perf: sort multi-column runs via row format and coalesce them

[Dandandan]

Which issue does this PR close?

Follow-up to #23202 (coalesce single-column sort runs). No specific issue.

Rationale for this change

#23202 coalesces buffered runs into fewer, larger runs to cut merge fan-in, but
only for single-column sorts. Multi-column runs were deliberately left one
run per batch, because coalescing them and sorting the larger run with
lexsort_to_indices regresses: lexsort's comparator walks the key columns
one at a time with per-comparison dynamic dispatch (and heap touches for
strings/dictionaries), and that cost grows super-linearly with run size.

The fix is to change how a multi-column run is sorted, not just its size: for
keys whose leading column is an expensive (variable-length or dictionary)
comparison, encode the key once into the Arrow row format and argsort the
row indices with a cheap memcmp. This is the same comparison the streaming
merge already uses (RowCursorStream), so it is a natural fit — and it removes
exactly the comparator cost that made coalescing regress. With that in place we
can coalesce all multi-column runs.

A microbenchmark (sort_indices, added here) isolates the index-computation
cost — lexsort_to_indices vs. row-encode+argsort — by key shape and run size
(speedup = lexsort / rowsort; >1 means row format is faster):

key shape	1k	64k	1M
i64 (single)	0.16×	0.17×	0.15×
utf8 high (single)	0.22×	0.31×	0.66×
utf8 tuple	1.86×	2.46×	3.38×
dict tuple	2.06×	2.66×	2.72×
utf8 view tuple	0.67×	1.10×	1.48×
mixed tuple (i64-leading)	0.49×	0.67×	0.78×

This is what motivates the gate: row format wins big for string/dict-leading
keys (and more as runs grow), but loses for single-column keys and for
primitive-leading keys (where lexsort short-circuits on the cheap first
column). So the row-format path is gated to multi-column keys with an expensive
leading column; everything else keeps lexsort.

End-to-end (cargo bench --bench sort, the SortExec cases, vs. current main):

benchmark	speedup
sort utf8 tuple 100k / 1M	1.34× / 1.19×
sort utf8 view tuple 100k / 1M	1.22× / 1.17×
sort utf8 dictionary tuple 100k / 1M	1.23× / 1.74×
sort mixed dictionary tuple 100k / 1M	1.47× / 1.81×
sort mixed tuple 100k / 1M	1.44× / 1.45×
sort mixed tuple with utf8 view 100k / 1M	1.51× / 1.40×
sort i64 100k / 1M	~unchanged (noise)

String/dict-leading tuples win via coalesce + row-format sort; the
primitive-leading mixed tuple wins ~1.4× purely from coalesce + lexsort on
larger runs (it now also gets coalesced). Single-column sorts are unchanged.

What changes are included in this PR?

• sorted_indices() in sorts/stream.rs: computes the sorted index permutation,
  choosing the Arrow row format for multi-column keys with an expensive leading
  column (use_row_format_sort) and lexsort_to_indices otherwise (single
  column, primitive-leading, or fetch/top-k).
• IncrementalSortIterator (the single point every in-memory sort run flows
  through) now calls sorted_indices(). No merge/cursor changes.
• in_mem_sort_stream now coalesces all runs, not just single-column ones.
• New microbenchmark datafusion/core/benches/sort_indices.rs.

Are these changes tested?

Existing sorts:: unit tests pass (NULLs, sort options, multi-column merge,
spill). Correctness of the row-format ordering is covered by the existing
multi-column sort tests; the new microbenchmark documents the perf rationale.

Are there any user-facing changes?

No (internal performance change only).

