GC & Deduplicate String View on Spill

[cetra3]

Which issue does this PR close?

• Closes External sort spill GC inflates deduplicated string views #23564

Rationale for this change

Spill files inflate string views via GC. I.e, a highly deduped string view can inflate considerably causing more memory and disk pressure.

What changes are included in this PR?

There is a new method added in the spill path to GC & Dedup string views

Are these changes tested?

Yes they are covered by existing tests.

However: I would like to make sure we run benchmarks with external sorting to see if this is a considerable regression. My instinct will tell me that, counterintuitively, this might be a performance increase when externally sorting since there is less bytes being moved around.

Are there any user-facing changes?

No this is an internal method change

[cetra3]

run benchmarks

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4969050947-1047-9mz5q 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 spill-dedup-view-arrays (25cc81b) to f755cb4 (merge-base) diff using: clickbench_partitioned
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-c4969050947-1048-jrc8t 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 spill-dedup-view-arrays (25cc81b) to f755cb4 (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-c4969050947-1049-jv4c6 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 spill-dedup-view-arrays (25cc81b) to f755cb4 (merge-base) diff using: tpch
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 spill-dedup-view-arrays
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃        spill-dedup-view-arrays ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │ 39.76 / 42.28 ±2.36 / 46.10 ms │ 38.31 / 39.36 ±1.13 / 41.55 ms │ +1.07x faster │
│ QQuery 2  │ 20.40 / 20.67 ±0.15 / 20.84 ms │ 19.38 / 19.75 ±0.21 / 19.97 ms │     no change │
│ QQuery 3  │ 32.53 / 33.62 ±1.17 / 35.85 ms │ 31.08 / 32.92 ±1.19 / 33.92 ms │     no change │
│ QQuery 4  │ 18.39 / 18.63 ±0.23 / 19.06 ms │ 17.77 / 18.19 ±0.69 / 19.57 ms │     no change │
│ QQuery 5  │ 41.46 / 43.24 ±1.42 / 45.74 ms │ 38.81 / 40.97 ±1.09 / 41.75 ms │ +1.06x faster │
│ QQuery 6  │ 16.97 / 18.09 ±1.77 / 21.61 ms │ 16.29 / 16.61 ±0.29 / 17.14 ms │ +1.09x faster │
│ QQuery 7  │ 50.62 / 51.41 ±0.78 / 52.72 ms │ 44.85 / 46.22 ±0.85 / 47.26 ms │ +1.11x faster │
│ QQuery 8  │ 46.31 / 46.75 ±0.27 / 47.05 ms │ 43.39 / 43.60 ±0.14 / 43.81 ms │ +1.07x faster │
│ QQuery 9  │ 53.80 / 56.20 ±1.64 / 58.08 ms │ 49.75 / 50.35 ±0.70 / 51.73 ms │ +1.12x faster │
│ QQuery 10 │ 46.12 / 47.10 ±1.29 / 49.53 ms │ 42.49 / 42.62 ±0.07 / 42.71 ms │ +1.11x faster │
│ QQuery 11 │ 14.76 / 15.23 ±0.65 / 16.51 ms │ 13.75 / 14.09 ±0.33 / 14.72 ms │ +1.08x faster │
│ QQuery 12 │ 25.58 / 27.38 ±2.28 / 31.51 ms │ 23.98 / 24.51 ±0.28 / 24.82 ms │ +1.12x faster │
│ QQuery 13 │ 34.22 / 35.84 ±1.47 / 38.41 ms │ 32.56 / 34.22 ±1.70 / 37.26 ms │     no change │
│ QQuery 14 │ 25.80 / 26.05 ±0.38 / 26.81 ms │ 23.89 / 24.05 ±0.12 / 24.22 ms │ +1.08x faster │
│ QQuery 15 │ 33.68 / 33.92 ±0.20 / 34.18 ms │ 31.49 / 31.64 ±0.22 / 32.08 ms │ +1.07x faster │
│ QQuery 16 │ 14.91 / 15.07 ±0.11 / 15.20 ms │ 13.92 / 14.27 ±0.20 / 14.54 ms │ +1.06x faster │
│ QQuery 17 │ 78.08 / 81.94 ±3.68 / 87.54 ms │ 73.34 / 74.14 ±0.46 / 74.62 ms │ +1.11x faster │
│ QQuery 18 │ 62.08 / 64.01 ±1.86 / 67.06 ms │ 60.27 / 61.21 ±0.60 / 62.16 ms │     no change │
│ QQuery 19 │ 34.21 / 34.48 ±0.19 / 34.70 ms │ 33.15 / 33.43 ±0.44 / 34.30 ms │     no change │
│ QQuery 20 │ 33.32 / 34.07 ±1.08 / 36.21 ms │ 32.50 / 32.78 ±0.24 / 33.16 ms │     no change │
│ QQuery 21 │ 56.18 / 58.90 ±1.57 / 61.01 ms │ 54.99 / 57.40 ±1.66 / 59.99 ms │     no change │
│ QQuery 22 │ 14.28 / 14.47 ±0.17 / 14.77 ms │ 14.15 / 14.29 ±0.08 / 14.39 ms │     no change │
└───────────┴────────────────────────────────┴────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                      ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                      │ 819.37ms │
│ Total Time (spill-dedup-view-arrays)   │ 766.60ms │
│ Average Time (HEAD)                    │  37.24ms │
│ Average Time (spill-dedup-view-arrays) │  34.85ms │
│ Queries Faster                         │       13 │
│ Queries Slower                         │        0 │
│ Queries with No Change                 │        9 │
│ Queries with Failure                   │        0 │
└────────────────────────────────────────┴──────────┘

