Add row-number late materialization for TopK by Dandandan · Pull Request #22617 · apache/datafusion

Dandandan · 2026-05-29T09:04:53Z

Which issue does this PR close?

None yet.

Rationale for this change

Large ORDER BY ... LIMIT queries can sort only key columns first and materialize non-key columns after the TopK rows are known. This reduces the amount of data read and moved through TopK execution, while preserving result order through row-number based selection.

What changes are included in this PR?

Adds a LateMaterialization physical optimizer rule enabled by default via datafusion.optimizer.enable_row_number_topk_late_materialization
Adds LateTopKMaterializationExec, including a generic fallback path and a Parquet/file row-selection pushdown path
Adds FileRowsSelection support to file scan extensions and Parquet access planning
Documents the new optimizer config and updates the optimizer rule reference
Adds optimizer and Parquet integration coverage

Are these changes tested?

Yes.

cargo fmt --all
cargo clippy --all-targets --all-features -- -D warnings
cargo check -p datafusion-datasource-parquet -p datafusion-physical-optimizer -p datafusion
cargo clippy -p datafusion-physical-optimizer -p datafusion-datasource-parquet -- -D warnings
cargo test -p datafusion --test core_integration late_materialization
cargo test -p datafusion --test parquet_integration file_rows_selection
git diff --check

Are there any user-facing changes?

Yes. The new optimizer rule is enabled by default and can be disabled with datafusion.optimizer.enable_row_number_topk_late_materialization = false.

Dandandan · 2026-05-29T09:05:51Z

run benchmarks

adriangbot · 2026-05-29T09:09:09Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4572940303-359-qhx52 6.12.68+ #1 SMP Wed Apr 1 02:23:28 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/late-materialization-topk (68e8de6) to 0da8961 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-29T09:09:09Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4572940303-360-5lwrn 6.12.68+ #1 SMP Wed Apr 1 02:23:28 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/late-materialization-topk (68e8de6) to 0da8961 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-29T09:15:03Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4572940303-361-9b7d8 6.12.68+ #1 SMP Wed Apr 1 02:23:28 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/late-materialization-topk (68e8de6) to 0da8961 (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

…ization-topk # Conflicts: # datafusion/core/src/optimizer_rule_reference.md # datafusion/datasource-parquet/src/opener/mod.rs

adriangbot · 2026-05-29T09:26:54Z

🤖 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_late-materialization-topk
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ codex_late-materialization-topk ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 38.11 / 39.02 ±1.00 / 40.89 ms │  38.28 / 39.41 ±0.98 / 40.86 ms │ no change │
│ QQuery 2  │ 19.66 / 19.90 ±0.22 / 20.31 ms │  19.83 / 19.95 ±0.13 / 20.19 ms │ no change │
│ QQuery 3  │ 33.18 / 34.70 ±1.28 / 36.62 ms │  33.20 / 34.84 ±1.16 / 36.17 ms │ no change │
│ QQuery 4  │ 17.33 / 17.59 ±0.20 / 17.87 ms │  17.23 / 17.54 ±0.33 / 18.13 ms │ no change │
│ QQuery 5  │ 41.58 / 42.20 ±0.36 / 42.56 ms │  41.86 / 42.60 ±0.47 / 43.22 ms │ no change │
│ QQuery 6  │ 16.15 / 16.54 ±0.31 / 17.09 ms │  16.23 / 16.37 ±0.10 / 16.49 ms │ no change │
│ QQuery 7  │ 45.40 / 48.39 ±2.13 / 50.78 ms │  45.39 / 48.18 ±2.33 / 52.23 ms │ no change │
│ QQuery 8  │ 44.69 / 45.07 ±0.40 / 45.81 ms │  45.20 / 45.96 ±0.93 / 47.66 ms │ no change │
│ QQuery 9  │ 49.19 / 49.73 ±0.55 / 50.47 ms │  49.10 / 50.30 ±0.94 / 51.42 ms │ no change │
│ QQuery 10 │ 62.76 / 63.10 ±0.26 / 63.46 ms │  63.53 / 63.97 ±0.61 / 65.12 ms │ no change │
│ QQuery 11 │ 13.22 / 13.37 ±0.16 / 13.67 ms │  13.40 / 13.88 ±0.75 / 15.36 ms │ no change │
│ QQuery 12 │ 24.21 / 24.85 ±0.66 / 26.12 ms │  23.88 / 24.72 ±0.84 / 26.05 ms │ no change │
│ QQuery 13 │ 33.69 / 35.46 ±1.43 / 37.58 ms │  33.40 / 34.08 ±0.64 / 34.99 ms │ no change │
│ QQuery 14 │ 25.22 / 25.36 ±0.07 / 25.42 ms │  25.50 / 25.61 ±0.09 / 25.72 ms │ no change │
│ QQuery 15 │ 31.45 / 31.72 ±0.29 / 32.17 ms │  31.67 / 32.25 ±0.69 / 33.58 ms │ no change │
│ QQuery 16 │ 14.43 / 14.67 ±0.13 / 14.77 ms │  14.46 / 15.06 ±0.66 / 16.31 ms │ no change │
│ QQuery 17 │ 73.57 / 74.65 ±0.69 / 75.59 ms │  73.86 / 74.34 ±0.38 / 75.01 ms │ no change │
│ QQuery 18 │ 62.37 / 63.99 ±1.70 / 67.06 ms │  61.33 / 64.59 ±3.52 / 71.44 ms │ no change │
│ QQuery 19 │ 33.94 / 34.29 ±0.44 / 35.09 ms │  34.02 / 34.65 ±0.45 / 35.23 ms │ no change │
│ QQuery 20 │ 37.18 / 37.34 ±0.15 / 37.62 ms │  37.60 / 37.74 ±0.17 / 38.04 ms │ no change │
│ QQuery 21 │ 56.52 / 56.89 ±0.30 / 57.27 ms │  56.38 / 58.45 ±1.18 / 59.71 ms │ no change │
│ QQuery 22 │ 22.96 / 24.17 ±1.36 / 26.81 ms │  23.31 / 24.62 ±1.62 / 27.81 ms │ no change │
└───────────┴────────────────────────────────┴─────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                              ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 813.00ms │
│ Total Time (codex_late-materialization-topk)   │ 819.11ms │
│ Average Time (HEAD)                            │  36.95ms │
│ Average Time (codex_late-materialization-topk) │  37.23ms │
│ Queries Faster                                 │        0 │
│ Queries Slower                                 │        0 │
│ Queries with No Change                         │       22 │
│ Queries with Failure                           │        0 │
└────────────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	5.6 GiB
Avg memory	5.1 GiB
CPU user	29.5s
CPU sys	2.3s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	5.5 GiB
Avg memory	5.0 GiB
CPU user	29.5s
CPU sys	2.4s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-05-29T09:28:20Z

