │  411.94 / 417.86 ±5.25 / 426.87 ms │  401.99 / 410.06 ±9.36 / 426.44 ms │    no change │
│ QQuery 9  │  595.18 / 602.10 ±6.71 / 610.79 ms │  584.40 / 594.08 ±6.49 / 603.32 ms │    no change │
│ QQuery 10 │  333.76 / 339.71 ±5.81 / 347.01 ms │ 331.06 / 342.81 ±14.39 / 369.59 ms │    no change │
│ QQuery 11 │     91.42 / 92.98 ±1.17 / 94.47 ms │     90.89 / 93.74 ±1.89 / 95.46 ms │    no change │
│ QQuery 12 │ 188.02 / 196.24 ±12.33 / 220.69 ms │  190.63 / 196.65 ±6.60 / 207.46 ms │    no change │
│ QQuery 13 │ 313.79 / 331.13 ±10.73 / 341.88 ms │  317.61 / 327.86 ±5.48 / 334.05 ms │    no change │
│ QQuery 14 │  185.33 / 190.45 ±4.88 / 198.46 ms │  190.50 / 195.50 ±3.71 / 200.12 ms │    no change │
│ QQuery 15 │  322.47 / 329.70 ±3.74 / 332.57 ms │  326.75 / 331.56 ±3.07 / 335.81 ms │    no change │
│ QQuery 16 │     68.40 / 70.30 ±2.00 / 73.90 ms │     69.52 / 74.60 ±5.69 / 85.60 ms │ 1.06x slower │
│ QQuery 17 │  696.96 / 707.28 ±8.61 / 719.72 ms │ 717.13 / 740.50 ±12.66 / 755.00 ms │    no change │
│ QQuery 18 │ 778.36 / 802.71 ±22.11 / 842.82 ms │ 775.65 / 793.07 ±19.02 / 828.97 ms │    no change │
│ QQuery 19 │ 263.42 / 278.75 ±23.78 / 325.84 ms │ 255.06 / 279.04 ±26.05 / 324.22 ms │    no change │
│ QQuery 20 │ 299.53 / 314.47 ±10.22 / 328.44 ms │  302.31 / 316.11 ±7.03 / 321.85 ms │    no change │
│ QQuery 21 │ 844.62 / 880.09 ±20.40 / 901.81 ms │ 855.38 / 887.81 ±20.54 / 913.03 ms │    no change │
│ QQuery 22 │     65.51 / 67.99 ±1.97 / 71.02 ms │     64.02 / 68.14 ±2.94 / 73.09 ms │    no change │
└───────────┴────────────────────────────────────┴────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                               ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                               │ 7516.05ms │
│ Total Time (codex_existence-semi-join-kernel)   │ 7534.29ms │
│ Average Time (HEAD)                             │  341.64ms │
│ Average Time (codex_existence-semi-join-kernel) │  342.47ms │
│ Queries Faster                                  │         0 │
│ Queries Slower                                  │         1 │
│ Queries with No Change                          │        21 │
│ Queries with Failure                            │         0 │
└─────────────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — base (merge-base)

Metric	Value
Wall time	40.0s
Peak memory	12.3 GiB
Avg memory	8.8 GiB
CPU user	390.4s
CPU sys	25.6s
Peak spill	0 B

tpch10 — branch

Metric	Value
Wall time	40.0s
Peak memory	12.3 GiB
Avg memory	8.6 GiB
CPU user	389.9s
CPU sys	25.9s
Peak spill	0 B

File an issue against this benchmark runner

Dandandan · 2026-06-09T05:34:59Z

run benchmark tpcds

adriangbot · 2026-06-09T05:36:57Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4656458499-500-gd6dw 6.12.68+ #1 SMP Sat May 2 07:49:07 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 codex/existence-semi-join-kernel (53bec28) to 883c38e (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-09T05:52:06Z