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.2 GiB
Avg memory	672.2 MiB
CPU user	23.5s
CPU sys	1.9s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.2 GiB
Avg memory	531.0 MiB
CPU user	22.2s
CPU sys	1.7s
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 spill-dedup-view-arrays
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃               spill-dedup-view-arrays ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.47 / 5.98 ±0.88 / 7.75 ms │           5.60 / 6.13 ±0.85 / 7.82 ms │     no change │
│ QQuery 2  │        80.96 / 81.51 ±0.44 / 82.19 ms │        80.62 / 81.03 ±0.28 / 81.36 ms │     no change │
│ QQuery 3  │        29.29 / 29.53 ±0.15 / 29.70 ms │        29.14 / 29.43 ±0.16 / 29.62 ms │     no change │
│ QQuery 4  │     475.66 / 480.38 ±5.78 / 489.39 ms │     485.60 / 488.82 ±4.76 / 498.17 ms │     no change │
│ QQuery 5  │        51.65 / 52.10 ±0.33 / 52.60 ms │        52.18 / 52.59 ±0.34 / 53.12 ms │     no change │
│ QQuery 6  │        36.81 / 37.28 ±0.46 / 38.17 ms │        37.05 / 37.20 ±0.17 / 37.45 ms │     no change │
│ QQuery 7  │        94.14 / 94.63 ±0.35 / 95.18 ms │        94.21 / 94.79 ±0.38 / 95.29 ms │     no change │
│ QQuery 8  │        36.90 / 37.29 ±0.34 / 37.90 ms │        37.08 / 37.57 ±0.55 / 38.62 ms │     no change │
│ QQuery 9  │        52.74 / 54.35 ±1.43 / 56.29 ms │        54.73 / 55.23 ±0.42 / 56.00 ms │     no change │
│ QQuery 10 │        62.62 / 62.97 ±0.32 / 63.38 ms │        63.03 / 63.39 ±0.23 / 63.71 ms │     no change │
│ QQuery 11 │     293.29 / 295.79 ±1.52 / 297.29 ms │     292.55 / 295.59 ±1.75 / 297.54 ms │     no change │
│ QQuery 12 │        28.56 / 28.93 ±0.33 / 29.53 ms │        28.40 / 28.77 ±0.50 / 29.72 ms │     no change │
│ QQuery 13 │     118.39 / 119.23 ±0.59 / 120.01 ms │     119.05 / 120.06 ±0.79 / 120.94 ms │     no change │
│ QQuery 14 │     414.53 / 419.74 ±4.20 / 425.27 ms │     418.00 / 420.22 ±2.50 / 424.67 ms │     no change │
│ QQuery 15 │        57.30 / 58.17 ±0.61 / 59.10 ms │        57.35 / 58.04 ±0.60 / 59.17 ms │     no change │
│ QQuery 16 │           6.78 / 6.89 ±0.15 / 7.17 ms │           6.83 / 6.96 ±0.12 / 7.17 ms │     no change │
│ QQuery 17 │        80.64 / 81.68 ±1.31 / 84.21 ms │        81.48 / 82.23 ±0.93 / 84.03 ms │     no change │
│ QQuery 18 │     124.33 / 125.52 ±1.47 / 128.35 ms │     124.33 / 124.98 ±0.61 / 125.96 ms │     no change │
│ QQuery 19 │        41.28 / 41.51 ±0.16 / 41.71 ms │        41.74 / 42.10 ±0.35 / 42.62 ms │     no change │
│ QQuery 20 │        35.46 / 35.86 ±0.28 / 36.17 ms │        35.54 / 35.98 ±0.29 / 36.34 ms │     no change │