[Dandandan]

run benchmark sort_tpch sort_tpch10 tpcds tpch10

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4821177817-725-qddlj 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing perf/sort-rowformat-multicol-runs (1b99d3e) to 3bb9314 (merge-base) diff using: sort_tpch
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4821177817-727-rmzsj 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing perf/sort-rowformat-multicol-runs (1b99d3e) to 3bb9314 (merge-base) diff using: tpcds
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4821177817-728-jxcxp 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing perf/sort-rowformat-multicol-runs (1b99d3e) to 3bb9314 (merge-base) diff using: tpch10
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4821177817-726-8mk6c 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing perf/sort-rowformat-multicol-runs (1b99d3e) to 3bb9314 (merge-base) diff using: sort_tpch10
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 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 perf_sort-rowformat-multicol-runs
--------------------
Benchmark sort_tpch1.json
--------------------
┏━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃                               HEAD ┃  perf_sort-rowformat-multicol-runs ┃        Change ┃
┡━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ Q1    │  111.39 / 112.10 ±0.82 / 113.70 ms │  112.38 / 113.00 ±0.85 / 114.68 ms │     no change │
│ Q2    │  100.52 / 101.31 ±1.04 / 103.36 ms │  100.32 / 100.99 ±0.90 / 102.70 ms │     no change │
│ Q3    │  646.65 / 651.67 ±4.15 / 657.31 ms │  645.54 / 650.10 ±3.30 / 654.37 ms │     no change │
│ Q4    │  180.17 / 185.30 ±6.85 / 198.10 ms │  160.48 / 163.03 ±3.11 / 169.10 ms │ +1.14x faster │
│ Q5    │  262.12 / 263.60 ±1.46 / 266.10 ms │  268.44 / 268.82 ±0.37 / 269.49 ms │     no change │
│ Q6    │  274.99 / 276.59 ±1.26 / 278.09 ms │ 280.28 / 286.34 ±11.39 / 309.11 ms │     no change │
│ Q7    │  444.89 / 447.68 ±2.01 / 451.15 ms │  446.82 / 448.44 ±2.12 / 452.61 ms │     no change │
│ Q8    │  313.60 / 320.93 ±9.40 / 338.21 ms │  308.46 / 319.29 ±9.95 / 336.24 ms │     no change │
│ Q9    │ 327.81 / 347.85 ±16.02 / 370.08 ms │ 331.74 / 352.72 ±15.78 / 374.20 ms │     no change │
│ Q10   │  467.87 / 473.81 ±3.43 / 477.22 ms │  471.35 / 477.48 ±4.22 / 481.13 ms │     no change │
│ Q11   │ 229.96 / 239.37 ±13.36 / 265.91 ms │ 227.09 / 243.23 ±25.07 / 292.96 ms │     no change │
└───────┴────────────────────────────────────┴────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                │ 3420.23ms │
│ Total Time (perf_sort-rowformat-multicol-runs)   │ 3423.44ms │
│ Average Time (HEAD)                              │  310.93ms │
│ Average Time (perf_sort-rowformat-multicol-runs) │  311.22ms │
│ Queries Faster                                   │         1 │
│ Queries Slower                                   │         0 │
│ Queries with No Change                           │        10 │
│ Queries with Failure                             │         0 │
└──────────────────────────────────────────────────┴───────────┘

Resource Usage

sort_tpch — base (merge-base)

Metric	Value
Wall time	20.0s
Peak memory	2.5 GiB
Avg memory	1.2 GiB
CPU user	64.2s
CPU sys	2.8s
Peak spill	0 B