🤖 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_late-materialization-topk
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       codex_late-materialization-topk ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.75 / 6.28 ±0.92 / 8.11 ms │           5.74 / 6.24 ±0.92 / 8.07 ms │     no change │
│ QQuery 2  │        79.43 / 80.16 ±0.57 / 81.05 ms │        79.94 / 80.20 ±0.17 / 80.43 ms │     no change │
│ QQuery 3  │        28.64 / 28.92 ±0.32 / 29.41 ms │        28.51 / 28.73 ±0.21 / 29.08 ms │     no change │
│ QQuery 4  │     498.73 / 506.02 ±6.17 / 513.82 ms │     502.03 / 506.01 ±4.77 / 514.73 ms │     no change │
│ QQuery 5  │        50.34 / 50.55 ±0.16 / 50.82 ms │        50.67 / 51.37 ±0.58 / 52.10 ms │     no change │
│ QQuery 6  │        35.19 / 35.61 ±0.32 / 36.11 ms │        34.44 / 35.49 ±0.56 / 36.00 ms │     no change │
│ QQuery 7  │     107.90 / 110.10 ±2.89 / 115.72 ms │     107.77 / 109.29 ±2.58 / 114.45 ms │     no change │
│ QQuery 8  │        35.96 / 36.49 ±0.45 / 37.15 ms │        35.88 / 36.41 ±0.45 / 37.15 ms │     no change │
│ QQuery 9  │        53.32 / 54.24 ±0.92 / 55.85 ms │        53.14 / 54.57 ±0.96 / 56.13 ms │     no change │
│ QQuery 10 │        79.81 / 80.18 ±0.21 / 80.36 ms │        79.71 / 80.11 ±0.24 / 80.37 ms │     no change │
│ QQuery 11 │     303.65 / 307.69 ±2.74 / 310.81 ms │     308.59 / 311.46 ±2.23 / 315.10 ms │     no change │
│ QQuery 12 │        28.17 / 28.58 ±0.67 / 29.92 ms │        27.90 / 28.35 ±0.32 / 28.78 ms │     no change │
│ QQuery 13 │     125.95 / 126.98 ±0.80 / 128.38 ms │     125.68 / 126.53 ±0.96 / 128.31 ms │     no change │
│ QQuery 14 │     504.47 / 506.32 ±1.41 / 508.21 ms │     498.75 / 502.32 ±2.23 / 505.27 ms │     no change │
│ QQuery 15 │        60.24 / 61.99 ±1.47 / 63.95 ms │        59.31 / 60.16 ±0.60 / 60.98 ms │     no change │
│ QQuery 16 │           6.42 / 6.54 ±0.19 / 6.92 ms │           6.43 / 6.62 ±0.15 / 6.83 ms │     no change │
│ QQuery 17 │        80.44 / 81.55 ±0.66 / 82.51 ms │        80.84 / 81.34 ±0.44 / 82.16 ms │     no change │
│ QQuery 18 │     149.97 / 151.87 ±2.33 / 156.38 ms │     151.14 / 151.82 ±0.53 / 152.54 ms │     no change │
│ QQuery 19 │        40.88 / 41.29 ±0.28 / 41.73 ms │        40.59 / 40.99 ±0.25 / 41.21 ms │     no change │
│ QQuery 20 │        34.98 / 35.55 ±0.35 / 36.06 ms │        34.58 / 34.98 ±0.42 / 35.78 ms │     no change │
│ QQuery 21 │        16.89 / 17.76 ±1.21 / 20.14 ms │        16.39 / 16.79 ±0.34 / 17.41 ms │ +1.06x faster │
│ QQuery 22 │        61.45 / 62.39 ±0.64 / 63.20 ms │        61.85 / 62.03 ±0.15 / 62.27 ms │     no change │
│ QQuery 23 │     476.70 / 484.35 ±4.25 / 489.66 ms │     475.13 / 482.38 ±5.76 / 491.00 ms │     no change │
│ QQuery 24 │     233.43 / 236.68 ±3.62 / 243.75 ms │     232.97 / 235.31 ±3.07 / 241.26 ms │     no change │
│ QQuery 25 │     113.04 / 114.38 ±1.11 / 116.37 ms │     112.62 / 113.95 ±0.90 / 115.40 ms │     no change │
│ QQuery 26 │        69.91 / 70.53 ±0.43 / 71.13 ms │        69.50 / 70.25 ±0.40 / 70.57 ms │     no change │
│ QQuery 27 │           6.66 / 6.79 ±0.12 / 7.02 ms │           6.52 / 6.63 ±0.11 / 6.83 ms │     no change │
│ QQuery 28 │        61.21 / 61.97 ±0.46 / 62.51 ms │        61.32 / 61.82 ±0.47 / 62.59 ms │     no change │
│ QQuery 29 │      98.54 / 100.63 ±1.98 / 104.17 ms │      97.75 / 100.84 ±2.59 / 105.03 ms │     no change │
│ QQuery 30 │        30.18 / 30.60 ±0.29 / 30.94 ms │        30.00 / 30.63 ±0.63 / 31.67 ms │     no change │
│ QQuery 31 │     111.18 / 113.20 ±3.23 / 119.62 ms │     110.67 / 112.11 ±1.83 / 115.73 ms │     no change │
│ QQuery 32 │        19.82 / 20.07 ±0.22 / 20.44 ms │        19.51 / 20.17 ±0.47 / 20.92 ms │     no change │
│ QQuery 33 │        38.11 / 38.48 ±0.21 / 38.68 ms │        37.77 / 38.05 ±0.16 / 38.29 ms │     no change │
│ QQuery 34 │          9.16 / 9.56 ±0.34 / 10.06 ms │           9.11 / 9.38 ±0.28 / 9.81 ms │     no change │
│ QQuery 35 │        79.84 / 80.29 ±0.35 / 80.85 ms │        80.19 / 80.41 ±0.21 / 80.80 ms │     no change │
│ QQuery 36 │           5.91 / 6.05 ±0.15 / 6.34 ms │           5.81 / 5.92 ±0.15 / 6.21 ms │     no change │
│ QQuery 37 │           6.75 / 6.86 ±0.07 / 6.94 ms │           6.65 / 6.69 ±0.05 / 6.79 ms │     no change │
│ QQuery 38 │        68.26 / 68.70 ±0.39 / 69.35 ms │        67.85 / 69.82 ±1.98 / 72.93 ms │     no change │
│ QQuery 39 │        97.17 / 97.99 ±0.50 / 98.52 ms │        96.16 / 97.06 ±0.76 / 97.86 ms │     no change │
│ QQuery 40 │        22.67 / 23.04 ±0.26 / 23.47 ms │        22.20 / 22.72 ±0.34 / 23.12 ms │     no change │
│ QQuery 41 │        11.48 / 11.54 ±0.08 / 11.69 ms │        11.29 / 11.44 ±0.16 / 11.74 ms │     no change │
│ QQuery 42 │        23.65 / 24.89 ±1.62 / 27.99 ms │        23.70 / 24.43 ±1.12 / 26.66 ms │     no change │
│ QQuery 43 │           4.71 / 4.77 ±0.06 / 4.86 ms │           4.86 / 5.43 ±1.03 / 7.48 ms │  1.14x slower │
│ QQuery 44 │        10.59 / 10.82 ±0.19 / 11.12 ms │        10.52 / 10.71 ±0.13 / 10.88 ms │     no change │
│ QQuery 45 │        39.89 / 40.59 ±0.63 / 41.43 ms │        39.24 / 40.44 ±1.06 / 41.97 ms │     no change │
│ QQuery 46 │        12.47 / 12.80 ±0.32 / 13.40 ms │        12.56 / 12.97 ±0.30 / 13.38 ms │     no change │
│ QQuery 47 │     225.41 / 229.42 ±3.29 / 233.51 ms │     226.11 / 227.92 ±1.58 / 229.86 ms │     no change │
│ QQuery 48 │     102.24 / 102.98 ±0.57 / 103.84 ms │     102.91 / 103.51 ±0.70 / 104.71 ms │     no change │
│ QQuery 49 │        78.68 / 80.85 ±2.16 / 84.75 ms │        78.61 / 79.18 ±0.31 / 79.43 ms │     no change │
│ QQuery 50 │        59.79 / 60.05 ±0.23 / 60.43 ms │        59.62 / 59.89 ±0.24 / 60.28 ms │     no change │
│ QQuery 51 │        92.58 / 93.19 ±0.50 / 93.82 ms │       92.12 / 96.04 ±3.36 / 100.25 ms │     no change │
│ QQuery 52 │        23.65 / 24.26 ±0.56 / 25.19 ms │        23.79 / 24.42 ±0.50 / 25.29 ms │     no change │
│ QQuery 53 │        29.49 / 29.65 ±0.18 / 29.95 ms │        29.05 / 29.37 ±0.17 / 29.53 ms │     no change │
│ QQuery 54 │        54.25 / 54.67 ±0.29 / 55.07 ms │        53.97 / 54.30 ±0.17 / 54.43 ms │     no change │
│ QQuery 55 │        23.39 / 23.60 ±0.18 / 23.84 ms │        23.08 / 23.67 ±0.35 / 24.11 ms │     no change │
│ QQuery 56 │        38.06 / 39.76 ±1.59 / 41.79 ms │        38.64 / 39.39 ±0.84 / 40.90 ms │     no change │
│ QQuery 57 │     174.23 / 176.20 ±1.34 / 178.42 ms │     174.78 / 176.88 ±1.40 / 178.55 ms │     no change │
│ QQuery 58 │     115.58 / 117.35 ±1.10 / 119.01 ms │     115.17 / 117.04 ±1.42 / 119.52 ms │     no change │
│ QQuery 59 │     116.91 / 118.06 ±0.69 / 118.99 ms │     115.69 / 116.80 ±1.42 / 119.53 ms │     no change │
│ QQuery 60 │        38.87 / 39.16 ±0.22 / 39.42 ms │        38.87 / 39.49 ±0.49 / 40.36 ms │     no change │
│ QQuery 61 │        12.55 / 12.71 ±0.21 / 13.13 ms │        12.64 / 12.83 ±0.22 / 13.25 ms │     no change │
│ QQuery 62 │        45.95 / 46.07 ±0.08 / 46.18 ms │        45.67 / 46.06 ±0.39 / 46.70 ms │     no change │
│ QQuery 63 │        29.44 / 29.76 ±0.27 / 30.14 ms │        29.19 / 29.40 ±0.17 / 29.69 ms │     no change │
│ QQuery 64 │     456.20 / 463.44 ±5.11 / 471.93 ms │     458.67 / 464.83 ±7.32 / 478.88 ms │     no change │
│ QQuery 65 │     146.31 / 151.46 ±4.35 / 157.80 ms │     146.57 / 151.31 ±4.39 / 158.75 ms │     no change │
│ QQuery 66 │        77.63 / 79.53 ±2.02 / 83.07 ms │        76.76 / 78.98 ±1.68 / 81.86 ms │     no change │
│ QQuery 67 │     240.86 / 246.22 ±4.59 / 252.46 ms │     243.46 / 245.14 ±0.92 / 246.09 ms │     no change │
│ QQuery 68 │        12.96 / 13.08 ±0.21 / 13.50 ms │        12.88 / 13.03 ±0.20 / 13.44 ms │     no change │
│ QQuery 69 │        75.61 / 76.08 ±0.35 / 76.65 ms │        75.86 / 76.36 ±0.29 / 76.76 ms │     no change │
│ QQuery 70 │     106.68 / 109.43 ±2.91 / 113.37 ms │     105.59 / 112.85 ±7.17 / 123.41 ms │     no change │
│ QQuery 71 │        35.69 / 35.96 ±0.16 / 36.17 ms │        35.55 / 35.87 ±0.22 / 36.20 ms │     no change │
│ QQuery 72 │ 2113.10 / 2185.55 ±44.19 / 2238.19 ms │ 2149.20 / 2201.73 ±41.74 / 2253.00 ms │     no change │
│ QQuery 73 │           8.99 / 9.22 ±0.26 / 9.72 ms │           9.05 / 9.23 ±0.24 / 9.71 ms │     no change │
│ QQuery 74 │     175.33 / 178.61 ±3.27 / 184.67 ms │     175.12 / 176.10 ±0.69 / 177.03 ms │     no change │
│ QQuery 75 │     145.75 / 147.68 ±1.52 / 150.26 ms │     146.10 / 147.01 ±0.74 / 148.14 ms │     no change │
│ QQuery 76 │        34.86 / 35.55 ±0.44 / 36.12 ms │        35.26 / 35.57 ±0.28 / 36.06 ms │     no change │
│ QQuery 77 │        59.83 / 60.07 ±0.16 / 60.33 ms │        59.86 / 60.19 ±0.29 / 60.65 ms │     no change │
│ QQuery 78 │     188.68 / 190.23 ±1.54 / 192.42 ms │     185.73 / 187.47 ±1.51 / 190.21 ms │     no change │
│ QQuery 79 │        66.90 / 67.25 ±0.35 / 67.75 ms │        66.86 / 67.30 ±0.36 / 67.84 ms │     no change │
│ QQuery 80 │     100.43 / 101.35 ±1.12 / 103.54 ms │      98.22 / 102.59 ±3.78 / 108.55 ms │     no change │
│ QQuery 81 │        23.67 / 23.92 ±0.17 / 24.13 ms │        23.89 / 24.13 ±0.17 / 24.41 ms │     no change │
│ QQuery 82 │        16.24 / 16.46 ±0.13 / 16.66 ms │        16.34 / 16.55 ±0.18 / 16.83 ms │     no change │
│ QQuery 83 │        36.19 / 36.42 ±0.12 / 36.57 ms │        35.86 / 36.19 ±0.27 / 36.61 ms │     no change │
│ QQuery 84 │        42.97 / 43.58 ±0.88 / 45.31 ms │        43.18 / 44.90 ±2.43 / 49.71 ms │     no change │
│ QQuery 85 │     134.74 / 137.58 ±2.17 / 140.68 ms │     134.98 / 136.18 ±0.86 / 137.39 ms │     no change │
│ QQuery 86 │        24.46 / 24.99 ±0.52 / 25.89 ms │        24.43 / 24.80 ±0.27 / 25.20 ms │     no change │
│ QQuery 87 │        68.92 / 70.05 ±1.19 / 72.10 ms │        68.63 / 69.93 ±1.73 / 73.28 ms │     no change │
│ QQuery 88 │        61.95 / 62.18 ±0.24 / 62.61 ms │        61.70 / 62.27 ±0.61 / 63.15 ms │     no change │
│ QQuery 89 │        35.78 / 36.15 ±0.32 / 36.53 ms │        35.27 / 35.42 ±0.09 / 35.53 ms │     no change │
│ QQuery 90 │        16.95 / 17.12 ±0.16 / 17.39 ms │        16.57 / 16.74 ±0.15 / 16.94 ms │     no change │
│ QQuery 91 │        51.28 / 52.06 ±0.54 / 52.87 ms │        51.05 / 51.91 ±1.18 / 54.25 ms │     no change │
│ QQuery 92 │        29.24 / 30.43 ±1.52 / 33.37 ms │        29.05 / 29.42 ±0.27 / 29.73 ms │     no change │
│ QQuery 93 │        50.19 / 51.17 ±0.58 / 51.98 ms │        50.29 / 50.78 ±0.30 / 51.16 ms │     no change │
│ QQuery 94 │        37.59 / 37.87 ±0.20 / 38.12 ms │        37.32 / 37.69 ±0.45 / 38.55 ms │     no change │
│ QQuery 95 │        84.27 / 85.12 ±0.71 / 86.02 ms │        83.90 / 85.09 ±1.39 / 87.68 ms │     no change │
│ QQuery 96 │        23.91 / 24.16 ±0.23 / 24.59 ms │        24.09 / 24.36 ±0.23 / 24.67 ms │     no change │
│ QQuery 97 │        45.58 / 45.91 ±0.20 / 46.18 ms │        45.85 / 46.07 ±0.11 / 46.20 ms │     no change │
│ QQuery 98 │        42.11 / 42.84 ±0.42 / 43.28 ms │        42.57 / 42.74 ±0.13 / 42.91 ms │     no change │
│ QQuery 99 │        68.84 / 69.42 ±0.50 / 70.13 ms │        68.56 / 69.16 ±0.32 / 69.42 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 10428.52ms │
│ Total Time (codex_late-materialization-topk)   │ 10427.47ms │
│ Average Time (HEAD)                            │   105.34ms │
│ Average Time (codex_late-materialization-topk) │   105.33ms │
│ Queries Faster                                 │          1 │
│ Queries Slower                                 │          1 │
│ Queries with No Change                         │         97 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	7.0 GiB
Avg memory	6.2 GiB
CPU user	235.7s
CPU sys	5.6s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	6.9 GiB
Avg memory	6.1 GiB
CPU user	234.9s
CPU sys	5.6s
Peak spill	0 B