│ QQuery 21 │        17.41 / 17.72 ±0.17 / 17.89 ms │        17.52 / 17.79 ±0.23 / 18.07 ms │     no change │
│ QQuery 22 │        62.37 / 63.71 ±1.48 / 66.54 ms │        62.46 / 63.76 ±2.02 / 67.75 ms │     no change │
│ QQuery 23 │     347.33 / 348.96 ±1.26 / 351.07 ms │     343.44 / 347.81 ±3.50 / 354.13 ms │     no change │
│ QQuery 24 │     225.96 / 229.60 ±5.94 / 241.39 ms │     226.43 / 229.68 ±2.35 / 232.42 ms │     no change │
│ QQuery 25 │     111.08 / 112.04 ±0.89 / 113.73 ms │     111.19 / 112.31 ±1.03 / 113.88 ms │     no change │
│ QQuery 26 │        57.66 / 59.07 ±2.17 / 63.39 ms │        57.63 / 59.87 ±3.34 / 66.36 ms │     no change │
│ QQuery 27 │           6.28 / 6.41 ±0.15 / 6.70 ms │           6.28 / 6.46 ±0.21 / 6.86 ms │     no change │
│ QQuery 28 │        61.46 / 62.21 ±0.71 / 63.42 ms │        61.97 / 62.63 ±0.86 / 64.32 ms │     no change │
│ QQuery 29 │       97.33 / 98.19 ±1.08 / 100.17 ms │       97.77 / 99.77 ±2.18 / 103.61 ms │     no change │
│ QQuery 30 │        32.25 / 34.38 ±2.99 / 40.23 ms │        32.51 / 33.45 ±0.75 / 34.65 ms │     no change │
│ QQuery 31 │     110.69 / 112.33 ±1.72 / 115.50 ms │     112.24 / 113.11 ±0.48 / 113.67 ms │     no change │
│ QQuery 32 │        20.50 / 20.66 ±0.27 / 21.20 ms │        20.33 / 20.68 ±0.33 / 21.29 ms │     no change │
│ QQuery 33 │        37.91 / 38.31 ±0.37 / 38.95 ms │        37.95 / 40.06 ±3.84 / 47.73 ms │     no change │
│ QQuery 34 │         9.85 / 10.01 ±0.18 / 10.35 ms │         9.82 / 10.01 ±0.21 / 10.40 ms │     no change │
│ QQuery 35 │        73.03 / 74.88 ±1.99 / 78.68 ms │        73.21 / 73.97 ±0.40 / 74.34 ms │     no change │
│ QQuery 36 │           5.94 / 6.09 ±0.15 / 6.37 ms │           5.88 / 6.04 ±0.18 / 6.40 ms │     no change │
│ QQuery 37 │           6.92 / 7.05 ±0.06 / 7.11 ms │           6.98 / 7.11 ±0.12 / 7.32 ms │     no change │
│ QQuery 38 │        62.21 / 62.80 ±0.30 / 63.08 ms │        62.92 / 63.11 ±0.15 / 63.34 ms │     no change │
│ QQuery 39 │        90.53 / 90.91 ±0.39 / 91.58 ms │        91.18 / 92.82 ±1.83 / 96.35 ms │     no change │
│ QQuery 40 │        23.21 / 24.50 ±0.70 / 25.06 ms │        23.40 / 23.80 ±0.27 / 24.13 ms │     no change │
│ QQuery 41 │        11.57 / 11.71 ±0.17 / 12.04 ms │        11.57 / 11.77 ±0.20 / 12.15 ms │     no change │
│ QQuery 42 │        23.72 / 24.08 ±0.29 / 24.60 ms │        23.92 / 24.11 ±0.15 / 24.36 ms │     no change │
│ QQuery 43 │           5.04 / 5.11 ±0.09 / 5.28 ms │           5.14 / 5.25 ±0.13 / 5.50 ms │     no change │