sort_tpch — branch

Metric	Value
Wall time	20.0s
Peak memory	2.5 GiB
Avg memory	1.2 GiB
CPU user	63.6s
CPU sys	2.8s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 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 perf_sort-rowformat-multicol-runs
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃     perf_sort-rowformat-multicol-runs ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.54 / 6.05 ±0.83 / 7.70 ms │           5.55 / 6.05 ±0.80 / 7.63 ms │    no change │
│ QQuery 2  │        81.24 / 81.87 ±0.49 / 82.69 ms │        81.27 / 81.59 ±0.19 / 81.80 ms │    no change │
│ QQuery 3  │        29.58 / 29.87 ±0.17 / 30.04 ms │        29.77 / 29.90 ±0.10 / 30.01 ms │    no change │
│ QQuery 4  │     489.31 / 495.81 ±3.89 / 500.61 ms │     492.56 / 495.59 ±3.45 / 502.23 ms │    no change │
│ QQuery 5  │        52.24 / 52.75 ±0.35 / 53.23 ms │        51.75 / 52.41 ±0.53 / 52.98 ms │    no change │
│ QQuery 6  │        36.35 / 37.06 ±0.45 / 37.56 ms │        36.65 / 36.94 ±0.29 / 37.46 ms │    no change │
│ QQuery 7  │        93.95 / 94.48 ±0.47 / 95.34 ms │        95.25 / 96.03 ±0.46 / 96.48 ms │    no change │
│ QQuery 8  │        37.37 / 38.12 ±0.40 / 38.50 ms │        37.01 / 37.53 ±0.66 / 38.75 ms │    no change │
│ QQuery 9  │        53.72 / 56.56 ±2.16 / 59.14 ms │        55.09 / 56.72 ±1.36 / 58.98 ms │    no change │
│ QQuery 10 │        63.14 / 63.58 ±0.29 / 63.93 ms │        63.77 / 64.14 ±0.31 / 64.60 ms │    no change │
│ QQuery 11 │     302.21 / 311.39 ±7.42 / 324.71 ms │     307.10 / 311.07 ±2.61 / 314.53 ms │    no change │
│ QQuery 12 │        28.89 / 29.27 ±0.54 / 30.32 ms │        28.96 / 29.26 ±0.22 / 29.63 ms │    no change │
│ QQuery 13 │     118.96 / 119.38 ±0.31 / 119.88 ms │     120.07 / 121.10 ±0.74 / 121.99 ms │    no change │
│ QQuery 14 │     412.92 / 417.51 ±3.13 / 421.29 ms │     413.41 / 417.43 ±2.60 / 420.76 ms │    no change │
│ QQuery 15 │        58.23 / 58.89 ±0.49 / 59.60 ms │        57.87 / 58.71 ±0.59 / 59.70 ms │    no change │
│ QQuery 16 │           6.81 / 6.99 ±0.18 / 7.35 ms │           6.75 / 6.93 ±0.21 / 7.33 ms │    no change │
│ QQuery 17 │        80.48 / 81.88 ±1.92 / 85.67 ms │        80.29 / 82.47 ±3.14 / 88.69 ms │    no change │
│ QQuery 18 │     122.78 / 125.17 ±1.52 / 127.29 ms │     124.51 / 126.08 ±1.07 / 127.13 ms │    no change │
│ QQuery 19 │        41.68 / 41.99 ±0.25 / 42.39 ms │        41.65 / 42.11 ±0.23 / 42.32 ms │    no change │
│ QQuery 20 │        35.37 / 36.20 ±0.76 / 37.63 ms │        35.51 / 36.31 ±0.58 / 37.08 ms │    no change │
│ QQuery 21 │        17.72 / 18.37 ±0.78 / 19.83 ms │        17.65 / 17.90 ±0.17 / 18.12 ms │    no change │
│ QQuery 22 │        62.09 / 62.73 ±0.56 / 63.66 ms │        62.76 / 64.13 ±1.24 / 65.79 ms │    no change │
│ QQuery 23 │     355.40 / 361.58 ±4.06 / 365.84 ms │    356.45 / 370.68 ±15.37 / 398.47 ms │    no change │
│ QQuery 24 │     225.19 / 229.97 ±5.19 / 238.45 ms │     227.40 / 229.13 ±2.39 / 233.70 ms │    no change │