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and codex_existence-semi-join-kernel
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃      codex_existence-semi-join-kernel ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │           5.71 / 6.21 ±0.90 / 8.00 ms │           5.73 / 6.23 ±0.87 / 7.97 ms │ no change │
│ QQuery 2  │        79.99 / 80.20 ±0.15 / 80.39 ms │        80.82 / 81.07 ±0.15 / 81.24 ms │ no change │
│ QQuery 3  │        28.55 / 28.82 ±0.20 / 29.04 ms │        29.01 / 29.29 ±0.18 / 29.56 ms │ no change │
│ QQuery 4  │     481.47 / 488.17 ±4.35 / 493.33 ms │     482.68 / 487.21 ±3.39 / 492.40 ms │ no change │
│ QQuery 5  │        51.49 / 52.33 ±0.69 / 53.37 ms │        51.04 / 51.53 ±0.28 / 51.88 ms │ no change │
│ QQuery 6  │        35.97 / 36.64 ±0.36 / 36.97 ms │        36.34 / 36.83 ±0.31 / 37.08 ms │ no change │
│ QQuery 7  │        95.56 / 96.10 ±0.68 / 97.42 ms │        95.33 / 95.95 ±0.33 / 96.29 ms │ no change │
│ QQuery 8  │        36.91 / 38.04 ±1.67 / 41.35 ms │        37.11 / 38.11 ±1.22 / 40.48 ms │ no change │
│ QQuery 9  │        52.35 / 55.54 ±1.97 / 57.89 ms │        53.04 / 54.75 ±1.87 / 58.06 ms │ no change │
│ QQuery 10 │        62.65 / 63.10 ±0.33 / 63.65 ms │        62.88 / 63.26 ±0.28 / 63.61 ms │ no change │
│ QQuery 11 │     296.43 / 303.05 ±7.00 / 316.34 ms │     297.14 / 299.72 ±2.04 / 303.28 ms │ no change │
│ QQuery 12 │        28.60 / 28.93 ±0.37 / 29.62 ms │        28.35 / 28.90 ±0.49 / 29.81 ms │ no change │
│ QQuery 13 │     119.40 / 120.02 ±0.58 / 120.91 ms │     119.64 / 120.33 ±0.62 / 121.44 ms │ no change │
│ QQuery 14 │     409.76 / 415.32 ±3.99 / 420.20 ms │     406.34 / 408.19 ±1.86 / 411.64 ms │ no change │
│ QQuery 15 │        59.54 / 59.84 ±0.29 / 60.39 ms │        59.23 / 60.06 ±0.60 / 61.07 ms │ no change │
│ QQuery 16 │           6.80 / 6.91 ±0.14 / 7.17 ms │           6.73 / 6.91 ±0.26 / 7.42 ms │ no change │
│ QQuery 17 │        81.59 / 82.57 ±1.20 / 84.86 ms │        81.04 / 82.36 ±1.40 / 85.06 ms │ no change │
│ QQuery 18 │     124.35 / 125.67 ±1.34 / 127.98 ms │     124.48 / 127.39 ±2.39 / 131.45 ms │ no change │
│ QQuery 19 │        41.60 / 42.00 ±0.27 / 42.42 ms │        41.63 / 42.04 ±0.24 / 42.35 ms │ no change │
│ QQuery 20 │        35.65 / 36.85 ±1.21 / 39.16 ms │        35.44 / 35.81 ±0.26 / 36.03 ms │ no change │
│ QQuery 21 │        16.94 / 17.28 ±0.24 / 17.66 ms │        17.23 / 17.71 ±0.61 / 18.90 ms │ no change │
│ QQuery 22 │        62.81 / 63.21 ±0.36 / 63.72 ms │        62.52 / 63.66 ±0.72 / 64.64 ms │ no change │
│ QQuery 23 │     320.38 / 321.92 ±1.58 / 324.83 ms │     321.01 / 324.03 ±2.21 / 327.74 ms │ no change │
│ QQuery 24 │     223.48 / 225.63 ±2.19 / 229.69 ms │     225.96 / 227.94 ±1.45 / 229.18 ms │ no change │
│ QQuery 25 │     111.18 / 112.02 ±0.67 / 113.13 ms │     111.61 / 113.71 ±1.74 / 115.99 ms │ no change │
│ QQuery 26 │        58.05 / 60.31 ±3.25 / 66.77 ms │        58.41 / 59.81 ±2.30 / 64.38 ms │ no change │
│ QQuery 27 │           6.70 / 6.80 ±0.10 / 6.97 ms │           6.74 / 6.83 ±0.14 / 7.11 ms │ no change │
│ QQuery 28 │        56.69 / 59.84 ±2.18 / 61.95 ms │        57.02 / 61.55 ±2.51 / 64.66 ms │ no change │
│ QQuery 29 │        98.78 / 98.92 ±0.12 / 99.09 ms │        98.29 / 98.80 ±0.49 / 99.69 ms │ no change │
│ QQuery 30 │        32.26 / 33.67 ±1.91 / 37.40 ms │        32.56 / 33.36 ±1.36 / 36.07 ms │ no change │
│ QQuery 31 │     111.60 / 112.46 ±0.78 / 113.91 ms │     112.13 / 113.18 ±0.81 / 114.58 ms │ no change │
│ QQuery 32 │        20.08 / 20.43 ±0.30 / 20.85 ms │        20.13 / 20.30 ±0.26 / 20.81 ms │ no change │
│ QQuery 33 │        38.27 / 38.60 ±0.24 / 38.93 ms │        38.34 / 38.58 ±0.16 / 38.82 ms │ no change │
│ QQuery 34 │          9.51 / 9.71 ±0.23 / 10.10 ms │           9.27 / 9.48 ±0.18 / 9.74 ms │ no change │