│ QQuery 44 │           9.23 / 9.42 ±0.11 / 9.57 ms │           9.70 / 9.76 ±0.06 / 9.84 ms │     no change │
│ QQuery 45 │        38.87 / 39.17 ±0.19 / 39.42 ms │        38.53 / 38.96 ±0.25 / 39.25 ms │     no change │
│ QQuery 46 │        11.69 / 11.97 ±0.20 / 12.14 ms │        11.94 / 12.14 ±0.11 / 12.25 ms │     no change │
│ QQuery 47 │     227.08 / 229.72 ±1.95 / 232.66 ms │     225.44 / 227.67 ±2.41 / 232.34 ms │     no change │
│ QQuery 48 │       96.61 / 98.03 ±1.53 / 100.87 ms │        96.65 / 97.55 ±0.87 / 99.21 ms │     no change │
│ QQuery 49 │        76.30 / 77.23 ±1.29 / 79.78 ms │        77.04 / 77.97 ±0.60 / 78.65 ms │     no change │
│ QQuery 50 │        58.36 / 59.50 ±0.70 / 60.25 ms │        58.98 / 62.57 ±3.87 / 69.80 ms │  1.05x slower │
│ QQuery 51 │        90.98 / 92.89 ±2.20 / 96.62 ms │        92.43 / 94.43 ±1.23 / 96.17 ms │     no change │
│ QQuery 52 │        23.48 / 23.94 ±0.25 / 24.25 ms │        24.16 / 24.42 ±0.29 / 24.97 ms │     no change │
│ QQuery 53 │        29.31 / 29.64 ±0.28 / 30.13 ms │        30.33 / 30.59 ±0.30 / 31.17 ms │     no change │
│ QQuery 54 │        55.08 / 57.50 ±3.44 / 64.32 ms │        56.63 / 59.05 ±2.98 / 64.81 ms │     no change │
│ QQuery 55 │        23.14 / 23.53 ±0.57 / 24.66 ms │        24.27 / 24.79 ±0.31 / 25.16 ms │  1.05x slower │
│ QQuery 56 │        39.19 / 39.89 ±0.58 / 40.90 ms │        39.72 / 40.17 ±0.50 / 41.13 ms │     no change │
│ QQuery 57 │     176.55 / 180.83 ±3.85 / 188.04 ms │     178.98 / 181.17 ±2.32 / 185.42 ms │     no change │
│ QQuery 58 │     115.91 / 117.99 ±1.62 / 120.84 ms │     116.98 / 119.13 ±2.41 / 123.58 ms │     no change │
│ QQuery 59 │     119.38 / 120.32 ±1.13 / 121.98 ms │     119.29 / 120.52 ±0.70 / 121.36 ms │     no change │
│ QQuery 60 │        39.38 / 40.29 ±0.65 / 41.21 ms │        40.08 / 40.57 ±0.65 / 41.81 ms │     no change │
│ QQuery 61 │        12.86 / 13.00 ±0.20 / 13.39 ms │        12.91 / 13.09 ±0.14 / 13.33 ms │     no change │
│ QQuery 62 │        46.65 / 48.11 ±2.37 / 52.82 ms │        47.21 / 48.35 ±1.90 / 52.14 ms │     no change │
│ QQuery 63 │        30.18 / 30.79 ±0.61 / 31.73 ms │        30.53 / 31.74 ±1.52 / 34.74 ms │     no change │
│ QQuery 64 │     408.48 / 415.90 ±7.68 / 429.92 ms │     419.87 / 421.64 ±1.59 / 424.32 ms │     no change │
│ QQuery 65 │     145.49 / 149.23 ±2.92 / 153.88 ms │     146.68 / 149.17 ±1.54 / 150.58 ms │     no change │
│ QQuery 66 │        79.88 / 82.67 ±3.89 / 90.37 ms │        81.41 / 82.44 ±0.97 / 84.26 ms │     no change │
│ QQuery 67 │     236.24 / 241.32 ±4.47 / 247.68 ms │     239.08 / 242.90 ±2.52 / 246.03 ms │     no change │