│ QQuery 25 │     109.97 / 111.80 ±1.57 / 114.47 ms │     110.67 / 112.48 ±1.63 / 115.52 ms │    no change │
│ QQuery 26 │        58.17 / 60.24 ±3.25 / 66.66 ms │        59.10 / 60.16 ±1.33 / 62.76 ms │    no change │
│ QQuery 27 │           6.40 / 6.52 ±0.17 / 6.86 ms │           6.45 / 6.61 ±0.16 / 6.92 ms │    no change │
│ QQuery 28 │        57.38 / 61.16 ±2.11 / 63.79 ms │        57.15 / 61.25 ±2.11 / 62.92 ms │    no change │
│ QQuery 29 │      97.80 / 101.91 ±3.74 / 107.75 ms │      97.55 / 102.03 ±5.93 / 113.75 ms │    no change │
│ QQuery 30 │        32.72 / 33.61 ±0.83 / 34.92 ms │        33.29 / 33.73 ±0.36 / 34.25 ms │    no change │
│ QQuery 31 │     113.03 / 113.25 ±0.19 / 113.56 ms │     112.08 / 113.13 ±0.66 / 114.14 ms │    no change │
│ QQuery 32 │        20.66 / 20.82 ±0.21 / 21.21 ms │        20.89 / 22.54 ±2.39 / 27.28 ms │ 1.08x slower │
│ QQuery 33 │        38.02 / 39.92 ±3.53 / 46.98 ms │        38.12 / 38.62 ±0.36 / 39.01 ms │    no change │
│ QQuery 34 │         9.81 / 10.39 ±0.64 / 11.57 ms │        10.03 / 10.27 ±0.19 / 10.53 ms │    no change │
│ QQuery 35 │        72.65 / 72.90 ±0.27 / 73.30 ms │        72.71 / 73.58 ±0.84 / 74.74 ms │    no change │
│ QQuery 36 │           5.85 / 5.95 ±0.17 / 6.30 ms │           5.86 / 5.98 ±0.23 / 6.44 ms │    no change │
│ QQuery 37 │           6.97 / 7.08 ±0.08 / 7.21 ms │           7.02 / 7.29 ±0.22 / 7.61 ms │    no change │
│ QQuery 38 │        62.52 / 62.86 ±0.22 / 63.15 ms │        63.28 / 63.85 ±0.48 / 64.62 ms │    no change │
│ QQuery 39 │        85.85 / 86.96 ±0.62 / 87.63 ms │        86.54 / 87.29 ±0.92 / 88.99 ms │    no change │
│ QQuery 40 │        23.43 / 23.76 ±0.21 / 23.98 ms │        23.68 / 24.06 ±0.32 / 24.58 ms │    no change │
│ QQuery 41 │        11.46 / 11.70 ±0.23 / 12.13 ms │        11.35 / 11.61 ±0.24 / 12.07 ms │    no change │
│ QQuery 42 │        24.20 / 24.82 ±0.42 / 25.29 ms │        24.12 / 24.33 ±0.24 / 24.72 ms │    no change │
│ QQuery 43 │           4.90 / 5.04 ±0.12 / 5.23 ms │           4.90 / 5.00 ±0.13 / 5.25 ms │    no change │
│ QQuery 44 │           9.41 / 9.56 ±0.14 / 9.78 ms │           9.36 / 9.47 ±0.07 / 9.56 ms │    no change │
│ QQuery 45 │        38.34 / 39.05 ±0.67 / 40.11 ms │        38.31 / 39.75 ±1.84 / 43.34 ms │    no change │
│ QQuery 46 │        11.59 / 12.45 ±1.24 / 14.89 ms │        11.79 / 12.38 ±0.67 / 13.68 ms │    no change │
│ QQuery 47 │     232.69 / 235.33 ±3.09 / 240.64 ms │     232.11 / 236.72 ±3.35 / 240.47 ms │    no change │
│ QQuery 48 │        96.47 / 96.85 ±0.29 / 97.30 ms │       98.10 / 98.98 ±0.87 / 100.43 ms │    no change │
│ QQuery 49 │        76.80 / 77.63 ±0.54 / 78.40 ms │        76.84 / 77.40 ±0.33 / 77.81 ms │    no change │
│ QQuery 50 │        59.21 / 59.90 ±0.91 / 61.61 ms │        59.27 / 59.90 ±0.60 / 60.98 ms │    no change │
│ QQuery 51 │      97.82 / 100.20 ±2.42 / 103.80 ms │        97.15 / 98.39 ±0.91 / 99.87 ms │    no change │
│ QQuery 52 │        24.13 / 24.48 ±0.18 / 24.64 ms │        24.35 / 25.02 ±1.00 / 27.01 ms │    no change │