File an issue against this benchmark runner

github-actions · 2026-05-29T09:30:05Z

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 v53.1.0 (current)
       Built [ 102.306s] (current)
     Parsing datafusion v53.1.0 (current)
      Parsed [   0.037s] (current)
    Building datafusion v53.1.0 (baseline)
       Built [  97.281s] (baseline)
     Parsing datafusion v53.1.0 (baseline)
      Parsed [   0.037s] (baseline)
    Checking datafusion v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   0.904s] 222 checks: 222 pass, 30 skip
     Summary no semver update required
    Finished [ 202.586s] datafusion
    Building datafusion-common v53.1.0 (current)
       Built [  32.364s] (current)
     Parsing datafusion-common v53.1.0 (current)
      Parsed [   0.062s] (current)
    Building datafusion-common v53.1.0 (baseline)
       Built [  32.630s] (baseline)
     Parsing datafusion-common v53.1.0 (baseline)
      Parsed [   0.064s] (baseline)
    Checking datafusion-common v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   0.975s] 222 checks: 221 pass, 1 fail, 0 warn, 30 skip

--- failure constructible_struct_adds_field: externally-constructible struct adds field ---

Description:
A pub struct constructible with a struct literal has a new pub field. Existing struct literals must be updated to include the new field.
        ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.47.0/src/lints/constructible_struct_adds_field.ron