│ QQuery 68 │        11.87 / 12.07 ±0.26 / 12.58 ms │        12.13 / 12.27 ±0.19 / 12.64 ms │     no change │
│ QQuery 69 │        57.04 / 57.63 ±0.54 / 58.56 ms │        58.74 / 60.35 ±2.82 / 65.99 ms │     no change │
│ QQuery 70 │     106.38 / 111.64 ±7.32 / 125.95 ms │     106.27 / 110.01 ±4.45 / 117.87 ms │     no change │
│ QQuery 71 │        35.71 / 36.03 ±0.26 / 36.49 ms │        36.06 / 36.31 ±0.16 / 36.47 ms │     no change │
│ QQuery 72 │ 2090.46 / 2210.07 ±80.08 / 2310.63 ms │ 2163.70 / 2219.16 ±48.84 / 2298.38 ms │     no change │
│ QQuery 73 │          9.55 / 9.82 ±0.25 / 10.25 ms │         9.66 / 12.36 ±3.92 / 19.98 ms │  1.26x slower │
│ QQuery 74 │     168.61 / 172.01 ±3.12 / 177.72 ms │     168.68 / 171.05 ±2.22 / 173.97 ms │     no change │
│ QQuery 75 │     149.63 / 152.55 ±4.16 / 160.80 ms │     149.87 / 151.10 ±0.85 / 152.34 ms │     no change │
│ QQuery 76 │        35.55 / 36.18 ±0.52 / 36.88 ms │        35.31 / 35.82 ±0.26 / 36.01 ms │     no change │
│ QQuery 77 │        61.76 / 62.13 ±0.23 / 62.40 ms │        61.58 / 62.02 ±0.45 / 62.87 ms │     no change │
│ QQuery 78 │     197.32 / 200.01 ±2.10 / 202.70 ms │     195.17 / 200.98 ±5.61 / 209.00 ms │     no change │
│ QQuery 79 │        67.50 / 72.05 ±6.50 / 84.64 ms │        67.75 / 69.85 ±3.53 / 76.88 ms │     no change │
│ QQuery 80 │      99.99 / 101.59 ±1.63 / 103.82 ms │      99.49 / 100.93 ±1.24 / 102.96 ms │     no change │
│ QQuery 81 │        25.63 / 25.91 ±0.17 / 26.15 ms │        25.96 / 28.24 ±3.72 / 35.66 ms │  1.09x slower │
│ QQuery 82 │        16.29 / 16.76 ±0.25 / 17.03 ms │        17.27 / 18.93 ±1.85 / 21.72 ms │  1.13x slower │
│ QQuery 83 │        40.66 / 44.96 ±4.96 / 54.29 ms │        40.60 / 41.08 ±0.65 / 42.34 ms │ +1.09x faster │
│ QQuery 84 │        30.85 / 31.18 ±0.21 / 31.45 ms │        30.88 / 31.07 ±0.10 / 31.15 ms │     no change │
│ QQuery 85 │     107.28 / 108.31 ±0.80 / 109.31 ms │     107.16 / 108.96 ±1.49 / 111.69 ms │     no change │
│ QQuery 86 │        25.24 / 25.45 ±0.27 / 25.90 ms │        24.94 / 26.25 ±1.00 / 27.93 ms │     no change │
│ QQuery 87 │        63.38 / 65.58 ±2.74 / 70.82 ms │        63.16 / 63.95 ±0.82 / 65.48 ms │     no change │
│ QQuery 88 │        62.82 / 63.50 ±0.55 / 64.15 ms │        63.80 / 64.33 ±0.48 / 65.03 ms │     no change │
│ QQuery 89 │        35.59 / 35.96 ±0.22 / 36.19 ms │        35.57 / 37.12 ±1.92 / 40.88 ms │     no change │
│ QQuery 90 │        17.20 / 17.37 ±0.16 / 17.67 ms │        17.37 / 17.55 ±0.20 / 17.89 ms │     no change │
│ QQuery 91 │        45.89 / 47.22 ±1.15 / 49.24 ms │        46.77 / 47.58 ±0.66 / 48.67 ms │     no change │