│ QQuery 53 │        29.94 / 32.10 ±3.34 / 38.69 ms │        29.87 / 30.97 ±1.50 / 33.94 ms │    no change │
│ QQuery 54 │        56.59 / 57.03 ±0.33 / 57.59 ms │        55.88 / 56.14 ±0.21 / 56.48 ms │    no change │
│ QQuery 55 │        23.58 / 23.88 ±0.21 / 24.17 ms │        23.60 / 23.81 ±0.17 / 24.02 ms │    no change │
│ QQuery 56 │        39.54 / 40.03 ±0.34 / 40.40 ms │        39.06 / 39.75 ±0.53 / 40.63 ms │    no change │
│ QQuery 57 │     177.98 / 180.31 ±2.82 / 185.76 ms │     177.51 / 179.47 ±1.77 / 182.80 ms │    no change │
│ QQuery 58 │     114.51 / 117.11 ±2.63 / 122.18 ms │     116.22 / 118.42 ±2.17 / 121.97 ms │    no change │
│ QQuery 59 │     117.60 / 118.37 ±0.55 / 119.22 ms │     117.48 / 118.25 ±0.47 / 118.93 ms │    no change │
│ QQuery 60 │        39.40 / 39.91 ±0.43 / 40.70 ms │        39.73 / 40.67 ±1.25 / 43.12 ms │    no change │
│ QQuery 61 │        12.34 / 12.57 ±0.20 / 12.94 ms │        12.65 / 12.76 ±0.13 / 13.01 ms │    no change │
│ QQuery 62 │        46.36 / 46.96 ±0.62 / 48.11 ms │        46.92 / 47.36 ±0.73 / 48.81 ms │    no change │
│ QQuery 63 │        29.99 / 30.73 ±0.47 / 31.33 ms │        30.04 / 30.53 ±0.54 / 31.45 ms │    no change │
│ QQuery 64 │     410.96 / 413.55 ±1.65 / 416.16 ms │     409.26 / 416.14 ±4.34 / 422.55 ms │    no change │
│ QQuery 65 │     147.78 / 154.22 ±5.66 / 161.43 ms │     149.37 / 151.49 ±1.83 / 154.08 ms │    no change │
│ QQuery 66 │        79.01 / 79.88 ±0.79 / 81.09 ms │        79.61 / 80.26 ±0.78 / 81.47 ms │    no change │
│ QQuery 67 │     240.31 / 244.07 ±3.71 / 250.27 ms │     239.76 / 243.16 ±3.74 / 248.95 ms │    no change │
│ QQuery 68 │        11.86 / 11.96 ±0.17 / 12.31 ms │        12.13 / 13.99 ±3.47 / 20.92 ms │ 1.17x slower │
│ QQuery 69 │        57.60 / 60.09 ±4.57 / 69.23 ms │        58.47 / 59.39 ±0.96 / 61.21 ms │    no change │
│ QQuery 70 │     104.90 / 107.15 ±2.76 / 111.56 ms │     105.32 / 107.04 ±1.35 / 108.75 ms │    no change │
│ QQuery 71 │        35.37 / 36.27 ±1.00 / 38.17 ms │        35.75 / 36.92 ±2.00 / 40.92 ms │    no change │
│ QQuery 72 │ 2019.92 / 2114.39 ±62.69 / 2216.98 ms │ 2083.26 / 2139.90 ±42.24 / 2179.49 ms │    no change │
│ QQuery 73 │          9.63 / 9.78 ±0.21 / 10.18 ms │          9.76 / 9.92 ±0.21 / 10.31 ms │    no change │
│ QQuery 74 │     170.08 / 171.50 ±1.28 / 173.14 ms │     171.79 / 174.05 ±2.19 / 178.04 ms │    no change │
│ QQuery 75 │     149.44 / 151.52 ±1.18 / 153.04 ms │     150.98 / 153.57 ±2.09 / 157.10 ms │    no change │
│ QQuery 76 │        36.33 / 37.38 ±1.63 / 40.59 ms │        35.64 / 35.98 ±0.24 / 36.39 ms │    no change │
│ QQuery 77 │        61.19 / 62.10 ±0.66 / 62.91 ms │        61.68 / 61.97 ±0.23 / 62.25 ms │    no change │
│ QQuery 78 │     185.55 / 187.67 ±1.84 / 190.21 ms │     186.26 / 189.96 ±3.16 / 195.02 ms │    no change │
│ QQuery 79 │        67.05 / 67.41 ±0.25 / 67.66 ms │        67.84 / 69.43 ±1.18 / 70.82 ms │    no change │
│ QQuery 80 │     100.11 / 100.41 ±0.34 / 101.06 ms │     100.27 / 101.68 ±1.32 / 103.76 ms │    no change │
