perf(aggregate): reuse dictionary hashes and group mappings by Rich-T-kid · Pull Request #26826 · apache/datafusion

Rich-T-kid · 2026-10-27T20:01:36Z

Which issue does this PR close?

Rationale for this change

Dictionary batches often share one values array while only their keys change. Partial aggregation was rehashing the same dictionary values for every batch.

What changes are included in this PR?

The non-streaming group table now caches dictionary value hashes while the values Arc is unchanged. A single dictionary group key can also map known keys directly to group indices.

Dictionary batch
       |
       v
single dictionary column + complete cached mapping?
       | yes                              | no
       v                                  v
keys -> cached group IDs           per-column hashing
                                   |                 |
                            eligible dictionary   fallback
                                   |                 |
                          cached value hashes   hash column
                                   |                 |
                          scatter through keys -----+
                                   |
                                   v
                         hash/equality group lookup

New keys, changed or oversized dictionaries, and partial emission use the normal fallback. Multi-column hashing keeps the same semantics as create_hashes. The benchmark fixtures are split into #27246.

What is the testing strategy for this PR?

Unit tests cover cached and uncached hash consistency, unseen keys, changed dictionaries, null groups, partial emission, and forced collisions.

Hosted cases from #27246 show alternating dictionary-only 1.30× faster; alternating dictionary + Int32 and fresh dictionaries flat; and shared dictionary + Int32 1.14× faster at 32K values and 1.28× at 100K.

Exact-head hosted suites with RLE dictionary reads enabled on both sides:

TPC-H: 796.01 → 799.79 ms (0.5% higher).
TPC-DS: 9.02 → 8.88 s (1.6% lower).
ClickBench: 38.26 → 38.16 s (0.3% lower).

Are there any user-facing changes?

No.

Rich-T-kid · 2026-10-27T20:02:17Z

run benchmark tpch tpcds clickbench_partitioned

env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

adriangbot · 2026-10-27T20:06:14Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6452254723-3959-9ccts 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 e1aa5f4 (e1aa5f4) to 4ede583 diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:11:27Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6452254723-3960-g5dwq 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 e1aa5f4 (e1aa5f4) to 4ede583 diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:14:22Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing e1aa5f4 (e1aa5f4) to 4ede583 diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃      HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │  40.39 ms │                                   40.23 ms │ no change │
│ QQuery 2  │  18.78 ms │                                   18.57 ms │ no change │
│ QQuery 3  │  28.27 ms │                                   28.01 ms │ no change │
│ QQuery 4  │  17.29 ms │                                   17.40 ms │ no change │
│ QQuery 5  │  35.77 ms │                                   35.65 ms │ no change │
│ QQuery 6  │  13.82 ms │                                   13.84 ms │ no change │
│ QQuery 7  │  41.66 ms │                                   40.89 ms │ no change │
│ QQuery 8  │  41.53 ms │                                   40.65 ms │ no change │
│ QQuery 9  │  51.68 ms │                                   50.92 ms │ no change │
│ QQuery 10 │  59.88 ms │                                   56.95 ms │ no change │
│ QQuery 11 │  13.74 ms │                                   13.35 ms │ no change │
│ QQuery 12 │  21.41 ms │                                   21.40 ms │ no change │
│ QQuery 13 │  47.29 ms │                                   46.19 ms │ no change │
│ QQuery 14 │  24.18 ms │                                   23.62 ms │ no change │
│ QQuery 15 │  31.37 ms │                                   30.64 ms │ no change │
│ QQuery 16 │  13.64 ms │                                   13.27 ms │ no change │
│ QQuery 17 │  75.07 ms │                                   73.86 ms │ no change │
│ QQuery 18 │ 113.72 ms │                                  113.07 ms │ no change │
│ QQuery 19 │  30.06 ms │                                   29.95 ms │ no change │
│ QQuery 20 │  33.74 ms │                                   33.35 ms │ no change │
│ QQuery 21 │  52.44 ms │                                   54.26 ms │ no change │
│ QQuery 22 │  15.69 ms │                                   15.64 ms │ no change │
└───────────┴───────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 821.43ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 811.70ms │
│ Average Time (HEAD)                                       │  37.34ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  36.90ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                              HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │    40.39 / 41.26 ±1.13 / 43.44 ms │             40.23 / 40.93 ±1.23 / 43.39 ms │    no change │
│ QQuery 2  │    18.78 / 19.30 ±0.69 / 20.55 ms │             18.57 / 18.80 ±0.22 / 19.23 ms │    no change │
│ QQuery 3  │    28.27 / 28.47 ±0.21 / 28.82 ms │             28.01 / 28.27 ±0.19 / 28.61 ms │    no change │
│ QQuery 4  │    17.29 / 17.64 ±0.54 / 18.70 ms │             17.40 / 17.65 ±0.43 / 18.50 ms │    no change │
│ QQuery 5  │    35.77 / 36.38 ±1.01 / 38.38 ms │             35.65 / 36.23 ±0.81 / 37.80 ms │    no change │
│ QQuery 6  │    13.82 / 13.92 ±0.11 / 14.13 ms │             13.84 / 13.90 ±0.04 / 13.95 ms │    no change │
│ QQuery 7  │    41.66 / 43.59 ±1.43 / 46.01 ms │             40.89 / 42.01 ±0.65 / 42.70 ms │    no change │
│ QQuery 8  │    41.53 / 42.08 ±0.55 / 42.89 ms │             40.65 / 42.42 ±1.50 / 44.15 ms │    no change │
│ QQuery 9  │    51.68 / 52.76 ±0.78 / 53.76 ms │             50.92 / 51.67 ±0.49 / 52.32 ms │    no change │
│ QQuery 10 │    59.88 / 60.07 ±0.22 / 60.36 ms │             56.95 / 57.29 ±0.27 / 57.57 ms │    no change │
│ QQuery 11 │    13.74 / 14.32 ±0.51 / 15.26 ms │             13.35 / 13.80 ±0.53 / 14.82 ms │    no change │
│ QQuery 12 │    21.41 / 21.79 ±0.40 / 22.52 ms │             21.40 / 21.66 ±0.19 / 21.93 ms │    no change │
│ QQuery 13 │    47.29 / 48.59 ±1.08 / 50.47 ms │             46.19 / 47.72 ±2.18 / 51.99 ms │    no change │
│ QQuery 14 │    24.18 / 24.93 ±1.18 / 27.27 ms │             23.62 / 23.79 ±0.18 / 24.13 ms │    no change │
│ QQuery 15 │    31.37 / 31.66 ±0.28 / 32.15 ms │             30.64 / 30.90 ±0.18 / 31.12 ms │    no change │
│ QQuery 16 │    13.64 / 13.70 ±0.04 / 13.76 ms │             13.27 / 13.50 ±0.11 / 13.57 ms │    no change │
│ QQuery 17 │    75.07 / 76.05 ±0.82 / 77.44 ms │             73.86 / 75.73 ±1.62 / 78.23 ms │    no change │
│ QQuery 18 │ 113.72 / 116.18 ±2.63 / 121.13 ms │          113.07 / 117.31 ±4.14 / 124.10 ms │    no change │
│ QQuery 19 │    30.06 / 31.52 ±2.27 / 36.03 ms │             29.95 / 30.43 ±0.70 / 31.81 ms │    no change │
│ QQuery 20 │    33.74 / 34.32 ±0.55 / 35.25 ms │             33.35 / 33.69 ±0.37 / 34.39 ms │    no change │
│ QQuery 21 │    52.44 / 54.35 ±2.35 / 58.81 ms │             54.26 / 57.53 ±4.60 / 66.46 ms │ 1.06x slower │
│ QQuery 22 │    15.69 / 15.85 ±0.20 / 16.24 ms │             15.64 / 16.00 ±0.28 / 16.47 ms │    no change │
└───────────┴───────────────────────────────────┴────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 838.74ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 831.23ms │
│ Average Time (HEAD)                                       │  38.12ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  37.78ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        1 │
│ Queries with No Change                                    │       21 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	40.1 MiB	30.0 MiB	-25.1%
Query 2	23.1 MiB	23.1 MiB	+0.0%
Query 3	11.5 MiB	11.5 MiB	+0.0%
Query 4	22.2 MiB	22.2 MiB	-0.0%
Query 5	32.8 MiB	42.8 MiB	+30.6%
Query 6	832 B	832 B	+0.0%
Query 7	107.6 MiB	107.6 MiB	+0.0%
Query 8	32.5 MiB	33.3 MiB	+2.2%
Query 9	155.6 MiB	155.6 MiB	+0.0%
Query 10	61.2 MiB	69.2 MiB	+13.1%
Query 11	61.7 MiB	41.5 MiB	-32.7%
Query 12	20.1 MiB	10.2 MiB	-49.2%
Query 13	52.3 MiB	62.3 MiB	+19.1%
Query 14	7.7 MiB	7.7 MiB	+0.0%
Query 15	12.9 MiB	12.8 MiB	-0.7%
Query 16	42.9 MiB	33.6 MiB	-21.8%
Query 17	150.3 MiB	150.3 MiB	+0.0%
Query 18	2.7 GiB	2.7 GiB	+0.0%
Query 19	228.6 KiB	228.6 KiB	+0.0%
Query 20	57.7 MiB	60.6 MiB	+5.1%
Query 21	37.7 MiB	37.7 MiB	+0.0%
Query 22	21.9 MiB	31.9 MiB	+45.6%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	2.7 GiB	1.4 GiB	−1.3 GiB	0.5×
tpch	changed (`e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc`)	2.7 GiB	2.0 GiB	−757.0 MiB	0.7×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.4 GiB
Avg memory	712.8 MiB
CPU user	21.7s
CPU sys	1.7s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	2.0 GiB
Avg memory	788.4 MiB
CPU user	21.6s
CPU sys	1.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:16:18Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6452254723-3961-mbgcz 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 e1aa5f4 (e1aa5f4) to 4ede583 diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:16:35Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing e1aa5f4 (e1aa5f4) to 4ede583 diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.26 ms │                                    5.69 ms │  1.08x slower │
│ QQuery 2  │   76.63 ms │                                   78.42 ms │     no change │
│ QQuery 3  │   28.62 ms │                                   29.69 ms │     no change │
│ QQuery 4  │  791.45 ms │                                  800.11 ms │     no change │
│ QQuery 5  │   49.97 ms │                                   51.77 ms │     no change │
│ QQuery 6  │   32.98 ms │                                   34.31 ms │     no change │
│ QQuery 7  │   72.12 ms │                                   73.85 ms │     no change │
│ QQuery 8  │   37.73 ms │                                   38.37 ms │     no change │
│ QQuery 9  │   49.91 ms │                                   52.17 ms │     no change │
│ QQuery 10 │   61.11 ms │                                   62.97 ms │     no change │
│ QQuery 11 │  447.20 ms │                                  460.44 ms │     no change │
│ QQuery 12 │   25.70 ms │                                   26.58 ms │     no change │
│ QQuery 13 │  111.38 ms │                                  112.83 ms │     no change │
│ QQuery 14 │  379.09 ms │                                  391.12 ms │     no change │
│ QQuery 15 │   50.99 ms │                                   52.48 ms │     no change │
│ QQuery 16 │    5.98 ms │                                    6.17 ms │     no change │
│ QQuery 17 │   79.17 ms │                                   79.88 ms │     no change │
│ QQuery 18 │  103.71 ms │                                  107.39 ms │     no change │
│ QQuery 19 │   40.80 ms │                                   41.38 ms │     no change │
│ QQuery 20 │   30.92 ms │                                   31.48 ms │     no change │
│ QQuery 21 │   16.32 ms │                                   17.24 ms │  1.06x slower │
│ QQuery 22 │   70.03 ms │                                   62.34 ms │ +1.12x faster │
│ QQuery 23 │  296.83 ms │                                  322.09 ms │  1.09x slower │
│ QQuery 24 │  187.78 ms │                                  201.68 ms │  1.07x slower │
│ QQuery 25 │  104.79 ms │                                  110.81 ms │  1.06x slower │
│ QQuery 26 │   46.56 ms │                                   47.48 ms │     no change │
│ QQuery 27 │    6.08 ms │                                    6.38 ms │  1.05x slower │
│ QQuery 28 │   60.63 ms │                                   61.75 ms │     no change │
│ QQuery 29 │   93.32 ms │                                   97.62 ms │     no change │
│ QQuery 30 │   36.06 ms │                                   38.71 ms │  1.07x slower │
│ QQuery 31 │  110.12 ms │                                  114.67 ms │     no change │
│ QQuery 32 │   19.42 ms │                                   21.23 ms │  1.09x slower │
│ QQuery 33 │   36.42 ms │                                   39.10 ms │  1.07x slower │
│ QQuery 34 │    9.49 ms │                                   10.26 ms │  1.08x slower │
│ QQuery 35 │   70.47 ms │                                   74.61 ms │  1.06x slower │
│ QQuery 36 │    5.69 ms │                                    6.35 ms │  1.12x slower │
│ QQuery 37 │    6.98 ms │                                    7.98 ms │  1.14x slower │
│ QQuery 38 │   62.94 ms │                                   63.06 ms │     no change │
│ QQuery 39 │   70.70 ms │                                   75.18 ms │  1.06x slower │
│ QQuery 40 │   22.01 ms │                                   24.29 ms │  1.10x slower │
│ QQuery 41 │   13.40 ms │                                   14.20 ms │  1.06x slower │
│ QQuery 42 │   23.77 ms │                                   24.39 ms │     no change │
│ QQuery 43 │    5.00 ms │                                    5.37 ms │  1.07x slower │
│ QQuery 44 │    8.37 ms │                                    8.55 ms │     no change │
│ QQuery 45 │   33.49 ms │                                   37.39 ms │  1.12x slower │
│ QQuery 46 │   11.00 ms │                                   11.85 ms │  1.08x slower │
│ QQuery 47 │  197.84 ms │                                  212.47 ms │  1.07x slower │
│ QQuery 48 │   90.19 ms │                                   92.68 ms │     no change │
│ QQuery 49 │   69.07 ms │                                   72.54 ms │  1.05x slower │
│ QQuery 50 │   57.15 ms │                                   59.57 ms │     no change │
│ QQuery 51 │   90.92 ms │                                   92.34 ms │     no change │
│ QQuery 52 │   23.44 ms │                                   24.50 ms │     no change │
│ QQuery 53 │   28.69 ms │                                   29.71 ms │     no change │
│ QQuery 54 │   51.80 ms │                                   54.21 ms │     no change │
│ QQuery 55 │   22.97 ms │                                   24.10 ms │     no change │
│ QQuery 56 │   38.31 ms │                                   38.59 ms │     no change │
│ QQuery 57 │  150.06 ms │                                  157.46 ms │     no change │
│ QQuery 58 │  102.28 ms │                                  104.54 ms │     no change │
│ QQuery 59 │  105.27 ms │                                  108.83 ms │     no change │
│ QQuery 60 │   42.09 ms │                                   42.65 ms │     no change │
│ QQuery 61 │    9.32 ms │                                    9.57 ms │     no change │
│ QQuery 62 │   37.48 ms │                                   38.40 ms │     no change │
│ QQuery 63 │   28.58 ms │                                   29.17 ms │     no change │
│ QQuery 64 │  358.64 ms │                                  375.02 ms │     no change │
│ QQuery 65 │  129.13 ms │                                  136.99 ms │  1.06x slower │
│ QQuery 66 │   73.17 ms │                                   77.25 ms │  1.06x slower │
│ QQuery 67 │  220.18 ms │                                  226.36 ms │     no change │
│ QQuery 68 │   11.04 ms │                                   12.07 ms │  1.09x slower │
│ QQuery 69 │   55.43 ms │                                   57.33 ms │     no change │
│ QQuery 70 │   97.51 ms │                                  102.45 ms │  1.05x slower │
│ QQuery 71 │   34.65 ms │                                   36.23 ms │     no change │
│ QQuery 72 │ 1386.27 ms │                                 1499.23 ms │  1.08x slower │
│ QQuery 73 │    9.13 ms │                                    9.83 ms │  1.08x slower │
│ QQuery 74 │  236.19 ms │                                  255.91 ms │  1.08x slower │
│ QQuery 75 │  136.92 ms │                                  142.73 ms │     no change │
│ QQuery 76 │   33.60 ms │                                   35.38 ms │  1.05x slower │
│ QQuery 77 │   58.91 ms │                                   62.15 ms │  1.05x slower │
│ QQuery 78 │  155.57 ms │                                  166.06 ms │  1.07x slower │
│ QQuery 79 │   66.68 ms │                                   68.92 ms │     no change │
│ QQuery 80 │   92.85 ms │                                   99.02 ms │  1.07x slower │
│ QQuery 81 │   27.49 ms │                                   30.40 ms │  1.11x slower │
│ QQuery 82 │   16.43 ms │                                   17.85 ms │  1.09x slower │
│ QQuery 83 │   31.23 ms │                                   34.35 ms │  1.10x slower │
│ QQuery 84 │   29.95 ms │                                   31.96 ms │  1.07x slower │
│ QQuery 85 │   97.68 ms │                                  103.13 ms │  1.06x slower │
│ QQuery 86 │   23.83 ms │                                   25.83 ms │  1.08x slower │
│ QQuery 87 │   63.76 ms │                                   67.76 ms │  1.06x slower │
│ QQuery 88 │   57.91 ms │                                   61.16 ms │  1.06x slower │
│ QQuery 89 │   34.72 ms │                                   37.02 ms │  1.07x slower │
│ QQuery 90 │   15.50 ms │                                   16.41 ms │  1.06x slower │
│ QQuery 91 │   39.72 ms │                                   40.56 ms │     no change │
│ QQuery 92 │   27.99 ms │                                   30.18 ms │  1.08x slower │
│ QQuery 93 │   48.33 ms │                                   49.78 ms │     no change │
│ QQuery 94 │   36.80 ms │                                   38.91 ms │  1.06x slower │
│ QQuery 95 │   77.30 ms │                                   82.42 ms │  1.07x slower │
│ QQuery 96 │   23.24 ms │                                   24.53 ms │  1.06x slower │
│ QQuery 97 │   49.41 ms │                                   52.92 ms │  1.07x slower │
│ QQuery 98 │   37.07 ms │                                   39.77 ms │  1.07x slower │
│ QQuery 99 │   55.50 ms │                                   58.92 ms │  1.06x slower │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9003.57ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9439.80ms │
│ Average Time (HEAD)                                       │   90.95ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   95.35ms │
│ Queries Faster                                            │         1 │
│ Queries Slower                                            │        49 │
│ Queries with No Change                                    │        49 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.26 / 5.97 ±1.12 / 8.19 ms │                5.69 / 6.41 ±1.07 / 8.52 ms │  1.07x slower │
│ QQuery 2  │        76.63 / 76.92 ±0.22 / 77.28 ms │             78.42 / 78.80 ±0.34 / 79.39 ms │     no change │
│ QQuery 3  │        28.62 / 28.84 ±0.22 / 29.25 ms │             29.69 / 29.87 ±0.20 / 30.26 ms │     no change │
│ QQuery 4  │     791.45 / 797.88 ±5.41 / 807.17 ms │         800.11 / 823.84 ±18.35 / 851.29 ms │     no change │
│ QQuery 5  │        49.97 / 50.28 ±0.28 / 50.75 ms │             51.77 / 52.55 ±0.55 / 53.11 ms │     no change │
│ QQuery 6  │        32.98 / 33.35 ±0.23 / 33.54 ms │             34.31 / 37.56 ±4.78 / 47.04 ms │  1.13x slower │
│ QQuery 7  │        72.12 / 74.80 ±2.09 / 78.22 ms │             73.85 / 75.49 ±0.87 / 76.43 ms │     no change │
│ QQuery 8  │        37.73 / 38.02 ±0.22 / 38.35 ms │             38.37 / 38.64 ±0.18 / 38.89 ms │     no change │
│ QQuery 9  │        49.91 / 51.52 ±1.82 / 54.63 ms │             52.17 / 54.05 ±1.18 / 55.83 ms │     no change │
│ QQuery 10 │        61.11 / 62.82 ±2.29 / 67.21 ms │             62.97 / 64.61 ±2.55 / 69.68 ms │     no change │
│ QQuery 11 │     447.20 / 455.52 ±5.14 / 461.54 ms │         460.44 / 477.78 ±12.44 / 498.64 ms │     no change │
│ QQuery 12 │        25.70 / 25.89 ±0.16 / 26.19 ms │             26.58 / 26.84 ±0.20 / 27.10 ms │     no change │
│ QQuery 13 │     111.38 / 112.51 ±1.64 / 115.77 ms │          112.83 / 114.13 ±0.95 / 115.58 ms │     no change │
│ QQuery 14 │     379.09 / 384.82 ±4.34 / 391.22 ms │          391.12 / 393.82 ±1.90 / 396.00 ms │     no change │
│ QQuery 15 │        50.99 / 52.64 ±1.80 / 55.13 ms │             52.48 / 53.07 ±0.71 / 54.31 ms │     no change │
│ QQuery 16 │           5.98 / 6.31 ±0.26 / 6.68 ms │                6.17 / 6.41 ±0.28 / 6.95 ms │     no change │
│ QQuery 17 │        79.17 / 80.29 ±1.23 / 82.54 ms │             79.88 / 80.72 ±0.77 / 81.72 ms │     no change │
│ QQuery 18 │     103.71 / 104.86 ±0.79 / 105.57 ms │          107.39 / 107.95 ±0.47 / 108.66 ms │     no change │
│ QQuery 19 │        40.80 / 41.18 ±0.41 / 41.97 ms │             41.38 / 41.70 ±0.34 / 42.32 ms │     no change │
│ QQuery 20 │        30.92 / 31.11 ±0.20 / 31.48 ms │             31.48 / 31.79 ±0.37 / 32.47 ms │     no change │
│ QQuery 21 │        16.32 / 16.60 ±0.28 / 17.11 ms │             17.24 / 17.58 ±0.32 / 18.07 ms │  1.06x slower │
│ QQuery 22 │        70.03 / 70.51 ±0.27 / 70.77 ms │             62.34 / 62.94 ±0.67 / 63.97 ms │ +1.12x faster │
│ QQuery 23 │     296.83 / 303.95 ±5.82 / 314.38 ms │          322.09 / 328.26 ±4.74 / 336.24 ms │  1.08x slower │
│ QQuery 24 │     187.78 / 189.61 ±1.63 / 192.47 ms │          201.68 / 208.83 ±4.90 / 214.86 ms │  1.10x slower │
│ QQuery 25 │     104.79 / 105.86 ±0.91 / 107.53 ms │          110.81 / 111.95 ±1.29 / 114.11 ms │  1.06x slower │
│ QQuery 26 │        46.56 / 47.73 ±1.44 / 50.34 ms │             47.48 / 47.77 ±0.26 / 48.19 ms │     no change │
│ QQuery 27 │           6.08 / 6.25 ±0.18 / 6.56 ms │                6.38 / 6.49 ±0.07 / 6.59 ms │     no change │
│ QQuery 28 │        60.63 / 61.17 ±0.43 / 61.88 ms │             61.75 / 62.25 ±0.30 / 62.64 ms │     no change │
│ QQuery 29 │        93.32 / 94.67 ±1.37 / 97.30 ms │           97.62 / 103.93 ±8.88 / 121.36 ms │  1.10x slower │
│ QQuery 30 │        36.06 / 37.27 ±1.46 / 39.92 ms │             38.71 / 39.08 ±0.41 / 39.77 ms │     no change │
│ QQuery 31 │     110.12 / 111.17 ±0.96 / 112.99 ms │          114.67 / 115.77 ±0.76 / 116.54 ms │     no change │
│ QQuery 32 │        19.42 / 19.72 ±0.31 / 20.20 ms │             21.23 / 22.91 ±2.67 / 28.23 ms │  1.16x slower │
│ QQuery 33 │        36.42 / 37.60 ±0.87 / 38.91 ms │             39.10 / 40.32 ±1.79 / 43.87 ms │  1.07x slower │
│ QQuery 34 │          9.49 / 9.64 ±0.19 / 10.00 ms │             10.26 / 10.39 ±0.11 / 10.54 ms │  1.08x slower │
│ QQuery 35 │        70.47 / 71.68 ±1.34 / 74.27 ms │             74.61 / 75.66 ±0.77 / 76.82 ms │  1.06x slower │
│ QQuery 36 │           5.69 / 5.86 ±0.25 / 6.35 ms │                6.35 / 6.57 ±0.14 / 6.77 ms │  1.12x slower │
│ QQuery 37 │           6.98 / 7.20 ±0.17 / 7.51 ms │                7.98 / 8.06 ±0.06 / 8.13 ms │  1.12x slower │
│ QQuery 38 │        62.94 / 63.40 ±0.27 / 63.72 ms │             63.06 / 66.50 ±2.04 / 69.49 ms │     no change │
│ QQuery 39 │        70.70 / 71.47 ±0.88 / 73.10 ms │             75.18 / 75.97 ±0.63 / 76.97 ms │  1.06x slower │
│ QQuery 40 │        22.01 / 22.74 ±0.78 / 24.25 ms │             24.29 / 25.00 ±0.68 / 26.28 ms │  1.10x slower │
│ QQuery 41 │        13.40 / 13.57 ±0.13 / 13.74 ms │             14.20 / 14.41 ±0.15 / 14.59 ms │  1.06x slower │
│ QQuery 42 │        23.77 / 24.09 ±0.48 / 25.04 ms │             24.39 / 24.82 ±0.48 / 25.76 ms │     no change │
│ QQuery 43 │           5.00 / 5.10 ±0.12 / 5.33 ms │                5.37 / 5.55 ±0.15 / 5.81 ms │  1.09x slower │
│ QQuery 44 │           8.37 / 8.47 ±0.12 / 8.71 ms │                8.55 / 8.90 ±0.27 / 9.27 ms │  1.05x slower │
│ QQuery 45 │        33.49 / 34.27 ±0.46 / 34.74 ms │             37.39 / 39.63 ±1.86 / 42.51 ms │  1.16x slower │
│ QQuery 46 │        11.00 / 11.16 ±0.11 / 11.31 ms │             11.85 / 12.06 ±0.22 / 12.34 ms │  1.08x slower │
│ QQuery 47 │     197.84 / 198.70 ±0.65 / 199.72 ms │          212.47 / 219.61 ±4.10 / 224.76 ms │  1.11x slower │
│ QQuery 48 │        90.19 / 91.44 ±1.22 / 93.49 ms │             92.68 / 94.05 ±1.08 / 95.65 ms │     no change │
│ QQuery 49 │        69.07 / 69.77 ±0.98 / 71.72 ms │             72.54 / 76.47 ±4.33 / 84.38 ms │  1.10x slower │
│ QQuery 50 │        57.15 / 57.51 ±0.43 / 58.33 ms │             59.57 / 60.52 ±0.52 / 61.00 ms │  1.05x slower │
│ QQuery 51 │        90.92 / 92.06 ±1.02 / 93.63 ms │             92.34 / 95.05 ±1.86 / 97.85 ms │     no change │
│ QQuery 52 │        23.44 / 23.87 ±0.31 / 24.34 ms │             24.50 / 24.87 ±0.26 / 25.20 ms │     no change │
│ QQuery 53 │        28.69 / 29.05 ±0.28 / 29.54 ms │             29.71 / 31.61 ±2.66 / 36.87 ms │  1.09x slower │
│ QQuery 54 │        51.80 / 52.34 ±0.29 / 52.64 ms │             54.21 / 54.68 ±0.50 / 55.61 ms │     no change │
│ QQuery 55 │        22.97 / 23.18 ±0.21 / 23.57 ms │             24.10 / 25.07 ±1.31 / 27.66 ms │  1.08x slower │
│ QQuery 56 │        38.31 / 40.67 ±3.20 / 46.89 ms │             38.59 / 39.53 ±0.76 / 40.90 ms │     no change │
│ QQuery 57 │     150.06 / 152.03 ±1.26 / 153.48 ms │          157.46 / 160.59 ±2.61 / 165.35 ms │  1.06x slower │
│ QQuery 58 │     102.28 / 104.32 ±3.05 / 110.34 ms │          104.54 / 106.71 ±1.82 / 109.89 ms │     no change │
│ QQuery 59 │     105.27 / 106.19 ±0.55 / 106.87 ms │          108.83 / 109.54 ±0.55 / 110.53 ms │     no change │
│ QQuery 60 │        42.09 / 43.40 ±0.89 / 44.36 ms │             42.65 / 43.33 ±0.45 / 43.87 ms │     no change │
│ QQuery 61 │          9.32 / 9.54 ±0.26 / 10.05 ms │               9.57 / 9.87 ±0.26 / 10.33 ms │     no change │
│ QQuery 62 │        37.48 / 38.24 ±0.61 / 39.13 ms │             38.40 / 39.76 ±1.84 / 43.40 ms │     no change │
│ QQuery 63 │        28.58 / 28.94 ±0.35 / 29.58 ms │             29.17 / 29.74 ±0.39 / 30.20 ms │     no change │
│ QQuery 64 │     358.64 / 362.39 ±2.03 / 364.26 ms │          375.02 / 388.55 ±8.01 / 399.97 ms │  1.07x slower │
│ QQuery 65 │     129.13 / 130.88 ±1.67 / 133.99 ms │          136.99 / 138.85 ±1.04 / 139.84 ms │  1.06x slower │
│ QQuery 66 │        73.17 / 75.17 ±3.32 / 81.75 ms │             77.25 / 78.60 ±0.92 / 79.78 ms │     no change │
│ QQuery 67 │     220.18 / 222.80 ±1.97 / 225.81 ms │          226.36 / 239.44 ±9.34 / 251.49 ms │  1.07x slower │
│ QQuery 68 │        11.04 / 11.26 ±0.23 / 11.70 ms │             12.07 / 12.17 ±0.10 / 12.34 ms │  1.08x slower │
│ QQuery 69 │        55.43 / 55.85 ±0.40 / 56.60 ms │             57.33 / 58.06 ±0.67 / 59.20 ms │     no change │
│ QQuery 70 │      97.51 / 100.32 ±2.92 / 105.74 ms │          102.45 / 102.85 ±0.61 / 104.06 ms │     no change │
│ QQuery 71 │        34.65 / 35.03 ±0.42 / 35.59 ms │             36.23 / 37.88 ±2.70 / 43.25 ms │  1.08x slower │
│ QQuery 72 │ 1386.27 / 1488.82 ±94.10 / 1658.55 ms │      1499.23 / 1542.41 ±26.50 / 1578.78 ms │     no change │
│ QQuery 73 │           9.13 / 9.35 ±0.19 / 9.68 ms │              9.83 / 10.10 ±0.15 / 10.28 ms │  1.08x slower │
│ QQuery 74 │     236.19 / 241.01 ±3.37 / 246.37 ms │          255.91 / 259.96 ±2.28 / 262.93 ms │  1.08x slower │
│ QQuery 75 │     136.92 / 137.57 ±0.53 / 138.34 ms │          142.73 / 143.83 ±0.75 / 145.08 ms │     no change │
│ QQuery 76 │        33.60 / 33.88 ±0.21 / 34.23 ms │             35.38 / 36.12 ±1.14 / 38.38 ms │  1.07x slower │
│ QQuery 77 │        58.91 / 59.37 ±0.38 / 59.90 ms │             62.15 / 62.86 ±0.77 / 64.28 ms │  1.06x slower │
│ QQuery 78 │     155.57 / 161.49 ±3.97 / 167.48 ms │          166.06 / 169.86 ±4.43 / 176.42 ms │  1.05x slower │
│ QQuery 79 │        66.68 / 67.19 ±0.58 / 68.26 ms │             68.92 / 70.66 ±1.95 / 74.42 ms │  1.05x slower │
│ QQuery 80 │        92.85 / 93.69 ±0.80 / 95.20 ms │           99.02 / 102.43 ±4.71 / 111.74 ms │  1.09x slower │
│ QQuery 81 │        27.49 / 28.23 ±0.66 / 29.40 ms │             30.40 / 30.74 ±0.36 / 31.41 ms │  1.09x slower │
│ QQuery 82 │        16.43 / 19.24 ±4.15 / 27.46 ms │             17.85 / 18.27 ±0.28 / 18.72 ms │ +1.05x faster │
│ QQuery 83 │        31.23 / 31.69 ±0.35 / 32.19 ms │             34.35 / 35.51 ±1.72 / 38.91 ms │  1.12x slower │
│ QQuery 84 │        29.95 / 30.50 ±0.40 / 31.05 ms │             31.96 / 32.29 ±0.25 / 32.65 ms │  1.06x slower │
│ QQuery 85 │        97.68 / 98.22 ±0.44 / 98.78 ms │          103.13 / 105.19 ±2.21 / 109.36 ms │  1.07x slower │
│ QQuery 86 │        23.83 / 24.46 ±0.59 / 25.55 ms │             25.83 / 26.53 ±0.49 / 27.32 ms │  1.08x slower │
│ QQuery 87 │        63.76 / 65.69 ±2.68 / 70.99 ms │             67.76 / 68.52 ±0.68 / 69.36 ms │     no change │
│ QQuery 88 │        57.91 / 58.52 ±0.42 / 59.17 ms │             61.16 / 62.89 ±2.13 / 66.99 ms │  1.07x slower │
│ QQuery 89 │        34.72 / 35.20 ±0.48 / 36.11 ms │             37.02 / 37.33 ±0.16 / 37.47 ms │  1.06x slower │
│ QQuery 90 │        15.50 / 15.84 ±0.23 / 16.13 ms │             16.41 / 17.27 ±1.24 / 19.75 ms │  1.09x slower │
│ QQuery 91 │        39.72 / 40.24 ±0.55 / 41.25 ms │             40.56 / 41.38 ±0.66 / 42.51 ms │     no change │
│ QQuery 92 │        27.99 / 28.51 ±0.38 / 29.14 ms │             30.18 / 30.91 ±0.49 / 31.47 ms │  1.08x slower │
│ QQuery 93 │        48.33 / 49.31 ±0.78 / 50.38 ms │             49.78 / 51.41 ±1.58 / 53.97 ms │     no change │
│ QQuery 94 │        36.80 / 37.31 ±0.40 / 38.00 ms │             38.91 / 39.45 ±0.36 / 40.00 ms │  1.06x slower │
│ QQuery 95 │        77.30 / 77.90 ±0.53 / 78.64 ms │             82.42 / 83.41 ±0.53 / 83.94 ms │  1.07x slower │
│ QQuery 96 │        23.24 / 24.19 ±1.75 / 27.68 ms │             24.53 / 24.81 ±0.27 / 25.25 ms │     no change │
│ QQuery 97 │        49.41 / 50.42 ±1.08 / 52.49 ms │             52.92 / 54.26 ±1.21 / 55.79 ms │  1.08x slower │
│ QQuery 98 │        37.07 / 37.61 ±0.43 / 38.23 ms │             39.77 / 40.37 ±0.53 / 41.34 ms │  1.07x slower │
│ QQuery 99 │        55.50 / 56.05 ±0.41 / 56.66 ms │             58.92 / 60.52 ±1.09 / 62.21 ms │  1.08x slower │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9222.75ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9674.31ms │
│ Average Time (HEAD)                                       │   93.16ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   97.72ms │
│ Queries Faster                                            │         2 │
│ Queries Slower                                            │        52 │
│ Queries with No Change                                    │        45 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc

tpcds — tpcds_sf1

Query	Base	Changed	Change
Query 1	198.0 KiB	137.1 KiB	-30.8%
Query 2	82.9 MiB	89.1 MiB	+7.4%
Query 3	2.1 MiB	1.9 MiB	-8.6%
Query 4	79.7 MiB	79.7 MiB	-0.0%
Query 5	12.5 MiB	12.5 MiB	+0.1%
Query 6	4.4 MiB	4.7 MiB	+6.9%
Query 7	3.6 MiB	3.7 MiB	+1.4%
Query 8	1.7 MiB	1.9 MiB	+11.5%
Query 9	32.1 KiB	32.1 KiB	+0.0%
Query 10	2.0 MiB	2.0 MiB	+0.0%
Query 11	62.0 MiB	62.0 MiB	+0.1%
Query 12	114.4 MiB	114.4 MiB	-0.0%
Query 13	13.0 MiB	13.7 MiB	+5.1%
Query 14	170.9 MiB	171.7 MiB	+0.4%
Query 15	4.2 MiB	4.0 MiB	-6.0%
Query 16	118.6 KiB	118.6 KiB	+0.0%
Query 17	40.5 MiB	40.6 MiB	+0.1%
Query 18	21.2 MiB	21.3 MiB	+0.2%
Query 19	2.1 MiB	2.1 MiB	+1.2%
Query 20	114.6 MiB	114.6 MiB	+0.0%
Query 21	2.5 MiB	3.6 MiB	+43.3%
Query 22	61.8 MiB	63.5 MiB	+2.8%
Query 23	212.5 MiB	213.2 MiB	+0.3%
Query 24	38.7 MiB	38.7 MiB	+0.0%
Query 25	99.9 MiB	99.9 MiB	+0.0%
Query 26	3.4 MiB	3.6 MiB	+4.9%
Query 27	1.6 MiB	1.6 MiB	+0.0%
Query 28	1.3 MiB	1.3 MiB	+0.0%
Query 29	102.9 MiB	104.7 MiB	+1.8%
Query 30	3.9 MiB	3.7 MiB	-5.1%
Query 31	13.4 MiB	13.4 MiB	+0.0%
Query 32	2.7 MiB	2.5 MiB	-8.1%
Query 33	4.2 MiB	4.5 MiB	+7.2%
Query 34	6.0 MiB	6.0 MiB	+0.0%
Query 35	16.6 MiB	16.6 MiB	+0.0%
Query 36	10.3 MiB	10.3 MiB	+0.0%
Query 37	3.3 MiB	3.3 MiB	+0.0%
Query 38	10.8 MiB	10.8 MiB	+0.4%
Query 39	42.1 MiB	53.0 MiB	+25.8%
Query 40	12.2 MiB	12.2 MiB	+0.0%
Query 41	2.1 MiB	2.1 MiB	+0.1%
Query 42	1.3 MiB	1.2 MiB	-2.0%
Query 43	275.0 KiB	275.4 KiB	+0.1%
Query 44	233.2 KiB	233.2 KiB	+0.0%
Query 45	4.5 MiB	4.9 MiB	+9.0%
Query 46	6.2 MiB	6.2 MiB	+0.0%
Query 47	128.5 MiB	132.6 MiB	+3.1%
Query 48	11.8 MiB	11.4 MiB	-3.2%
Query 49	51.6 MiB	52.8 MiB	+2.4%
Query 50	16.9 MiB	17.7 MiB	+4.4%
Query 51	177.7 MiB	178.0 MiB	+0.2%
Query 52	1.5 MiB	1.5 MiB	+0.1%
Query 53	30.4 MiB	30.4 MiB	+0.0%
Query 54	2.2 MiB	2.2 MiB	+0.0%
Query 55	1.5 MiB	1.5 MiB	+0.8%
Query 56	3.0 MiB	3.0 MiB	-0.3%
Query 57	120.8 MiB	120.8 MiB	-0.0%
Query 58	4.4 MiB	4.2 MiB	-4.8%
Query 59	11.7 MiB	11.7 MiB	+0.0%
Query 60	7.2 MiB	7.5 MiB	+4.6%
Query 61	2.4 KiB	2.4 KiB	+0.7%
Query 62	3.6 MiB	3.3 MiB	-8.9%
Query 63	30.4 MiB	20.6 MiB	-32.4%
Query 64	83.7 MiB	93.7 MiB	+11.9%
Query 65	29.7 MiB	29.7 MiB	+0.0%
Query 66	14.9 MiB	14.9 MiB	+0.0%
Query 67	137.9 MiB	138.1 MiB	+0.2%
Query 68	6.2 MiB	6.2 MiB	+0.0%
Query 69	3.7 MiB	3.5 MiB	-4.8%
Query 70	44.0 MiB	44.0 MiB	+0.0%
Query 71	30.4 MiB	40.4 MiB	+32.9%
Query 72	97.2 MiB	97.2 MiB	+0.0%
Query 73	6.0 MiB	6.0 MiB	+0.0%
Query 74	34.6 MiB	34.6 MiB	+0.0%
Query 75	20.3 MiB	21.3 MiB	+5.2%
Query 76	7.9 MiB	7.9 MiB	+0.0%
Query 77	4.1 MiB	4.1 MiB	+0.0%
Query 78	108.2 MiB	111.5 MiB	+3.1%
Query 79	8.7 MiB	8.7 MiB	+0.0%
Query 80	41.2 MiB	42.0 MiB	+2.0%
Query 81	4.1 MiB	4.1 MiB	+0.5%
Query 82	4.9 MiB	4.9 MiB	+0.0%
Query 83	9.0 MiB	9.0 MiB	+0.0%
Query 84	11.9 MiB	11.9 MiB	+0.0%
Query 85	17.0 MiB	17.0 MiB	+0.0%
Query 86	82.1 MiB	82.1 MiB	+0.0%
Query 87	12.9 MiB	13.1 MiB	+1.7%
Query 88	1.7 MiB	1.6 MiB	-9.9%
Query 89	112.2 MiB	112.2 MiB	-0.0%
Query 90	351.7 KiB	345.5 KiB	-1.8%
Query 91	31.0 MiB	21.0 MiB	-32.2%
Query 92	2.7 MiB	2.7 MiB	+0.0%
Query 93	15.1 MiB	15.2 MiB	+0.3%
Query 94	3.7 MiB	3.4 MiB	-7.6%
Query 95	17.4 MiB	18.4 MiB	+5.4%
Query 96	224.7 KiB	224.7 KiB	+0.0%
Query 97	33.1 MiB	33.1 MiB	+0.0%
Query 98	114.7 MiB	114.7 MiB	-0.0%
Query 99	1.7 MiB	1.6 MiB	-1.3%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpcds	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	212.5 MiB	2.0 GiB	1.8 GiB	9.5×
tpcds	changed (`e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc`)	213.2 MiB	1.9 GiB	1.7 GiB	9.3×

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	50.0s
Peak memory	2.0 GiB
Avg memory	1.4 GiB
CPU user	181.8s
CPU sys	5.3s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	50.0s
Peak memory	1.9 GiB
Avg memory	1.4 GiB
CPU user	192.2s
CPU sys	5.9s
Peak spill	0 B

File an issue against this benchmark runner

Rich-T-kid · 2026-10-27T20:22:36Z

run benchmark tpch tpcds clickbench_partitioned

env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

adriangbot · 2026-10-27T20:24:31Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6452661322-3965-np5lm 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:25:50Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6452661322-3966-4g845 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:26:30Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6452661322-3967-q7l5b 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:31:14Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃      HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │  40.81 ms │                                   40.80 ms │ no change │
│ QQuery 2  │  19.06 ms │                                   19.00 ms │ no change │
│ QQuery 3  │  28.95 ms │                                   29.00 ms │ no change │
│ QQuery 4  │  17.65 ms │                                   17.70 ms │ no change │
│ QQuery 5  │  36.45 ms │                                   36.60 ms │ no change │
│ QQuery 6  │  13.84 ms │                                   14.07 ms │ no change │
│ QQuery 7  │  43.57 ms │                                   43.36 ms │ no change │
│ QQuery 8  │  41.58 ms │                                   41.96 ms │ no change │
│ QQuery 9  │  52.20 ms │                                   52.33 ms │ no change │
│ QQuery 10 │  60.67 ms │                                   59.98 ms │ no change │
│ QQuery 11 │  14.11 ms │                                   14.37 ms │ no change │
│ QQuery 12 │  21.71 ms │                                   21.85 ms │ no change │
│ QQuery 13 │  48.69 ms │                                   48.92 ms │ no change │
│ QQuery 14 │  24.55 ms │                                   24.56 ms │ no change │
│ QQuery 15 │  31.67 ms │                                   31.68 ms │ no change │
│ QQuery 16 │  13.65 ms │                                   13.83 ms │ no change │
│ QQuery 17 │  78.37 ms │                                   78.64 ms │ no change │
│ QQuery 18 │ 119.18 ms │                                  118.26 ms │ no change │
│ QQuery 19 │  30.69 ms │                                   30.96 ms │ no change │
│ QQuery 20 │  34.45 ms │                                   35.10 ms │ no change │
│ QQuery 21 │  54.38 ms │                                   56.48 ms │ no change │
│ QQuery 22 │  16.28 ms │                                   16.23 ms │ no change │
└───────────┴───────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 842.52ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 845.66ms │
│ Average Time (HEAD)                                       │  38.30ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  38.44ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                              HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │    40.81 / 42.21 ±1.73 / 45.06 ms │             40.80 / 41.99 ±1.14 / 43.69 ms │    no change │
│ QQuery 2  │    19.06 / 19.86 ±0.59 / 20.66 ms │             19.00 / 19.82 ±1.09 / 21.95 ms │    no change │
│ QQuery 3  │    28.95 / 29.72 ±1.12 / 31.87 ms │             29.00 / 29.75 ±1.08 / 31.86 ms │    no change │
│ QQuery 4  │    17.65 / 18.03 ±0.65 / 19.32 ms │             17.70 / 17.80 ±0.08 / 17.94 ms │    no change │
│ QQuery 5  │    36.45 / 37.49 ±1.00 / 38.88 ms │             36.60 / 36.88 ±0.22 / 37.24 ms │    no change │
│ QQuery 6  │    13.84 / 14.34 ±0.76 / 15.86 ms │             14.07 / 14.58 ±0.84 / 16.24 ms │    no change │
│ QQuery 7  │    43.57 / 45.13 ±1.13 / 46.69 ms │             43.36 / 46.94 ±3.12 / 52.31 ms │    no change │
│ QQuery 8  │    41.58 / 41.96 ±0.37 / 42.58 ms │             41.96 / 42.88 ±1.14 / 45.13 ms │    no change │
│ QQuery 9  │    52.20 / 53.31 ±1.23 / 55.56 ms │             52.33 / 53.66 ±1.10 / 55.58 ms │    no change │
│ QQuery 10 │    60.67 / 61.18 ±0.45 / 61.80 ms │             59.98 / 60.77 ±0.95 / 62.53 ms │    no change │
│ QQuery 11 │    14.11 / 14.40 ±0.33 / 15.05 ms │             14.37 / 14.43 ±0.06 / 14.54 ms │    no change │
│ QQuery 12 │    21.71 / 23.31 ±1.67 / 26.32 ms │             21.85 / 22.46 ±0.68 / 23.72 ms │    no change │
│ QQuery 13 │    48.69 / 49.58 ±0.96 / 51.40 ms │             48.92 / 49.29 ±0.37 / 49.99 ms │    no change │
│ QQuery 14 │    24.55 / 24.62 ±0.06 / 24.69 ms │             24.56 / 24.66 ±0.06 / 24.71 ms │    no change │
│ QQuery 15 │    31.67 / 31.93 ±0.36 / 32.62 ms │             31.68 / 32.20 ±0.36 / 32.79 ms │    no change │
│ QQuery 16 │    13.65 / 13.90 ±0.19 / 14.23 ms │             13.83 / 13.95 ±0.11 / 14.16 ms │    no change │
│ QQuery 17 │    78.37 / 80.11 ±2.47 / 85.00 ms │             78.64 / 80.10 ±1.15 / 81.73 ms │    no change │
│ QQuery 18 │ 119.18 / 123.30 ±4.29 / 130.91 ms │          118.26 / 121.04 ±3.02 / 126.35 ms │    no change │
│ QQuery 19 │    30.69 / 30.97 ±0.23 / 31.23 ms │             30.96 / 33.37 ±3.35 / 39.75 ms │ 1.08x slower │
│ QQuery 20 │    34.45 / 34.97 ±0.51 / 35.84 ms │             35.10 / 35.47 ±0.38 / 35.96 ms │    no change │
│ QQuery 21 │    54.38 / 57.69 ±3.26 / 63.84 ms │             56.48 / 58.76 ±3.23 / 65.12 ms │    no change │
│ QQuery 22 │    16.28 / 16.39 ±0.06 / 16.47 ms │             16.23 / 16.43 ±0.17 / 16.70 ms │    no change │
└───────────┴───────────────────────────────────┴────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 864.38ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 867.23ms │
│ Average Time (HEAD)                                       │  39.29ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  39.42ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        1 │
│ Queries with No Change                                    │       21 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: d49dd6d5614860d21328eb63633fd9d8e08fbd37