│ QQuery 92 │        29.03 / 29.74 ±0.75 / 31.16 ms │        29.29 / 29.90 ±0.72 / 31.27 ms │     no change │
│ QQuery 93 │        49.75 / 50.47 ±0.79 / 51.93 ms │        50.03 / 50.64 ±0.47 / 51.46 ms │     no change │
│ QQuery 94 │        38.22 / 38.91 ±0.50 / 39.67 ms │        38.28 / 38.70 ±0.34 / 39.21 ms │     no change │
│ QQuery 95 │        81.96 / 83.58 ±1.69 / 86.66 ms │        82.66 / 84.66 ±1.75 / 86.85 ms │     no change │
│ QQuery 96 │        24.39 / 24.71 ±0.29 / 25.14 ms │        24.47 / 24.63 ±0.15 / 24.89 ms │     no change │
│ QQuery 97 │        46.64 / 47.40 ±0.62 / 48.39 ms │        47.06 / 47.37 ±0.25 / 47.61 ms │     no change │
│ QQuery 98 │        42.10 / 42.91 ±0.44 / 43.35 ms │        42.63 / 43.44 ±0.50 / 43.92 ms │     no change │
│ QQuery 99 │        70.09 / 70.46 ±0.32 / 71.05 ms │        69.93 / 71.81 ±2.11 / 75.67 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                      ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                      │ 10007.10ms │
│ Total Time (spill-dedup-view-arrays)   │ 10055.70ms │
│ Average Time (HEAD)                    │   101.08ms │
│ Average Time (spill-dedup-view-arrays) │   101.57ms │
│ Queries Faster                         │          1 │
│ Queries Slower                         │          5 │
│ Queries with No Change                 │         93 │
│ Queries with Failure                   │          0 │
└────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	2.0 GiB
Avg memory	1.3 GiB
CPU user	228.6s
CPU sys	5.8s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	2.2 GiB
Avg memory	1.5 GiB
CPU user	222.2s
CPU sys	5.9s
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 spill-dedup-view-arrays
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃               spill-dedup-view-arrays ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.25 / 4.02 ±5.44 / 14.89 ms │          1.22 / 3.99 ±5.44 / 14.88 ms │     no change │
│ QQuery 1  │        12.70 / 12.84 ±0.09 / 12.95 ms │        12.40 / 12.77 ±0.27 / 13.23 ms │     no change │
│ QQuery 2  │        35.80 / 36.36 ±0.36 / 36.79 ms │        35.67 / 35.95 ±0.36 / 36.61 ms │     no change │
│ QQuery 3  │        30.82 / 31.49 ±0.68 / 32.75 ms │        30.59 / 30.80 ±0.19 / 31.15 ms │     no change │
│ QQuery 4  │     219.80 / 228.85 ±4.65 / 232.90 ms │     220.22 / 227.50 ±4.02 / 230.88 ms │     no change │
│ QQuery 5  │     272.53 / 275.50 ±2.40 / 279.69 ms │     268.71 / 274.43 ±4.67 / 282.29 ms │     no change │
│ QQuery 6  │           1.26 / 1.41 ±0.24 / 1.88 ms │           1.24 / 1.41 ±0.24 / 1.88 ms │     no change │