Failed in:
  field OptimizerOptions.enable_row_number_topk_late_materialization in /home/runner/work/datafusion/datafusion/datafusion/common/src/config.rs:1085

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [  67.438s] datafusion-common
    Building datafusion-datasource v53.1.0 (current)
       Built [  36.220s] (current)
     Parsing datafusion-datasource v53.1.0 (current)
      Parsed [   0.034s] (current)
    Building datafusion-datasource v53.1.0 (baseline)
       Built [  35.083s] (baseline)
     Parsing datafusion-datasource v53.1.0 (baseline)
      Parsed [   0.035s] (baseline)
    Checking datafusion-datasource v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   0.352s] 222 checks: 222 pass, 30 skip
     Summary no semver update required
    Finished [  73.144s] datafusion-datasource
    Building datafusion-datasource-parquet v53.1.0 (current)
       Built [  41.397s] (current)
     Parsing datafusion-datasource-parquet v53.1.0 (current)
      Parsed [   0.030s] (current)
    Building datafusion-datasource-parquet v53.1.0 (baseline)
       Built [  41.507s] (baseline)
     Parsing datafusion-datasource-parquet v53.1.0 (baseline)
      Parsed [   0.031s] (baseline)
    Checking datafusion-datasource-parquet v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   0.193s] 222 checks: 222 pass, 30 skip
     Summary no semver update required
    Finished [  84.852s] datafusion-datasource-parquet
    Building datafusion-physical-optimizer v53.1.0 (current)
       Built [  36.088s] (current)
     Parsing datafusion-physical-optimizer v53.1.0 (current)
      Parsed [   0.025s] (current)
    Building datafusion-physical-optimizer v53.1.0 (baseline)
       Built [  35.879s] (baseline)
     Parsing datafusion-physical-optimizer v53.1.0 (baseline)
      Parsed [   0.024s] (baseline)
    Checking datafusion-physical-optimizer v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   0.160s] 222 checks: 222 pass, 30 skip
     Summary no semver update required
    Finished [  74.695s] datafusion-physical-optimizer

adriangbot · 2026-05-29T09:33:58Z

🤖 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_late-materialization-topk
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃          codex_late-materialization-topk ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │           1.14 / 4.58 ±6.79 / 18.17 ms │             1.16 / 4.70 ±7.03 / 18.76 ms │     no change │
│ QQuery 1  │         12.55 / 12.99 ±0.23 / 13.20 ms │           13.02 / 13.13 ±0.11 / 13.32 ms │     no change │
│ QQuery 2  │         35.61 / 35.90 ±0.27 / 36.37 ms │           36.31 / 36.56 ±0.27 / 37.06 ms │     no change │
│ QQuery 3  │         30.90 / 31.55 ±0.72 / 32.52 ms │           30.73 / 31.28 ±0.52 / 32.25 ms │     no change │
│ QQuery 4  │      219.42 / 223.36 ±2.16 / 225.88 ms │        218.78 / 222.79 ±2.44 / 225.48 ms │     no change │
│ QQuery 5  │      270.08 / 272.82 ±2.03 / 276.01 ms │        270.27 / 273.06 ±1.96 / 275.44 ms │     no change │
│ QQuery 6  │            1.20 / 1.34 ±0.22 / 1.77 ms │              1.18 / 1.32 ±0.21 / 1.74 ms │     no change │
│ QQuery 7  │         13.80 / 13.98 ±0.20 / 14.36 ms │           13.88 / 13.99 ±0.09 / 14.09 ms │     no change │
│ QQuery 8  │      316.60 / 321.00 ±3.18 / 324.51 ms │        317.98 / 322.99 ±3.45 / 328.83 ms │     no change │
│ QQuery 9  │      447.67 / 455.13 ±5.00 / 462.17 ms │        445.95 / 450.81 ±5.70 / 461.95 ms │     no change │
│ QQuery 10 │         68.70 / 69.95 ±0.91 / 71.16 ms │           69.41 / 71.35 ±2.47 / 76.03 ms │     no change │
│ QQuery 11 │         80.34 / 82.37 ±1.61 / 85.00 ms │           80.07 / 81.51 ±1.27 / 83.73 ms │     no change │
│ QQuery 12 │      262.86 / 266.87 ±4.47 / 272.51 ms │        265.55 / 271.46 ±4.77 / 278.62 ms │     no change │
│ QQuery 13 │      368.58 / 379.26 ±5.71 / 385.72 ms │       371.12 / 381.17 ±12.82 / 403.31 ms │     no change │
│ QQuery 14 │      276.79 / 281.74 ±5.03 / 289.86 ms │        275.92 / 282.45 ±5.26 / 291.74 ms │     no change │
│ QQuery 15 │      259.88 / 264.83 ±3.28 / 268.61 ms │       261.42 / 271.40 ±10.30 / 291.12 ms │     no change │
│ QQuery 16 │      606.93 / 613.04 ±7.66 / 627.86 ms │        604.31 / 613.74 ±7.27 / 622.88 ms │     no change │
│ QQuery 17 │     606.61 / 623.85 ±19.88 / 661.92 ms │        608.18 / 612.55 ±5.43 / 621.91 ms │     no change │
│ QQuery 18 │ 1264.23 / 1367.26 ±106.45 / 1496.16 ms │    1251.29 / 1266.47 ±13.87 / 1291.83 ms │ +1.08x faster │
│ QQuery 19 │         27.72 / 32.43 ±9.11 / 50.65 ms │          27.87 / 33.56 ±10.02 / 53.57 ms │     no change │
│ QQuery 20 │      515.02 / 522.70 ±9.61 / 541.35 ms │       517.23 / 524.39 ±11.86 / 547.90 ms │     no change │
│ QQuery 21 │      590.73 / 595.50 ±3.00 / 598.84 ms │        590.70 / 597.49 ±4.67 / 603.19 ms │     no change │
│ QQuery 22 │   1052.95 / 1054.77 ±1.07 / 1055.79 ms │    1054.57 / 1070.28 ±10.08 / 1083.32 ms │     no change │
│ QQuery 23 │  3148.83 / 3164.20 ±11.55 / 3179.17 ms │ 19838.41 / 19980.50 ±90.02 / 20088.59 ms │  6.31x slower │
│ QQuery 24 │         41.70 / 43.06 ±1.69 / 46.38 ms │       593.67 / 671.28 ±62.71 / 730.38 ms │ 15.59x slower │
│ QQuery 25 │      110.34 / 111.02 ±0.54 / 111.82 ms │        111.73 / 113.85 ±1.78 / 116.68 ms │     no change │
│ QQuery 26 │         42.03 / 45.93 ±5.00 / 55.82 ms │           41.97 / 42.37 ±0.42 / 43.17 ms │ +1.08x faster │
│ QQuery 27 │      661.50 / 670.94 ±7.28 / 679.32 ms │        679.16 / 686.64 ±6.06 / 696.41 ms │     no change │
│ QQuery 28 │  3009.02 / 3027.21 ±10.63 / 3040.40 ms │    3017.58 / 3043.51 ±14.30 / 3057.96 ms │     no change │
│ QQuery 29 │         39.59 / 42.97 ±3.91 / 49.58 ms │           40.06 / 41.40 ±2.00 / 45.35 ms │     no change │
│ QQuery 30 │      296.87 / 301.13 ±2.60 / 304.03 ms │        303.32 / 309.45 ±6.07 / 320.91 ms │     no change │
│ QQuery 31 │      279.64 / 286.04 ±4.36 / 291.01 ms │        288.24 / 293.09 ±4.62 / 301.78 ms │     no change │
│ QQuery 32 │      914.38 / 926.09 ±7.18 / 935.42 ms │       898.90 / 921.52 ±16.93 / 950.74 ms │     no change │
│ QQuery 33 │  1413.55 / 1437.88 ±23.85 / 1475.58 ms │    1423.09 / 1445.46 ±21.96 / 1484.82 ms │     no change │
│ QQuery 34 │   1424.79 / 1436.14 ±8.25 / 1450.13 ms │    1450.88 / 1462.91 ±12.37 / 1479.16 ms │     no change │
│ QQuery 35 │     276.89 / 304.00 ±44.52 / 392.69 ms │       275.08 / 287.35 ±11.85 / 307.71 ms │ +1.06x faster │
│ QQuery 36 │         64.79 / 67.35 ±1.55 / 69.36 ms │           64.31 / 69.29 ±4.21 / 74.97 ms │     no change │
│ QQuery 37 │         34.73 / 38.66 ±4.14 / 45.57 ms │           35.42 / 37.36 ±1.76 / 40.67 ms │     no change │
│ QQuery 38 │         40.38 / 41.62 ±0.78 / 42.29 ms │           40.53 / 45.64 ±5.90 / 56.63 ms │  1.10x slower │
│ QQuery 39 │      134.02 / 150.24 ±8.29 / 157.56 ms │        138.76 / 147.85 ±7.28 / 158.70 ms │     no change │
│ QQuery 40 │         13.68 / 14.28 ±0.52 / 14.92 ms │           13.78 / 14.02 ±0.33 / 14.66 ms │     no change │
│ QQuery 41 │         13.29 / 13.62 ±0.26 / 13.97 ms │           13.33 / 15.39 ±3.51 / 22.40 ms │  1.13x slower │
│ QQuery 42 │         12.85 / 14.88 ±3.51 / 21.88 ms │           12.99 / 13.14 ±0.14 / 13.38 ms │ +1.13x faster │
└───────────┴────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 19664.48ms │
│ Total Time (codex_late-materialization-topk)   │ 37090.49ms │
│ Average Time (HEAD)                            │   457.31ms │
│ Average Time (codex_late-materialization-topk) │   862.57ms │
│ Queries Faster                                 │          4 │
│ Queries Slower                                 │          4 │
│ Queries with No Change                         │         35 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	100.0s
Peak memory	31.0 GiB
Avg memory	23.6 GiB
CPU user	1023.9s
CPU sys	62.6s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	190.0s
Peak memory	30.4 GiB
Avg memory	22.1 GiB
CPU user	1102.3s
CPU sys	74.7s
Peak spill	0 B