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	30.1 MiB	30.2 MiB	+0.2%
Query 2	23.1 MiB	23.1 MiB	+0.0%
Query 3	11.5 MiB	11.5 MiB	+0.0%
Query 4	22.3 MiB	22.2 MiB	-0.2%
Query 5	42.7 MiB	32.9 MiB	-23.1%
Query 6	832 B	832 B	+0.0%
Query 7	107.6 MiB	107.6 MiB	+0.0%
Query 8	34.6 MiB	34.7 MiB	+0.2%
Query 9	155.6 MiB	155.6 MiB	+0.0%
Query 10	61.2 MiB	66.3 MiB	+8.3%
Query 11	61.7 MiB	72.1 MiB	+16.8%
Query 12	20.1 MiB	20.1 MiB	+0.0%
Query 13	52.5 MiB	52.6 MiB	+0.2%
Query 14	7.7 MiB	7.7 MiB	+0.0%
Query 15	12.8 MiB	12.8 MiB	-0.0%
Query 16	33.3 MiB	43.5 MiB	+30.8%
Query 17	150.3 MiB	150.3 MiB	+0.0%
Query 18	2.7 GiB	2.7 GiB	-0.0%
Query 19	228.6 KiB	228.6 KiB	+0.0%
Query 20	61.8 MiB	61.0 MiB	-1.3%
Query 21	37.7 MiB	37.7 MiB	+0.0%
Query 22	31.8 MiB	32.1 MiB	+1.2%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	2.7 GiB	2.0 GiB	−696.5 MiB	0.7×
tpch	changed (`d49dd6d5614860d21328eb63633fd9d8e08fbd37`)	2.7 GiB	1.4 GiB	−1.3 GiB	0.5×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	2.0 GiB
Avg memory	803.8 MiB
CPU user	21.8s
CPU sys	1.7s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.4 GiB
Avg memory	694.2 MiB
CPU user	21.7s
CPU sys	1.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:33:15Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.26 ms │                                    5.16 ms │     no change │
│ QQuery 2  │   76.38 ms │                                   76.81 ms │     no change │
│ QQuery 3  │   29.07 ms │                                   28.64 ms │     no change │
│ QQuery 4  │  848.25 ms │                                  730.51 ms │ +1.16x faster │
│ QQuery 5  │   49.81 ms │                                   50.70 ms │     no change │
│ QQuery 6  │   33.48 ms │                                   32.91 ms │     no change │
│ QQuery 7  │   71.47 ms │                                   71.77 ms │     no change │
│ QQuery 8  │   38.98 ms │                                   37.49 ms │     no change │
│ QQuery 9  │   51.69 ms │                                   51.63 ms │     no change │
│ QQuery 10 │   61.19 ms │                                   61.23 ms │     no change │
│ QQuery 11 │  474.56 ms │                                  412.40 ms │ +1.15x faster │
│ QQuery 12 │   25.98 ms │                                   25.66 ms │     no change │
│ QQuery 13 │  111.89 ms │                                  112.01 ms │     no change │
│ QQuery 14 │  379.70 ms │                                  381.89 ms │     no change │
│ QQuery 15 │   52.61 ms │                                   50.92 ms │     no change │
│ QQuery 16 │    6.55 ms │                                    5.98 ms │ +1.10x faster │
│ QQuery 17 │   78.89 ms │                                   78.21 ms │     no change │
│ QQuery 18 │  103.45 ms │                                  107.78 ms │     no change │
│ QQuery 19 │   42.06 ms │                                   41.13 ms │     no change │
│ QQuery 20 │   31.44 ms │                                   30.91 ms │     no change │
│ QQuery 21 │   16.75 ms │                                   17.44 ms │     no change │
│ QQuery 22 │   74.72 ms │                                   61.54 ms │ +1.21x faster │
│ QQuery 23 │  300.43 ms │                                  300.97 ms │     no change │
│ QQuery 24 │  185.63 ms │                                  192.11 ms │     no change │
│ QQuery 25 │  105.51 ms │                                  106.48 ms │     no change │
│ QQuery 26 │   45.61 ms │                                   46.64 ms │     no change │
│ QQuery 27 │    5.78 ms │                                    6.39 ms │  1.11x slower │
│ QQuery 28 │   56.78 ms │                                   58.13 ms │     no change │
│ QQuery 29 │   96.63 ms │                                   94.77 ms │     no change │
│ QQuery 30 │   36.03 ms │                                   37.11 ms │     no change │
│ QQuery 31 │  112.74 ms │                                  110.55 ms │     no change │
│ QQuery 32 │   20.62 ms │                                   19.22 ms │ +1.07x faster │
│ QQuery 33 │   38.67 ms │                                   36.87 ms │     no change │
│ QQuery 34 │   10.38 ms │                                    9.57 ms │ +1.09x faster │
│ QQuery 35 │   71.35 ms │                                   70.78 ms │     no change │
│ QQuery 36 │    5.84 ms │                                    5.84 ms │     no change │
│ QQuery 37 │    7.13 ms │                                    7.83 ms │  1.10x slower │
│ QQuery 38 │   63.48 ms │                                   62.04 ms │     no change │
│ QQuery 39 │   70.42 ms │                                   71.68 ms │     no change │
│ QQuery 40 │   22.35 ms │                                   24.63 ms │  1.10x slower │
│ QQuery 41 │   13.39 ms │                                   14.02 ms │     no change │
│ QQuery 42 │   24.48 ms │                                   24.62 ms │     no change │
│ QQuery 43 │    5.59 ms │                                    5.06 ms │ +1.10x faster │
│ QQuery 44 │    9.41 ms │                                    8.51 ms │ +1.11x faster │
│ QQuery 45 │   39.53 ms │                                   34.25 ms │ +1.15x faster │
│ QQuery 46 │   11.60 ms │                                   10.87 ms │ +1.07x faster │
│ QQuery 47 │  211.05 ms │                                  199.24 ms │ +1.06x faster │
│ QQuery 48 │   90.88 ms │                                   89.92 ms │     no change │
│ QQuery 49 │   69.90 ms │                                   68.75 ms │     no change │
│ QQuery 50 │   58.31 ms │                                   57.05 ms │     no change │
│ QQuery 51 │   90.24 ms │                                   90.33 ms │     no change │
│ QQuery 52 │   23.91 ms │                                   24.24 ms │     no change │
│ QQuery 53 │   30.05 ms │                                   29.36 ms │     no change │
│ QQuery 54 │   53.86 ms │                                   52.73 ms │     no change │
│ QQuery 55 │   23.79 ms │                                   24.46 ms │     no change │
│ QQuery 56 │   39.01 ms │                                   38.42 ms │     no change │
│ QQuery 57 │  158.65 ms │                                  149.85 ms │ +1.06x faster │
│ QQuery 58 │  104.71 ms │                                  101.08 ms │     no change │
│ QQuery 59 │  109.43 ms │                                  107.70 ms │     no change │
│ QQuery 60 │   43.30 ms │                                   43.42 ms │     no change │
│ QQuery 61 │    9.78 ms │                                    9.76 ms │     no change │
│ QQuery 62 │   39.14 ms │                                   38.05 ms │     no change │
│ QQuery 63 │   29.99 ms │                                   29.17 ms │     no change │
│ QQuery 64 │  378.84 ms │                                  367.00 ms │     no change │
│ QQuery 65 │  136.15 ms │                                  132.43 ms │     no change │
│ QQuery 66 │   74.80 ms │                                   73.52 ms │     no change │
│ QQuery 67 │  226.00 ms │                                  217.53 ms │     no change │
│ QQuery 68 │   11.14 ms │                                   12.42 ms │  1.11x slower │
│ QQuery 69 │   55.74 ms │                                   57.55 ms │     no change │
│ QQuery 70 │   97.95 ms │                                  100.84 ms │     no change │
│ QQuery 71 │   35.13 ms │                                   34.74 ms │     no change │
│ QQuery 72 │ 1656.75 ms │                                 1589.51 ms │     no change │
│ QQuery 73 │   10.05 ms │                                    9.68 ms │     no change │
│ QQuery 74 │  239.96 ms │                                  232.58 ms │     no change │
│ QQuery 75 │  137.00 ms │                                  139.40 ms │     no change │
│ QQuery 76 │   33.80 ms │                                   33.56 ms │     no change │
│ QQuery 77 │   58.37 ms │                                   59.56 ms │     no change │
│ QQuery 78 │  156.37 ms │                                  156.34 ms │     no change │
│ QQuery 79 │   66.75 ms │                                   66.08 ms │     no change │
│ QQuery 80 │   92.97 ms │                                   93.61 ms │     no change │
│ QQuery 81 │   27.44 ms │                                   27.61 ms │     no change │
│ QQuery 82 │   16.35 ms │                                   17.61 ms │  1.08x slower │
│ QQuery 83 │   31.65 ms │                                   33.19 ms │     no change │
│ QQuery 84 │   32.18 ms │                                   31.24 ms │     no change │
│ QQuery 85 │   99.62 ms │                                  100.13 ms │     no change │
│ QQuery 86 │   24.62 ms │                                   24.93 ms │     no change │
│ QQuery 87 │   68.90 ms │                                   62.78 ms │ +1.10x faster │
│ QQuery 88 │   59.01 ms │                                   58.73 ms │     no change │
│ QQuery 89 │   36.20 ms │                                   36.98 ms │     no change │
│ QQuery 90 │   16.25 ms │                                   16.49 ms │     no change │
│ QQuery 91 │   39.55 ms │                                   41.36 ms │     no change │
│ QQuery 92 │   27.66 ms │                                   30.72 ms │  1.11x slower │
│ QQuery 93 │   51.30 ms │                                   50.46 ms │     no change │
│ QQuery 94 │   37.76 ms │                                   38.76 ms │     no change │
│ QQuery 95 │   77.55 ms │                                   80.17 ms │     no change │
│ QQuery 96 │   23.32 ms │                                   23.91 ms │     no change │
│ QQuery 97 │   50.42 ms │                                   51.17 ms │     no change │
│ QQuery 98 │   39.02 ms │                                   38.09 ms │     no change │
│ QQuery 99 │   56.33 ms │                                   56.66 ms │     no change │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9492.48ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9182.48ms │
│ Average Time (HEAD)                                       │   95.88ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   92.75ms │
│ Queries Faster                                            │        13 │
│ Queries Slower                                            │         6 │
│ Queries with No Change                                    │        80 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.26 / 5.84 ±1.07 / 7.98 ms │                5.16 / 5.93 ±1.17 / 8.25 ms │     no change │
│ QQuery 2  │        76.38 / 77.28 ±0.52 / 77.90 ms │             76.81 / 77.40 ±0.58 / 78.14 ms │     no change │
│ QQuery 3  │        29.07 / 29.78 ±0.41 / 30.34 ms │             28.64 / 28.89 ±0.19 / 29.17 ms │     no change │
│ QQuery 4  │    848.25 / 875.93 ±16.53 / 899.52 ms │         730.51 / 777.28 ±32.35 / 831.96 ms │ +1.13x faster │
│ QQuery 5  │        49.81 / 50.42 ±0.34 / 50.75 ms │             50.70 / 51.40 ±0.63 / 52.19 ms │     no change │
│ QQuery 6  │        33.48 / 34.02 ±0.44 / 34.62 ms │             32.91 / 33.73 ±0.59 / 34.55 ms │     no change │
│ QQuery 7  │        71.47 / 73.40 ±1.18 / 74.73 ms │             71.77 / 73.51 ±1.67 / 76.54 ms │     no change │
│ QQuery 8  │        38.98 / 39.16 ±0.22 / 39.53 ms │             37.49 / 38.33 ±0.77 / 39.69 ms │     no change │
│ QQuery 9  │        51.69 / 52.82 ±1.15 / 54.87 ms │             51.63 / 52.95 ±0.76 / 53.99 ms │     no change │
│ QQuery 10 │        61.19 / 61.56 ±0.29 / 61.90 ms │             61.23 / 61.66 ±0.31 / 62.04 ms │     no change │
│ QQuery 11 │    474.56 / 494.94 ±15.07 / 520.23 ms │         412.40 / 433.65 ±15.15 / 456.30 ms │ +1.14x faster │
│ QQuery 12 │        25.98 / 27.61 ±1.77 / 30.92 ms │             25.66 / 26.47 ±0.74 / 27.76 ms │     no change │
│ QQuery 13 │     111.89 / 113.95 ±2.13 / 117.94 ms │          112.01 / 115.02 ±1.88 / 117.00 ms │     no change │
│ QQuery 14 │     379.70 / 387.89 ±4.32 / 391.67 ms │          381.89 / 387.80 ±4.42 / 393.90 ms │     no change │
│ QQuery 15 │        52.61 / 54.34 ±1.51 / 56.49 ms │             50.92 / 52.85 ±2.84 / 58.49 ms │     no change │
│ QQuery 16 │           6.55 / 6.82 ±0.19 / 7.02 ms │                5.98 / 6.11 ±0.22 / 6.56 ms │ +1.12x faster │
│ QQuery 17 │        78.89 / 81.77 ±3.20 / 87.48 ms │             78.21 / 81.68 ±4.33 / 90.10 ms │     no change │
│ QQuery 18 │     103.45 / 105.41 ±1.06 / 106.53 ms │          107.78 / 110.07 ±1.67 / 112.57 ms │     no change │
│ QQuery 19 │        42.06 / 42.49 ±0.44 / 43.27 ms │             41.13 / 42.58 ±0.77 / 43.14 ms │     no change │
│ QQuery 20 │        31.44 / 32.34 ±0.62 / 33.03 ms │             30.91 / 31.48 ±0.32 / 31.84 ms │     no change │
│ QQuery 21 │        16.75 / 17.53 ±1.19 / 19.89 ms │             17.44 / 17.56 ±0.07 / 17.62 ms │     no change │
│ QQuery 22 │        74.72 / 77.60 ±1.96 / 80.64 ms │             61.54 / 63.56 ±1.75 / 66.46 ms │ +1.22x faster │
│ QQuery 23 │     300.43 / 310.29 ±9.07 / 321.30 ms │          300.97 / 315.23 ±8.20 / 324.17 ms │     no change │
│ QQuery 24 │     185.63 / 196.96 ±6.03 / 203.32 ms │          192.11 / 196.66 ±3.62 / 201.89 ms │     no change │
│ QQuery 25 │     105.51 / 108.82 ±2.87 / 113.49 ms │          106.48 / 109.90 ±2.22 / 113.00 ms │     no change │
│ QQuery 26 │        45.61 / 46.20 ±0.46 / 46.87 ms │             46.64 / 49.00 ±1.71 / 51.47 ms │  1.06x slower │
│ QQuery 27 │           5.78 / 6.55 ±1.13 / 8.78 ms │                6.39 / 6.50 ±0.12 / 6.71 ms │     no change │
│ QQuery 28 │        56.78 / 59.70 ±2.15 / 62.12 ms │             58.13 / 60.09 ±1.08 / 61.16 ms │     no change │
│ QQuery 29 │       96.63 / 98.44 ±1.32 / 100.46 ms │            94.77 / 96.92 ±1.97 / 100.34 ms │     no change │
│ QQuery 30 │        36.03 / 38.10 ±1.31 / 39.89 ms │             37.11 / 40.38 ±3.26 / 46.47 ms │  1.06x slower │
│ QQuery 31 │     112.74 / 113.75 ±1.26 / 116.23 ms │          110.55 / 113.47 ±2.49 / 118.02 ms │     no change │
│ QQuery 32 │        20.62 / 21.88 ±1.86 / 25.55 ms │             19.22 / 19.69 ±0.61 / 20.77 ms │ +1.11x faster │
│ QQuery 33 │        38.67 / 39.23 ±0.37 / 39.81 ms │             36.87 / 38.20 ±0.78 / 38.99 ms │     no change │
│ QQuery 34 │        10.38 / 10.47 ±0.07 / 10.59 ms │               9.57 / 9.91 ±0.23 / 10.28 ms │ +1.06x faster │
│ QQuery 35 │        71.35 / 72.59 ±2.02 / 76.61 ms │             70.78 / 74.40 ±3.11 / 78.78 ms │     no change │
│ QQuery 36 │           5.84 / 6.15 ±0.29 / 6.64 ms │                5.84 / 6.04 ±0.21 / 6.42 ms │     no change │
│ QQuery 37 │           7.13 / 7.33 ±0.16 / 7.58 ms │                7.83 / 8.05 ±0.16 / 8.31 ms │  1.10x slower │
│ QQuery 38 │        63.48 / 66.31 ±1.63 / 68.28 ms │             62.04 / 63.79 ±1.26 / 65.86 ms │     no change │
│ QQuery 39 │        70.42 / 73.12 ±2.48 / 76.62 ms │             71.68 / 75.26 ±2.50 / 78.40 ms │     no change │
│ QQuery 40 │        22.35 / 22.81 ±0.34 / 23.27 ms │             24.63 / 26.05 ±0.81 / 27.07 ms │  1.14x slower │
│ QQuery 41 │        13.39 / 13.54 ±0.17 / 13.86 ms │             14.02 / 14.22 ±0.15 / 14.48 ms │  1.05x slower │
│ QQuery 42 │        24.48 / 25.05 ±0.64 / 26.21 ms │             24.62 / 24.84 ±0.33 / 25.48 ms │     no change │
│ QQuery 43 │           5.59 / 5.74 ±0.08 / 5.81 ms │                5.06 / 5.37 ±0.26 / 5.81 ms │ +1.07x faster │
│ QQuery 44 │           9.41 / 9.57 ±0.15 / 9.80 ms │                8.51 / 8.82 ±0.26 / 9.18 ms │ +1.08x faster │
│ QQuery 45 │        39.53 / 41.06 ±1.74 / 43.47 ms │             34.25 / 35.84 ±1.28 / 37.80 ms │ +1.15x faster │
│ QQuery 46 │        11.60 / 11.94 ±0.24 / 12.33 ms │             10.87 / 10.99 ±0.08 / 11.11 ms │ +1.09x faster │
│ QQuery 47 │     211.05 / 214.56 ±1.93 / 216.06 ms │          199.24 / 206.96 ±6.81 / 216.99 ms │     no change │
│ QQuery 48 │        90.88 / 92.40 ±1.13 / 94.36 ms │             89.92 / 92.19 ±1.78 / 95.07 ms │     no change │
│ QQuery 49 │        69.90 / 70.92 ±0.99 / 72.24 ms │             68.75 / 70.34 ±1.63 / 73.11 ms │     no change │
│ QQuery 50 │        58.31 / 59.48 ±0.86 / 60.83 ms │             57.05 / 57.93 ±0.60 / 58.92 ms │     no change │
│ QQuery 51 │        90.24 / 94.57 ±2.81 / 97.69 ms │            90.33 / 96.58 ±4.80 / 102.19 ms │     no change │
│ QQuery 52 │        23.91 / 24.72 ±0.56 / 25.23 ms │             24.24 / 24.71 ±0.33 / 25.25 ms │     no change │
│ QQuery 53 │        30.05 / 30.33 ±0.22 / 30.67 ms │             29.36 / 29.85 ±0.28 / 30.17 ms │     no change │
│ QQuery 54 │        53.86 / 56.36 ±3.79 / 63.85 ms │             52.73 / 53.67 ±0.73 / 54.75 ms │     no change │
│ QQuery 55 │        23.79 / 24.26 ±0.28 / 24.67 ms │             24.46 / 25.37 ±1.23 / 27.75 ms │     no change │
│ QQuery 56 │        39.01 / 39.75 ±0.58 / 40.76 ms │             38.42 / 39.06 ±0.42 / 39.43 ms │     no change │
│ QQuery 57 │     158.65 / 159.88 ±1.02 / 161.68 ms │          149.85 / 153.45 ±2.16 / 156.17 ms │     no change │
│ QQuery 58 │     104.71 / 106.79 ±1.54 / 109.33 ms │          101.08 / 103.78 ±3.39 / 110.39 ms │     no change │
│ QQuery 59 │     109.43 / 110.32 ±0.82 / 111.64 ms │          107.70 / 109.11 ±1.07 / 110.63 ms │     no change │
│ QQuery 60 │        43.30 / 43.78 ±0.41 / 44.41 ms │             43.42 / 44.56 ±1.27 / 47.04 ms │     no change │
│ QQuery 61 │          9.78 / 9.92 ±0.12 / 10.06 ms │               9.76 / 9.97 ±0.23 / 10.37 ms │     no change │
│ QQuery 62 │        39.14 / 39.74 ±0.57 / 40.82 ms │             38.05 / 38.73 ±0.52 / 39.58 ms │     no change │
│ QQuery 63 │        29.99 / 30.49 ±0.26 / 30.71 ms │             29.17 / 29.70 ±0.47 / 30.38 ms │     no change │
│ QQuery 64 │     378.84 / 383.66 ±2.86 / 386.66 ms │          367.00 / 374.97 ±6.71 / 386.03 ms │     no change │
│ QQuery 65 │     136.15 / 140.02 ±4.37 / 148.20 ms │          132.43 / 136.12 ±2.47 / 139.93 ms │     no change │
│ QQuery 66 │        74.80 / 80.18 ±4.43 / 88.34 ms │             73.52 / 75.04 ±2.22 / 79.28 ms │ +1.07x faster │
│ QQuery 67 │     226.00 / 240.87 ±7.92 / 248.00 ms │          217.53 / 225.71 ±6.09 / 233.54 ms │ +1.07x faster │
│ QQuery 68 │        11.14 / 11.53 ±0.27 / 11.88 ms │             12.42 / 12.62 ±0.22 / 13.02 ms │  1.09x slower │
│ QQuery 69 │        55.74 / 56.80 ±0.88 / 58.25 ms │             57.55 / 58.47 ±0.51 / 59.13 ms │     no change │
│ QQuery 70 │      97.95 / 101.77 ±4.75 / 111.15 ms │          100.84 / 103.19 ±2.05 / 106.92 ms │     no change │
│ QQuery 71 │        35.13 / 35.32 ±0.16 / 35.55 ms │             34.74 / 35.52 ±0.79 / 36.86 ms │     no change │
│ QQuery 72 │ 1656.75 / 1671.63 ±12.48 / 1689.66 ms │      1589.51 / 1616.54 ±27.62 / 1662.47 ms │     no change │
│ QQuery 73 │        10.05 / 10.20 ±0.12 / 10.39 ms │               9.68 / 9.87 ±0.15 / 10.08 ms │     no change │
│ QQuery 74 │    239.96 / 249.79 ±10.12 / 264.52 ms │         232.58 / 244.38 ±11.01 / 261.43 ms │     no change │
│ QQuery 75 │     137.00 / 142.18 ±3.48 / 145.86 ms │          139.40 / 140.18 ±0.79 / 141.57 ms │     no change │
│ QQuery 76 │        33.80 / 34.66 ±0.77 / 35.62 ms │             33.56 / 34.48 ±1.02 / 36.33 ms │     no change │
│ QQuery 77 │        58.37 / 60.80 ±2.39 / 64.20 ms │             59.56 / 60.89 ±1.20 / 62.39 ms │     no change │
│ QQuery 78 │     156.37 / 164.51 ±6.87 / 173.81 ms │          156.34 / 162.43 ±6.86 / 174.22 ms │     no change │
│ QQuery 79 │        66.75 / 69.03 ±2.02 / 72.48 ms │             66.08 / 69.92 ±4.12 / 77.69 ms │     no change │
│ QQuery 80 │        92.97 / 95.96 ±2.22 / 98.11 ms │             93.61 / 95.92 ±1.90 / 98.82 ms │     no change │
│ QQuery 81 │        27.44 / 27.74 ±0.24 / 28.17 ms │             27.61 / 29.06 ±0.86 / 30.19 ms │     no change │
│ QQuery 82 │        16.35 / 16.92 ±0.83 / 18.56 ms │             17.61 / 19.13 ±2.39 / 23.88 ms │  1.13x slower │
│ QQuery 83 │        31.65 / 32.28 ±0.65 / 33.40 ms │             33.19 / 33.72 ±0.43 / 34.28 ms │     no change │
│ QQuery 84 │        32.18 / 32.45 ±0.25 / 32.91 ms │             31.24 / 31.84 ±0.49 / 32.44 ms │     no change │
│ QQuery 85 │      99.62 / 103.16 ±2.77 / 107.50 ms │          100.13 / 101.06 ±0.89 / 102.23 ms │     no change │
│ QQuery 86 │        24.62 / 25.54 ±0.65 / 26.26 ms │             24.93 / 25.29 ±0.19 / 25.43 ms │     no change │
│ QQuery 87 │        68.90 / 70.07 ±1.62 / 73.17 ms │             62.78 / 65.27 ±1.71 / 67.39 ms │ +1.07x faster │
│ QQuery 88 │        59.01 / 60.00 ±0.64 / 60.64 ms │             58.73 / 59.60 ±0.59 / 60.28 ms │     no change │
│ QQuery 89 │        36.20 / 36.86 ±0.55 / 37.69 ms │             36.98 / 37.26 ±0.33 / 37.86 ms │     no change │
│ QQuery 90 │        16.25 / 16.48 ±0.19 / 16.75 ms │             16.49 / 17.47 ±1.37 / 20.19 ms │  1.06x slower │
│ QQuery 91 │        39.55 / 40.55 ±1.29 / 43.08 ms │             41.36 / 42.39 ±1.10 / 44.39 ms │     no change │
│ QQuery 92 │        27.66 / 28.51 ±0.66 / 29.47 ms │             30.72 / 31.14 ±0.23 / 31.38 ms │  1.09x slower │
│ QQuery 93 │        51.30 / 51.62 ±0.23 / 51.99 ms │             50.46 / 51.07 ±0.57 / 52.14 ms │     no change │
│ QQuery 94 │        37.76 / 38.98 ±0.63 / 39.56 ms │             38.76 / 39.48 ±0.44 / 40.09 ms │     no change │
│ QQuery 95 │        77.55 / 79.68 ±2.44 / 84.16 ms │             80.17 / 83.79 ±2.22 / 86.26 ms │  1.05x slower │
│ QQuery 96 │        23.32 / 23.95 ±0.63 / 25.15 ms │             23.91 / 24.34 ±0.25 / 24.66 ms │     no change │
│ QQuery 97 │        50.42 / 51.93 ±0.88 / 52.90 ms │             51.17 / 52.24 ±1.65 / 55.53 ms │     no change │
│ QQuery 98 │        39.02 / 39.59 ±0.39 / 39.93 ms │             38.09 / 38.39 ±0.27 / 38.85 ms │     no change │
│ QQuery 99 │        56.33 / 57.72 ±0.78 / 58.54 ms │             56.66 / 57.39 ±0.57 / 58.24 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9739.73ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9461.66ms │
│ Average Time (HEAD)                                       │   98.38ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   95.57ms │
│ Queries Faster                                            │        13 │
│ Queries Slower                                            │        10 │
│ Queries with No Change                                    │        76 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: d49dd6d5614860d21328eb63633fd9d8e08fbd37

tpcds — tpcds_sf1

Query	Base	Changed	Change
Query 1	198.0 KiB	197.5 KiB	-0.3%
Query 2	70.8 MiB	88.1 MiB	+24.4%
Query 3	1.9 MiB	1.7 MiB	-11.1%
Query 4	79.7 MiB	79.7 MiB	+0.0%
Query 5	13.0 MiB	13.0 MiB	-0.0%
Query 6	4.7 MiB	4.4 MiB	-6.2%
Query 7	3.7 MiB	3.7 MiB	+1.0%
Query 8	1.7 MiB	2.2 MiB	+27.7%
Query 9	32.1 KiB	32.1 KiB	+0.0%
Query 10	2.0 MiB	2.0 MiB	+0.0%
Query 11	62.0 MiB	62.0 MiB	+0.0%
Query 12	114.4 MiB	114.4 MiB	+0.0%
Query 13	13.7 MiB	13.7 MiB	+0.0%
Query 14	171.0 MiB	171.6 MiB	+0.4%
Query 15	4.2 MiB	4.3 MiB	+2.8%
Query 16	118.6 KiB	118.6 KiB	+0.0%
Query 17	40.6 MiB	41.0 MiB	+1.2%
Query 18	21.2 MiB	21.2 MiB	+0.0%
Query 19	2.1 MiB	1.8 MiB	-14.4%
Query 20	114.6 MiB	114.6 MiB	-0.0%
Query 21	3.3 MiB	3.7 MiB	+11.1%
Query 22	61.8 MiB	63.5 MiB	+2.8%
Query 23	211.9 MiB	211.5 MiB	-0.1%
Query 24	38.7 MiB	38.7 MiB	+0.0%
Query 25	99.9 MiB	99.9 MiB	+0.0%
Query 26	3.5 MiB	3.5 MiB	-0.0%
Query 27	1.6 MiB	1.6 MiB	+0.0%
Query 28	1.5 MiB	1.3 MiB	-9.1%
Query 29	104.7 MiB	104.7 MiB	+0.0%
Query 30	3.8 MiB	3.7 MiB	-2.6%
Query 31	13.4 MiB	13.4 MiB	+0.0%
Query 32	2.5 MiB	2.7 MiB	+8.9%
Query 33	5.5 MiB	4.2 MiB	-22.9%
Query 34	6.0 MiB	6.0 MiB	+0.0%
Query 35	16.6 MiB	17.5 MiB	+5.3%
Query 36	10.3 MiB	10.3 MiB	+0.0%
Query 37	3.3 MiB	3.3 MiB	+0.0%
Query 38	10.8 MiB	10.8 MiB	+0.2%
Query 39	53.0 MiB	43.6 MiB	-17.7%
Query 40	12.2 MiB	12.2 MiB	+0.0%
Query 41	2.1 MiB	2.1 MiB	+0.1%
Query 42	1.1 MiB	1.3 MiB	+21.9%
Query 43	257.8 KiB	327.1 KiB	+26.9%
Query 44	233.2 KiB	233.2 KiB	+0.0%
Query 45	4.5 MiB	4.5 MiB	+0.0%
Query 46	6.2 MiB	6.2 MiB	+0.0%
Query 47	132.5 MiB	128.5 MiB	-3.1%
Query 48	11.4 MiB	11.8 MiB	+3.3%
Query 49	52.8 MiB	42.8 MiB	-18.9%
Query 50	17.3 MiB	16.9 MiB	-2.5%
Query 51	177.7 MiB	177.7 MiB	+0.0%
Query 52	1.5 MiB	1.5 MiB	+3.0%
Query 53	30.6 MiB	30.4 MiB	-0.5%
Query 54	2.1 MiB	2.1 MiB	+0.0%
Query 55	1.5 MiB	1.3 MiB	-14.5%
Query 56	3.5 MiB	3.0 MiB	-12.4%
Query 57	120.8 MiB	120.8 MiB	-0.0%
Query 58	4.2 MiB	4.2 MiB	+0.0%
Query 59	11.7 MiB	11.7 MiB	+0.0%
Query 60	7.2 MiB	7.2 MiB	+0.2%
Query 61	2.4 KiB	2.5 KiB	+3.3%
Query 62	3.7 MiB	3.6 MiB	-0.1%
Query 63	20.6 MiB	20.6 MiB	-0.0%
Query 64	94.8 MiB	84.7 MiB	-10.6%
Query 65	29.7 MiB	29.7 MiB	+0.0%
Query 66	14.9 MiB	14.9 MiB	+0.0%
Query 67	137.5 MiB	140.2 MiB	+2.0%
Query 68	6.2 MiB	6.2 MiB	+0.0%
Query 69	3.7 MiB	3.8 MiB	+3.2%
Query 70	44.0 MiB	44.0 MiB	-0.0%
Query 71	40.4 MiB	40.4 MiB	+0.1%
Query 72	97.2 MiB	97.2 MiB	+0.0%
Query 73	6.0 MiB	6.0 MiB	+0.0%
Query 74	34.6 MiB	34.6 MiB	+0.0%
Query 75	20.7 MiB	20.2 MiB	-2.8%
Query 76	7.9 MiB	7.9 MiB	+0.0%
Query 77	4.1 MiB	4.1 MiB	+0.0%
Query 78	111.1 MiB	111.5 MiB	+0.4%
Query 79	8.7 MiB	9.3 MiB	+7.5%
Query 80	42.8 MiB	42.0 MiB	-1.9%
Query 81	4.0 MiB	4.2 MiB	+2.8%
Query 82	4.9 MiB	4.9 MiB	+0.0%
Query 83	9.7 MiB	9.0 MiB	-7.0%
Query 84	11.9 MiB	11.9 MiB	+0.0%
Query 85	17.0 MiB	17.0 MiB	+0.0%
Query 86	82.1 MiB	82.1 MiB	-0.0%
Query 87	12.9 MiB	13.1 MiB	+1.7%
Query 88	1.6 MiB	1.7 MiB	+6.8%
Query 89	112.2 MiB	112.2 MiB	-0.0%
Query 90	348.6 KiB	363.7 KiB	+4.3%
Query 91	21.1 MiB	11.4 MiB	-46.3%
Query 92	2.7 MiB	2.7 MiB	+0.0%
Query 93	15.2 MiB	15.4 MiB	+1.4%
Query 94	3.4 MiB	3.7 MiB	+8.2%
Query 95	18.3 MiB	17.5 MiB	-4.8%
Query 96	224.7 KiB	224.7 KiB	+0.0%
Query 97	33.1 MiB	33.1 MiB	+0.0%
Query 98	124.4 MiB	114.7 MiB	-7.8%
Query 99	1.9 MiB	1.9 MiB	+2.1%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpcds	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	211.9 MiB	1.9 GiB	1.7 GiB	9.4×
tpcds	changed (`d49dd6d5614860d21328eb63633fd9d8e08fbd37`)	211.5 MiB	1.9 GiB	1.7 GiB	9.2×