│ QQuery 7  │        13.53 / 13.67 ±0.16 / 13.91 ms │        13.40 / 13.61 ±0.15 / 13.78 ms │     no change │
│ QQuery 8  │     318.06 / 326.57 ±5.60 / 332.74 ms │     323.03 / 331.22 ±4.58 / 336.78 ms │     no change │
│ QQuery 9  │     457.93 / 463.57 ±3.65 / 468.97 ms │    455.97 / 468.33 ±10.04 / 484.29 ms │     no change │
│ QQuery 10 │        70.47 / 71.30 ±0.73 / 72.40 ms │        70.60 / 75.63 ±8.38 / 92.35 ms │  1.06x slower │
│ QQuery 11 │        81.92 / 82.76 ±0.50 / 83.44 ms │        80.64 / 82.29 ±2.61 / 87.47 ms │     no change │
│ QQuery 12 │     271.52 / 276.53 ±4.62 / 284.48 ms │     268.29 / 279.93 ±8.78 / 294.33 ms │     no change │
│ QQuery 13 │    378.37 / 393.95 ±14.04 / 418.05 ms │    375.01 / 383.14 ±10.56 / 403.19 ms │     no change │
│ QQuery 14 │     285.60 / 291.61 ±4.83 / 299.89 ms │     279.54 / 292.76 ±7.92 / 300.60 ms │     no change │
│ QQuery 15 │     273.96 / 280.09 ±3.87 / 284.58 ms │     274.98 / 285.61 ±7.05 / 292.92 ms │     no change │
│ QQuery 16 │    627.08 / 639.96 ±12.19 / 656.35 ms │    624.06 / 634.64 ±15.43 / 665.17 ms │     no change │
│ QQuery 17 │    628.06 / 637.35 ±12.65 / 661.47 ms │     627.76 / 636.06 ±5.87 / 643.34 ms │     no change │
│ QQuery 18 │ 1277.71 / 1297.36 ±16.85 / 1326.15 ms │ 1273.05 / 1309.98 ±25.78 / 1349.45 ms │     no change │
│ QQuery 19 │        28.87 / 34.42 ±9.45 / 53.27 ms │        27.47 / 27.74 ±0.21 / 28.02 ms │ +1.24x faster │
│ QQuery 20 │    515.66 / 527.27 ±11.49 / 543.78 ms │     519.96 / 525.18 ±5.83 / 536.00 ms │     no change │
│ QQuery 21 │     515.86 / 522.39 ±6.74 / 534.31 ms │     514.76 / 523.97 ±8.11 / 537.60 ms │     no change │
│ QQuery 22 │    993.42 / 999.55 ±6.77 / 1009.95 ms │   981.34 / 996.05 ±13.43 / 1014.97 ms │     no change │
│ QQuery 23 │ 3062.01 / 3098.23 ±38.93 / 3173.76 ms │ 3068.63 / 3120.26 ±37.76 / 3162.40 ms │     no change │
│ QQuery 24 │        41.41 / 42.73 ±1.37 / 45.20 ms │        40.81 / 42.03 ±0.89 / 43.45 ms │     no change │
│ QQuery 25 │     113.66 / 114.04 ±0.41 / 114.76 ms │     111.03 / 111.93 ±0.78 / 113.33 ms │     no change │
│ QQuery 26 │        41.52 / 42.20 ±0.47 / 43.00 ms │        41.58 / 43.34 ±2.05 / 46.74 ms │     no change │
│ QQuery 27 │    670.59 / 679.48 ±10.59 / 700.12 ms │     668.73 / 681.20 ±7.41 / 690.93 ms │     no change │
│ QQuery 28 │ 3019.28 / 3046.35 ±23.56 / 3084.36 ms │ 3060.72 / 3075.89 ±18.86 / 3112.02 ms │     no change │
│ QQuery 29 │        40.87 / 42.24 ±2.20 / 46.63 ms │       40.88 / 48.06 ±12.64 / 73.28 ms │  1.14x slower │