File an issue against this benchmark runner

AdamGS · 2026-05-29T09:39:04Z

Is this selective late materialization 👀 ?

Dandandan · 2026-05-29T10:52:01Z

Is this selective late materialization 👀 ?

That would be the idea!

Dandandan · 2026-05-29T11:39:38Z

run benchmarks

adriangbot · 2026-05-29T11:42:20Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4574493650-365-4k8pt 6.12.68+ #1 SMP Wed Apr 1 02:23:28 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/late-materialization-topk (c5da797) to 32a1fe5 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-29T11:43:01Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4574493650-366-wbmr9 6.12.68+ #1 SMP Wed Apr 1 02:23:28 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/late-materialization-topk (c5da797) to 32a1fe5 (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-29T11:43:03Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4574493650-367-mvvkt 6.12.68+ #1 SMP Wed Apr 1 02:23:28 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/late-materialization-topk (c5da797) to 32a1fe5 (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-05-29T11:58:03Z

🤖 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_late-materialization-topk
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ codex_late-materialization-topk ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 38.07 / 39.14 ±1.14 / 41.06 ms │  38.18 / 39.32 ±0.98 / 40.87 ms │ no change │
│ QQuery 2  │ 18.34 / 18.80 ±0.37 / 19.48 ms │  18.24 / 18.44 ±0.16 / 18.64 ms │ no change │
│ QQuery 3  │ 31.15 / 33.12 ±0.99 / 33.84 ms │  30.33 / 32.20 ±2.20 / 35.99 ms │ no change │
│ QQuery 4  │ 17.01 / 17.15 ±0.10 / 17.26 ms │  16.82 / 17.37 ±0.57 / 18.45 ms │ no change │
│ QQuery 5  │ 37.84 / 39.76 ±1.91 / 43.33 ms │  36.79 / 38.59 ±1.40 / 39.94 ms │ no change │
│ QQuery 6  │ 15.94 / 16.15 ±0.23 / 16.51 ms │  15.85 / 16.60 ±1.05 / 18.66 ms │ no change │
│ QQuery 7  │ 44.61 / 45.45 ±0.61 / 46.28 ms │  43.82 / 45.75 ±2.00 / 49.26 ms │ no change │
│ QQuery 8  │ 42.48 / 43.21 ±0.76 / 44.48 ms │  42.43 / 42.66 ±0.24 / 43.03 ms │ no change │
│ QQuery 9  │ 48.79 / 49.80 ±0.80 / 51.23 ms │  49.36 / 50.36 ±0.64 / 50.92 ms │ no change │
│ QQuery 10 │ 41.59 / 41.81 ±0.27 / 42.32 ms │  41.70 / 42.08 ±0.22 / 42.36 ms │ no change │
│ QQuery 11 │ 12.61 / 13.26 ±0.71 / 14.58 ms │  12.81 / 13.01 ±0.16 / 13.30 ms │ no change │
│ QQuery 12 │ 23.71 / 24.07 ±0.26 / 24.50 ms │  23.77 / 24.05 ±0.33 / 24.68 ms │ no change │
│ QQuery 13 │ 32.22 / 34.03 ±1.91 / 37.53 ms │  31.91 / 33.90 ±1.74 / 36.68 ms │ no change │
│ QQuery 14 │ 23.25 / 24.19 ±1.13 / 26.23 ms │  23.31 / 23.67 ±0.22 / 23.99 ms │ no change │
│ QQuery 15 │ 30.90 / 31.01 ±0.08 / 31.10 ms │  30.93 / 31.39 ±0.61 / 32.61 ms │ no change │
│ QQuery 16 │ 13.74 / 14.06 ±0.32 / 14.64 ms │  13.93 / 14.11 ±0.22 / 14.50 ms │ no change │
│ QQuery 17 │ 71.87 / 72.53 ±0.51 / 73.36 ms │  71.89 / 72.82 ±0.86 / 74.44 ms │ no change │
│ QQuery 18 │ 58.95 / 60.16 ±0.76 / 60.94 ms │  58.25 / 60.03 ±1.55 / 62.20 ms │ no change │
│ QQuery 19 │ 32.67 / 33.11 ±0.54 / 34.07 ms │  32.80 / 33.64 ±1.64 / 36.93 ms │ no change │
│ QQuery 20 │ 31.62 / 31.85 ±0.20 / 32.13 ms │  31.93 / 32.24 ±0.26 / 32.69 ms │ no change │
│ QQuery 21 │ 54.70 / 55.68 ±0.55 / 56.40 ms │  55.38 / 56.48 ±0.84 / 57.88 ms │ no change │
│ QQuery 22 │ 13.61 / 13.82 ±0.14 / 13.99 ms │  13.52 / 13.74 ±0.14 / 13.90 ms │ no change │
└───────────┴────────────────────────────────┴─────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                              ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 752.15ms │
│ Total Time (codex_late-materialization-topk)   │ 752.43ms │
│ Average Time (HEAD)                            │  34.19ms │
│ Average Time (codex_late-materialization-topk) │  34.20ms │
│ Queries Faster                                 │        0 │
│ Queries Slower                                 │        0 │
│ Queries with No Change                         │       22 │
│ Queries with Failure                           │        0 │
└────────────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	5.6 GiB
Avg memory	4.9 GiB
CPU user	29.4s
CPU sys	2.2s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	5.7 GiB
Avg memory	4.9 GiB
CPU user	29.4s
CPU sys	2.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-05-29T12:00:04Z