│ QQuery 35 │        71.00 / 72.75 ±1.31 / 75.00 ms │        71.83 / 73.75 ±1.78 / 76.53 ms │ no change │
│ QQuery 36 │           5.86 / 6.03 ±0.19 / 6.40 ms │           5.89 / 6.02 ±0.21 / 6.44 ms │ no change │
│ QQuery 37 │           6.90 / 7.08 ±0.13 / 7.22 ms │           6.81 / 6.99 ±0.28 / 7.55 ms │ no change │
│ QQuery 38 │        62.94 / 64.01 ±1.08 / 66.06 ms │        63.47 / 63.79 ±0.27 / 64.06 ms │ no change │
│ QQuery 39 │     450.39 / 456.74 ±4.18 / 461.43 ms │     452.39 / 459.46 ±4.45 / 465.74 ms │ no change │
│ QQuery 40 │        23.08 / 24.70 ±2.62 / 29.93 ms │        23.31 / 24.54 ±2.03 / 28.57 ms │ no change │
│ QQuery 41 │        11.36 / 11.54 ±0.17 / 11.84 ms │        11.31 / 11.47 ±0.15 / 11.75 ms │ no change │
│ QQuery 42 │        23.72 / 24.85 ±1.58 / 27.93 ms │        23.81 / 25.27 ±2.58 / 30.42 ms │ no change │
│ QQuery 43 │           4.92 / 5.04 ±0.11 / 5.18 ms │           4.97 / 5.14 ±0.20 / 5.52 ms │ no change │
│ QQuery 44 │           9.09 / 9.11 ±0.02 / 9.14 ms │           9.09 / 9.19 ±0.09 / 9.34 ms │ no change │
│ QQuery 45 │        38.30 / 38.59 ±0.18 / 38.80 ms │        38.69 / 39.03 ±0.33 / 39.46 ms │ no change │
│ QQuery 46 │        11.57 / 11.83 ±0.19 / 12.14 ms │        11.69 / 11.95 ±0.25 / 12.41 ms │ no change │
│ QQuery 47 │     227.07 / 230.24 ±2.55 / 233.80 ms │     229.22 / 232.21 ±3.03 / 236.09 ms │ no change │
│ QQuery 48 │        96.37 / 96.99 ±0.37 / 97.45 ms │        97.11 / 97.82 ±0.85 / 99.47 ms │ no change │
│ QQuery 49 │        78.16 / 80.39 ±4.07 / 88.53 ms │        77.56 / 78.46 ±0.87 / 80.06 ms │ no change │
│ QQuery 50 │        59.75 / 60.40 ±0.35 / 60.73 ms │        59.38 / 59.98 ±0.48 / 60.74 ms │ no change │
│ QQuery 51 │       93.61 / 96.12 ±3.43 / 102.84 ms │        91.73 / 94.05 ±1.69 / 96.59 ms │ no change │
│ QQuery 52 │        23.78 / 24.25 ±0.34 / 24.82 ms │        24.01 / 24.29 ±0.40 / 25.08 ms │ no change │
│ QQuery 53 │        29.10 / 29.49 ±0.33 / 29.97 ms │        29.65 / 29.77 ±0.11 / 29.97 ms │ no change │
│ QQuery 54 │        55.29 / 55.56 ±0.26 / 56.00 ms │        54.95 / 55.55 ±0.43 / 56.11 ms │ no change │
│ QQuery 55 │        23.32 / 23.95 ±0.53 / 24.66 ms │        23.27 / 24.08 ±1.22 / 26.51 ms │ no change │
│ QQuery 56 │        38.63 / 40.92 ±3.49 / 47.83 ms │        38.68 / 40.26 ±2.25 / 44.73 ms │ no change │
│ QQuery 57 │     177.31 / 179.30 ±1.82 / 182.59 ms │     176.95 / 179.26 ±1.66 / 182.03 ms │ no change │
│ QQuery 58 │     113.72 / 115.60 ±2.07 / 119.60 ms │     114.89 / 116.02 ±1.13 / 118.15 ms │ no change │
│ QQuery 59 │     116.85 / 118.75 ±1.97 / 122.03 ms │     117.30 / 118.78 ±1.99 / 122.63 ms │ no change │
│ QQuery 60 │        39.50 / 40.19 ±0.49 / 40.99 ms │        39.42 / 40.17 ±0.63 / 41.33 ms │ no change │
│ QQuery 61 │        13.01 / 13.16 ±0.13 / 13.40 ms │        12.99 / 13.10 ±0.15 / 13.38 ms │ no change │
│ QQuery 62 │        45.87 / 46.66 ±0.63 / 47.48 ms │        45.94 / 46.19 ±0.19 / 46.50 ms │ no change │
│ QQuery 63 │        29.44 / 29.90 ±0.34 / 30.43 ms │        29.78 / 29.90 ±0.13 / 30.13 ms │ no change │
│ QQuery 64 │     398.09 / 406.40 ±8.59 / 417.75 ms │     397.48 / 400.95 ±5.05 / 410.93 ms │ no change │
│ QQuery 65 │     148.72 / 152.60 ±3.55 / 158.29 ms │     148.96 / 153.88 ±3.83 / 160.02 ms │ no change │
│ QQuery 66 │        79.49 / 80.15 ±0.55 / 80.82 ms │        79.06 / 79.75 ±0.48 / 80.37 ms │ no change │
│ QQuery 67 │     247.45 / 250.96 ±4.56 / 259.66 ms │     245.64 / 250.62 ±3.14 / 255.12 ms │ no change │
│ QQuery 68 │        11.73 / 11.92 ±0.21 / 12.30 ms │        11.95 / 12.22 ±0.15 / 12.40 ms │ no change │
│ QQuery 69 │        56.93 / 57.29 ±0.33 / 57.87 ms │        56.64 / 57.18 ±0.43 / 57.92 ms │ no change │
│ QQuery 70 │     104.14 / 108.32 ±5.57 / 118.37 ms │     104.46 / 106.53 ±2.60 / 111.37 ms │ no change │