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	50.0s
Peak memory	1.9 GiB
Avg memory	1.4 GiB
CPU user	199.1s
CPU sys	5.7s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	50.0s
Peak memory	1.9 GiB
Avg memory	1.4 GiB
CPU user	192.3s
CPU sys	5.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:34:12Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing e1aa5f4 (e1aa5f4) to 4ede583 diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃        HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 0  │     1.14 ms │                                    1.13 ms │    no change │
│ QQuery 1  │    10.80 ms │                                   11.11 ms │    no change │
│ QQuery 2  │    36.17 ms │                                   35.69 ms │    no change │
│ QQuery 3  │    30.82 ms │                                   31.28 ms │    no change │
│ QQuery 4  │   230.48 ms │                                  237.01 ms │    no change │
│ QQuery 5  │  1467.64 ms │                                 1491.31 ms │    no change │
│ QQuery 6  │     1.16 ms │                                    1.17 ms │    no change │
│ QQuery 7  │    12.06 ms │                                   12.32 ms │    no change │
│ QQuery 8  │   330.97 ms │                                  335.77 ms │    no change │
│ QQuery 9  │   467.95 ms │                                  472.92 ms │    no change │
│ QQuery 10 │   197.33 ms │                                  199.89 ms │    no change │
│ QQuery 11 │   208.43 ms │                                  209.93 ms │    no change │
│ QQuery 12 │  1503.49 ms │                                 1516.12 ms │    no change │
│ QQuery 13 │  2036.41 ms │                                 1964.48 ms │    no change │
│ QQuery 14 │  1578.00 ms │                                 1667.91 ms │ 1.06x slower │
│ QQuery 15 │   285.68 ms │                                  284.88 ms │    no change │
│ QQuery 16 │  3527.19 ms │                                 3623.24 ms │    no change │
│ QQuery 17 │  3673.63 ms │                                 3667.34 ms │    no change │
│ QQuery 18 │  7064.88 ms │                                 6800.53 ms │    no change │
│ QQuery 19 │    26.58 ms │                                   26.80 ms │    no change │
│ QQuery 20 │  1042.27 ms │                                 1048.38 ms │    no change │
│ QQuery 21 │  1349.53 ms │                                 1359.83 ms │    no change │
│ QQuery 22 │  3228.98 ms │                                 3221.52 ms │    no change │
│ QQuery 23 │  9041.74 ms │                                 9198.98 ms │    no change │
│ QQuery 24 │    94.75 ms │                                   96.48 ms │    no change │
│ QQuery 25 │   349.45 ms │                                  349.39 ms │    no change │
│ QQuery 26 │    97.43 ms │                                   96.10 ms │    no change │
│ QQuery 27 │  1143.59 ms │                                 1129.82 ms │    no change │
│ QQuery 28 │  2807.12 ms │                                 2747.76 ms │    no change │
│ QQuery 29 │    39.53 ms │                                   38.75 ms │    no change │
│ QQuery 30 │   536.26 ms │                                  539.05 ms │    no change │
│ QQuery 31 │   546.36 ms │                                  539.95 ms │    no change │
│ QQuery 32 │  1008.59 ms │                                 1023.93 ms │    no change │
│ QQuery 33 │ 13780.06 ms │                                13431.55 ms │    no change │
│ QQuery 34 │ 13782.38 ms │                                13762.29 ms │    no change │
│ QQuery 35 │   296.92 ms │                                  303.46 ms │    no change │
│ QQuery 36 │   176.71 ms │                                  175.99 ms │    no change │
│ QQuery 37 │    96.70 ms │                                   97.82 ms │    no change │
│ QQuery 38 │    79.89 ms │                                   79.34 ms │    no change │
│ QQuery 39 │   400.18 ms │                                  387.95 ms │    no change │
│ QQuery 40 │    14.42 ms │                                   14.45 ms │    no change │
│ QQuery 41 │    13.47 ms │                                   13.88 ms │    no change │
│ QQuery 42 │    13.00 ms │                                   13.53 ms │    no change │
└───────────┴─────────────┴────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 72630.14ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 72261.02ms │
│ Average Time (HEAD)                                       │  1689.07ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  1680.49ms │
│ Queries Faster                                            │          0 │
│ Queries Slower                                            │          1 │
│ Queries with No Change                                    │         42 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                        HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │                1.14 / 4.18 ±5.98 / 16.14 ms │               1.13 / 4.27 ±6.10 / 16.47 ms │     no change │
│ QQuery 1  │              10.80 / 10.99 ±0.14 / 11.20 ms │             11.11 / 11.23 ±0.09 / 11.34 ms │     no change │
│ QQuery 2  │              36.17 / 36.83 ±1.07 / 38.97 ms │             35.69 / 35.99 ±0.28 / 36.44 ms │     no change │
│ QQuery 3  │              30.82 / 31.53 ±1.08 / 33.67 ms │             31.28 / 32.27 ±1.46 / 35.16 ms │     no change │
│ QQuery 4  │           230.48 / 234.14 ±2.26 / 236.54 ms │          237.01 / 240.52 ±2.11 / 243.16 ms │     no change │
│ QQuery 5  │       1467.64 / 1484.99 ±11.03 / 1497.97 ms │      1491.31 / 1533.94 ±54.06 / 1632.03 ms │     no change │
│ QQuery 6  │                 1.16 / 1.32 ±0.24 / 1.80 ms │                1.17 / 1.33 ±0.24 / 1.79 ms │     no change │
│ QQuery 7  │              12.06 / 12.13 ±0.05 / 12.19 ms │             12.32 / 12.44 ±0.10 / 12.55 ms │     no change │
│ QQuery 8  │           330.97 / 337.01 ±5.04 / 343.89 ms │          335.77 / 342.35 ±5.76 / 352.32 ms │     no change │
│ QQuery 9  │           467.95 / 480.37 ±9.25 / 493.27 ms │         472.92 / 494.92 ±16.75 / 523.40 ms │     no change │
│ QQuery 10 │           197.33 / 199.70 ±2.51 / 203.99 ms │          199.89 / 201.38 ±1.72 / 204.61 ms │     no change │
│ QQuery 11 │           208.43 / 209.44 ±0.87 / 210.60 ms │          209.93 / 211.23 ±0.84 / 212.32 ms │     no change │
│ QQuery 12 │       1503.49 / 1535.07 ±25.89 / 1572.52 ms │      1516.12 / 1564.18 ±35.33 / 1616.44 ms │     no change │
│ QQuery 13 │      2036.41 / 2254.22 ±336.04 / 2922.97 ms │     1964.48 / 2335.35 ±308.32 / 2729.20 ms │     no change │
│ QQuery 14 │       1578.00 / 1685.09 ±80.99 / 1823.81 ms │      1667.91 / 1695.99 ±30.93 / 1745.75 ms │     no change │
│ QQuery 15 │          285.68 / 299.88 ±14.50 / 325.36 ms │         284.88 / 296.94 ±17.93 / 332.58 ms │     no change │
│ QQuery 16 │       3527.19 / 3672.89 ±93.87 / 3821.62 ms │     3623.24 / 3823.62 ±136.69 / 4028.06 ms │     no change │
│ QQuery 17 │       3673.63 / 3733.76 ±49.39 / 3813.53 ms │     3667.34 / 3878.87 ±128.71 / 4052.94 ms │     no change │
│ QQuery 18 │      7064.88 / 7245.16 ±235.14 / 7700.36 ms │     6800.53 / 7240.54 ±265.34 / 7621.10 ms │     no change │
│ QQuery 19 │              26.58 / 29.36 ±4.35 / 38.02 ms │            26.80 / 36.64 ±15.43 / 67.01 ms │  1.25x slower │
│ QQuery 20 │        1042.27 / 1054.90 ±8.77 / 1067.41 ms │      1048.38 / 1071.18 ±30.03 / 1129.71 ms │     no change │
│ QQuery 21 │        1349.53 / 1363.17 ±8.87 / 1377.19 ms │       1359.83 / 1376.97 ±9.07 / 1386.81 ms │     no change │
│ QQuery 22 │       3228.98 / 3249.21 ±16.65 / 3271.17 ms │      3221.52 / 3295.26 ±59.82 / 3373.24 ms │     no change │
│ QQuery 23 │      9041.74 / 9283.64 ±165.92 / 9531.68 ms │      9198.98 / 9329.24 ±94.54 / 9439.52 ms │     no change │
│ QQuery 24 │           94.75 / 135.78 ±48.58 / 196.87 ms │          96.48 / 143.80 ±69.12 / 275.93 ms │  1.06x slower │
│ QQuery 25 │           349.45 / 352.10 ±1.36 / 353.22 ms │          349.39 / 355.17 ±7.77 / 370.55 ms │     no change │
│ QQuery 26 │             97.43 / 98.91 ±1.10 / 100.11 ms │            96.10 / 99.29 ±4.08 / 107.12 ms │     no change │
│ QQuery 27 │        1143.59 / 1148.20 ±3.00 / 1152.32 ms │       1129.82 / 1146.92 ±9.21 / 1157.10 ms │     no change │
│ QQuery 28 │       2807.12 / 2849.63 ±27.94 / 2890.02 ms │      2747.76 / 2787.28 ±24.10 / 2821.54 ms │     no change │
│ QQuery 29 │              39.53 / 43.83 ±8.33 / 60.49 ms │             38.75 / 39.08 ±0.19 / 39.34 ms │ +1.12x faster │
│ QQuery 30 │           536.26 / 546.65 ±6.13 / 555.06 ms │          539.05 / 548.76 ±6.67 / 557.69 ms │     no change │
│ QQuery 31 │           546.36 / 553.11 ±7.46 / 567.49 ms │          539.95 / 545.52 ±4.30 / 550.77 ms │     no change │
│ QQuery 32 │       1008.59 / 1115.50 ±72.87 / 1233.51 ms │      1023.93 / 1083.95 ±50.75 / 1154.90 ms │     no change │
│ QQuery 33 │ 13780.06 / 20866.61 ±13226.03 / 47312.57 ms │  13431.55 / 14250.97 ±528.68 / 15007.84 ms │ +1.46x faster │
│ QQuery 34 │   13782.38 / 14561.46 ±717.08 / 15639.69 ms │  13762.29 / 14494.61 ±518.20 / 15230.09 ms │     no change │
│ QQuery 35 │          296.92 / 328.39 ±39.87 / 399.86 ms │        303.46 / 395.00 ±154.14 / 702.59 ms │  1.20x slower │
│ QQuery 36 │           176.71 / 184.60 ±7.48 / 197.92 ms │         175.99 / 190.33 ±23.74 / 237.72 ms │     no change │
│ QQuery 37 │            96.70 / 103.60 ±6.15 / 114.79 ms │           97.82 / 101.79 ±4.38 / 110.02 ms │     no change │
│ QQuery 38 │              79.89 / 83.44 ±2.11 / 85.82 ms │             79.34 / 82.00 ±2.05 / 84.21 ms │     no change │
│ QQuery 39 │           400.18 / 414.50 ±9.40 / 429.15 ms │          387.95 / 397.29 ±9.45 / 410.36 ms │     no change │
│ QQuery 40 │              14.42 / 14.79 ±0.41 / 15.55 ms │             14.45 / 16.28 ±1.53 / 18.94 ms │  1.10x slower │
│ QQuery 41 │              13.47 / 15.09 ±2.59 / 20.25 ms │             13.88 / 17.86 ±7.55 / 32.96 ms │  1.18x slower │
│ QQuery 42 │              13.00 / 14.41 ±2.35 / 19.10 ms │             13.53 / 15.61 ±3.44 / 22.46 ms │  1.08x slower │
└───────────┴─────────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 81879.57ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 75778.17ms │
│ Average Time (HEAD)                                       │  1904.18ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  1762.28ms │
│ Queries Faster                                            │          2 │
│ Queries Slower                                            │          6 │
│ Queries with No Change                                    │         35 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc

clickbench_partitioned

Query	Base	Changed	Change
Query 0	0 B	0 B	0.0%
Query 1	104 B	104 B	+0.0%
Query 2	936 B	936 B	+0.0%
Query 3	312 B	312 B	+0.0%
Query 4	815.1 MiB	832.3 MiB	+2.1%
Query 5	1.4 GiB	1.5 GiB	+3.6%
Query 6	0 B	0 B	0.0%
Query 7	50.0 MiB	50.3 MiB	+0.5%
Query 8	943.6 MiB	944.0 MiB	+0.0%
Query 9	552.2 MiB	552.2 MiB	+0.0%
Query 10	72.8 MiB	72.7 MiB	-0.1%
Query 11	78.9 MiB	76.7 MiB	-2.8%
Query 12	1.6 GiB	1.6 GiB	-1.2%
Query 13	1.9 GiB	1.9 GiB	-1.3%
Query 14	1.9 GiB	1.9 GiB	+0.0%
Query 15	1.2 GiB	1.2 GiB	+2.4%
Query 16	3.2 GiB	3.4 GiB	+5.5%
Query 17	3.4 GiB	3.3 GiB	-2.4%
Query 18	7.6 GiB	7.3 GiB	-3.0%
Query 19	0 B	0 B	0.0%
Query 20	104 B	104 B	+0.0%
Query 21	1.3 MiB	1.6 MiB	+16.8%
Query 22	2.4 MiB	2.6 MiB	+8.6%
Query 23	12.7 MiB	12.9 MiB	+1.9%
Query 24	6.1 MiB	6.9 MiB	+12.7%
Query 25	6.9 MiB	8.1 MiB	+17.6%
Query 26	7.1 MiB	7.9 MiB	+12.5%
Query 27	2.2 MiB	2.5 MiB	+9.7%
Query 28	1.9 GiB	2.0 GiB	+2.4%
Query 29	624 B	624 B	+0.0%
Query 30	639.4 MiB	692.6 MiB	+8.3%
Query 31	1.6 GiB	1.5 GiB	-5.9%
Query 32	7.1 GiB	7.2 GiB	+0.6%
Query 33	12.7 GiB	12.7 GiB	+0.2%
Query 34	13.8 GiB	12.7 GiB	-7.7%
Query 35	675.3 MiB	669.7 MiB	-0.8%
Query 36	474.1 MiB	475.6 MiB	+0.3%
Query 37	8.1 MiB	8.6 MiB	+6.2%
Query 38	5.3 MiB	5.3 MiB	+0.0%
Query 39	450.8 MiB	478.9 MiB	+6.2%
Query 40	2.3 MiB	2.3 MiB	-3.2%
Query 41	3.1 MiB	3.1 MiB	+0.0%
Query 42	1.7 MiB	1.8 MiB	+6.1%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
clickbench_partitioned	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	13.8 GiB	16.0 GiB	2.2 GiB	1.2×
clickbench_partitioned	changed (`e1aa5f4bed1c2880b1043b0d87502e5c984fa9cc`)	12.7 GiB	14.1 GiB	1.4 GiB	1.1×

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	410.1s
Peak memory	16.0 GiB
Avg memory	6.6 GiB
CPU user	3295.8s
CPU sys	241.7s
Peak spill	12.2 GiB

clickbench_partitioned — branch

Metric	Value
Wall time	380.1s
Peak memory	14.1 GiB
Avg memory	6.3 GiB
CPU user	3285.7s
CPU sys	203.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-10-27T20:46:27Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃        HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │     1.24 ms │                                    1.23 ms │     no change │
│ QQuery 1  │    11.39 ms │                                   11.48 ms │     no change │
│ QQuery 2  │    36.00 ms │                                   36.24 ms │     no change │
│ QQuery 3  │    30.84 ms │                                   32.19 ms │     no change │
│ QQuery 4  │   250.17 ms │                                  259.56 ms │     no change │
│ QQuery 5  │  1554.87 ms │                                 1688.98 ms │  1.09x slower │
│ QQuery 6  │     1.35 ms │                                    1.29 ms │     no change │
│ QQuery 7  │    13.48 ms │                                   12.89 ms │     no change │
│ QQuery 8  │   369.87 ms │                                  370.98 ms │     no change │
│ QQuery 9  │   537.62 ms │                                  535.14 ms │     no change │
│ QQuery 10 │   204.04 ms │                                  200.65 ms │     no change │
│ QQuery 11 │   211.91 ms │                                  213.47 ms │     no change │
│ QQuery 12 │  1629.48 ms │                                 1653.94 ms │     no change │
│ QQuery 13 │  2421.02 ms │                                 2102.30 ms │ +1.15x faster │
│ QQuery 14 │  1766.15 ms │                                 1771.41 ms │     no change │
│ QQuery 15 │   324.18 ms │                                  340.62 ms │  1.05x slower │
│ QQuery 16 │  3813.31 ms │                                 3929.59 ms │     no change │
│ QQuery 17 │  3958.39 ms │                                 4102.62 ms │     no change │
│ QQuery 18 │  6952.31 ms │                                 7161.41 ms │     no change │
│ QQuery 19 │    27.38 ms │                                   26.61 ms │     no change │
│ QQuery 20 │  1065.79 ms │                                 1071.70 ms │     no change │
│ QQuery 21 │  1391.73 ms │                                 1413.93 ms │     no change │
│ QQuery 22 │  3262.20 ms │                                 3306.94 ms │     no change │
│ QQuery 23 │  9309.46 ms │                                 9508.51 ms │     no change │
│ QQuery 24 │   101.84 ms │                                   97.65 ms │     no change │
│ QQuery 25 │   353.30 ms │                                  353.21 ms │     no change │
│ QQuery 26 │    96.95 ms │                                   97.87 ms │     no change │
│ QQuery 27 │  1175.64 ms │                                 1175.20 ms │     no change │
│ QQuery 28 │  2818.59 ms │                                 2922.48 ms │     no change │
│ QQuery 29 │    40.14 ms │                                   40.73 ms │     no change │
│ QQuery 30 │   554.76 ms │                                  578.14 ms │     no change │
│ QQuery 31 │   556.13 ms │                                  548.69 ms │     no change │
│ QQuery 32 │  1132.26 ms │                                 1098.89 ms │     no change │
│ QQuery 33 │ 14715.69 ms │                                14171.04 ms │     no change │
│ QQuery 34 │ 15011.51 ms │                                14579.01 ms │     no change │
│ QQuery 35 │   344.75 ms │                                  342.47 ms │     no change │
│ QQuery 36 │   192.27 ms │                                  191.39 ms │     no change │
│ QQuery 37 │    98.50 ms │                                  104.12 ms │  1.06x slower │
│ QQuery 38 │    85.78 ms │                                   78.85 ms │ +1.09x faster │
│ QQuery 39 │   434.83 ms │                                  400.76 ms │ +1.09x faster │
│ QQuery 40 │    15.97 ms │                                   15.10 ms │ +1.06x faster │
│ QQuery 41 │    14.85 ms │                                   14.49 ms │     no change │
│ QQuery 42 │    14.43 ms │                                   14.46 ms │     no change │
└───────────┴─────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 76902.36ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 76578.25ms │
│ Average Time (HEAD)                                       │  1788.43ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  1780.89ms │
│ Queries Faster                                            │          4 │
│ Queries Slower                                            │          3 │
│ Queries with No Change                                    │         36 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                      HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │              1.24 / 4.56 ±6.42 / 17.40 ms │               1.23 / 4.81 ±6.62 / 18.04 ms │  1.06x slower │
│ QQuery 1  │            11.39 / 11.54 ±0.13 / 11.78 ms │             11.48 / 11.67 ±0.16 / 11.86 ms │     no change │
│ QQuery 2  │            36.00 / 36.43 ±0.41 / 37.04 ms │             36.24 / 36.55 ±0.30 / 37.10 ms │     no change │
│ QQuery 3  │            30.84 / 31.73 ±0.60 / 32.54 ms │             32.19 / 32.69 ±0.51 / 33.67 ms │     no change │
│ QQuery 4  │         250.17 / 267.15 ±9.92 / 281.00 ms │          259.56 / 270.29 ±6.57 / 279.89 ms │     no change │
│ QQuery 5  │     1554.87 / 1667.36 ±72.28 / 1755.81 ms │      1688.98 / 1703.75 ±14.27 / 1730.86 ms │     no change │
│ QQuery 6  │               1.35 / 1.61 ±0.40 / 2.42 ms │                1.29 / 1.43 ±0.22 / 1.87 ms │ +1.13x faster │
│ QQuery 7  │            13.48 / 17.86 ±8.13 / 34.10 ms │             12.89 / 13.13 ±0.13 / 13.26 ms │ +1.36x faster │
│ QQuery 8  │        369.87 / 382.96 ±11.37 / 403.44 ms │         370.98 / 385.50 ±13.16 / 408.54 ms │     no change │
│ QQuery 9  │         537.62 / 544.29 ±5.20 / 549.45 ms │          535.14 / 542.94 ±7.16 / 552.17 ms │     no change │
│ QQuery 10 │         204.04 / 206.76 ±1.59 / 208.70 ms │          200.65 / 205.63 ±5.29 / 215.08 ms │     no change │
│ QQuery 11 │         211.91 / 213.64 ±1.18 / 214.98 ms │          213.47 / 214.70 ±1.12 / 216.20 ms │     no change │
│ QQuery 12 │     1629.48 / 1741.26 ±73.26 / 1851.64 ms │      1653.94 / 1708.64 ±47.88 / 1769.51 ms │     no change │
│ QQuery 13 │    2421.02 / 2702.60 ±271.76 / 3037.20 ms │     2102.30 / 2479.87 ±334.68 / 3094.71 ms │ +1.09x faster │
│ QQuery 14 │     1766.15 / 1834.88 ±49.09 / 1899.70 ms │      1771.41 / 1872.15 ±51.11 / 1910.89 ms │     no change │
│ QQuery 15 │        324.18 / 344.26 ±18.20 / 374.68 ms │         340.62 / 369.46 ±20.68 / 396.33 ms │  1.07x slower │
│ QQuery 16 │    3813.31 / 3938.51 ±105.32 / 4116.87 ms │     3929.59 / 4064.22 ±114.89 / 4268.20 ms │     no change │
│ QQuery 17 │    3958.39 / 4184.28 ±129.41 / 4321.73 ms │      4102.62 / 4192.06 ±82.27 / 4334.57 ms │     no change │
│ QQuery 18 │    6952.31 / 7665.23 ±419.46 / 8162.61 ms │     7161.41 / 7425.32 ±219.89 / 7766.80 ms │     no change │
│ QQuery 19 │            27.38 / 32.25 ±5.75 / 41.77 ms │             26.61 / 27.48 ±0.98 / 29.17 ms │ +1.17x faster │
│ QQuery 20 │     1065.79 / 1088.95 ±15.37 / 1104.44 ms │      1071.70 / 1085.88 ±10.18 / 1102.78 ms │     no change │
│ QQuery 21 │     1391.73 / 1406.13 ±14.66 / 1430.25 ms │      1413.93 / 1430.49 ±17.85 / 1463.07 ms │     no change │
│ QQuery 22 │     3262.20 / 3389.02 ±93.77 / 3534.14 ms │      3306.94 / 3373.73 ±50.16 / 3461.65 ms │     no change │
│ QQuery 23 │    9309.46 / 9567.68 ±193.63 / 9804.41 ms │     9508.51 / 9646.63 ±178.94 / 9988.42 ms │     no change │
│ QQuery 24 │        101.84 / 149.15 ±77.08 / 302.52 ms │            97.65 / 99.10 ±1.04 / 100.72 ms │ +1.51x faster │
│ QQuery 25 │         353.30 / 356.51 ±1.89 / 358.44 ms │          353.21 / 357.76 ±4.99 / 366.61 ms │     no change │
│ QQuery 26 │           96.95 / 99.40 ±1.56 / 101.75 ms │             97.87 / 98.57 ±0.53 / 99.15 ms │     no change │
│ QQuery 27 │     1175.64 / 1205.67 ±21.36 / 1236.82 ms │      1175.20 / 1212.64 ±50.29 / 1310.46 ms │     no change │
│ QQuery 28 │     2818.59 / 2922.65 ±53.25 / 2957.81 ms │      2922.48 / 2945.87 ±31.62 / 3008.16 ms │     no change │
│ QQuery 29 │            40.14 / 43.01 ±4.71 / 52.36 ms │             40.73 / 40.98 ±0.21 / 41.33 ms │     no change │
│ QQuery 30 │        554.76 / 567.63 ±11.79 / 589.53 ms │         578.14 / 588.15 ±13.28 / 614.22 ms │     no change │
│ QQuery 31 │        556.13 / 573.35 ±14.56 / 600.31 ms │         548.69 / 567.49 ±11.38 / 583.75 ms │     no change │
│ QQuery 32 │     1132.26 / 1188.28 ±57.99 / 1298.51 ms │      1098.89 / 1218.91 ±63.09 / 1279.77 ms │     no change │
│ QQuery 33 │ 14715.69 / 15022.68 ±347.44 / 15683.19 ms │ 14171.04 / 15250.71 ±1026.77 / 16989.54 ms │     no change │
│ QQuery 34 │ 15011.51 / 15685.87 ±539.53 / 16206.31 ms │  14579.01 / 15472.00 ±575.12 / 16350.90 ms │     no change │
│ QQuery 35 │        344.75 / 407.59 ±80.14 / 565.99 ms │         342.47 / 384.69 ±67.82 / 519.70 ms │ +1.06x faster │
│ QQuery 36 │        192.27 / 221.76 ±25.01 / 256.30 ms │          191.39 / 198.98 ±6.05 / 209.81 ms │ +1.11x faster │
│ QQuery 37 │         98.50 / 105.27 ±10.59 / 126.34 ms │          104.12 / 110.47 ±8.61 / 127.21 ms │     no change │
│ QQuery 38 │            85.78 / 87.93 ±1.27 / 89.40 ms │             78.85 / 84.68 ±3.66 / 88.79 ms │     no change │
│ QQuery 39 │        434.83 / 451.44 ±12.53 / 466.06 ms │         400.76 / 420.86 ±15.08 / 443.28 ms │ +1.07x faster │
│ QQuery 40 │            15.97 / 16.82 ±0.51 / 17.56 ms │             15.10 / 15.51 ±0.24 / 15.84 ms │ +1.08x faster │
│ QQuery 41 │            14.85 / 17.84 ±4.82 / 27.40 ms │             14.49 / 18.97 ±5.43 / 27.47 ms │  1.06x slower │
│ QQuery 42 │            14.43 / 14.65 ±0.17 / 14.87 ms │             14.46 / 14.86 ±0.36 / 15.34 ms │     no change │
└───────────┴───────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 80418.43ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 80200.18ms │
│ Average Time (HEAD)                                       │  1870.20ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  1865.12ms │
│ Queries Faster                                            │          9 │
│ Queries Slower                                            │          3 │
│ Queries with No Change                                    │         31 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: d49dd6d5614860d21328eb63633fd9d8e08fbd37

clickbench_partitioned

Query	Base	Changed	Change
Query 0	0 B	0 B	0.0%
Query 1	104 B	104 B	+0.0%
Query 2	936 B	936 B	+0.0%
Query 3	312 B	312 B	+0.0%
Query 4	831.8 MiB	833.1 MiB	+0.2%
Query 5	1.4 GiB	1.4 GiB	-0.2%
Query 6	0 B	0 B	0.0%
Query 7	50.3 MiB	50.3 MiB	-0.0%
Query 8	944.3 MiB	966.1 MiB	+2.3%
Query 9	552.0 MiB	594.2 MiB	+7.6%
Query 10	70.9 MiB	75.1 MiB	+5.9%
Query 11	76.6 MiB	78.6 MiB	+2.6%
Query 12	1.6 GiB	1.6 GiB	+0.4%
Query 13	2.0 GiB	2.0 GiB	+0.0%
Query 14	1.9 GiB	1.9 GiB	-1.1%
Query 15	1.2 GiB	1.2 GiB	-2.8%
Query 16	3.3 GiB	3.4 GiB	+2.1%
Query 17	3.3 GiB	3.5 GiB	+5.0%
Query 18	7.4 GiB	7.4 GiB	+0.3%
Query 19	0 B	0 B	0.0%
Query 20	104 B	104 B	+0.0%
Query 21	1.5 MiB	1.5 MiB	-0.6%
Query 22	2.5 MiB	2.4 MiB	-4.8%
Query 23	12.9 MiB	12.1 MiB	-5.9%
Query 24	6.9 MiB	6.9 MiB	+0.1%
Query 25	8.1 MiB	8.1 MiB	-0.0%
Query 26	7.8 MiB	8.2 MiB	+5.4%
Query 27	2.7 MiB	2.5 MiB	-8.2%
Query 28	2.0 GiB	2.1 GiB	+5.9%
Query 29	624 B	624 B	+0.0%
Query 30	653.5 MiB	682.9 MiB	+4.5%
Query 31	1.5 GiB	1.5 GiB	-0.0%
Query 32	7.2 GiB	7.1 GiB	-0.1%
Query 33	12.7 GiB	12.7 GiB	-0.1%
Query 34	13.3 GiB	12.7 GiB	-4.0%
Query 35	668.7 MiB	678.7 MiB	+1.5%
Query 36	474.1 MiB	475.6 MiB	+0.3%
Query 37	8.6 MiB	8.9 MiB	+3.2%
Query 38	5.3 MiB	5.0 MiB	-4.9%
Query 39	460.4 MiB	478.9 MiB	+4.0%
Query 40	2.3 MiB	2.3 MiB	+0.7%
Query 41	3.1 MiB	3.1 MiB	+0.0%
Query 42	1.8 MiB	1.7 MiB	-6.6%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
clickbench_partitioned	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	13.3 GiB	13.0 GiB	−285.9 MiB	1.0×
clickbench_partitioned	changed (`d49dd6d5614860d21328eb63633fd9d8e08fbd37`)	12.7 GiB	13.8 GiB	1.0 GiB	1.1×

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	405.1s
Peak memory	13.0 GiB
Avg memory	6.0 GiB
CPU user	3486.1s
CPU sys	221.4s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	405.1s
Peak memory	13.8 GiB
Avg memory	6.2 GiB
CPU user	3471.2s
CPU sys	219.6s
Peak spill	0 B

File an issue against this benchmark runner

Rich-T-kid · 2026-10-27T20:47:54Z

run benchmark dictionary_group_values

env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

adriangbot · 2026-10-27T20:51:39Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6453176160-3969-rwqx7 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark dictionary_group_values
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T21:14:22Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing d49dd6d (d49dd6d) to 4ede583 diff