│ QQuery 81 │        25.98 / 26.12 ±0.13 / 26.34 ms │        25.82 / 26.10 ±0.28 / 26.45 ms │    no change │
│ QQuery 82 │        16.56 / 16.71 ±0.12 / 16.87 ms │        16.45 / 16.64 ±0.12 / 16.83 ms │    no change │
│ QQuery 83 │        40.68 / 40.84 ±0.20 / 41.22 ms │        40.91 / 43.75 ±3.12 / 47.93 ms │ 1.07x slower │
│ QQuery 84 │        30.61 / 32.94 ±4.12 / 41.16 ms │        30.97 / 31.97 ±0.73 / 33.14 ms │    no change │
│ QQuery 85 │     106.48 / 108.12 ±1.54 / 110.80 ms │     107.85 / 109.34 ±1.20 / 111.31 ms │    no change │
│ QQuery 86 │        25.18 / 25.60 ±0.26 / 26.01 ms │        25.19 / 26.30 ±1.76 / 29.79 ms │    no change │
│ QQuery 87 │        63.09 / 65.74 ±2.86 / 71.29 ms │        63.71 / 65.76 ±2.22 / 69.90 ms │    no change │
│ QQuery 88 │        63.46 / 64.32 ±0.52 / 64.98 ms │        64.13 / 64.55 ±0.51 / 65.53 ms │    no change │
│ QQuery 89 │        36.13 / 36.56 ±0.40 / 37.10 ms │        36.19 / 36.57 ±0.35 / 37.23 ms │    no change │
│ QQuery 90 │        17.21 / 17.41 ±0.25 / 17.87 ms │        17.26 / 17.47 ±0.24 / 17.93 ms │    no change │
│ QQuery 91 │        46.12 / 46.26 ±0.13 / 46.48 ms │        46.78 / 47.44 ±0.61 / 48.48 ms │    no change │
│ QQuery 92 │        29.16 / 30.93 ±2.18 / 35.20 ms │        29.92 / 31.60 ±1.43 / 34.14 ms │    no change │
│ QQuery 93 │        50.13 / 51.32 ±0.82 / 52.42 ms │        50.35 / 51.12 ±0.71 / 52.42 ms │    no change │
│ QQuery 94 │        38.65 / 39.35 ±0.56 / 40.15 ms │        38.41 / 38.92 ±0.39 / 39.38 ms │    no change │
│ QQuery 95 │        81.46 / 81.96 ±0.35 / 82.42 ms │        81.64 / 83.44 ±1.54 / 85.95 ms │    no change │
│ QQuery 96 │        24.41 / 24.67 ±0.22 / 25.01 ms │        24.56 / 25.65 ±1.63 / 28.87 ms │    no change │
│ QQuery 97 │        55.72 / 56.71 ±1.01 / 58.50 ms │        55.26 / 55.68 ±0.51 / 56.64 ms │    no change │
│ QQuery 98 │        42.62 / 43.12 ±0.67 / 44.44 ms │        42.67 / 43.48 ±0.49 / 44.21 ms │    no change │
│ QQuery 99 │        70.21 / 70.89 ±0.49 / 71.60 ms │        70.72 / 72.08 ±2.39 / 76.85 ms │    no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                │  9961.50ms │
│ Total Time (perf_sort-rowformat-multicol-runs)   │ 10022.89ms │
│ Average Time (HEAD)                              │   100.62ms │
│ Average Time (perf_sort-rowformat-multicol-runs) │   101.24ms │
│ Queries Faster                                   │          0 │
│ Queries Slower                                   │          3 │
│ Queries with No Change                           │         96 │
│ Queries with Failure                             │          0 │
└──────────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	2.1 GiB
Avg memory	1.4 GiB
CPU user	228.7s
CPU sys	5.9s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	2.3 GiB
Avg memory	1.5 GiB
CPU user	228.3s
CPU sys	6.0s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 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 perf_sort-rowformat-multicol-runs
--------------------