│ QQuery 71 │        35.94 / 36.51 ±0.33 / 36.91 ms │        35.52 / 36.06 ±0.46 / 36.78 ms │ no change │
│ QQuery 72 │ 2042.65 / 2165.64 ±91.59 / 2304.62 ms │ 2133.48 / 2193.43 ±31.57 / 2218.78 ms │ no change │
│ QQuery 73 │          9.18 / 9.57 ±0.41 / 10.34 ms │          9.25 / 9.66 ±0.29 / 10.00 ms │ no change │
│ QQuery 74 │     169.44 / 172.30 ±2.31 / 176.49 ms │     170.36 / 175.03 ±4.45 / 180.62 ms │ no change │
│ QQuery 75 │     147.19 / 149.85 ±4.03 / 157.86 ms │     147.38 / 149.77 ±1.44 / 151.64 ms │ no change │
│ QQuery 76 │        35.21 / 36.88 ±2.44 / 41.73 ms │        35.43 / 38.38 ±4.34 / 46.99 ms │ no change │
│ QQuery 77 │        61.59 / 62.15 ±0.53 / 62.84 ms │        61.44 / 63.85 ±4.31 / 72.47 ms │ no change │
│ QQuery 78 │     184.71 / 190.22 ±4.33 / 196.22 ms │     186.61 / 189.00 ±1.38 / 190.37 ms │ no change │
│ QQuery 79 │        67.58 / 67.80 ±0.15 / 68.02 ms │        67.46 / 68.16 ±0.51 / 68.94 ms │ no change │
│ QQuery 80 │     100.81 / 103.43 ±2.78 / 108.79 ms │     100.43 / 104.28 ±4.72 / 113.46 ms │ no change │
│ QQuery 81 │        25.68 / 27.72 ±2.88 / 33.45 ms │        25.95 / 26.55 ±0.43 / 27.23 ms │ no change │
│ QQuery 82 │        16.62 / 16.84 ±0.24 / 17.15 ms │        16.35 / 16.76 ±0.28 / 17.19 ms │ no change │
│ QQuery 83 │        39.58 / 40.12 ±0.32 / 40.46 ms │        39.93 / 40.13 ±0.21 / 40.49 ms │ no change │
│ QQuery 84 │        34.53 / 34.83 ±0.22 / 35.15 ms │        34.72 / 35.01 ±0.21 / 35.32 ms │ no change │
│ QQuery 85 │     108.62 / 111.92 ±3.65 / 118.56 ms │     109.07 / 111.55 ±2.40 / 114.63 ms │ no change │
│ QQuery 86 │        25.40 / 25.86 ±0.24 / 26.07 ms │        25.15 / 25.52 ±0.22 / 25.81 ms │ no change │
│ QQuery 87 │        63.82 / 64.68 ±0.66 / 65.83 ms │        63.22 / 64.43 ±0.81 / 65.43 ms │ no change │
│ QQuery 88 │        61.71 / 62.76 ±1.21 / 64.82 ms │        61.02 / 62.41 ±0.83 / 63.61 ms │ no change │
│ QQuery 89 │        35.58 / 36.79 ±1.39 / 39.47 ms │        36.00 / 36.77 ±0.42 / 37.17 ms │ no change │
│ QQuery 90 │        17.07 / 17.27 ±0.16 / 17.54 ms │        17.20 / 17.39 ±0.15 / 17.65 ms │ no change │
│ QQuery 91 │        44.92 / 45.20 ±0.26 / 45.59 ms │        44.82 / 45.04 ±0.20 / 45.39 ms │ no change │
│ QQuery 92 │        29.74 / 30.03 ±0.46 / 30.92 ms │        29.51 / 30.12 ±0.42 / 30.73 ms │ no change │
│ QQuery 93 │        50.58 / 51.27 ±0.40 / 51.73 ms │        50.62 / 51.30 ±0.42 / 51.73 ms │ no change │
│ QQuery 94 │        39.36 / 40.53 ±1.45 / 43.28 ms │        39.16 / 40.90 ±2.57 / 46.00 ms │ no change │
│ QQuery 95 │        85.39 / 86.59 ±1.55 / 89.52 ms │        83.82 / 84.87 ±0.94 / 86.33 ms │ no change │
│ QQuery 96 │        23.83 / 24.03 ±0.26 / 24.53 ms │        24.42 / 24.57 ±0.20 / 24.94 ms │ no change │
│ QQuery 97 │        45.61 / 46.21 ±0.47 / 47.02 ms │        46.18 / 47.81 ±2.88 / 53.54 ms │ no change │
│ QQuery 98 │        42.55 / 45.08 ±3.32 / 51.59 ms │        43.12 / 43.61 ±0.34 / 44.08 ms │ no change │
│ QQuery 99 │        69.49 / 69.93 ±0.37 / 70.38 ms │        69.51 / 70.23 ±0.60 / 71.24 ms │ no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                               │ 10300.86ms │
│ Total Time (codex_existence-semi-join-kernel)   │ 10325.08ms │
│ Average Time (HEAD)                             │   104.05ms │
│ Average Time (codex_existence-semi-join-kernel) │   104.29ms │
│ Queries Faster                                  │          0 │
│ Queries Slower                                  │          0 │
│ Queries with No Change                          │         99 │
│ Queries with Failure                            │          0 │
└─────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	5.8 GiB
Avg memory	5.1 GiB
CPU user	235.7s
CPU sys	6.4s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	6.0 GiB
Avg memory	5.2 GiB
CPU user	237.0s
CPU sys	6.4s
Peak spill	0 B