🤖 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_late-materialization-topk
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       codex_late-materialization-topk ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.81 / 6.27 ±0.85 / 7.97 ms │           6.06 / 6.53 ±0.84 / 8.21 ms │     no change │
│ QQuery 2  │        83.34 / 83.77 ±0.29 / 84.15 ms │        81.98 / 82.10 ±0.21 / 82.51 ms │     no change │
│ QQuery 3  │        29.11 / 29.44 ±0.33 / 30.05 ms │        29.53 / 29.63 ±0.07 / 29.75 ms │     no change │
│ QQuery 4  │     485.26 / 491.12 ±4.68 / 499.37 ms │     498.28 / 502.55 ±3.86 / 508.80 ms │     no change │
│ QQuery 5  │        51.99 / 52.60 ±0.35 / 52.93 ms │        51.94 / 52.51 ±0.56 / 53.56 ms │     no change │
│ QQuery 6  │        36.70 / 37.20 ±0.51 / 38.16 ms │        35.96 / 36.87 ±0.49 / 37.35 ms │     no change │
│ QQuery 7  │        95.15 / 95.75 ±0.52 / 96.62 ms │       96.51 / 98.52 ±3.04 / 104.57 ms │     no change │
│ QQuery 8  │        37.47 / 39.03 ±2.63 / 44.28 ms │        37.17 / 37.86 ±0.44 / 38.56 ms │     no change │
│ QQuery 9  │        53.27 / 54.30 ±1.17 / 55.81 ms │        55.13 / 56.39 ±1.36 / 58.82 ms │     no change │
│ QQuery 10 │        68.47 / 69.04 ±0.33 / 69.45 ms │        68.80 / 69.17 ±0.20 / 69.36 ms │     no change │
│ QQuery 11 │     306.39 / 307.60 ±0.66 / 308.36 ms │     299.98 / 309.29 ±8.28 / 323.63 ms │     no change │
│ QQuery 12 │        28.73 / 29.40 ±0.40 / 29.97 ms │        29.07 / 29.33 ±0.20 / 29.60 ms │     no change │
│ QQuery 13 │     119.66 / 121.13 ±2.11 / 125.33 ms │     120.61 / 122.29 ±1.58 / 125.10 ms │     no change │
│ QQuery 14 │     504.34 / 508.97 ±4.38 / 517.08 ms │     506.99 / 511.74 ±3.74 / 516.80 ms │     no change │
│ QQuery 15 │        60.02 / 61.51 ±1.52 / 64.43 ms │        60.09 / 60.73 ±0.53 / 61.42 ms │     no change │
│ QQuery 16 │           6.90 / 7.07 ±0.26 / 7.58 ms │          6.90 / 8.31 ±2.33 / 12.94 ms │  1.18x slower │
│ QQuery 17 │        81.44 / 82.89 ±1.54 / 85.85 ms │        82.01 / 83.14 ±1.56 / 86.20 ms │     no change │
│ QQuery 18 │     126.31 / 127.13 ±0.49 / 127.85 ms │     125.98 / 127.37 ±1.14 / 128.93 ms │     no change │
│ QQuery 19 │        42.09 / 43.80 ±2.71 / 49.16 ms │        42.22 / 42.75 ±0.28 / 42.99 ms │     no change │
│ QQuery 20 │        35.82 / 36.41 ±0.63 / 37.58 ms │        36.86 / 37.26 ±0.29 / 37.68 ms │     no change │
│ QQuery 21 │        17.19 / 17.48 ±0.24 / 17.80 ms │        17.43 / 17.73 ±0.22 / 18.05 ms │     no change │
│ QQuery 22 │        62.67 / 63.69 ±0.59 / 64.26 ms │        62.58 / 63.39 ±0.92 / 65.10 ms │     no change │
│ QQuery 23 │     482.19 / 488.69 ±9.83 / 508.17 ms │     484.19 / 489.04 ±4.58 / 497.11 ms │     no change │
│ QQuery 24 │     224.88 / 230.35 ±5.69 / 239.43 ms │     225.40 / 230.93 ±7.62 / 245.90 ms │     no change │
│ QQuery 25 │     114.93 / 116.82 ±3.05 / 122.88 ms │     116.14 / 119.12 ±2.80 / 124.27 ms │     no change │
│ QQuery 26 │        58.28 / 60.59 ±3.50 / 67.53 ms │        58.98 / 59.82 ±0.51 / 60.58 ms │     no change │
│ QQuery 27 │           6.77 / 6.90 ±0.10 / 7.07 ms │           6.66 / 6.81 ±0.14 / 7.08 ms │     no change │
│ QQuery 28 │        57.32 / 61.24 ±1.97 / 62.36 ms │        57.70 / 60.19 ±2.07 / 62.56 ms │     no change │
│ QQuery 29 │      98.28 / 102.20 ±4.38 / 108.89 ms │      98.92 / 103.10 ±3.92 / 109.74 ms │     no change │
│ QQuery 30 │        33.21 / 33.77 ±0.49 / 34.67 ms │        33.05 / 33.40 ±0.49 / 34.36 ms │     no change │
│ QQuery 31 │     113.22 / 114.02 ±0.63 / 114.99 ms │     111.93 / 114.52 ±3.62 / 121.69 ms │     no change │
│ QQuery 32 │        21.33 / 23.54 ±3.46 / 30.42 ms │        20.69 / 22.36 ±2.56 / 27.43 ms │     no change │
│ QQuery 33 │        39.59 / 39.87 ±0.22 / 40.23 ms │        39.42 / 39.85 ±0.30 / 40.16 ms │     no change │
│ QQuery 34 │         9.93 / 10.41 ±0.60 / 11.56 ms │        10.23 / 10.82 ±0.83 / 12.46 ms │     no change │
│ QQuery 35 │        77.93 / 78.71 ±0.66 / 79.77 ms │        78.34 / 79.11 ±0.68 / 80.32 ms │     no change │
│ QQuery 36 │           5.93 / 6.07 ±0.12 / 6.30 ms │           6.16 / 6.27 ±0.13 / 6.53 ms │     no change │
│ QQuery 37 │           6.96 / 7.03 ±0.04 / 7.07 ms │           6.88 / 6.98 ±0.08 / 7.07 ms │     no change │
│ QQuery 38 │        64.29 / 66.40 ±2.12 / 70.37 ms │        64.16 / 67.35 ±3.50 / 72.06 ms │     no change │
│ QQuery 39 │      99.26 / 100.93 ±1.27 / 103.01 ms │     100.23 / 100.94 ±0.59 / 101.63 ms │     no change │
│ QQuery 40 │        23.77 / 24.00 ±0.18 / 24.25 ms │        24.00 / 24.26 ±0.14 / 24.43 ms │     no change │
│ QQuery 41 │        11.69 / 11.79 ±0.12 / 12.02 ms │        11.56 / 11.72 ±0.21 / 12.13 ms │     no change │
│ QQuery 42 │        24.42 / 25.09 ±0.38 / 25.46 ms │        24.29 / 24.72 ±0.47 / 25.46 ms │     no change │
│ QQuery 43 │           4.79 / 4.90 ±0.15 / 5.21 ms │           4.84 / 5.01 ±0.16 / 5.30 ms │     no change │
│ QQuery 44 │        10.73 / 11.81 ±2.06 / 15.92 ms │        11.27 / 13.86 ±2.87 / 17.63 ms │  1.17x slower │
│ QQuery 45 │        40.17 / 46.99 ±6.18 / 55.21 ms │        40.86 / 41.46 ±0.40 / 41.90 ms │ +1.13x faster │
│ QQuery 46 │        12.34 / 12.75 ±0.30 / 13.16 ms │        12.49 / 13.02 ±0.35 / 13.40 ms │     no change │
│ QQuery 47 │     237.07 / 242.83 ±3.46 / 247.49 ms │     242.61 / 245.62 ±2.30 / 249.22 ms │     no change │
│ QQuery 48 │       98.21 / 99.77 ±2.23 / 104.18 ms │        97.98 / 98.65 ±0.61 / 99.67 ms │     no change │
│ QQuery 49 │        79.38 / 81.63 ±2.72 / 86.97 ms │        80.06 / 81.79 ±2.79 / 87.34 ms │     no change │
│ QQuery 50 │        59.73 / 60.43 ±0.38 / 60.91 ms │        59.69 / 60.52 ±0.47 / 61.05 ms │     no change │
│ QQuery 51 │       94.23 / 98.18 ±4.33 / 105.18 ms │      95.56 / 100.07 ±4.40 / 108.08 ms │     no change │
│ QQuery 52 │        24.69 / 25.09 ±0.38 / 25.77 ms │        24.40 / 24.84 ±0.40 / 25.59 ms │     no change │