Run configuration

run benchmark dictionary_group_values
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "d49dd6d5614860d21328eb63633fd9d8e08fbd37"

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

group                                                                             HEAD                                    rich-T-kid_aggregate-dictionary-hash-cache
-----                                                                             ----                                    ------------------------------------------
dict_intern_emit/intern_emit/size_65536_card_1000_null_0.00                       1.00    477.9±2.34µs 130.8 MElem/sec    1.03    491.0±1.92µs 127.3 MElem/sec
dict_intern_emit/intern_emit/size_65536_card_20_null_0.00                         1.00    411.0±2.88µs 152.1 MElem/sec    1.04    429.4±2.75µs 145.6 MElem/sec
dict_intern_emit/intern_emit/size_65536_card_300_null_0.00                        1.00    440.5±2.96µs 141.9 MElem/sec    1.03    453.9±4.36µs 137.7 MElem/sec
dict_intern_emit/intern_emit/size_65536_card_65536_null_0.00                      1.03      6.2±0.01ms 10.1 MElem/sec     1.00      6.0±0.01ms 10.4 MElem/sec
dict_intern_emit/intern_emit/size_65536_card_75_null_0.00                         1.00    433.8±2.49µs 144.1 MElem/sec    1.02    443.4±2.78µs 141.0 MElem/sec
dict_intern_emit/intern_emit/size_8192_card_1000_null_0.00                        1.01    112.3±0.71µs 69.6 MElem/sec     1.00    111.4±5.95µs 70.1 MElem/sec
dict_intern_emit/intern_emit/size_8192_card_20_null_0.00                          1.00     55.2±0.40µs 141.5 MElem/sec    1.04     57.4±0.24µs 136.1 MElem/sec
dict_intern_emit/intern_emit/size_8192_card_300_null_0.00                         1.00     72.2±0.54µs 108.2 MElem/sec    1.01     73.2±0.32µs 106.7 MElem/sec
dict_intern_emit/intern_emit/size_8192_card_75_null_0.00                          1.00     60.9±0.44µs 128.3 MElem/sec    1.02     62.0±0.37µs 125.9 MElem/sec
dict_intern_emit/intern_emit/size_8192_card_8192_null_0.00                        1.03    529.7±1.96µs 14.7 MElem/sec     1.00    512.4±2.55µs 15.2 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_65536_card_1000_null_0.10     1.00      2.3±0.01ms 106.6 MElem/sec    1.02      2.4±0.01ms 105.0 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_65536_card_20_null_0.10       1.00      2.2±0.01ms 114.8 MElem/sec    1.02      2.2±0.01ms 112.3 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_65536_card_300_null_0.10      1.00      2.3±0.01ms 109.8 MElem/sec    1.01      2.3±0.00ms 108.3 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_65536_card_65536_null_0.10    1.05     19.1±0.09ms 13.1 MElem/sec     1.00     18.2±0.07ms 13.7 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_65536_card_75_null_0.10       1.00      2.3±0.01ms 110.8 MElem/sec    1.01      2.3±0.01ms 109.3 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_8192_card_1000_null_0.10      1.00    400.5±1.55µs 78.0 MElem/sec     1.00    402.2±2.05µs 77.7 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_8192_card_20_null_0.10        1.00    263.9±1.99µs 118.4 MElem/sec    1.03    270.9±1.29µs 115.4 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_8192_card_300_null_0.10       1.00    308.6±2.88µs 101.3 MElem/sec    1.01    311.8±1.96µs 100.2 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_8192_card_75_null_0.10        1.05  298.4±189.04µs 104.7 MElem/sec    1.00    285.1±1.42µs 109.6 MElem/sec
dict_repeated_intern_emit/repeated_intern_emit/size_8192_card_8192_null_0.10      1.03   1306.9±7.76µs 23.9 MElem/sec     1.00   1268.0±3.25µs 24.6 MElem/sec
dict_scalar_append_equal/scalar_append_equal/size_65536_card_1000_null_0.10       1.00   1360.3±3.99µs 45.9 MElem/sec     1.04   1421.3±8.39µs 44.0 MElem/sec
dict_scalar_append_equal/scalar_append_equal/size_65536_card_20_null_0.10         1.01   1269.3±3.30µs 49.2 MElem/sec     1.00   1262.3±9.32µs 49.5 MElem/sec
dict_scalar_append_equal/scalar_append_equal/size_65536_card_300_null_0.10        1.00   1266.1±3.76µs 49.4 MElem/sec     1.02   1294.7±8.27µs 48.3 MElem/sec
dict_scalar_append_equal/scalar_append_equal/size_65536_card_65536_null_0.10      1.01      7.6±0.02ms  8.3 MElem/sec     1.00      7.5±0.02ms  8.4 MElem/sec
dict_scalar_append_equal/scalar_append_equal/size_65536_card_75_null_0.10         1.00   1251.9±3.12µs 49.9 MElem/sec     1.06   1321.1±8.66µs 47.3 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_100000                    1.48     18.0±0.07ms 13.9 MElem/sec     1.00     12.2±0.06ms 20.4 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_32768                     1.32      8.1±0.01ms 31.0 MElem/sec     1.00      6.1±0.01ms 40.9 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_500000                    1.86     60.7±1.55ms  4.1 MElem/sec     1.00     32.6±1.49ms  7.7 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_65536                     1.44     14.3±0.03ms 17.5 MElem/sec     1.00      9.9±0.05ms 25.2 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_8192                      1.20      2.8±0.01ms 88.7 MElem/sec     1.00      2.3±0.01ms 106.6 MElem/sec
dict_take_n/take_n/size_65536_card_1000_null_0.10                                 1.00      2.5±0.01ms 100.5 MElem/sec    1.02      2.5±0.01ms 98.8 MElem/sec
dict_take_n/take_n/size_65536_card_20_null_0.10                                   1.00      2.2±0.01ms 116.2 MElem/sec    1.03      2.2±0.01ms 113.2 MElem/sec
dict_take_n/take_n/size_65536_card_300_null_0.10                                  1.00      2.3±0.01ms 109.1 MElem/sec    1.02      2.3±0.01ms 107.0 MElem/sec
dict_take_n/take_n/size_65536_card_65536_null_0.10                                1.03     26.1±0.04ms  9.6 MElem/sec     1.00     25.4±0.08ms  9.8 MElem/sec
dict_take_n/take_n/size_65536_card_75_null_0.10                                   1.00      2.3±0.01ms 111.1 MElem/sec    1.01      2.3±0.01ms 110.3 MElem/sec

Resource Usage

dictionary_group_values — base (merge-base)

Metric	Value
Wall time	675.2s
Peak memory	900.7 MiB
Avg memory	60.5 MiB
CPU user	408.6s
CPU sys	23.3s
Peak spill	0 B

dictionary_group_values — branch

Metric	Value
Wall time	675.1s
Peak memory	949.8 MiB
Avg memory	69.6 MiB
CPU user	416.7s
CPU sys	18.5s
Peak spill	0 B

File an issue against this benchmark runner

Rich-T-kid · 2026-10-27T22:28:26Z

+    fn create_hashes(&mut self, cols: &[ArrayRef], hashes: &mut [u64]) -> Result<()> {
+        if !self.has_dictionary {
+            create_hashes(cols, &self.random_state, hashes)?;


this gets re-used across all 3 PR's I think we should move it up into the hash util PR

Rich-T-kid · 2026-10-27T23:10:06Z

run benchmark dictionary_group_values

env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

Rich-T-kid · 2026-10-27T23:10:07Z

run benchmark tpch tpcds clickbench_partitioned

env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

adriangbot · 2026-10-27T23:13:07Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6455848174-3972-xhsnq 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 0044519 (0044519) to 4ede583 diff

Run configuration

run benchmark dictionary_group_values
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T23:13:54Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6455848444-3973-sqhb6 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 0044519 (0044519) to 4ede583 diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T23:14:02Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6455848444-3974-r4cll 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 0044519 (0044519) to 4ede583 diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T23:14:57Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6455848444-3975-2tqtv 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 0044519 (0044519) to 4ede583 diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-10-27T23:20:40Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing 0044519 (0044519) to 4ede583 diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃      HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │  40.85 ms │                                   40.83 ms │ no change │
│ QQuery 2  │  19.18 ms │                                   19.03 ms │ no change │
│ QQuery 3  │  28.77 ms │                                   28.76 ms │ no change │
│ QQuery 4  │  17.57 ms │                                   17.58 ms │ no change │
│ QQuery 5  │  36.13 ms │                                   36.05 ms │ no change │
│ QQuery 6  │  14.09 ms │                                   13.92 ms │ no change │
│ QQuery 7  │  42.62 ms │                                   42.94 ms │ no change │
│ QQuery 8  │  41.94 ms │                                   41.72 ms │ no change │
│ QQuery 9  │  52.06 ms │                                   51.86 ms │ no change │
│ QQuery 10 │  60.40 ms │                                   59.50 ms │ no change │
│ QQuery 11 │  14.07 ms │                                   14.12 ms │ no change │
│ QQuery 12 │  21.89 ms │                                   21.48 ms │ no change │
│ QQuery 13 │  48.25 ms │                                   48.79 ms │ no change │
│ QQuery 14 │  24.65 ms │                                   24.61 ms │ no change │
│ QQuery 15 │  31.74 ms │                                   31.94 ms │ no change │
│ QQuery 16 │  13.65 ms │                                   13.84 ms │ no change │
│ QQuery 17 │  79.35 ms │                                   77.76 ms │ no change │
│ QQuery 18 │ 118.15 ms │                                  117.09 ms │ no change │
│ QQuery 19 │  30.82 ms │                                   30.48 ms │ no change │
│ QQuery 20 │  34.07 ms │                                   34.44 ms │ no change │
│ QQuery 21 │  54.90 ms │                                   55.75 ms │ no change │
│ QQuery 22 │  16.19 ms │                                   16.25 ms │ no change │
└───────────┴───────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 841.34ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 838.72ms │
│ Average Time (HEAD)                                       │  38.24ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  38.12ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                              HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │    40.85 / 42.02 ±1.36 / 43.94 ms │             40.83 / 42.70 ±1.45 / 44.47 ms │ no change │
│ QQuery 2  │    19.18 / 19.58 ±0.48 / 20.44 ms │             19.03 / 19.25 ±0.25 / 19.69 ms │ no change │
│ QQuery 3  │    28.77 / 29.59 ±1.21 / 31.99 ms │             28.76 / 28.95 ±0.12 / 29.11 ms │ no change │
│ QQuery 4  │    17.57 / 17.67 ±0.10 / 17.87 ms │             17.58 / 17.95 ±0.62 / 19.19 ms │ no change │
│ QQuery 5  │    36.13 / 36.45 ±0.24 / 36.79 ms │             36.05 / 36.39 ±0.26 / 36.68 ms │ no change │
│ QQuery 6  │    14.09 / 14.72 ±0.72 / 16.07 ms │             13.92 / 14.18 ±0.38 / 14.94 ms │ no change │
│ QQuery 7  │    42.62 / 45.12 ±1.41 / 46.61 ms │             42.94 / 44.78 ±1.31 / 46.43 ms │ no change │
│ QQuery 8  │    41.94 / 42.15 ±0.18 / 42.43 ms │             41.72 / 41.85 ±0.15 / 42.13 ms │ no change │
│ QQuery 9  │    52.06 / 53.00 ±0.98 / 54.51 ms │             51.86 / 52.92 ±0.59 / 53.66 ms │ no change │
│ QQuery 10 │    60.40 / 61.24 ±0.66 / 62.30 ms │             59.50 / 59.60 ±0.12 / 59.82 ms │ no change │
│ QQuery 11 │    14.07 / 14.26 ±0.12 / 14.38 ms │             14.12 / 14.25 ±0.14 / 14.51 ms │ no change │
│ QQuery 12 │    21.89 / 22.00 ±0.13 / 22.21 ms │             21.48 / 21.78 ±0.19 / 22.06 ms │ no change │
│ QQuery 13 │    48.25 / 49.49 ±1.61 / 52.68 ms │             48.79 / 49.90 ±1.78 / 53.45 ms │ no change │
│ QQuery 14 │    24.65 / 24.72 ±0.06 / 24.80 ms │             24.61 / 24.81 ±0.12 / 24.94 ms │ no change │
│ QQuery 15 │    31.74 / 32.13 ±0.38 / 32.79 ms │             31.94 / 32.56 ±0.95 / 34.42 ms │ no change │
│ QQuery 16 │    13.65 / 13.79 ±0.08 / 13.88 ms │             13.84 / 14.01 ±0.09 / 14.11 ms │ no change │
│ QQuery 17 │    79.35 / 80.70 ±1.27 / 82.89 ms │             77.76 / 79.80 ±1.94 / 82.34 ms │ no change │
│ QQuery 18 │ 118.15 / 122.22 ±6.81 / 135.78 ms │          117.09 / 124.20 ±7.55 / 138.47 ms │ no change │
│ QQuery 19 │    30.82 / 31.02 ±0.15 / 31.23 ms │             30.48 / 30.74 ±0.30 / 31.30 ms │ no change │
│ QQuery 20 │    34.07 / 34.72 ±0.34 / 35.04 ms │             34.44 / 34.51 ±0.05 / 34.58 ms │ no change │
│ QQuery 21 │    54.90 / 58.71 ±6.41 / 71.48 ms │             55.75 / 57.74 ±1.59 / 60.55 ms │ no change │
│ QQuery 22 │    16.19 / 16.57 ±0.21 / 16.84 ms │             16.25 / 16.44 ±0.15 / 16.62 ms │ no change │
└───────────┴───────────────────────────────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 861.88ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 859.31ms │
│ Average Time (HEAD)                                       │  39.18ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  39.06ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: 0044519d3b609ae6fbb0cfc59500b322ea368d39

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	20.1 MiB	30.1 MiB	+49.8%
Query 2	23.1 MiB	23.1 MiB	+0.0%
Query 3	11.5 MiB	11.5 MiB	+0.0%
Query 4	22.2 MiB	32.2 MiB	+45.0%
Query 5	42.8 MiB	42.8 MiB	+0.0%
Query 6	832 B	832 B	+0.0%
Query 7	107.6 MiB	107.6 MiB	+0.0%
Query 8	35.5 MiB	33.1 MiB	-6.6%
Query 9	155.6 MiB	155.6 MiB	+0.0%
Query 10	62.4 MiB	66.3 MiB	+6.3%
Query 11	61.4 MiB	71.9 MiB	+17.2%
Query 12	20.1 MiB	20.0 MiB	-0.3%
Query 13	62.3 MiB	43.0 MiB	-31.1%
Query 14	7.7 MiB	7.7 MiB	+0.0%
Query 15	12.8 MiB	12.8 MiB	-0.0%
Query 16	43.1 MiB	33.8 MiB	-21.7%
Query 17	150.3 MiB	150.3 MiB	+0.0%
Query 18	2.7 GiB	2.7 GiB	-0.0%
Query 19	228.6 KiB	228.6 KiB	+0.0%
Query 20	58.6 MiB	59.2 MiB	+1.1%
Query 21	37.7 MiB	37.7 MiB	+0.0%
Query 22	22.1 MiB	31.8 MiB	+43.4%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	2.7 GiB	2.0 GiB	−677.4 MiB	0.8×
tpch	changed (`0044519d3b609ae6fbb0cfc59500b322ea368d39`)	2.7 GiB	2.0 GiB	−666.5 MiB	0.8×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	2.0 GiB
Avg memory	807.8 MiB
CPU user	21.9s
CPU sys	1.6s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	2.0 GiB
Avg memory	813.6 MiB
CPU user	21.8s
CPU sys	1.5s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-10-27T23:22:05Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing 0044519 (0044519) to 4ede583 diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "4ede5833f68dad5bc8169492dedc5749fcc4b8d5"
changed:
  ref: "0044519d3b609ae6fbb0cfc59500b322ea368d39"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.28 ms │                                    5.30 ms │     no change │
│ QQuery 2  │   76.60 ms │                                   77.13 ms │     no change │
│ QQuery 3  │   28.91 ms │                                   28.63 ms │     no change │
│ QQuery 4  │  808.96 ms │                                  735.65 ms │ +1.10x faster │
│ QQuery 5  │   49.96 ms │                                   49.96 ms │     no change │
│ QQuery 6  │   33.25 ms │                                   33.15 ms │     no change │
│ QQuery 7  │   72.26 ms │                                   73.08 ms │     no change │
│ QQuery 8  │   37.83 ms │                                   38.13 ms │     no change │
│ QQuery 9  │   51.66 ms │                                   51.30 ms │     no change │
│ QQuery 10 │   61.32 ms │                                   61.37 ms │     no change │
│ QQuery 11 │  463.01 ms │                                  421.54 ms │ +1.10x faster │
│ QQuery 12 │   25.70 ms │                                   26.03 ms │     no change │
│ QQuery 13 │  111.29 ms │                                  112.07 ms │     no change │
│ QQuery 14 │  385.25 ms │                                  385.72 ms │     no change │
│ QQuery 15 │   51.31 ms │                                   50.96 ms │     no change │
│ QQuery 16 │    6.16 ms │                                    6.32 ms │     no change │
│ QQuery 17 │   79.69 ms │                                   79.58 ms │     no change │
│ QQuery 18 │  105.60 ms │                                  105.42 ms │     no change │
│ QQuery 19 │   40.85 ms │                                   41.29 ms │     no change │
│ QQuery 20 │   30.88 ms │                                   31.52 ms │     no change │
│ QQuery 21 │   17.06 ms │                                   17.19 ms │     no change │
│ QQuery 22 │   70.99 ms │                                   61.20 ms │ +1.16x faster │
│ QQuery 23 │  300.87 ms │                                  301.28 ms │     no change │
│ QQuery 24 │  188.43 ms │                                  190.42 ms │     no change │
│ QQuery 25 │  107.42 ms │                                  105.53 ms │     no change │
│ QQuery 26 │   46.12 ms │                                   46.78 ms │     no change │
│ QQuery 27 │    5.90 ms │                                    5.98 ms │     no change │
│ QQuery 28 │   56.64 ms │                                   60.31 ms │  1.06x slower │
│ QQuery 29 │   95.87 ms │                                   94.06 ms │     no change │
│ QQuery 30 │   36.34 ms │                                   36.14 ms │     no change │
│ QQuery 31 │  110.42 ms │                                  111.11 ms │     no change │
│ QQuery 32 │   19.90 ms │                                   19.91 ms │     no change │
│ QQuery 33 │   37.43 ms │                                   37.44 ms │     no change │
│ QQuery 34 │    9.40 ms │                                    9.51 ms │     no change │
│ QQuery 35 │   70.45 ms │                                   71.69 ms │     no change │
│ QQuery 36 │    5.81 ms │                                    6.02 ms │     no change │
│ QQuery 37 │    7.15 ms │                                    7.34 ms │     no change │
│ QQuery 38 │   63.86 ms │                                   64.15 ms │     no change │
│ QQuery 39 │   71.21 ms │                                   71.92 ms │     no change │
│ QQuery 40 │   22.33 ms │                                   23.18 ms │     no change │
│ QQuery 41 │   13.79 ms │                                   13.50 ms │     no change │
│ QQuery 42 │   23.84 ms │                                   23.87 ms │     no change │
│ QQuery 43 │    5.10 ms │                                    4.91 ms │     no change │
│ QQuery 44 │    8.58 ms │                                    8.22 ms │     no change │
│ QQuery 45 │   34.68 ms │                                   34.67 ms │     no change │
│ QQuery 46 │   11.37 ms │                                   11.10 ms │     no change │
│ QQuery 47 │  201.54 ms │                                  202.26 ms │     no change │
│ QQuery 48 │   89.99 ms │                                   90.74 ms │     no change │
│ QQuery 49 │   69.24 ms │                                   69.82 ms │     no change │
│ QQuery 50 │   58.06 ms │                                   57.13 ms │     no change │
│ QQuery 51 │   90.44 ms │                                   92.39 ms │     no change │
│ QQuery 52 │   24.01 ms │                                   23.91 ms │     no change │
│ QQuery 53 │   29.21 ms │                                   29.05 ms │     no change │
│ QQuery 54 │   52.69 ms │                                   52.87 ms │     no change │
│ QQuery 55 │   23.39 ms │                                   23.43 ms │     no change │
│ QQuery 56 │   38.20 ms │                                   38.63 ms │     no change │
│ QQuery 57 │  151.73 ms │                                  150.67 ms │     no change │
│ QQuery 58 │  102.30 ms │                                  103.36 ms │     no change │
│ QQuery 59 │  106.34 ms │                                  106.93 ms │     no change │
│ QQuery 60 │   42.75 ms │                                   42.50 ms │     no change │
│ QQuery 61 │    9.47 ms │                                    9.43 ms │     no change │
│ QQuery 62 │   38.25 ms │                                   38.44 ms │     no change │
│ QQuery 63 │   29.26 ms │                                   29.37 ms │     no change │
│ QQuery 64 │  366.93 ms │                                  364.72 ms │     no change │
│ QQuery 65 │  132.26 ms │                                  128.56 ms │     no change │
│ QQuery 66 │   72.87 ms │                                   73.64 ms │     no change │
│ QQuery 67 │  226.95 ms │                                  215.75 ms │     no change │
│ QQuery 68 │   11.38 ms │                                   11.23 ms │     no change │
│ QQuery 69 │   56.33 ms │                                   56.54 ms │     no change │
│ QQuery 70 │   99.19 ms │                                   98.29 ms │     no change │
│ QQuery 71 │   34.59 ms │                                   34.85 ms │     no change │
│ QQuery 72 │ 1514.18 ms │                                 1553.28 ms │     no change │
│ QQuery 73 │    9.31 ms │                                    9.12 ms │     no change │
│ QQuery 74 │  241.08 ms │                                  232.19 ms │     no change │
│ QQuery 75 │  138.07 ms │                                  137.72 ms │     no change │
│ QQuery 76 │   33.79 ms │                                   34.11 ms │     no change │
│ QQuery 77 │   58.63 ms │                                   59.73 ms │     no change │
│ QQuery 78 │  158.60 ms │                                  160.51 ms │     no change │
│ QQuery 79 │   66.63 ms │                                   66.91 ms │     no change │
│ QQuery 80 │   94.10 ms │                                   94.86 ms │     no change │
│ QQuery 81 │   28.08 ms │                                   28.43 ms │     no change │
│ QQuery 82 │   17.08 ms │                                   17.14 ms │     no change │
│ QQuery 83 │   32.11 ms │                                   32.11 ms │     no change │
│ QQuery 84 │   30.64 ms │                                   31.17 ms │     no change │
│ QQuery 85 │   97.46 ms │                                   98.72 ms │     no change │
│ QQuery 86 │   24.61 ms │                                   24.63 ms │     no change │
│ QQuery 87 │   64.34 ms │                                   64.57 ms │     no change │
│ QQuery 88 │   59.07 ms │                                   58.35 ms │     no change │
│ QQuery 89 │   35.52 ms │                                   35.63 ms │     no change │
│ QQuery 90 │   16.04 ms │                                   16.11 ms │     no change │
│ QQuery 91 │   39.86 ms │                                   40.06 ms │     no change │
│ QQuery 92 │   28.56 ms │                                   28.70 ms │     no change │
│ QQuery 93 │   47.83 ms │                                   48.40 ms │     no change │
│ QQuery 94 │   37.40 ms │                                   37.48 ms │     no change │
│ QQuery 95 │   79.05 ms │                                   78.40 ms │     no change │
│ QQuery 96 │   23.69 ms │                                   23.76 ms │     no change │
│ QQuery 97 │   51.53 ms │                                   50.04 ms │     no change │
│ QQuery 98 │   38.29 ms │                                   37.34 ms │     no change │
│ QQuery 99 │   56.36 ms │                                   56.15 ms │     no change │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9246.00ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9151.66ms │
│ Average Time (HEAD)                                       │   93.39ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   92.44ms │
│ Queries Faster                                            │         3 │
│ Queries Slower                                            │         1 │
│ Queries with No Change                                    │        95 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.28 / 5.88 ±1.06 / 7.99 ms │                5.30 / 5.91 ±1.08 / 8.06 ms │     no change │
│ QQuery 2  │        76.60 / 77.00 ±0.31 / 77.52 ms │             77.13 / 77.42 ±0.15 / 77.58 ms │     no change │
│ QQuery 3  │        28.91 / 29.03 ±0.07 / 29.10 ms │             28.63 / 28.94 ±0.27 / 29.41 ms │     no change │
│ QQuery 4  │     808.96 / 818.02 ±9.17 / 833.98 ms │          735.65 / 740.30 ±3.60 / 745.39 ms │ +1.10x faster │
│ QQuery 5  │        49.96 / 50.68 ±0.47 / 51.27 ms │             49.96 / 50.64 ±0.81 / 52.16 ms │     no change │
│ QQuery 6  │        33.25 / 35.60 ±3.72 / 42.98 ms │             33.15 / 33.49 ±0.23 / 33.87 ms │ +1.06x faster │
│ QQuery 7  │        72.26 / 73.54 ±1.15 / 75.69 ms │             73.08 / 73.30 ±0.19 / 73.67 ms │     no change │
│ QQuery 8  │        37.83 / 37.92 ±0.06 / 38.00 ms │             38.13 / 38.48 ±0.35 / 39.13 ms │     no change │
│ QQuery 9  │        51.66 / 53.42 ±0.94 / 54.24 ms │             51.30 / 53.80 ±1.77 / 56.46 ms │     no change │
│ QQuery 10 │        61.32 / 63.40 ±2.42 / 68.11 ms │             61.37 / 61.72 ±0.40 / 62.35 ms │     no change │
│ QQuery 11 │     463.01 / 469.74 ±4.66 / 475.00 ms │          421.54 / 429.43 ±6.50 / 438.90 ms │ +1.09x faster │
│ QQuery 12 │        25.70 / 26.01 ±0.27 / 26.38 ms │             26.03 / 27.80 ±2.15 / 31.85 ms │  1.07x slower │
│ QQuery 13 │     111.29 / 113.26 ±1.96 / 116.80 ms │          112.07 / 113.55 ±1.31 / 115.74 ms │     no change │
│ QQuery 14 │     385.25 / 387.84 ±2.47 / 391.77 ms │          385.72 / 389.69 ±4.25 / 396.80 ms │     no change │
│ QQuery 15 │        51.31 / 53.16 ±2.92 / 58.92 ms │             50.96 / 51.97 ±0.71 / 52.94 ms │     no change │
│ QQuery 16 │           6.16 / 6.31 ±0.25 / 6.82 ms │               6.32 / 7.45 ±1.73 / 10.89 ms │  1.18x slower │
│ QQuery 17 │        79.69 / 80.54 ±0.81 / 81.99 ms │             79.58 / 81.28 ±2.63 / 86.43 ms │     no change │
│ QQuery 18 │     105.60 / 107.33 ±2.18 / 111.63 ms │          105.42 / 105.92 ±0.60 / 107.08 ms │     no change │
│ QQuery 19 │        40.85 / 41.31 ±0.46 / 42.07 ms │             41.29 / 41.79 ±0.53 / 42.74 ms │     no change │
│ QQuery 20 │        30.88 / 31.48 ±0.31 / 31.71 ms │             31.52 / 31.82 ±0.16 / 31.97 ms │     no change │
│ QQuery 21 │        17.06 / 17.35 ±0.19 / 17.53 ms │             17.19 / 17.45 ±0.32 / 18.00 ms │     no change │
│ QQuery 22 │        70.99 / 72.67 ±1.13 / 74.13 ms │             61.20 / 61.94 ±0.62 / 62.78 ms │ +1.17x faster │
│ QQuery 23 │     300.87 / 306.09 ±5.34 / 315.64 ms │          301.28 / 305.38 ±3.34 / 311.31 ms │     no change │
│ QQuery 24 │     188.43 / 191.07 ±1.50 / 192.92 ms │          190.42 / 195.43 ±7.98 / 211.26 ms │     no change │
│ QQuery 25 │     107.42 / 108.39 ±0.93 / 110.05 ms │          105.53 / 107.99 ±2.45 / 112.58 ms │     no change │
│ QQuery 26 │        46.12 / 47.77 ±2.32 / 52.34 ms │             46.78 / 46.89 ±0.09 / 47.00 ms │     no change │
│ QQuery 27 │           5.90 / 6.00 ±0.15 / 6.30 ms │                5.98 / 6.15 ±0.16 / 6.43 ms │     no change │
│ QQuery 28 │        56.64 / 59.09 ±1.78 / 60.94 ms │             60.31 / 61.07 ±0.46 / 61.65 ms │     no change │
│ QQuery 29 │        95.87 / 97.00 ±0.93 / 97.99 ms │             94.06 / 95.34 ±1.54 / 98.26 ms │     no change │
│ QQuery 30 │        36.34 / 37.59 ±1.99 / 41.56 ms │             36.14 / 36.53 ±0.26 / 36.89 ms │     no change │
│ QQuery 31 │     110.42 / 111.73 ±1.58 / 114.76 ms │          111.11 / 112.47 ±0.88 / 113.55 ms │     no change │
│ QQuery 32 │        19.90 / 20.29 ±0.43 / 21.09 ms │             19.91 / 20.18 ±0.16 / 20.31 ms │     no change │
│ QQuery 33 │        37.43 / 38.00 ±0.60 / 39.10 ms │             37.44 / 38.56 ±0.70 / 39.35 ms │     no change │
│ QQuery 34 │           9.40 / 9.58 ±0.20 / 9.96 ms │               9.51 / 9.78 ±0.29 / 10.31 ms │     no change │
│ QQuery 35 │        70.45 / 71.06 ±0.47 / 71.80 ms │             71.69 / 72.64 ±1.11 / 74.79 ms │     no change │
│ QQuery 36 │           5.81 / 5.92 ±0.18 / 6.28 ms │                6.02 / 6.15 ±0.20 / 6.54 ms │     no change │
│ QQuery 37 │           7.15 / 7.32 ±0.14 / 7.55 ms │                7.34 / 7.73 ±0.34 / 8.16 ms │  1.06x slower │
│ QQuery 38 │        63.86 / 64.44 ±0.49 / 65.20 ms │             64.15 / 64.78 ±0.55 / 65.71 ms │     no change │
│ QQuery 39 │        71.21 / 71.82 ±0.46 / 72.45 ms │             71.92 / 72.88 ±0.61 / 73.70 ms │     no change │
│ QQuery 40 │        22.33 / 22.86 ±0.43 / 23.59 ms │             23.18 / 24.22 ±0.92 / 25.89 ms │  1.06x slower │
│ QQuery 41 │        13.79 / 14.07 ±0.21 / 14.38 ms │             13.50 / 14.55 ±1.65 / 17.85 ms │     no change │
│ QQuery 42 │        23.84 / 24.25 ±0.35 / 24.78 ms │             23.87 / 24.35 ±0.31 / 24.69 ms │     no change │
│ QQuery 43 │           5.10 / 5.29 ±0.21 / 5.69 ms │                4.91 / 5.06 ±0.26 / 5.59 ms │     no change │
│ QQuery 44 │           8.58 / 8.86 ±0.19 / 9.11 ms │                8.22 / 8.34 ±0.10 / 8.52 ms │ +1.06x faster │
│ QQuery 45 │        34.68 / 35.72 ±1.44 / 38.59 ms │             34.67 / 34.93 ±0.18 / 35.21 ms │     no change │
│ QQuery 46 │        11.37 / 11.56 ±0.11 / 11.68 ms │             11.10 / 11.35 ±0.26 / 11.83 ms │     no change │
│ QQuery 47 │     201.54 / 203.00 ±0.95 / 204.44 ms │          202.26 / 205.64 ±4.98 / 215.51 ms │     no change │
│ QQuery 48 │        89.99 / 92.61 ±1.64 / 95.13 ms │             90.74 / 91.56 ±0.65 / 92.44 ms │     no change │
│ QQuery 49 │        69.24 / 69.53 ±0.27 / 70.04 ms │             69.82 / 70.51 ±0.73 / 71.54 ms │     no change │
│ QQuery 50 │        58.06 / 59.46 ±1.66 / 62.24 ms │             57.13 / 57.83 ±0.42 / 58.40 ms │     no change │
│ QQuery 51 │       90.44 / 93.20 ±3.87 / 100.81 ms │            92.39 / 96.02 ±3.54 / 102.58 ms │     no change │
│ QQuery 52 │        24.01 / 24.14 ±0.13 / 24.33 ms │             23.91 / 24.15 ±0.23 / 24.55 ms │     no change │
│ QQuery 53 │        29.21 / 29.44 ±0.20 / 29.75 ms │             29.05 / 29.19 ±0.12 / 29.37 ms │     no change │
│ QQuery 54 │        52.69 / 53.21 ±0.61 / 54.36 ms │             52.87 / 53.37 ±0.40 / 54.08 ms │     no change │
│ QQuery 55 │        23.39 / 23.78 ±0.42 / 24.59 ms │             23.43 / 23.59 ±0.13 / 23.78 ms │     no change │
│ QQuery 56 │        38.20 / 38.54 ±0.37 / 39.25 ms │             38.63 / 39.86 ±1.74 / 43.29 ms │     no change │
│ QQuery 57 │     151.73 / 152.95 ±1.40 / 155.46 ms │          150.67 / 153.08 ±2.73 / 158.37 ms │     no change │
│ QQuery 58 │     102.30 / 104.10 ±1.46 / 106.33 ms │          103.36 / 104.95 ±1.43 / 106.96 ms │     no change │
│ QQuery 59 │     106.34 / 106.69 ±0.21 / 106.91 ms │          106.93 / 107.22 ±0.30 / 107.60 ms │     no change │
│ QQuery 60 │        42.75 / 43.66 ±0.99 / 45.54 ms │             42.50 / 42.88 ±0.39 / 43.54 ms │     no change │
│ QQuery 61 │          9.47 / 9.71 ±0.20 / 10.05 ms │                9.43 / 9.62 ±0.19 / 9.97 ms │     no change │
│ QQuery 62 │        38.25 / 38.50 ±0.24 / 38.91 ms │             38.44 / 39.58 ±1.70 / 42.93 ms │     no change │
│ QQuery 63 │        29.26 / 29.60 ±0.31 / 30.12 ms │             29.37 / 29.67 ±0.18 / 29.88 ms │     no change │
│ QQuery 64 │     366.93 / 368.60 ±1.64 / 370.72 ms │          364.72 / 369.54 ±4.53 / 375.10 ms │     no change │
│ QQuery 65 │     132.26 / 134.37 ±1.74 / 136.60 ms │          128.56 / 132.53 ±3.07 / 136.53 ms │     no change │
│ QQuery 66 │        72.87 / 74.67 ±2.48 / 79.59 ms │             73.64 / 74.58 ±1.08 / 76.38 ms │     no change │
│ QQuery 67 │     226.95 / 229.27 ±2.14 / 232.08 ms │          215.75 / 222.09 ±3.58 / 226.73 ms │     no change │
│ QQuery 68 │        11.38 / 11.57 ±0.18 / 11.88 ms │             11.23 / 11.64 ±0.45 / 12.52 ms │     no change │
│ QQuery 69 │        56.33 / 56.58 ±0.23 / 57.00 ms │             56.54 / 56.72 ±0.13 / 56.87 ms │     no change │
│ QQuery 70 │      99.19 / 101.20 ±2.28 / 104.98 ms │           98.29 / 101.21 ±3.36 / 107.13 ms │     no change │
│ QQuery 71 │        34.59 / 34.80 ±0.12 / 34.97 ms │             34.85 / 35.06 ±0.21 / 35.42 ms │     no change │
│ QQuery 72 │ 1514.18 / 1558.62 ±50.15 / 1654.49 ms │      1553.28 / 1654.09 ±88.21 / 1812.25 ms │  1.06x slower │
│ QQuery 73 │           9.31 / 9.50 ±0.22 / 9.91 ms │               9.12 / 9.52 ±0.58 / 10.68 ms │     no change │
│ QQuery 74 │     241.08 / 246.25 ±4.37 / 253.97 ms │          232.19 / 237.93 ±5.10 / 245.21 ms │     no change │
│ QQuery 75 │     138.07 / 139.67 ±1.29 / 141.44 ms │          137.72 / 140.56 ±2.97 / 146.15 ms │     no change │
│ QQuery 76 │        33.79 / 34.19 ±0.39 / 34.74 ms │             34.11 / 34.32 ±0.11 / 34.43 ms │     no change │
│ QQuery 77 │        58.63 / 58.85 ±0.15 / 59.07 ms │             59.73 / 61.26 ±2.56 / 66.38 ms │     no change │
│ QQuery 78 │     158.60 / 161.90 ±2.86 / 166.52 ms │          160.51 / 163.55 ±4.06 / 171.55 ms │     no change │
│ QQuery 79 │        66.63 / 69.32 ±3.19 / 75.45 ms │             66.91 / 67.18 ±0.26 / 67.64 ms │     no change │
│ QQuery 80 │        94.10 / 94.69 ±0.90 / 96.47 ms │             94.86 / 96.07 ±1.04 / 97.81 ms │     no change │
│ QQuery 81 │        28.08 / 28.20 ±0.16 / 28.52 ms │             28.43 / 28.72 ±0.21 / 28.96 ms │     no change │
│ QQuery 82 │        17.08 / 17.69 ±0.74 / 19.11 ms │             17.14 / 17.44 ±0.37 / 18.15 ms │     no change │
│ QQuery 83 │        32.11 / 32.33 ±0.19 / 32.60 ms │             32.11 / 34.50 ±3.66 / 41.76 ms │  1.07x slower │
│ QQuery 84 │        30.64 / 32.28 ±2.95 / 38.17 ms │             31.17 / 32.53 ±1.57 / 35.21 ms │     no change │
│ QQuery 85 │      97.46 / 100.17 ±2.89 / 105.77 ms │            98.72 / 99.64 ±0.88 / 100.79 ms │     no change │
│ QQuery 86 │        24.61 / 24.86 ±0.16 / 25.09 ms │             24.63 / 24.86 ±0.15 / 24.99 ms │     no change │
│ QQuery 87 │        64.34 / 65.96 ±1.98 / 69.73 ms │             64.57 / 66.76 ±3.09 / 72.87 ms │     no change │
│ QQuery 88 │        59.07 / 60.09 ±1.31 / 62.69 ms │             58.35 / 58.95 ±0.40 / 59.44 ms │     no change │
│ QQuery 89 │        35.52 / 35.99 ±0.44 / 36.72 ms │             35.63 / 35.80 ±0.14 / 35.93 ms │     no change │
│ QQuery 90 │        16.04 / 16.17 ±0.12 / 16.38 ms │             16.11 / 16.26 ±0.10 / 16.36 ms │     no change │
│ QQuery 91 │        39.86 / 40.73 ±0.73 / 41.93 ms │             40.06 / 40.31 ±0.22 / 40.69 ms │     no change │
│ QQuery 92 │        28.56 / 28.94 ±0.34 / 29.50 ms │             28.70 / 30.06 ±1.41 / 32.76 ms │     no change │
│ QQuery 93 │        47.83 / 49.13 ±1.14 / 51.21 ms │             48.40 / 49.92 ±1.90 / 53.59 ms │     no change │
│ QQuery 94 │        37.40 / 39.04 ±1.48 / 41.81 ms │             37.48 / 37.72 ±0.14 / 37.88 ms │     no change │
│ QQuery 95 │        79.05 / 79.70 ±0.73 / 81.07 ms │             78.40 / 78.99 ±0.47 / 79.73 ms │     no change │
│ QQuery 96 │        23.69 / 23.86 ±0.17 / 24.15 ms │             23.76 / 24.73 ±1.22 / 27.03 ms │     no change │
│ QQuery 97 │        51.53 / 52.43 ±0.81 / 53.59 ms │             50.04 / 52.64 ±2.26 / 56.29 ms │     no change │
│ QQuery 98 │        38.29 / 39.08 ±1.11 / 41.25 ms │             37.34 / 37.77 ±0.31 / 38.17 ms │     no change │
│ QQuery 99 │        56.36 / 56.84 ±0.39 / 57.44 ms │             56.15 / 56.57 ±0.36 / 57.12 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9406.91ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9377.60ms │
│ Average Time (HEAD)                                       │   95.02ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   94.72ms │
│ Queries Faster                                            │         5 │
│ Queries Slower                                            │         6 │
│ Queries with No Change                                    │        88 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 4ede5833f68dad5bc8169492dedc5749fcc4b8d5 | Changed: 0044519d3b609ae6fbb0cfc59500b322ea368d39