Benchmark tpch_sf10.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃     perf_sort-rowformat-multicol-runs ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │     309.59 / 311.77 ±1.72 / 313.29 ms │     312.28 / 314.62 ±1.39 / 316.14 ms │ no change │
│ QQuery 2  │     107.85 / 109.93 ±1.73 / 112.42 ms │     105.35 / 108.64 ±3.02 / 113.89 ms │ no change │
│ QQuery 3  │     234.09 / 236.05 ±2.00 / 238.80 ms │     233.64 / 236.92 ±3.42 / 243.36 ms │ no change │
│ QQuery 4  │     111.04 / 112.98 ±1.06 / 114.25 ms │     112.86 / 115.26 ±1.48 / 116.85 ms │ no change │
│ QQuery 5  │     357.01 / 364.72 ±7.01 / 374.13 ms │     350.76 / 358.51 ±7.86 / 372.72 ms │ no change │
│ QQuery 6  │     124.46 / 126.38 ±2.42 / 131.04 ms │     127.10 / 129.84 ±4.78 / 139.38 ms │ no change │
│ QQuery 7  │     460.59 / 469.47 ±8.27 / 481.38 ms │     468.10 / 481.41 ±7.89 / 489.01 ms │ no change │
│ QQuery 8  │     378.74 / 385.36 ±4.56 / 390.28 ms │     381.31 / 387.71 ±3.25 / 390.21 ms │ no change │
│ QQuery 9  │     546.06 / 551.16 ±2.93 / 554.90 ms │    547.80 / 564.24 ±12.57 / 579.74 ms │ no change │
│ QQuery 10 │    305.23 / 317.21 ±12.77 / 333.89 ms │     301.18 / 312.34 ±5.96 / 318.84 ms │ no change │
│ QQuery 11 │       86.76 / 94.92 ±8.39 / 107.48 ms │        87.73 / 91.77 ±3.44 / 96.81 ms │ no change │
│ QQuery 12 │     176.78 / 180.29 ±3.61 / 187.11 ms │     179.74 / 186.89 ±8.49 / 200.94 ms │ no change │
│ QQuery 13 │    291.60 / 305.93 ±15.39 / 334.66 ms │     294.06 / 302.48 ±6.52 / 313.31 ms │ no change │
│ QQuery 14 │     170.39 / 176.89 ±6.34 / 187.46 ms │     171.98 / 177.38 ±6.34 / 187.19 ms │ no change │
│ QQuery 15 │     307.88 / 309.97 ±2.32 / 314.42 ms │     310.05 / 314.65 ±4.31 / 322.66 ms │ no change │
│ QQuery 16 │        71.54 / 74.57 ±2.32 / 77.46 ms │        71.06 / 75.03 ±2.64 / 77.96 ms │ no change │
│ QQuery 17 │     630.00 / 641.09 ±9.49 / 657.09 ms │     623.46 / 635.08 ±9.23 / 643.27 ms │ no change │
│ QQuery 18 │ 1434.22 / 1467.48 ±20.57 / 1495.82 ms │ 1411.11 / 1454.96 ±25.00 / 1485.26 ms │ no change │
│ QQuery 19 │    245.47 / 266.51 ±20.66 / 296.90 ms │    251.17 / 267.14 ±22.01 / 309.83 ms │ no change │
│ QQuery 20 │     274.90 / 290.98 ±8.36 / 297.70 ms │     284.59 / 293.60 ±8.63 / 303.97 ms │ no change │
│ QQuery 21 │     655.33 / 663.32 ±6.04 / 671.77 ms │    658.86 / 670.16 ±11.14 / 690.38 ms │ no change │
│ QQuery 22 │        61.20 / 64.31 ±3.06 / 69.83 ms │        61.49 / 64.62 ±2.12 / 66.75 ms │ no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                │ 7521.28ms │
│ Total Time (perf_sort-rowformat-multicol-runs)   │ 7543.26ms │
│ Average Time (HEAD)                              │  341.88ms │
│ Average Time (perf_sort-rowformat-multicol-runs) │  342.88ms │
│ Queries Faster                                   │         0 │
│ Queries Slower                                   │         0 │
│ Queries with No Change                           │        22 │
│ Queries with Failure                             │         0 │
└──────────────────────────────────────────────────┴───────────┘