│ QQuery 53 │        29.80 / 30.04 ±0.24 / 30.49 ms │        30.17 / 30.42 ±0.14 / 30.58 ms │     no change │
│ QQuery 54 │        56.11 / 56.38 ±0.24 / 56.68 ms │        56.01 / 56.49 ±0.35 / 56.94 ms │     no change │
│ QQuery 55 │        23.82 / 24.26 ±0.28 / 24.61 ms │        24.12 / 24.50 ±0.34 / 24.99 ms │     no change │
│ QQuery 56 │        39.16 / 39.81 ±0.65 / 41.05 ms │        39.53 / 40.85 ±1.04 / 42.60 ms │     no change │
│ QQuery 57 │     177.33 / 181.34 ±3.97 / 188.79 ms │     179.47 / 181.57 ±1.84 / 183.91 ms │     no change │
│ QQuery 58 │     118.81 / 121.16 ±3.07 / 126.64 ms │     118.76 / 120.44 ±2.44 / 125.25 ms │     no change │
│ QQuery 59 │     121.07 / 124.87 ±3.75 / 129.45 ms │     118.95 / 121.18 ±4.03 / 129.23 ms │     no change │
│ QQuery 60 │        39.91 / 40.57 ±0.37 / 41.03 ms │        40.22 / 41.08 ±0.63 / 42.13 ms │     no change │
│ QQuery 61 │        12.71 / 12.89 ±0.14 / 13.13 ms │        12.98 / 13.12 ±0.16 / 13.41 ms │     no change │
│ QQuery 62 │        46.62 / 47.12 ±0.38 / 47.73 ms │        46.75 / 47.13 ±0.32 / 47.53 ms │     no change │
│ QQuery 63 │        30.01 / 30.15 ±0.11 / 30.33 ms │        30.20 / 30.46 ±0.20 / 30.82 ms │     no change │
│ QQuery 64 │     398.09 / 403.25 ±7.62 / 418.29 ms │     395.00 / 403.23 ±6.81 / 414.69 ms │     no change │
│ QQuery 65 │     147.34 / 153.30 ±4.08 / 157.92 ms │     150.30 / 156.62 ±7.06 / 169.01 ms │     no change │
│ QQuery 66 │        79.84 / 81.43 ±2.35 / 85.98 ms │        80.92 / 85.12 ±7.32 / 99.74 ms │     no change │
│ QQuery 67 │     248.61 / 255.61 ±6.20 / 266.10 ms │     258.68 / 266.61 ±5.70 / 275.05 ms │     no change │
│ QQuery 68 │        11.73 / 11.94 ±0.22 / 12.32 ms │        12.34 / 12.63 ±0.21 / 12.86 ms │  1.06x slower │
│ QQuery 69 │        62.90 / 63.08 ±0.15 / 63.27 ms │        63.48 / 63.83 ±0.31 / 64.32 ms │     no change │
│ QQuery 70 │     105.11 / 114.66 ±8.17 / 124.09 ms │     108.37 / 115.41 ±7.93 / 130.37 ms │     no change │
│ QQuery 71 │        35.92 / 36.29 ±0.20 / 36.50 ms │        36.21 / 36.59 ±0.34 / 37.03 ms │     no change │
│ QQuery 72 │ 2192.50 / 2252.18 ±44.05 / 2307.84 ms │ 2147.65 / 2216.48 ±58.35 / 2299.64 ms │     no change │
│ QQuery 73 │          9.68 / 9.88 ±0.23 / 10.21 ms │          9.71 / 9.90 ±0.15 / 10.16 ms │     no change │
│ QQuery 74 │     178.80 / 184.42 ±4.47 / 191.33 ms │     179.80 / 185.15 ±5.13 / 192.42 ms │     no change │
│ QQuery 75 │     148.44 / 150.15 ±1.26 / 152.36 ms │     150.51 / 152.36 ±1.48 / 154.02 ms │     no change │
│ QQuery 76 │        35.77 / 36.21 ±0.32 / 36.63 ms │        36.42 / 37.06 ±0.50 / 37.60 ms │     no change │
│ QQuery 77 │        60.79 / 61.31 ±0.47 / 61.95 ms │        61.75 / 62.62 ±0.60 / 63.56 ms │     no change │
│ QQuery 78 │     190.42 / 196.58 ±5.57 / 203.72 ms │     193.94 / 200.50 ±5.18 / 207.39 ms │     no change │
│ QQuery 79 │        67.58 / 71.64 ±4.25 / 78.66 ms │        68.66 / 70.84 ±2.68 / 76.00 ms │     no change │
│ QQuery 80 │     101.25 / 103.31 ±2.12 / 107.17 ms │     102.55 / 104.58 ±1.76 / 107.09 ms │     no change │
│ QQuery 81 │        26.26 / 26.44 ±0.16 / 26.74 ms │        26.34 / 26.54 ±0.24 / 27.00 ms │     no change │
│ QQuery 82 │        16.56 / 16.81 ±0.23 / 17.20 ms │        16.80 / 16.92 ±0.09 / 17.03 ms │     no change │
│ QQuery 83 │        41.00 / 44.91 ±4.67 / 52.47 ms │        41.18 / 46.13 ±4.84 / 52.24 ms │     no change │
│ QQuery 84 │        34.97 / 35.67 ±0.48 / 36.35 ms │        35.75 / 36.06 ±0.26 / 36.47 ms │     no change │
│ QQuery 85 │     109.10 / 110.56 ±1.25 / 112.81 ms │     110.11 / 111.71 ±0.94 / 112.49 ms │     no change │
│ QQuery 86 │        25.70 / 27.14 ±2.24 / 31.52 ms │        25.67 / 27.01 ±2.24 / 31.48 ms │     no change │
│ QQuery 87 │        65.57 / 67.37 ±1.31 / 69.26 ms │        66.24 / 68.10 ±1.59 / 70.12 ms │     no change │
│ QQuery 88 │        62.63 / 63.12 ±0.31 / 63.56 ms │        63.07 / 63.65 ±0.43 / 64.28 ms │     no change │
│ QQuery 89 │        36.06 / 36.36 ±0.25 / 36.64 ms │        36.42 / 36.78 ±0.33 / 37.22 ms │     no change │
│ QQuery 90 │        17.04 / 17.27 ±0.17 / 17.55 ms │        17.35 / 17.57 ±0.12 / 17.71 ms │     no change │
│ QQuery 91 │        45.02 / 48.26 ±3.30 / 53.84 ms │        46.08 / 46.31 ±0.20 / 46.65 ms │     no change │
│ QQuery 92 │        30.18 / 31.31 ±0.71 / 32.00 ms │        31.04 / 31.44 ±0.31 / 31.88 ms │     no change │
│ QQuery 93 │        52.05 / 52.68 ±0.66 / 53.95 ms │        52.01 / 52.72 ±0.79 / 53.73 ms │     no change │
│ QQuery 94 │        39.31 / 39.46 ±0.14 / 39.69 ms │        39.46 / 39.63 ±0.22 / 40.06 ms │     no change │
│ QQuery 95 │        84.70 / 87.45 ±2.76 / 92.06 ms │        85.98 / 88.14 ±3.14 / 94.11 ms │     no change │
│ QQuery 96 │        24.15 / 24.46 ±0.22 / 24.71 ms │        24.82 / 24.95 ±0.18 / 25.31 ms │     no change │
│ QQuery 97 │        46.05 / 46.97 ±0.75 / 48.18 ms │        46.83 / 47.65 ±0.57 / 48.34 ms │     no change │
│ QQuery 98 │        42.59 / 43.75 ±0.99 / 45.29 ms │        43.97 / 44.22 ±0.29 / 44.74 ms │     no change │
│ QQuery 99 │        70.71 / 72.86 ±2.45 / 75.90 ms │        70.31 / 73.58 ±5.68 / 84.94 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 10445.09ms │
│ Total Time (codex_late-materialization-topk)   │ 10467.56ms │
│ Average Time (HEAD)                            │   105.51ms │
│ Average Time (codex_late-materialization-topk) │   105.73ms │
│ Queries Faster                                 │          1 │
│ Queries Slower                                 │          3 │
│ Queries with No Change                         │         95 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	6.9 GiB
Avg memory	6.1 GiB
CPU user	241.8s
CPU sys	7.1s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	6.9 GiB
Avg memory	6.1 GiB
CPU user	240.0s
CPU sys	7.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-05-29T12:02:48Z