│ QQuery 30 │    307.57 / 318.44 ±18.14 / 354.52 ms │     299.62 / 307.09 ±7.17 / 319.99 ms │     no change │
│ QQuery 31 │     278.20 / 290.83 ±7.95 / 301.96 ms │    277.20 / 293.43 ±12.97 / 309.85 ms │     no change │
│ QQuery 32 │   923.31 / 957.06 ±30.85 / 1003.81 ms │    949.60 / 971.42 ±17.08 / 990.64 ms │     no change │
│ QQuery 33 │ 1477.83 / 1499.48 ±15.84 / 1515.48 ms │ 1434.93 / 1489.19 ±35.53 / 1530.22 ms │     no change │
│ QQuery 34 │ 1497.68 / 1529.92 ±17.91 / 1546.25 ms │ 1491.65 / 1528.99 ±25.64 / 1561.06 ms │     no change │
│ QQuery 35 │    278.59 / 306.51 ±29.30 / 358.79 ms │    283.22 / 335.27 ±74.30 / 482.55 ms │  1.09x slower │
│ QQuery 36 │        66.18 / 72.47 ±4.66 / 78.49 ms │       65.94 / 77.11 ±10.67 / 96.44 ms │  1.06x slower │
│ QQuery 37 │        35.48 / 37.67 ±2.20 / 41.91 ms │        35.49 / 37.25 ±1.92 / 40.79 ms │     no change │
│ QQuery 38 │        41.60 / 42.64 ±1.12 / 44.74 ms │       41.43 / 48.82 ±13.17 / 75.13 ms │  1.14x slower │
│ QQuery 39 │     143.52 / 151.58 ±8.57 / 167.87 ms │    138.54 / 151.49 ±10.32 / 163.38 ms │     no change │
│ QQuery 40 │        14.10 / 14.89 ±0.46 / 15.53 ms │        14.07 / 14.42 ±0.34 / 15.05 ms │     no change │
│ QQuery 41 │        13.71 / 13.99 ±0.32 / 14.58 ms │        13.55 / 14.05 ±0.31 / 14.52 ms │     no change │
│ QQuery 42 │        13.11 / 13.49 ±0.32 / 13.98 ms │        13.11 / 13.40 ±0.17 / 13.65 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                      ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                      │ 19763.05ms │
│ Total Time (spill-dedup-view-arrays)   │ 19857.62ms │
│ Average Time (HEAD)                    │   459.61ms │
│ Average Time (spill-dedup-view-arrays) │   461.81ms │
│ Queries Faster                         │          1 │
│ Queries Slower                         │          5 │
│ Queries with No Change                 │         37 │
│ Queries with Failure                   │          0 │
└────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	100.0s
Peak memory	11.4 GiB
Avg memory	4.6 GiB
CPU user	1018.4s
CPU sys	68.9s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	100.0s
Peak memory	10.3 GiB
Avg memory	4.1 GiB
CPU user	1014.1s
CPU sys	71.7s
Peak spill	0 B

---

File an issue against this benchmark runner

[cetra3]

Those benchmarks don't actually stress this path at all. In fact I don't think there are any standard benchmarks that will sweat this particular change. There is a sort_tpch benchmark but that has a high cardinality l_comment string view column, which, once again, doesn't hit this path or show any benefits here.