File an issue against this benchmark runner

SubhamSinghal · 2026-06-09T15:04:16Z

Is it similar to this PR: #21775

github-actions · 2026-08-09T02:13:35Z

Thank you for your contribution. Unfortunately, this pull request is stale because it has been open 60 days with no activity. Please remove the stale label or comment or this will be closed in 7 days.

github-actions Bot added the physical-plan Changes to the physical-plan crate label Jun 8, 2026

github-actions Bot added the auto detected api change Auto detected API change label Jun 8, 2026

Dandandan force-pushed the codex/existence-semi-join-kernel branch from fb316c6 to 72363e8 Compare June 8, 2026 19:37

Optimize existence-only right semi joins

53bec28

Dandandan force-pushed the codex/existence-semi-join-kernel branch from 72363e8 to 53bec28 Compare June 8, 2026 20:02

This was referenced Jun 10, 2026

Optimize semi, anti-joins to avoid dynamic dispatch for comparison #22877

Closed

Optimize mark joins using existence probes #22880

Open

github-actions Bot added the Stale PR has not had any activity for some time label Aug 9, 2026

Uh oh!

Conversation

Dandandan commented Jun 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

github-actions Bot commented Jun 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Dandandan commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

mbutrovich commented Jun 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Rationale for this change

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

Dandandan commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

adriangbot commented Jun 8, 2026

Uh oh!

Dandandan commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

SubhamSinghal commented Jun 9, 2026

Uh oh!

github-actions Bot commented Aug 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Dandandan commented Jun 8, 2026 •

edited

Loading

github-actions Bot commented Jun 8, 2026 •

edited

Loading

mbutrovich commented Jun 8, 2026 •

edited

Loading