🤖 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_late-materialization-topk
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃       codex_late-materialization-topk ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.15 / 3.27 ±4.14 / 11.55 ms │          1.16 / 3.28 ±4.18 / 11.65 ms │     no change │
│ QQuery 1  │        12.58 / 12.77 ±0.13 / 12.92 ms │        12.47 / 12.60 ±0.10 / 12.74 ms │     no change │
│ QQuery 2  │        35.44 / 35.83 ±0.35 / 36.39 ms │        35.39 / 35.61 ±0.17 / 35.89 ms │     no change │
│ QQuery 3  │        30.61 / 31.08 ±0.56 / 32.14 ms │        30.43 / 30.72 ±0.39 / 31.46 ms │     no change │
│ QQuery 4  │     218.95 / 223.62 ±3.35 / 227.92 ms │     215.64 / 222.58 ±4.49 / 228.85 ms │     no change │
│ QQuery 5  │     270.05 / 272.44 ±2.61 / 276.59 ms │     268.02 / 271.05 ±3.81 / 278.44 ms │     no change │
│ QQuery 6  │           1.18 / 1.33 ±0.22 / 1.75 ms │           1.21 / 1.35 ±0.21 / 1.77 ms │     no change │
│ QQuery 7  │        13.83 / 13.97 ±0.11 / 14.12 ms │        13.68 / 13.78 ±0.06 / 13.87 ms │     no change │
│ QQuery 8  │     316.75 / 319.78 ±2.39 / 322.61 ms │     313.51 / 320.45 ±6.42 / 329.77 ms │     no change │
│ QQuery 9  │    448.14 / 458.56 ±12.60 / 480.91 ms │    432.62 / 453.31 ±10.75 / 461.75 ms │     no change │
│ QQuery 10 │        69.57 / 70.56 ±0.97 / 72.04 ms │        70.13 / 71.19 ±0.87 / 72.25 ms │     no change │
│ QQuery 11 │        80.48 / 82.15 ±1.15 / 83.89 ms │        81.52 / 85.05 ±5.71 / 96.34 ms │     no change │
│ QQuery 12 │     264.63 / 268.13 ±3.03 / 271.71 ms │     261.48 / 267.97 ±5.47 / 275.35 ms │     no change │
│ QQuery 13 │     361.91 / 376.79 ±9.71 / 389.98 ms │     357.29 / 368.89 ±6.96 / 376.29 ms │     no change │
│ QQuery 14 │     279.13 / 283.80 ±5.51 / 294.14 ms │     277.12 / 280.12 ±4.09 / 288.20 ms │     no change │
│ QQuery 15 │     265.94 / 271.91 ±7.18 / 285.30 ms │     262.71 / 268.75 ±6.61 / 277.05 ms │     no change │
│ QQuery 16 │     606.70 / 613.11 ±6.24 / 624.54 ms │     605.82 / 618.30 ±9.82 / 632.04 ms │     no change │
│ QQuery 17 │    611.31 / 621.56 ±11.81 / 644.36 ms │     609.77 / 619.09 ±9.50 / 636.08 ms │     no change │
│ QQuery 18 │ 1231.25 / 1262.38 ±17.71 / 1286.34 ms │ 1233.90 / 1272.84 ±28.64 / 1306.02 ms │     no change │
│ QQuery 19 │        27.46 / 32.57 ±8.60 / 49.65 ms │        28.14 / 31.19 ±4.91 / 40.83 ms │     no change │
│ QQuery 20 │     518.99 / 523.41 ±5.80 / 534.68 ms │     523.58 / 525.94 ±2.73 / 531.07 ms │     no change │
│ QQuery 21 │     589.87 / 596.48 ±4.91 / 602.49 ms │     604.18 / 607.04 ±2.93 / 610.67 ms │     no change │
│ QQuery 22 │  1049.90 / 1062.75 ±8.97 / 1072.28 ms │  1061.51 / 1071.48 ±8.02 / 1081.02 ms │     no change │
│ QQuery 23 │ 3162.41 / 3199.94 ±29.57 / 3243.60 ms │ 3160.92 / 3212.25 ±41.85 / 3272.75 ms │     no change │
│ QQuery 24 │        41.32 / 46.89 ±5.45 / 55.65 ms │        41.47 / 46.89 ±5.73 / 54.81 ms │     no change │
│ QQuery 25 │     110.45 / 114.22 ±4.89 / 123.86 ms │     110.55 / 111.23 ±0.96 / 113.11 ms │     no change │
│ QQuery 26 │        41.32 / 42.34 ±0.95 / 43.98 ms │        41.82 / 46.86 ±9.40 / 65.64 ms │  1.11x slower │
│ QQuery 27 │     666.37 / 675.01 ±6.56 / 685.31 ms │    664.74 / 680.49 ±16.44 / 710.63 ms │     no change │
│ QQuery 28 │  3024.36 / 3036.04 ±6.67 / 3043.03 ms │ 3031.41 / 3053.22 ±19.90 / 3087.60 ms │     no change │
│ QQuery 29 │      39.92 / 61.50 ±28.90 / 115.38 ms │        39.68 / 40.00 ±0.29 / 40.40 ms │ +1.54x faster │
│ QQuery 30 │     296.52 / 305.35 ±5.79 / 313.35 ms │     296.13 / 305.62 ±9.03 / 316.65 ms │     no change │
│ QQuery 31 │     277.00 / 292.54 ±8.69 / 300.78 ms │    273.76 / 286.88 ±10.42 / 297.97 ms │     no change │
│ QQuery 32 │   916.45 / 976.09 ±36.37 / 1013.33 ms │    923.90 / 945.62 ±26.89 / 997.58 ms │     no change │
│ QQuery 33 │ 1425.27 / 1457.42 ±19.90 / 1483.76 ms │ 1430.13 / 1465.32 ±22.24 / 1489.24 ms │     no change │
│ QQuery 34 │ 1446.35 / 1523.66 ±69.41 / 1612.38 ms │ 1451.00 / 1475.18 ±18.85 / 1502.26 ms │     no change │
│ QQuery 35 │    269.15 / 317.63 ±63.83 / 440.39 ms │    274.65 / 292.67 ±21.11 / 334.10 ms │ +1.09x faster │
│ QQuery 36 │      66.37 / 75.68 ±15.11 / 105.84 ms │        64.69 / 69.44 ±6.50 / 82.13 ms │ +1.09x faster │
│ QQuery 37 │        34.73 / 35.63 ±0.54 / 36.30 ms │        34.90 / 38.02 ±4.37 / 46.47 ms │  1.07x slower │
│ QQuery 38 │        39.10 / 40.38 ±0.88 / 41.52 ms │        40.61 / 47.14 ±6.58 / 58.83 ms │  1.17x slower │
│ QQuery 39 │     147.05 / 153.54 ±5.06 / 161.09 ms │    139.22 / 149.89 ±10.94 / 170.15 ms │     no change │
│ QQuery 40 │        13.50 / 13.98 ±0.70 / 15.36 ms │        13.49 / 13.67 ±0.30 / 14.27 ms │     no change │
│ QQuery 41 │        13.19 / 13.27 ±0.06 / 13.37 ms │        13.63 / 18.11 ±6.23 / 29.94 ms │  1.36x slower │
│ QQuery 42 │        12.75 / 12.85 ±0.09 / 13.01 ms │        12.53 / 16.05 ±6.69 / 29.41 ms │  1.25x slower │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                              ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                              │ 19832.21ms │
│ Total Time (codex_late-materialization-topk)   │ 19767.14ms │
│ Average Time (HEAD)                            │   461.21ms │
│ Average Time (codex_late-materialization-topk) │   459.70ms │
│ Queries Faster                                 │          3 │
│ Queries Slower                                 │          5 │
│ Queries with No Change                         │         35 │
│ Queries with Failure                           │          0 │
└────────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	100.0s
Peak memory	30.6 GiB
Avg memory	23.2 GiB
CPU user	1016.7s
CPU sys	73.7s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	100.0s
Peak memory	30.3 GiB
Avg memory	23.2 GiB
CPU user	1016.6s
CPU sys	73.0s
Peak spill	0 B

File an issue against this benchmark runner

Add row-number late materialization for TopK

68e8de6

github-actions Bot added documentation Improvements or additions to documentation optimizer Optimizer rules core Core DataFusion crate common Related to common crate datasource Changes to the datasource crate labels May 29, 2026

Merge remote-tracking branch 'upstream/main' into codex/late-material…

cc18034

…ization-topk # Conflicts: # datafusion/core/src/optimizer_rule_reference.md # datafusion/datasource-parquet/src/opener/mod.rs

github-actions Bot added the development-process Related to development process of DataFusion label May 29, 2026

github-actions Bot added the auto detected api change Auto detected API change label May 29, 2026

Support filtered TopK late materialization

c5da797

Conversation

Dandandan commented May 29, 2026

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!

Dandandan commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

github-actions Bot commented May 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

AdamGS commented May 29, 2026

Uh oh!

Dandandan commented May 29, 2026

Uh oh!

Dandandan commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

adriangbot commented May 29, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

github-actions Bot commented May 29, 2026 •

edited

Loading