tpcds — tpcds_sf1

Query	Base	Changed	Change
Query 1	197.5 KiB	139.9 KiB	-29.2%
Query 2	71.8 MiB	80.9 MiB	+12.8%
Query 3	1.7 MiB	1.5 MiB	-11.2%
Query 4	79.7 MiB	79.7 MiB	+0.0%
Query 5	12.5 MiB	13.0 MiB	+4.2%
Query 6	4.4 MiB	4.4 MiB	+0.2%
Query 7	3.7 MiB	3.6 MiB	-1.6%
Query 8	2.1 MiB	2.1 MiB	+0.1%
Query 9	32.1 KiB	32.1 KiB	+0.0%
Query 10	2.0 MiB	2.0 MiB	+0.0%
Query 11	62.0 MiB	62.0 MiB	+0.0%
Query 12	114.4 MiB	114.4 MiB	-0.0%
Query 13	13.0 MiB	13.7 MiB	+5.1%
Query 14	170.9 MiB	170.9 MiB	+0.0%
Query 15	4.3 MiB	4.2 MiB	-1.8%
Query 16	118.6 KiB	118.6 KiB	+0.0%
Query 17	40.6 MiB	40.6 MiB	+0.0%
Query 18	21.2 MiB	21.2 MiB	+0.2%
Query 19	2.1 MiB	1.9 MiB	-9.1%
Query 20	124.3 MiB	114.6 MiB	-7.8%
Query 21	3.3 MiB	3.7 MiB	+11.0%
Query 22	61.8 MiB	63.4 MiB	+2.7%
Query 23	212.6 MiB	212.2 MiB	-0.2%
Query 24	38.7 MiB	38.7 MiB	+0.0%
Query 25	99.9 MiB	99.9 MiB	+0.0%
Query 26	3.5 MiB	3.9 MiB	+10.6%
Query 27	1.6 MiB	1.6 MiB	+0.0%
Query 28	1.5 MiB	1.3 MiB	-9.1%
Query 29	104.7 MiB	104.7 MiB	+0.0%
Query 30	3.8 MiB	3.7 MiB	-2.3%
Query 31	13.4 MiB	13.4 MiB	+0.0%
Query 32	2.5 MiB	2.7 MiB	+8.9%
Query 33	5.8 MiB	4.2 MiB	-27.1%
Query 34	6.0 MiB	6.0 MiB	+0.0%
Query 35	17.5 MiB	17.5 MiB	+0.0%
Query 36	10.3 MiB	10.3 MiB	+0.0%
Query 37	3.3 MiB	3.3 MiB	+0.0%
Query 38	10.8 MiB	10.5 MiB	-2.3%
Query 39	42.4 MiB	52.6 MiB	+23.9%
Query 40	12.2 MiB	12.2 MiB	+0.0%
Query 41	2.1 MiB	2.1 MiB	+0.1%
Query 42	1.3 MiB	1.5 MiB	+14.7%
Query 43	275.0 KiB	275.5 KiB	+0.2%
Query 44	233.2 KiB	233.2 KiB	+0.0%
Query 45	4.5 MiB	4.5 MiB	+0.0%
Query 46	6.2 MiB	6.2 MiB	+0.0%
Query 47	132.6 MiB	128.5 MiB	-3.1%
Query 48	11.8 MiB	11.8 MiB	+0.0%
Query 49	42.9 MiB	43.2 MiB	+0.6%
Query 50	16.9 MiB	17.7 MiB	+4.4%
Query 51	177.7 MiB	178.0 MiB	+0.2%
Query 52	1.5 MiB	1.3 MiB	-15.6%
Query 53	20.6 MiB	20.6 MiB	+0.0%
Query 54	2.1 MiB	2.1 MiB	+0.0%
Query 55	1.3 MiB	1.4 MiB	+11.5%
Query 56	3.0 MiB	3.2 MiB	+7.0%
Query 57	120.8 MiB	120.8 MiB	+0.0%
Query 58	4.4 MiB	4.2 MiB	-4.8%
Query 59	11.7 MiB	11.7 MiB	+0.0%
Query 60	7.2 MiB	7.2 MiB	+0.2%
Query 61	2.4 KiB	2.5 KiB	+3.3%
Query 62	3.7 MiB	3.6 MiB	-0.1%
Query 63	30.6 MiB	20.6 MiB	-32.7%
Query 64	84.7 MiB	82.9 MiB	-2.1%
Query 65	29.7 MiB	29.7 MiB	-0.0%
Query 66	14.9 MiB	14.9 MiB	+0.0%
Query 67	140.3 MiB	137.9 MiB	-1.8%
Query 68	6.2 MiB	6.2 MiB	+0.0%
Query 69	3.8 MiB	3.5 MiB	-7.6%
Query 70	44.0 MiB	44.0 MiB	-0.0%
Query 71	40.4 MiB	40.4 MiB	+0.1%
Query 72	97.2 MiB	97.2 MiB	+0.0%
Query 73	6.0 MiB	6.0 MiB	+0.0%
Query 74	34.6 MiB	34.6 MiB	+0.0%
Query 75	21.4 MiB	21.3 MiB	-0.3%
Query 76	7.9 MiB	7.9 MiB	+0.0%
Query 77	4.0 MiB	4.1 MiB	+2.0%
Query 78	111.1 MiB	109.9 MiB	-1.1%
Query 79	8.0 MiB	8.7 MiB	+8.1%
Query 80	41.7 MiB	41.6 MiB	-0.3%
Query 81	4.0 MiB	4.1 MiB	+0.3%
Query 82	4.9 MiB	4.9 MiB	+0.0%
Query 83	8.6 MiB	9.7 MiB	+13.6%
Query 84	11.9 MiB	11.9 MiB	+0.0%
Query 85	17.0 MiB	17.0 MiB	+0.0%
Query 86	82.1 MiB	92.1 MiB	+12.2%
Query 87	12.9 MiB	13.0 MiB	+0.5%
Query 88	1.5 MiB	1.5 MiB	-1.3%
Query 89	112.2 MiB	112.2 MiB	-0.0%
Query 90	351.7 KiB	351.7 KiB	+0.0%
Query 91	31.0 MiB	21.1 MiB	-31.9%
Query 92	2.7 MiB	2.7 MiB	+0.0%
Query 93	15.1 MiB	14.9 MiB	-1.7%
Query 94	3.7 MiB	3.4 MiB	-7.6%
Query 95	18.4 MiB	17.5 MiB	-5.0%
Query 96	224.7 KiB	224.7 KiB	+0.0%
Query 97	33.1 MiB	33.1 MiB	+0.2%
Query 98	114.7 MiB	124.8 MiB	+8.7%
Query 99	1.6 MiB	1.9 MiB	+17.3%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpcds	base (`4ede5833f68dad5bc8169492dedc5749fcc4b8d5`)	212.6 MiB	2.0 GiB	1.8 GiB	9.8×
tpcds	changed (`0044519d3b609ae6fbb0cfc59500b322ea368d39`)	212.2 MiB	1.9 GiB	1.7 GiB	9.1×

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	50.0s
Peak memory	2.0 GiB
Avg memory	1.5 GiB
CPU user	186.2s
CPU sys	5.4s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	50.0s
Peak memory	1.9 GiB
Avg memory	1.3 GiB
CPU user	188.9s
CPU sys	5.5s
Peak spill	0 B