Resource Usage

tpch10 — base (merge-base)

Metric	Value
Wall time	40.0s
Peak memory	4.4 GiB
Avg memory	1.6 GiB
CPU user	383.1s
CPU sys	20.0s
Peak spill	0 B

tpch10 — branch

Metric	Value
Wall time	40.0s
Peak memory	4.6 GiB
Avg memory	1.5 GiB
CPU user	383.6s
CPU sys	20.4s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 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 perf_sort-rowformat-multicol-runs
--------------------
Benchmark sort_tpch10.json
--------------------
┏━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃                                   HEAD ┃      perf_sort-rowformat-multicol-runs ┃        Change ┃
┡━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ Q1    │   1621.88 / 1630.19 ±6.05 / 1639.75 ms │   1629.45 / 1631.44 ±2.06 / 1635.14 ms │     no change │
│ Q2    │  1411.17 / 1445.79 ±22.36 / 1471.53 ms │  1413.23 / 1449.72 ±28.44 / 1493.88 ms │     no change │
│ Q3    │  9082.02 / 9182.58 ±61.25 / 9268.03 ms │  8907.46 / 9023.69 ±66.44 / 9090.09 ms │     no change │
│ Q4    │  2751.63 / 2800.38 ±25.63 / 2824.78 ms │  2595.07 / 2616.70 ±19.50 / 2650.91 ms │ +1.07x faster │
│ Q5    │  2671.61 / 2692.21 ±12.53 / 2706.68 ms │  2704.59 / 2724.44 ±14.90 / 2747.07 ms │     no change │
│ Q6    │  2815.87 / 2833.41 ±10.89 / 2848.69 ms │  2871.73 / 2890.33 ±11.93 / 2905.63 ms │     no change │
│ Q7    │  5586.74 / 5625.05 ±27.06 / 5663.47 ms │  5610.65 / 5639.08 ±30.71 / 5680.08 ms │     no change │
│ Q8    │  3964.03 / 4046.10 ±97.76 / 4220.66 ms │ 3881.33 / 4043.66 ±120.05 / 4188.84 ms │     no change │
│ Q9    │  4352.00 / 4383.90 ±25.47 / 4421.04 ms │  4318.43 / 4359.96 ±36.97 / 4409.99 ms │     no change │
│ Q10   │ 6829.08 / 7139.75 ±186.90 / 7380.30 ms │ 6712.30 / 7043.15 ±217.18 / 7384.78 ms │     no change │
│ Q11   │  3725.21 / 3780.09 ±77.61 / 3933.06 ms │  3655.91 / 3691.13 ±46.93 / 3784.14 ms │     no change │
└───────┴────────────────────────────────────────┴────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                │ 45559.45ms │
│ Total Time (perf_sort-rowformat-multicol-runs)   │ 45113.31ms │
│ Average Time (HEAD)                              │  4141.77ms │
│ Average Time (perf_sort-rowformat-multicol-runs) │  4101.21ms │
│ Queries Faster                                   │          1 │
│ Queries Slower                                   │          0 │
│ Queries with No Change                           │         10 │
│ Queries with Failure                             │          0 │
└──────────────────────────────────────────────────┴────────────┘

Resource Usage

sort_tpch10 — base (merge-base)

Metric	Value
Wall time	230.1s
Peak memory	19.2 GiB
Avg memory	7.4 GiB
CPU user	908.3s
CPU sys	69.4s
Peak spill	0 B

sort_tpch10 — branch

Metric	Value
Wall time	230.1s
Peak memory	19.3 GiB
Avg memory	7.3 GiB
CPU user	883.0s
CPU sys	68.2s
Peak spill	0 B

---

File an issue against this benchmark runner