File an issue against this benchmark runner

Rich-T-kid · 2026-11-08T18:29:16Z

run benchmark tpch tpcds clickbench_partitioned

env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

adriangbot · 2026-11-08T18:31:49Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6725365684-4625-tkfqf 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 8caaa35 (8caaa35) to 5191bcc diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-08T18:32:19Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6725365684-4624-24xwt 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 8caaa35 (8caaa35) to 5191bcc diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-08T18:32:43Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6725365684-4626-4wck6 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 8caaa35 (8caaa35) to 5191bcc diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-08T18:40:29Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing 8caaa35 (8caaa35) to 5191bcc diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 39.63 ms │                                   39.85 ms │ no change │
│ QQuery 2  │ 19.06 ms │                                   19.25 ms │ no change │
│ QQuery 3  │ 29.06 ms │                                   29.02 ms │ no change │
│ QQuery 4  │ 17.31 ms │                                   17.48 ms │ no change │
│ QQuery 5  │ 36.50 ms │                                   36.58 ms │ no change │
│ QQuery 6  │ 14.00 ms │                                   13.94 ms │ no change │
│ QQuery 7  │ 41.99 ms │                                   42.47 ms │ no change │
│ QQuery 8  │ 41.22 ms │                                   41.19 ms │ no change │
│ QQuery 9  │ 50.06 ms │                                   51.06 ms │ no change │
│ QQuery 10 │ 59.83 ms │                                   58.91 ms │ no change │
│ QQuery 11 │ 14.15 ms │                                   14.08 ms │ no change │
│ QQuery 12 │ 20.88 ms │                                   20.90 ms │ no change │
│ QQuery 13 │ 48.83 ms │                                   48.66 ms │ no change │
│ QQuery 14 │ 24.36 ms │                                   24.58 ms │ no change │
│ QQuery 15 │ 31.65 ms │                                   31.84 ms │ no change │
│ QQuery 16 │ 13.66 ms │                                   13.78 ms │ no change │
│ QQuery 17 │ 74.07 ms │                                   72.30 ms │ no change │
│ QQuery 18 │ 66.86 ms │                                   67.22 ms │ no change │
│ QQuery 19 │ 29.45 ms │                                   29.47 ms │ no change │
│ QQuery 20 │ 34.06 ms │                                   34.32 ms │ no change │
│ QQuery 21 │ 56.19 ms │                                   55.02 ms │ no change │
│ QQuery 22 │ 16.12 ms │                                   16.06 ms │ no change │
└───────────┴──────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 778.94ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 777.99ms │
│ Average Time (HEAD)                                       │  35.41ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  35.36ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │ 39.63 / 41.40 ±1.16 / 42.63 ms │             39.85 / 40.71 ±1.11 / 42.88 ms │    no change │
│ QQuery 2  │ 19.06 / 19.36 ±0.22 / 19.71 ms │             19.25 / 19.45 ±0.17 / 19.76 ms │    no change │
│ QQuery 3  │ 29.06 / 29.15 ±0.08 / 29.30 ms │             29.02 / 29.39 ±0.29 / 29.76 ms │    no change │
│ QQuery 4  │ 17.31 / 17.47 ±0.17 / 17.75 ms │             17.48 / 18.40 ±0.71 / 19.04 ms │ 1.05x slower │
│ QQuery 5  │ 36.50 / 37.64 ±1.44 / 40.42 ms │             36.58 / 36.85 ±0.40 / 37.65 ms │    no change │
│ QQuery 6  │ 14.00 / 14.30 ±0.26 / 14.77 ms │             13.94 / 14.02 ±0.04 / 14.06 ms │    no change │
│ QQuery 7  │ 41.99 / 43.28 ±1.02 / 44.49 ms │             42.47 / 44.01 ±1.16 / 45.36 ms │    no change │
│ QQuery 8  │ 41.22 / 41.44 ±0.18 / 41.73 ms │             41.19 / 41.57 ±0.51 / 42.57 ms │    no change │
│ QQuery 9  │ 50.06 / 51.30 ±0.83 / 52.16 ms │             51.06 / 52.65 ±1.07 / 54.38 ms │    no change │
│ QQuery 10 │ 59.83 / 60.18 ±0.33 / 60.75 ms │             58.91 / 59.35 ±0.35 / 59.97 ms │    no change │
│ QQuery 11 │ 14.15 / 14.29 ±0.13 / 14.52 ms │             14.08 / 14.22 ±0.10 / 14.34 ms │    no change │
│ QQuery 12 │ 20.88 / 21.25 ±0.37 / 21.96 ms │             20.90 / 21.46 ±0.57 / 22.55 ms │    no change │
│ QQuery 13 │ 48.83 / 50.06 ±1.39 / 52.57 ms │             48.66 / 49.62 ±1.43 / 52.44 ms │    no change │
│ QQuery 14 │ 24.36 / 24.53 ±0.14 / 24.68 ms │             24.58 / 24.75 ±0.11 / 24.90 ms │    no change │
│ QQuery 15 │ 31.65 / 32.24 ±0.78 / 33.63 ms │             31.84 / 32.41 ±0.58 / 33.11 ms │    no change │
│ QQuery 16 │ 13.66 / 13.84 ±0.12 / 13.99 ms │             13.78 / 13.88 ±0.08 / 14.00 ms │    no change │
│ QQuery 17 │ 74.07 / 75.05 ±0.75 / 76.08 ms │             72.30 / 72.83 ±0.43 / 73.54 ms │    no change │
│ QQuery 18 │ 66.86 / 68.43 ±1.70 / 71.71 ms │             67.22 / 69.27 ±1.51 / 71.13 ms │    no change │
│ QQuery 19 │ 29.45 / 30.65 ±1.36 / 32.58 ms │             29.47 / 30.20 ±0.87 / 31.86 ms │    no change │
│ QQuery 20 │ 34.06 / 34.53 ±0.25 / 34.74 ms │             34.32 / 34.71 ±0.31 / 35.06 ms │    no change │
│ QQuery 21 │ 56.19 / 57.29 ±0.95 / 58.92 ms │             55.02 / 55.95 ±0.59 / 56.59 ms │    no change │
│ QQuery 22 │ 16.12 / 16.19 ±0.05 / 16.25 ms │             16.06 / 16.20 ±0.08 / 16.29 ms │    no change │
└───────────┴────────────────────────────────┴────────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 793.86ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 791.90ms │
│ Average Time (HEAD)                                       │  36.08ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  36.00ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        1 │
│ Queries with No Change                                    │       21 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 5191bcc46d7e617bb8d11277fdc180b388141b11 | Changed: 8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	30.2 MiB	30.1 MiB	-0.1%
Query 2	23.1 MiB	23.1 MiB	+0.0%
Query 3	11.5 MiB	11.5 MiB	+0.0%
Query 4	22.3 MiB	22.2 MiB	-0.3%
Query 5	42.9 MiB	42.9 MiB	-0.0%
Query 6	832 B	832 B	+0.0%
Query 7	107.7 MiB	107.6 MiB	-0.1%
Query 8	39.8 MiB	39.8 MiB	-0.1%
Query 9	155.6 MiB	155.6 MiB	+0.0%
Query 10	56.3 MiB	53.3 MiB	-5.3%
Query 11	62.1 MiB	51.5 MiB	-17.0%
Query 12	22.5 MiB	12.6 MiB	-43.9%
Query 13	62.6 MiB	62.3 MiB	-0.4%
Query 14	7.7 MiB	7.7 MiB	+0.0%
Query 15	12.8 MiB	12.7 MiB	-0.8%
Query 16	33.6 MiB	42.9 MiB	+27.6%
Query 17	150.3 MiB	150.3 MiB	+0.0%
Query 18	127.3 MiB	127.0 MiB	-0.3%
Query 19	228.6 KiB	228.6 KiB	+0.0%
Query 20	64.4 MiB	61.7 MiB	-4.2%
Query 21	37.7 MiB	37.7 MiB	+0.0%
Query 22	31.9 MiB	32.0 MiB	+0.0%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`5191bcc46d7e617bb8d11277fdc180b388141b11`)	155.6 MiB	1.2 GiB	1.0 GiB	7.6×
tpch	changed (`8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e`)	155.6 MiB	1.1 GiB	963.4 MiB	7.2×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.2 GiB
Avg memory	478.5 MiB
CPU user	20.8s
CPU sys	1.6s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.1 GiB
Avg memory	465.2 MiB
CPU user	20.6s
CPU sys	1.8s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-11-08T18:41:30Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing 8caaa35 (8caaa35) to 5191bcc diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    5.26 ms │                                    5.29 ms │     no change │
│ QQuery 2  │   74.98 ms │                                   75.36 ms │     no change │
│ QQuery 3  │   27.92 ms │                                   28.05 ms │     no change │
│ QQuery 4  │  544.98 ms │                                  478.80 ms │ +1.14x faster │
│ QQuery 5  │   48.71 ms │                                   49.81 ms │     no change │
│ QQuery 6  │   32.07 ms │                                   32.23 ms │     no change │
│ QQuery 7  │   68.97 ms │                                   68.88 ms │     no change │
│ QQuery 8  │   37.16 ms │                                   36.69 ms │     no change │
│ QQuery 9  │   50.16 ms │                                   51.38 ms │     no change │
│ QQuery 10 │   57.11 ms │                                   57.48 ms │     no change │
│ QQuery 11 │  360.64 ms │                                  315.00 ms │ +1.14x faster │
│ QQuery 12 │   24.54 ms │                                   25.08 ms │     no change │
│ QQuery 13 │  110.11 ms │                                  109.74 ms │     no change │
│ QQuery 14 │  367.09 ms │                                  367.26 ms │     no change │
│ QQuery 15 │   50.23 ms │                                   50.04 ms │     no change │
│ QQuery 16 │    5.89 ms │                                    5.81 ms │     no change │
│ QQuery 17 │   74.95 ms │                                   75.13 ms │     no change │
│ QQuery 18 │  101.15 ms │                                  100.39 ms │     no change │
│ QQuery 19 │   40.03 ms │                                   40.14 ms │     no change │
│ QQuery 20 │   29.84 ms │                                   30.60 ms │     no change │
│ QQuery 21 │   15.89 ms │                                   16.14 ms │     no change │
│ QQuery 22 │   67.47 ms │                                   58.44 ms │ +1.15x faster │
│ QQuery 23 │  282.56 ms │                                  283.99 ms │     no change │
│ QQuery 24 │  180.94 ms │                                  181.10 ms │     no change │
│ QQuery 25 │  102.23 ms │                                  102.18 ms │     no change │
│ QQuery 26 │   43.34 ms │                                   42.93 ms │     no change │
│ QQuery 27 │    5.80 ms │                                    5.92 ms │     no change │
│ QQuery 28 │   60.09 ms │                                   57.78 ms │     no change │
│ QQuery 29 │   90.83 ms │                                   91.42 ms │     no change │
│ QQuery 30 │   35.01 ms │                                   34.20 ms │     no change │
│ QQuery 31 │  106.04 ms │                                  107.88 ms │     no change │
│ QQuery 32 │   18.90 ms │                                   18.96 ms │     no change │
│ QQuery 33 │   35.36 ms │                                   35.08 ms │     no change │
│ QQuery 34 │    8.97 ms │                                    8.90 ms │     no change │
│ QQuery 35 │   67.77 ms │                                   67.78 ms │     no change │
│ QQuery 36 │    5.75 ms │                                    5.66 ms │     no change │
│ QQuery 37 │    7.07 ms │                                    7.00 ms │     no change │
│ QQuery 38 │   62.35 ms │                                   62.69 ms │     no change │
│ QQuery 39 │   70.00 ms │                                   69.95 ms │     no change │
│ QQuery 40 │   21.82 ms │                                   21.58 ms │     no change │
│ QQuery 41 │   13.27 ms │                                   13.16 ms │     no change │
│ QQuery 42 │   22.92 ms │                                   22.89 ms │     no change │
│ QQuery 43 │    4.69 ms │                                    4.63 ms │     no change │
│ QQuery 44 │    8.24 ms │                                    8.19 ms │     no change │
│ QQuery 45 │   33.90 ms │                                   33.75 ms │     no change │
│ QQuery 46 │   10.59 ms │                                   10.62 ms │     no change │
│ QQuery 47 │  197.52 ms │                                  195.78 ms │     no change │
│ QQuery 48 │   88.55 ms │                                   88.55 ms │     no change │
│ QQuery 49 │   67.63 ms │                                   67.79 ms │     no change │
│ QQuery 50 │   55.36 ms │                                   55.93 ms │     no change │
│ QQuery 51 │   91.27 ms │                                   92.67 ms │     no change │
│ QQuery 52 │   23.52 ms │                                   23.40 ms │     no change │
│ QQuery 53 │   28.43 ms │                                   28.04 ms │     no change │
│ QQuery 54 │   51.86 ms │                                   51.90 ms │     no change │
│ QQuery 55 │   23.38 ms │                                   22.91 ms │     no change │
│ QQuery 56 │   37.56 ms │                                   36.96 ms │     no change │
│ QQuery 57 │  150.91 ms │                                  149.36 ms │     no change │
│ QQuery 58 │  103.90 ms │                                  101.27 ms │     no change │
│ QQuery 59 │  106.50 ms │                                  105.06 ms │     no change │
│ QQuery 60 │   37.85 ms │                                   37.07 ms │     no change │
│ QQuery 61 │   10.28 ms │                                   10.04 ms │     no change │
│ QQuery 62 │   36.84 ms │                                   36.72 ms │     no change │
│ QQuery 63 │   28.47 ms │                                   28.56 ms │     no change │
│ QQuery 64 │  350.17 ms │                                  345.59 ms │     no change │
│ QQuery 65 │   74.51 ms │                                   74.43 ms │     no change │
│ QQuery 66 │   70.88 ms │                                   70.31 ms │     no change │
│ QQuery 67 │  215.43 ms │                                  204.17 ms │ +1.06x faster │
│ QQuery 68 │   10.92 ms │                                   10.74 ms │     no change │
│ QQuery 69 │   52.64 ms │                                   52.32 ms │     no change │
│ QQuery 70 │   95.96 ms │                                   96.79 ms │     no change │
│ QQuery 71 │   34.46 ms │                                   34.31 ms │     no change │
│ QQuery 72 │ 1521.15 ms │                                 1594.20 ms │     no change │
│ QQuery 73 │    8.53 ms │                                    8.69 ms │     no change │
│ QQuery 74 │  234.84 ms │                                  221.70 ms │ +1.06x faster │
│ QQuery 75 │  135.14 ms │                                  135.09 ms │     no change │
│ QQuery 76 │   33.39 ms │                                   33.85 ms │     no change │
│ QQuery 77 │   58.16 ms │                                   58.62 ms │     no change │
│ QQuery 78 │  254.04 ms │                                  252.97 ms │     no change │
│ QQuery 79 │   65.33 ms │                                   65.72 ms │     no change │
│ QQuery 80 │   90.92 ms │                                   91.82 ms │     no change │
│ QQuery 81 │   26.88 ms │                                   26.91 ms │     no change │
│ QQuery 82 │   16.34 ms │                                   16.42 ms │     no change │
│ QQuery 83 │   30.92 ms │                                   31.42 ms │     no change │
│ QQuery 84 │   29.60 ms │                                   30.57 ms │     no change │
│ QQuery 85 │   92.75 ms │                                   92.55 ms │     no change │
│ QQuery 86 │   23.17 ms │                                   23.04 ms │     no change │
│ QQuery 87 │   62.62 ms │                                   63.30 ms │     no change │
│ QQuery 88 │   57.61 ms │                                   56.96 ms │     no change │
│ QQuery 89 │   34.55 ms │                                   34.34 ms │     no change │
│ QQuery 90 │   15.55 ms │                                   15.72 ms │     no change │
│ QQuery 91 │   37.90 ms │                                   38.03 ms │     no change │
│ QQuery 92 │   27.32 ms │                                   28.38 ms │     no change │
│ QQuery 93 │   47.38 ms │                                   47.51 ms │     no change │
│ QQuery 94 │   36.57 ms │                                   36.71 ms │     no change │
│ QQuery 95 │   77.31 ms │                                   77.79 ms │     no change │
│ QQuery 96 │   22.89 ms │                                   23.07 ms │     no change │
│ QQuery 97 │   48.50 ms │                                   50.21 ms │     no change │
│ QQuery 98 │   37.19 ms │                                   37.23 ms │     no change │
│ QQuery 99 │   54.62 ms │                                   54.99 ms │     no change │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 8719.54ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 8644.47ms │
│ Average Time (HEAD)                                       │   88.08ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   87.32ms │
│ Queries Faster                                            │         5 │
│ Queries Slower                                            │         0 │
│ Queries with No Change                                    │        94 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           5.26 / 5.86 ±1.10 / 8.07 ms │                5.29 / 5.93 ±1.10 / 8.14 ms │     no change │
│ QQuery 2  │        74.98 / 75.80 ±0.93 / 77.53 ms │             75.36 / 75.77 ±0.26 / 76.04 ms │     no change │
│ QQuery 3  │        27.92 / 28.29 ±0.22 / 28.58 ms │             28.05 / 28.44 ±0.21 / 28.60 ms │     no change │
│ QQuery 4  │     544.98 / 551.86 ±5.93 / 562.81 ms │          478.80 / 486.67 ±6.70 / 496.22 ms │ +1.13x faster │
│ QQuery 5  │        48.71 / 49.11 ±0.35 / 49.60 ms │             49.81 / 50.98 ±1.91 / 54.78 ms │     no change │
│ QQuery 6  │        32.07 / 32.36 ±0.26 / 32.82 ms │             32.23 / 32.88 ±0.58 / 33.81 ms │     no change │
│ QQuery 7  │        68.97 / 70.11 ±1.14 / 72.32 ms │             68.88 / 69.28 ±0.30 / 69.67 ms │     no change │
│ QQuery 8  │        37.16 / 37.45 ±0.27 / 37.77 ms │             36.69 / 36.91 ±0.30 / 37.50 ms │     no change │
│ QQuery 9  │        50.16 / 53.03 ±1.70 / 54.75 ms │             51.38 / 52.48 ±0.56 / 52.98 ms │     no change │
│ QQuery 10 │        57.11 / 57.44 ±0.22 / 57.79 ms │             57.48 / 57.99 ±0.32 / 58.39 ms │     no change │
│ QQuery 11 │     360.64 / 363.27 ±2.04 / 366.52 ms │          315.00 / 318.71 ±2.79 / 322.59 ms │ +1.14x faster │
│ QQuery 12 │        24.54 / 24.99 ±0.47 / 25.80 ms │             25.08 / 25.49 ±0.40 / 26.20 ms │     no change │
│ QQuery 13 │     110.11 / 110.60 ±0.54 / 111.58 ms │          109.74 / 110.32 ±0.35 / 110.79 ms │     no change │
│ QQuery 14 │     367.09 / 372.69 ±3.47 / 377.26 ms │          367.26 / 373.77 ±4.31 / 380.65 ms │     no change │
│ QQuery 15 │        50.23 / 51.09 ±0.52 / 51.82 ms │             50.04 / 51.09 ±1.45 / 53.93 ms │     no change │
│ QQuery 16 │           5.89 / 6.02 ±0.16 / 6.33 ms │                5.81 / 5.98 ±0.20 / 6.37 ms │     no change │
│ QQuery 17 │        74.95 / 75.52 ±0.42 / 76.04 ms │             75.13 / 78.19 ±4.02 / 86.14 ms │     no change │
│ QQuery 18 │     101.15 / 102.22 ±1.16 / 103.77 ms │          100.39 / 101.14 ±0.89 / 102.36 ms │     no change │
│ QQuery 19 │        40.03 / 40.39 ±0.64 / 41.66 ms │             40.14 / 41.65 ±1.50 / 44.40 ms │     no change │
│ QQuery 20 │        29.84 / 30.14 ±0.25 / 30.52 ms │             30.60 / 31.08 ±0.73 / 32.50 ms │     no change │
│ QQuery 21 │        15.89 / 16.06 ±0.19 / 16.39 ms │             16.14 / 16.32 ±0.14 / 16.57 ms │     no change │
│ QQuery 22 │        67.47 / 68.90 ±1.02 / 70.49 ms │             58.44 / 58.97 ±0.42 / 59.70 ms │ +1.17x faster │
│ QQuery 23 │     282.56 / 285.44 ±2.43 / 289.32 ms │          283.99 / 287.61 ±2.88 / 292.09 ms │     no change │
│ QQuery 24 │     180.94 / 185.33 ±4.30 / 192.96 ms │          181.10 / 183.29 ±1.62 / 185.85 ms │     no change │
│ QQuery 25 │     102.23 / 104.96 ±2.55 / 108.96 ms │          102.18 / 104.60 ±2.49 / 108.91 ms │     no change │
│ QQuery 26 │        43.34 / 44.63 ±1.64 / 47.86 ms │             42.93 / 43.30 ±0.30 / 43.84 ms │     no change │
│ QQuery 27 │           5.80 / 5.97 ±0.14 / 6.21 ms │                5.92 / 6.01 ±0.09 / 6.17 ms │     no change │
│ QQuery 28 │        60.09 / 60.81 ±0.49 / 61.33 ms │             57.78 / 60.02 ±1.16 / 61.21 ms │     no change │
│ QQuery 29 │       90.83 / 93.43 ±3.89 / 101.14 ms │             91.42 / 92.87 ±1.35 / 94.74 ms │     no change │
│ QQuery 30 │        35.01 / 35.92 ±1.00 / 37.62 ms │             34.20 / 34.52 ±0.32 / 35.14 ms │     no change │
│ QQuery 31 │     106.04 / 107.61 ±0.90 / 108.73 ms │          107.88 / 109.12 ±1.33 / 111.64 ms │     no change │
│ QQuery 32 │        18.90 / 19.20 ±0.35 / 19.81 ms │             18.96 / 19.23 ±0.37 / 19.96 ms │     no change │
│ QQuery 33 │        35.36 / 36.82 ±1.65 / 39.86 ms │             35.08 / 35.54 ±0.30 / 35.99 ms │     no change │
│ QQuery 34 │          8.97 / 9.66 ±0.91 / 11.45 ms │                8.90 / 9.14 ±0.19 / 9.49 ms │ +1.06x faster │
│ QQuery 35 │        67.77 / 68.44 ±0.56 / 69.33 ms │             67.78 / 68.08 ±0.27 / 68.44 ms │     no change │
│ QQuery 36 │           5.75 / 5.86 ±0.16 / 6.18 ms │                5.66 / 5.81 ±0.25 / 6.30 ms │     no change │
│ QQuery 37 │           7.07 / 7.19 ±0.14 / 7.45 ms │                7.00 / 7.25 ±0.23 / 7.68 ms │     no change │
│ QQuery 38 │        62.35 / 62.77 ±0.42 / 63.55 ms │             62.69 / 63.10 ±0.50 / 64.05 ms │     no change │
│ QQuery 39 │        70.00 / 71.26 ±1.75 / 74.70 ms │             69.95 / 70.78 ±0.54 / 71.48 ms │     no change │
│ QQuery 40 │        21.82 / 22.19 ±0.43 / 22.99 ms │             21.58 / 22.11 ±0.53 / 22.88 ms │     no change │
│ QQuery 41 │        13.27 / 13.41 ±0.14 / 13.67 ms │             13.16 / 13.30 ±0.12 / 13.51 ms │     no change │
│ QQuery 42 │        22.92 / 23.16 ±0.26 / 23.67 ms │             22.89 / 23.58 ±0.74 / 24.99 ms │     no change │
│ QQuery 43 │           4.69 / 4.86 ±0.16 / 5.13 ms │                4.63 / 4.71 ±0.13 / 4.96 ms │     no change │
│ QQuery 44 │           8.24 / 8.35 ±0.08 / 8.49 ms │                8.19 / 8.29 ±0.12 / 8.53 ms │     no change │
│ QQuery 45 │        33.90 / 34.51 ±0.48 / 35.31 ms │             33.75 / 34.70 ±1.31 / 37.30 ms │     no change │
│ QQuery 46 │        10.59 / 10.71 ±0.07 / 10.81 ms │             10.62 / 10.90 ±0.17 / 11.06 ms │     no change │
│ QQuery 47 │     197.52 / 202.28 ±6.05 / 214.11 ms │          195.78 / 199.90 ±2.74 / 203.37 ms │     no change │
│ QQuery 48 │        88.55 / 89.36 ±0.94 / 90.97 ms │             88.55 / 88.98 ±0.37 / 89.50 ms │     no change │
│ QQuery 49 │        67.63 / 68.92 ±0.79 / 70.00 ms │             67.79 / 68.23 ±0.34 / 68.74 ms │     no change │
│ QQuery 50 │        55.36 / 56.07 ±0.53 / 56.83 ms │             55.93 / 57.39 ±2.28 / 61.91 ms │     no change │
│ QQuery 51 │        91.27 / 94.36 ±1.92 / 96.29 ms │             92.67 / 93.61 ±0.68 / 94.40 ms │     no change │
│ QQuery 52 │        23.52 / 23.69 ±0.24 / 24.17 ms │             23.40 / 24.81 ±2.23 / 29.23 ms │     no change │
│ QQuery 53 │        28.43 / 28.58 ±0.16 / 28.89 ms │             28.04 / 28.24 ±0.19 / 28.56 ms │     no change │
│ QQuery 54 │        51.86 / 52.26 ±0.25 / 52.61 ms │             51.90 / 53.18 ±2.16 / 57.48 ms │     no change │
│ QQuery 55 │        23.38 / 23.53 ±0.11 / 23.70 ms │             22.91 / 23.25 ±0.30 / 23.67 ms │     no change │
│ QQuery 56 │        37.56 / 38.77 ±1.33 / 41.32 ms │             36.96 / 37.41 ±0.43 / 38.14 ms │     no change │
│ QQuery 57 │     150.91 / 151.35 ±0.51 / 152.31 ms │          149.36 / 150.97 ±1.39 / 153.33 ms │     no change │
│ QQuery 58 │     103.90 / 105.25 ±1.44 / 107.89 ms │          101.27 / 102.07 ±0.78 / 103.52 ms │     no change │
│ QQuery 59 │     106.50 / 107.38 ±0.74 / 108.65 ms │          105.06 / 106.32 ±1.25 / 108.68 ms │     no change │
│ QQuery 60 │        37.85 / 38.87 ±1.04 / 40.81 ms │             37.07 / 37.80 ±0.51 / 38.53 ms │     no change │
│ QQuery 61 │        10.28 / 10.41 ±0.13 / 10.65 ms │             10.04 / 10.22 ±0.23 / 10.66 ms │     no change │
│ QQuery 62 │        36.84 / 37.74 ±0.64 / 38.53 ms │             36.72 / 37.32 ±0.37 / 37.81 ms │     no change │
│ QQuery 63 │        28.47 / 28.59 ±0.09 / 28.71 ms │             28.56 / 29.34 ±1.43 / 32.20 ms │     no change │
│ QQuery 64 │     350.17 / 352.22 ±2.23 / 356.47 ms │          345.59 / 348.64 ±2.71 / 353.41 ms │     no change │
│ QQuery 65 │        74.51 / 75.64 ±1.25 / 77.90 ms │             74.43 / 76.26 ±2.54 / 81.05 ms │     no change │
│ QQuery 66 │        70.88 / 72.51 ±1.66 / 74.66 ms │             70.31 / 71.91 ±2.18 / 76.17 ms │     no change │
│ QQuery 67 │     215.43 / 221.01 ±4.33 / 228.49 ms │          204.17 / 207.53 ±3.68 / 214.31 ms │ +1.06x faster │
│ QQuery 68 │        10.92 / 11.09 ±0.31 / 11.70 ms │             10.74 / 11.10 ±0.33 / 11.57 ms │     no change │
│ QQuery 69 │        52.64 / 53.06 ±0.31 / 53.46 ms │             52.32 / 52.53 ±0.18 / 52.79 ms │     no change │
│ QQuery 70 │        95.96 / 96.63 ±0.59 / 97.71 ms │            96.79 / 98.37 ±2.55 / 103.44 ms │     no change │
│ QQuery 71 │        34.46 / 36.26 ±2.25 / 40.25 ms │             34.31 / 34.84 ±0.59 / 35.73 ms │     no change │
│ QQuery 72 │ 1521.15 / 1581.25 ±52.70 / 1678.09 ms │       1594.20 / 1603.63 ±7.82 / 1615.26 ms │     no change │
│ QQuery 73 │           8.53 / 8.85 ±0.39 / 9.62 ms │                8.69 / 8.89 ±0.31 / 9.50 ms │     no change │
│ QQuery 74 │     234.84 / 240.15 ±2.86 / 242.95 ms │          221.70 / 225.49 ±3.18 / 229.75 ms │ +1.06x faster │
│ QQuery 75 │     135.14 / 136.05 ±1.25 / 138.51 ms │          135.09 / 136.93 ±1.00 / 137.88 ms │     no change │
│ QQuery 76 │        33.39 / 34.25 ±1.15 / 36.50 ms │             33.85 / 36.20 ±3.23 / 42.48 ms │  1.06x slower │
│ QQuery 77 │        58.16 / 58.34 ±0.15 / 58.60 ms │             58.62 / 59.12 ±0.58 / 60.24 ms │     no change │
│ QQuery 78 │    254.04 / 264.37 ±10.78 / 279.16 ms │         252.97 / 267.99 ±14.62 / 288.57 ms │     no change │
│ QQuery 79 │        65.33 / 65.70 ±0.37 / 66.39 ms │             65.72 / 68.54 ±4.23 / 76.92 ms │     no change │
│ QQuery 80 │        90.92 / 93.65 ±1.94 / 96.95 ms │             91.82 / 93.08 ±1.21 / 95.03 ms │     no change │
│ QQuery 81 │        26.88 / 28.40 ±2.58 / 33.54 ms │             26.91 / 27.17 ±0.24 / 27.60 ms │     no change │
│ QQuery 82 │        16.34 / 16.56 ±0.30 / 17.12 ms │             16.42 / 16.54 ±0.11 / 16.69 ms │     no change │
│ QQuery 83 │        30.92 / 31.05 ±0.18 / 31.41 ms │             31.42 / 31.65 ±0.18 / 31.91 ms │     no change │
│ QQuery 84 │        29.60 / 30.83 ±2.30 / 35.42 ms │             30.57 / 32.42 ±2.57 / 37.47 ms │  1.05x slower │
│ QQuery 85 │        92.75 / 94.22 ±1.68 / 97.43 ms │             92.55 / 93.46 ±0.67 / 94.52 ms │     no change │
│ QQuery 86 │        23.17 / 23.35 ±0.16 / 23.61 ms │             23.04 / 23.30 ±0.37 / 24.02 ms │     no change │
│ QQuery 87 │        62.62 / 63.04 ±0.28 / 63.44 ms │             63.30 / 64.64 ±1.67 / 67.39 ms │     no change │
│ QQuery 88 │        57.61 / 58.99 ±1.69 / 62.25 ms │             56.96 / 57.71 ±0.55 / 58.60 ms │     no change │
│ QQuery 89 │        34.55 / 35.04 ±0.54 / 36.05 ms │             34.34 / 34.59 ±0.19 / 34.85 ms │     no change │
│ QQuery 90 │        15.55 / 15.65 ±0.09 / 15.80 ms │             15.72 / 15.87 ±0.12 / 16.07 ms │     no change │
│ QQuery 91 │        37.90 / 38.15 ±0.28 / 38.68 ms │             38.03 / 38.15 ±0.14 / 38.43 ms │     no change │
│ QQuery 92 │        27.32 / 27.76 ±0.40 / 28.40 ms │             28.38 / 29.53 ±1.38 / 32.04 ms │  1.06x slower │
│ QQuery 93 │        47.38 / 48.39 ±1.41 / 51.11 ms │             47.51 / 49.26 ±1.33 / 51.08 ms │     no change │
│ QQuery 94 │        36.57 / 37.81 ±0.83 / 39.09 ms │             36.71 / 37.17 ±0.36 / 37.63 ms │     no change │
│ QQuery 95 │        77.31 / 78.08 ±0.84 / 79.34 ms │             77.79 / 78.57 ±0.50 / 79.33 ms │     no change │
│ QQuery 96 │        22.89 / 23.07 ±0.24 / 23.55 ms │             23.07 / 23.26 ±0.18 / 23.57 ms │     no change │
│ QQuery 97 │        48.50 / 48.98 ±0.34 / 49.33 ms │             50.21 / 52.15 ±2.38 / 56.58 ms │  1.06x slower │
│ QQuery 98 │        37.19 / 38.52 ±1.28 / 40.61 ms │             37.23 / 37.68 ±0.59 / 38.84 ms │     no change │
│ QQuery 99 │        54.62 / 55.32 ±0.63 / 56.42 ms │             54.99 / 55.33 ±0.36 / 55.97 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 8897.34ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 8776.43ms │
│ Average Time (HEAD)                                       │   89.87ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   88.65ms │
│ Queries Faster                                            │         6 │
│ Queries Slower                                            │         4 │
│ Queries with No Change                                    │        89 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 5191bcc46d7e617bb8d11277fdc180b388141b11 | Changed: 8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e

tpcds — tpcds_sf1

Query	Base	Changed	Change
Query 1	198.0 KiB	139.1 KiB	-29.8%
Query 2	62.7 MiB	78.9 MiB	+25.8%
Query 3	1.6 MiB	1.7 MiB	+1.8%
Query 4	87.8 MiB	87.8 MiB	+0.0%
Query 5	12.5 MiB	12.5 MiB	+0.0%
Query 6	4.1 MiB	4.2 MiB	+1.7%
Query 7	3.9 MiB	3.6 MiB	-7.1%
Query 8	2.1 MiB	1.7 MiB	-20.9%
Query 9	32.1 KiB	32.1 KiB	+0.0%
Query 10	2.0 MiB	2.0 MiB	+0.0%
Query 11	58.9 MiB	58.9 MiB	+0.0%
Query 12	114.4 MiB	124.1 MiB	+8.5%
Query 13	13.7 MiB	13.7 MiB	+0.0%
Query 14	171.2 MiB	171.3 MiB	+0.1%
Query 15	4.1 MiB	4.1 MiB	+0.6%
Query 16	118.6 KiB	118.6 KiB	+0.0%
Query 17	40.1 MiB	40.1 MiB	+0.0%
Query 18	21.2 MiB	21.2 MiB	-0.1%
Query 19	1.9 MiB	2.1 MiB	+8.2%
Query 20	114.6 MiB	114.6 MiB	+0.0%
Query 21	2.5 MiB	2.2 MiB	-11.8%
Query 22	61.8 MiB	61.8 MiB	+0.0%
Query 23	213.0 MiB	212.1 MiB	-0.4%
Query 24	38.7 MiB	38.7 MiB	+0.0%
Query 25	99.9 MiB	99.9 MiB	+0.0%
Query 26	3.7 MiB	3.4 MiB	-8.0%
Query 27	1.6 MiB	1.6 MiB	+0.0%
Query 28	1.3 MiB	1.5 MiB	+10.0%
Query 29	104.7 MiB	104.8 MiB	+0.0%
Query 30	3.7 MiB	3.8 MiB	+3.7%
Query 31	13.4 MiB	13.4 MiB	+0.0%
Query 32	2.7 MiB	2.7 MiB	+0.1%
Query 33	4.0 MiB	4.0 MiB	+0.0%
Query 34	6.0 MiB	6.0 MiB	+0.0%
Query 35	18.4 MiB	16.6 MiB	-9.6%
Query 36	10.3 MiB	10.3 MiB	+0.0%
Query 37	3.3 MiB	3.3 MiB	+0.0%
Query 38	10.8 MiB	10.8 MiB	-0.0%
Query 39	41.5 MiB	52.9 MiB	+27.2%
Query 40	12.2 MiB	12.2 MiB	+0.0%
Query 41	2.1 MiB	2.1 MiB	+0.0%
Query 42	1.3 MiB	1.7 MiB	+31.7%
Query 43	275.0 KiB	275.4 KiB	+0.1%
Query 44	233.2 KiB	233.2 KiB	+0.0%
Query 45	4.5 MiB	4.5 MiB	+0.0%
Query 46	6.2 MiB	6.2 MiB	+0.0%
Query 47	138.2 MiB	128.5 MiB	-7.0%
Query 48	11.8 MiB	11.8 MiB	-0.0%
Query 49	52.4 MiB	52.4 MiB	-0.0%
Query 50	16.9 MiB	17.3 MiB	+2.2%
Query 51	177.7 MiB	177.7 MiB	+0.0%
Query 52	1.8 MiB	1.5 MiB	-14.9%
Query 53	30.4 MiB	30.6 MiB	+0.5%
Query 54	2.2 MiB	2.1 MiB	-4.2%
Query 55	1.5 MiB	1.6 MiB	+10.5%
Query 56	2.9 MiB	2.8 MiB	-3.4%
Query 57	120.8 MiB	128.1 MiB	+6.0%
Query 58	4.4 MiB	4.2 MiB	-4.9%
Query 59	11.7 MiB	11.7 MiB	+0.0%
Query 60	5.3 MiB	5.2 MiB	-0.9%
Query 61	1.2 MiB	1.2 MiB	+0.0%
Query 62	3.7 MiB	3.7 MiB	+0.0%
Query 63	20.6 MiB	30.6 MiB	+48.6%
Query 64	106.2 MiB	115.3 MiB	+8.5%
Query 65	29.7 MiB	29.7 MiB	-0.0%
Query 66	14.9 MiB	11.9 MiB	-20.7%
Query 67	138.8 MiB	137.0 MiB	-1.3%
Query 68	6.2 MiB	6.2 MiB	+0.0%
Query 69	3.7 MiB	3.6 MiB	-1.2%
Query 70	44.0 MiB	44.0 MiB	+0.0%
Query 71	40.4 MiB	40.4 MiB	+0.0%
Query 72	97.2 MiB	97.2 MiB	+0.0%
Query 73	6.0 MiB	6.0 MiB	+0.0%
Query 74	36.0 MiB	36.0 MiB	+0.0%
Query 75	22.2 MiB	21.6 MiB	-2.5%
Query 76	4.3 MiB	4.2 MiB	-1.9%
Query 77	4.0 MiB	4.1 MiB	+2.0%
Query 78	335.7 MiB	335.7 MiB	+0.0%
Query 79	9.3 MiB	8.0 MiB	-14.0%
Query 80	42.4 MiB	41.6 MiB	-1.9%
Query 81	4.1 MiB	4.1 MiB	-0.1%
Query 82	4.9 MiB	4.9 MiB	+0.0%
Query 83	9.7 MiB	8.6 MiB	-11.9%
Query 84	11.9 MiB	11.9 MiB	+0.0%
Query 85	17.0 MiB	17.0 MiB	+0.0%
Query 86	82.1 MiB	91.9 MiB	+12.0%
Query 87	12.9 MiB	12.9 MiB	+0.0%
Query 88	1.7 MiB	1.7 MiB	-0.4%
Query 89	112.2 MiB	112.2 MiB	+0.0%
Query 90	357.5 KiB	351.7 KiB	-1.6%
Query 91	21.5 MiB	31.4 MiB	+45.9%
Query 92	2.7 MiB	2.7 MiB	+0.0%
Query 93	15.1 MiB	15.4 MiB	+1.7%
Query 94	3.7 MiB	3.7 MiB	+0.0%
Query 95	24.0 MiB	22.6 MiB	-5.8%
Query 96	224.7 KiB	224.7 KiB	+0.0%
Query 97	33.1 MiB	33.1 MiB	+0.0%
Query 98	114.7 MiB	114.7 MiB	-0.0%
Query 99	1.4 MiB	1.4 MiB	+3.0%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpcds	base (`5191bcc46d7e617bb8d11277fdc180b388141b11`)	335.7 MiB	2.1 GiB	1.8 GiB	6.4×
tpcds	changed (`8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e`)	335.7 MiB	1.9 GiB	1.5 GiB	5.7×

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	45.0s
Peak memory	2.1 GiB
Avg memory	1.5 GiB
CPU user	178.4s
CPU sys	5.2s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	45.0s
Peak memory	1.9 GiB
Avg memory	1.3 GiB
CPU user	182.7s
CPU sys	5.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-11-08T18:48:32Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing 8caaa35 (8caaa35) to 5191bcc diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
baseline:
  ref: "5191bcc46d7e617bb8d11277fdc180b388141b11"
changed:
  ref: "8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.13 ms │                                    1.14 ms │     no change │
│ QQuery 1  │   10.90 ms │                                   10.94 ms │     no change │
│ QQuery 2  │   34.79 ms │                                   34.53 ms │     no change │
│ QQuery 3  │   30.21 ms │                                   30.13 ms │     no change │
│ QQuery 4  │  227.89 ms │                                  227.68 ms │     no change │
│ QQuery 5  │  779.53 ms │                                  727.44 ms │ +1.07x faster │
│ QQuery 6  │    1.13 ms │                                    1.16 ms │     no change │
│ QQuery 7  │   12.14 ms │                                   12.11 ms │     no change │
│ QQuery 8  │  327.25 ms │                                  322.53 ms │     no change │
│ QQuery 9  │  464.40 ms │                                  459.84 ms │     no change │
│ QQuery 10 │  199.95 ms │                                  198.56 ms │     no change │
│ QQuery 11 │  210.71 ms │                                  209.07 ms │     no change │
│ QQuery 12 │  790.61 ms │                                  727.85 ms │ +1.09x faster │
│ QQuery 13 │ 1070.08 ms │                                 1022.60 ms │     no change │
│ QQuery 14 │  832.36 ms │                                  840.02 ms │     no change │
│ QQuery 15 │  282.65 ms │                                  281.26 ms │     no change │
│ QQuery 16 │ 1254.71 ms │                                 1281.42 ms │     no change │
│ QQuery 17 │ 1245.30 ms │                                 1235.57 ms │     no change │
│ QQuery 18 │ 1979.65 ms │                                 1980.12 ms │     no change │
│ QQuery 19 │   26.60 ms │                                   26.61 ms │     no change │
│ QQuery 20 │ 1033.18 ms │                                 1052.81 ms │     no change │
│ QQuery 21 │ 1347.22 ms │                                 1357.97 ms │     no change │
│ QQuery 22 │ 3186.34 ms │                                 3198.45 ms │     no change │
│ QQuery 23 │ 9106.94 ms │                                 9046.64 ms │     no change │
│ QQuery 24 │   94.80 ms │                                   95.24 ms │     no change │
│ QQuery 25 │  351.21 ms │                                  349.08 ms │     no change │
│ QQuery 26 │   96.81 ms │                                   96.82 ms │     no change │
│ QQuery 27 │ 1129.56 ms │                                 1134.13 ms │     no change │
│ QQuery 28 │ 2528.46 ms │                                 2433.88 ms │     no change │
│ QQuery 29 │   37.63 ms │                                   37.20 ms │     no change │
│ QQuery 30 │  542.37 ms │                                  541.67 ms │     no change │
│ QQuery 31 │  533.45 ms │                                  528.35 ms │     no change │
│ QQuery 32 │  986.54 ms │                                 1015.24 ms │     no change │
│ QQuery 33 │ 3576.46 ms │                                 3210.48 ms │ +1.11x faster │
│ QQuery 34 │ 3603.11 ms │                                 3260.85 ms │ +1.10x faster │
│ QQuery 35 │  292.44 ms │                                  289.14 ms │     no change │
│ QQuery 36 │  134.95 ms │                                  129.66 ms │     no change │
│ QQuery 37 │   95.22 ms │                                   94.92 ms │     no change │
│ QQuery 38 │   81.49 ms │                                   80.44 ms │     no change │
│ QQuery 39 │  293.63 ms │                                  277.38 ms │ +1.06x faster │
│ QQuery 40 │   13.46 ms │                                   13.97 ms │     no change │
│ QQuery 41 │   12.90 ms │                                   13.02 ms │     no change │
│ QQuery 42 │   12.24 ms │                                   12.62 ms │     no change │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 38872.40ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 37900.55ms │
│ Average Time (HEAD)                                       │   904.01ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   881.41ms │
│ Queries Faster                                            │          5 │
│ Queries Slower                                            │          0 │
│ Queries with No Change                                    │         38 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │            1.13 / 1.73 ±1.10 / 3.93 ms │                1.14 / 1.76 ±1.12 / 4.00 ms │     no change │
│ QQuery 1  │         10.90 / 10.99 ±0.08 / 11.13 ms │             10.94 / 11.19 ±0.24 / 11.58 ms │     no change │
│ QQuery 2  │         34.79 / 35.06 ±0.17 / 35.29 ms │             34.53 / 34.82 ±0.20 / 35.15 ms │     no change │
│ QQuery 3  │         30.21 / 30.83 ±0.58 / 31.80 ms │             30.13 / 30.51 ±0.26 / 30.90 ms │     no change │
│ QQuery 4  │      227.89 / 233.03 ±3.52 / 238.06 ms │          227.68 / 235.38 ±5.84 / 244.69 ms │     no change │
│ QQuery 5  │     779.53 / 796.59 ±11.69 / 807.60 ms │          727.44 / 740.89 ±8.38 / 748.85 ms │ +1.08x faster │
│ QQuery 6  │            1.13 / 1.30 ±0.24 / 1.76 ms │                1.16 / 1.32 ±0.22 / 1.76 ms │     no change │
│ QQuery 7  │         12.14 / 12.27 ±0.14 / 12.54 ms │             12.11 / 12.27 ±0.12 / 12.47 ms │     no change │
│ QQuery 8  │      327.25 / 332.12 ±3.80 / 337.00 ms │          322.53 / 330.57 ±5.80 / 338.55 ms │     no change │
│ QQuery 9  │      464.40 / 471.42 ±6.96 / 484.30 ms │         459.84 / 475.43 ±16.68 / 502.49 ms │     no change │
│ QQuery 10 │      199.95 / 205.60 ±9.35 / 224.06 ms │          198.56 / 203.59 ±8.74 / 221.05 ms │     no change │
│ QQuery 11 │      210.71 / 211.90 ±1.04 / 213.48 ms │          209.07 / 209.86 ±0.82 / 211.28 ms │     no change │
│ QQuery 12 │     790.61 / 811.64 ±16.68 / 828.02 ms │         727.85 / 748.43 ±13.93 / 765.75 ms │ +1.08x faster │
│ QQuery 13 │  1070.08 / 1093.71 ±15.69 / 1112.86 ms │      1022.60 / 1057.41 ±22.66 / 1086.39 ms │     no change │
│ QQuery 14 │     832.36 / 846.66 ±10.90 / 858.13 ms │          840.02 / 849.86 ±8.15 / 859.61 ms │     no change │
│ QQuery 15 │      282.65 / 287.48 ±6.22 / 299.61 ms │          281.26 / 290.74 ±7.62 / 299.91 ms │     no change │
│ QQuery 16 │  1254.71 / 1291.29 ±28.56 / 1326.76 ms │      1281.42 / 1301.87 ±17.33 / 1329.18 ms │     no change │
│ QQuery 17 │  1245.30 / 1262.28 ±12.04 / 1281.57 ms │      1235.57 / 1261.49 ±14.28 / 1277.48 ms │     no change │
│ QQuery 18 │  1979.65 / 2000.49 ±15.90 / 2016.32 ms │      1980.12 / 2005.66 ±19.46 / 2027.21 ms │     no change │
│ QQuery 19 │         26.60 / 31.50 ±8.34 / 48.11 ms │             26.61 / 27.15 ±0.58 / 28.22 ms │ +1.16x faster │
│ QQuery 20 │   1033.18 / 1041.41 ±4.31 / 1045.89 ms │       1052.81 / 1055.17 ±2.36 / 1058.74 ms │     no change │
│ QQuery 21 │  1347.22 / 1362.28 ±11.04 / 1380.59 ms │       1357.97 / 1362.86 ±3.40 / 1368.33 ms │     no change │
│ QQuery 22 │  3186.34 / 3231.66 ±26.14 / 3258.94 ms │      3198.45 / 3256.57 ±35.64 / 3305.25 ms │     no change │
│ QQuery 23 │ 9106.94 / 9378.86 ±181.24 / 9629.61 ms │     9046.64 / 9255.63 ±221.27 / 9665.10 ms │     no change │
│ QQuery 24 │         94.80 / 96.27 ±0.98 / 97.90 ms │          95.24 / 121.56 ±50.49 / 222.52 ms │  1.26x slower │
│ QQuery 25 │      351.21 / 353.97 ±2.23 / 357.38 ms │          349.08 / 351.74 ±3.64 / 358.91 ms │     no change │
│ QQuery 26 │         96.81 / 97.93 ±0.93 / 99.58 ms │             96.82 / 97.85 ±0.98 / 99.18 ms │     no change │
│ QQuery 27 │  1129.56 / 1145.09 ±20.80 / 1185.98 ms │       1134.13 / 1139.81 ±5.19 / 1149.29 ms │     no change │
│ QQuery 28 │  2528.46 / 2565.31 ±36.08 / 2630.72 ms │      2433.88 / 2478.93 ±27.77 / 2516.81 ms │     no change │
│ QQuery 29 │         37.63 / 37.95 ±0.32 / 38.37 ms │             37.20 / 37.55 ±0.37 / 38.02 ms │     no change │
│ QQuery 30 │      542.37 / 549.15 ±5.53 / 556.40 ms │          541.67 / 546.08 ±4.59 / 553.62 ms │     no change │
│ QQuery 31 │     533.45 / 551.75 ±14.16 / 575.18 ms │         528.35 / 543.84 ±16.37 / 574.95 ms │     no change │
│ QQuery 32 │   986.54 / 1055.32 ±66.00 / 1177.97 ms │      1015.24 / 1077.62 ±51.55 / 1167.17 ms │     no change │
│ QQuery 33 │  3576.46 / 3658.03 ±87.55 / 3817.15 ms │      3210.48 / 3278.97 ±51.57 / 3359.32 ms │ +1.12x faster │
│ QQuery 34 │  3603.11 / 3685.04 ±73.27 / 3818.90 ms │      3260.85 / 3385.30 ±77.18 / 3500.22 ms │ +1.09x faster │
│ QQuery 35 │     292.44 / 306.06 ±18.18 / 338.65 ms │          289.14 / 296.57 ±7.07 / 306.98 ms │     no change │
│ QQuery 36 │     134.95 / 149.03 ±10.63 / 161.27 ms │          129.66 / 138.98 ±6.50 / 147.28 ms │ +1.07x faster │
│ QQuery 37 │        95.22 / 99.84 ±2.84 / 102.61 ms │           94.92 / 101.83 ±7.43 / 115.85 ms │     no change │
│ QQuery 38 │         81.49 / 83.79 ±2.14 / 87.66 ms │             80.44 / 84.75 ±6.23 / 97.02 ms │     no change │
│ QQuery 39 │      293.63 / 302.04 ±5.37 / 310.23 ms │         277.38 / 299.06 ±14.72 / 319.31 ms │     no change │
│ QQuery 40 │         13.46 / 15.58 ±2.35 / 19.99 ms │             13.97 / 14.81 ±0.56 / 15.58 ms │     no change │
│ QQuery 41 │        12.90 / 18.34 ±10.21 / 38.75 ms │             13.02 / 14.97 ±3.46 / 21.88 ms │ +1.22x faster │
│ QQuery 42 │         12.24 / 12.82 ±0.60 / 13.88 ms │             12.62 / 14.19 ±2.97 / 20.12 ms │  1.11x slower │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 39765.42ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 38784.81ms │
│ Average Time (HEAD)                                       │   924.78ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   901.97ms │
│ Queries Faster                                            │          7 │
│ Queries Slower                                            │          2 │
│ Queries with No Change                                    │         34 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 5191bcc46d7e617bb8d11277fdc180b388141b11 | Changed: 8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e

clickbench_partitioned

Query	Base	Changed	Change
Query 0	0 B	0 B	0.0%
Query 1	104 B	104 B	+0.0%
Query 2	936 B	936 B	+0.0%
Query 3	312 B	312 B	+0.0%
Query 4	827.5 MiB	831.1 MiB	+0.4%
Query 5	1.5 GiB	1.3 GiB	-8.3%
Query 6	0 B	0 B	0.0%
Query 7	60.0 MiB	40.3 MiB	-32.8%
Query 8	944.6 MiB	942.6 MiB	-0.2%
Query 9	552.2 MiB	552.2 MiB	+0.0%
Query 10	70.8 MiB	71.8 MiB	+1.4%
Query 11	76.9 MiB	77.7 MiB	+0.9%
Query 12	1.6 GiB	1.5 GiB	-7.1%
Query 13	1.6 GiB	1.5 GiB	-5.2%
Query 14	1.6 GiB	1.6 GiB	-2.0%
Query 15	1.2 GiB	1.2 GiB	-1.3%
Query 16	3.3 GiB	3.3 GiB	+0.3%
Query 17	3.2 GiB	3.2 GiB	-1.1%
Query 18	7.3 GiB	7.5 GiB	+2.4%
Query 19	0 B	0 B	0.0%
Query 20	104 B	104 B	+0.0%
Query 21	1.4 MiB	1.5 MiB	+8.4%
Query 22	2.4 MiB	2.3 MiB	-3.2%
Query 23	11.0 MiB	12.2 MiB	+10.8%
Query 24	6.9 MiB	6.7 MiB	-2.6%
Query 25	8.1 MiB	8.1 MiB	+0.0%
Query 26	7.3 MiB	8.1 MiB	+9.8%
Query 27	2.5 MiB	2.5 MiB	+0.0%
Query 28	1.9 GiB	1.8 GiB	-3.5%
Query 29	624 B	624 B	+0.0%
Query 30	665.0 MiB	646.8 MiB	-2.7%
Query 31	1.5 GiB	1.5 GiB	-3.0%
Query 32	7.2 GiB	7.1 GiB	-0.3%
Query 33	9.9 GiB	9.6 GiB	-3.8%
Query 34	9.9 GiB	9.9 GiB	-0.7%
Query 35	665.4 MiB	674.7 MiB	+1.4%
Query 36	105.4 MiB	100.1 MiB	-5.0%
Query 37	8.1 MiB	8.0 MiB	-0.7%
Query 38	5.2 MiB	5.6 MiB	+7.2%
Query 39	213.4 MiB	213.4 MiB	+0.0%
Query 40	2.3 MiB	2.3 MiB	+0.1%
Query 41	3.3 MiB	3.1 MiB	-8.3%
Query 42	1.7 MiB	1.7 MiB	+0.4%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
clickbench_partitioned	base (`5191bcc46d7e617bb8d11277fdc180b388141b11`)	9.9 GiB	12.6 GiB	2.7 GiB	1.3×
clickbench_partitioned	changed (`8caaa35c2c4fce9f1c602e27fe114be0cf0acd0e`)	9.9 GiB	13.3 GiB	3.4 GiB	1.3×

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	200.0s
Peak memory	12.6 GiB
Avg memory	5.4 GiB
CPU user	1947.8s
CPU sys	124.2s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	195.1s
Peak memory	13.3 GiB
Avg memory	5.7 GiB
CPU user	1875.5s
CPU sys	125.3s
Peak spill	0 B

File an issue against this benchmark runner

Rich-T-kid · 2026-11-09T18:34:18Z

run benchmark dictionary_group_values

env:
  BENCH_FILTER: "dict_(alternating_values_(single|multi)|shared_values_multi|fresh_values_per_batch)"
baseline:
  ref: "b14a6523cd8be7d5033ea4651ad1487e92a8e27d"
changed:
  ref: "4b19df3cf2916d7ef7eccccd7d668a6a210cf0ea"

adriangbot · 2026-11-09T18:37:52Z

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

Benchmarks requested: dictionary_group_values

Runner log (last 40 lines)

2026-11-09T18:37:05.740600Z  INFO runner starting benchmark runner bench_type=Datafusion, pr_url=https://github.com/apache/datafusion/pull/26826, benchmarks=dictionary_group_values
2026-11-09T18:37:05.816886Z  INFO benchmark_controller::runner::bench_datafusion === Cloning PR branch ===
2026-11-09T18:37:05.816907Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["clone", "--depth=200", "https://github.com/apache/datafusion.git", "/workspace/datafusion-branch"], cwd="/"
2026-11-09T18:37:15.828162Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["fetch", "origin", "refs/pull/26826/head:rich-T-kid/aggregate-dictionary-hash-cache", "main"], cwd="/workspace/datafusion-branch"
2026-11-09T18:37:20.829844Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["checkout", "rich-T-kid/aggregate-dictionary-hash-cache"], cwd="/workspace/datafusion-branch"
2026-11-09T18:37:25.832147Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["merge-base", "HEAD", "origin/main"], cwd="/workspace/datafusion-branch"
2026-11-09T18:37:30.833726Z  INFO benchmark_controller::runner::bench_datafusion === Checking out custom changed ref === changed_ref=4b19df3cf2916d7ef7eccccd7d668a6a210cf0ea
2026-11-09T18:37:30.833741Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["fetch", "origin", "refs/pull/26826/head"], cwd="/workspace/datafusion-branch"
2026-11-09T18:37:35.836238Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["fetch", "origin"], cwd="/workspace/datafusion-branch"
2026-11-09T18:37:40.838292Z  INFO benchmark_controller::runner::shell running command cmd=git, args=["-c", "advice.detachedHead=false", "checkout", "4b19df3cf2916d7ef7eccccd7d668a6a210cf0ea"], cwd="/workspace/datafusion-branch"
2026-11-09T18:37:45.839993Z ERROR runner benchmark failed error=git checkout: git -c advice.detachedHead=false checkout 4b19df3cf2916d7ef7eccccd7d668a6a210cf0ea exited with code 128
stdout:

stderr:
fatal: reference is not a tree: 4b19df3cf2916d7ef7eccccd7d668a6a210cf0ea

Kubernetes message

Job has reached the specified backoff limit

File an issue against this benchmark runner

Rich-T-kid · 2026-11-09T18:38:27Z

run benchmark dictionary_group_values

env:
  BENCH_FILTER: "dict_shared_values_arc"

adriangbot · 2026-11-09T18:39:38Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6748266850-4708-74vgf 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark dictionary_group_values
env:
  BENCH_FILTER: "dict_shared_values_arc"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-09T18:43:59Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark dictionary_group_values
env:
  BENCH_FILTER: "dict_shared_values_arc"

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

group                                                             HEAD                                   rich-T-kid_aggregate-dictionary-hash-cache
-----                                                             ----                                   ------------------------------------------
dict_shared_values_arc/shared_values_arc/size_8192_card_100000    1.79     18.6±0.24ms 13.5 MElem/sec    1.00     10.4±0.10ms 24.1 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_32768     1.62      7.9±0.02ms 31.6 MElem/sec    1.00      4.9±0.01ms 51.0 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_500000    2.60     71.8±2.37ms  3.5 MElem/sec    1.00     27.6±2.56ms  9.1 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_65536     1.81     13.9±0.07ms 18.0 MElem/sec    1.00      7.7±0.03ms 32.6 MElem/sec
dict_shared_values_arc/shared_values_arc/size_8192_card_8192      2.47      2.8±0.02ms 90.1 MElem/sec    1.00  1121.2±239.88µs 223.0 MElem/sec

Resource Usage

dictionary_group_values — base (merge-base)

Metric	Value
Wall time	130.0s
Peak memory	100.0 MiB
Avg memory	23.3 MiB
CPU user	60.4s
CPU sys	1.6s
Peak spill	0 B

dictionary_group_values — branch

Metric	Value
Wall time	120.0s
Peak memory	125.5 MiB
Avg memory	26.5 MiB
CPU user	61.1s
CPU sys	1.5s
Peak spill	0 B

File an issue against this benchmark runner

Rich-T-kid · 2026-11-09T18:50:01Z

run benchmark tpch tpcds clickbench_partitioned

env:
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"

adriangbot · 2026-11-09T18:52:26Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6748503722-4709-f6drr 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-09T18:53:24Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6748503722-4710-fq75w 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-09T18:53:38Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c6748503722-4711-5ttnz 6.12.105+ #1 SMP Sat Sep 19 07:55:25 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 rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"

Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-11-09T18:56:16Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark tpch
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃     HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 39.70 ms │                                   39.76 ms │ no change │
│ QQuery 2  │ 18.87 ms │                                   18.87 ms │ no change │
│ QQuery 3  │ 28.67 ms │                                   28.64 ms │ no change │
│ QQuery 4  │ 17.29 ms │                                   17.19 ms │ no change │
│ QQuery 5  │ 35.71 ms │                                   35.41 ms │ no change │
│ QQuery 6  │ 13.76 ms │                                   13.73 ms │ no change │
│ QQuery 7  │ 42.02 ms │                                   41.63 ms │ no change │
│ QQuery 8  │ 40.67 ms │                                   40.76 ms │ no change │
│ QQuery 9  │ 49.25 ms │                                   50.34 ms │ no change │
│ QQuery 10 │ 58.40 ms │                                   57.56 ms │ no change │
│ QQuery 11 │ 13.79 ms │                                   13.83 ms │ no change │
│ QQuery 12 │ 20.41 ms │                                   20.42 ms │ no change │
│ QQuery 13 │ 46.85 ms │                                   46.47 ms │ no change │
│ QQuery 14 │ 23.95 ms │                                   24.03 ms │ no change │
│ QQuery 15 │ 31.08 ms │                                   30.99 ms │ no change │
│ QQuery 16 │ 13.31 ms │                                   13.39 ms │ no change │
│ QQuery 17 │ 69.22 ms │                                   68.98 ms │ no change │
│ QQuery 18 │ 64.54 ms │                                   64.24 ms │ no change │
│ QQuery 19 │ 29.15 ms │                                   28.90 ms │ no change │
│ QQuery 20 │ 33.75 ms │                                   33.84 ms │ no change │
│ QQuery 21 │ 53.91 ms │                                   53.42 ms │ no change │
│ QQuery 22 │ 15.72 ms │                                   15.59 ms │ no change │
└───────────┴──────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 760.04ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 757.99ms │
│ Average Time (HEAD)                                       │  34.55ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  34.45ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Query     ┃                           HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃    Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ QQuery 1  │ 39.70 / 41.14 ±1.23 / 42.94 ms │             39.76 / 41.48 ±1.30 / 43.17 ms │ no change │
│ QQuery 2  │ 18.87 / 19.66 ±1.13 / 21.87 ms │             18.87 / 19.21 ±0.24 / 19.49 ms │ no change │
│ QQuery 3  │ 28.67 / 29.45 ±1.10 / 31.63 ms │             28.64 / 29.68 ±1.28 / 32.13 ms │ no change │
│ QQuery 4  │ 17.29 / 18.00 ±0.78 / 19.06 ms │             17.19 / 17.49 ±0.51 / 18.52 ms │ no change │
│ QQuery 5  │ 35.71 / 36.35 ±1.04 / 38.42 ms │             35.41 / 35.87 ±0.28 / 36.17 ms │ no change │
│ QQuery 6  │ 13.76 / 13.99 ±0.22 / 14.37 ms │             13.73 / 14.23 ±0.58 / 15.26 ms │ no change │
│ QQuery 7  │ 42.02 / 43.07 ±0.83 / 44.33 ms │             41.63 / 42.25 ±0.53 / 43.13 ms │ no change │
│ QQuery 8  │ 40.67 / 41.09 ±0.30 / 41.45 ms │             40.76 / 42.23 ±1.86 / 45.87 ms │ no change │
│ QQuery 9  │ 49.25 / 50.69 ±0.91 / 52.05 ms │             50.34 / 51.25 ±1.01 / 53.20 ms │ no change │
│ QQuery 10 │ 58.40 / 58.90 ±0.42 / 59.59 ms │             57.56 / 57.70 ±0.20 / 58.10 ms │ no change │
│ QQuery 11 │ 13.79 / 13.91 ±0.11 / 14.09 ms │             13.83 / 13.94 ±0.13 / 14.16 ms │ no change │
│ QQuery 12 │ 20.41 / 20.98 ±0.53 / 21.97 ms │             20.42 / 20.89 ±0.54 / 21.90 ms │ no change │
│ QQuery 13 │ 46.85 / 47.05 ±0.12 / 47.17 ms │             46.47 / 47.00 ±0.56 / 48.06 ms │ no change │
│ QQuery 14 │ 23.95 / 24.04 ±0.06 / 24.14 ms │             24.03 / 24.08 ±0.07 / 24.21 ms │ no change │
│ QQuery 15 │ 31.08 / 31.79 ±1.32 / 34.42 ms │             30.99 / 32.51 ±0.78 / 33.13 ms │ no change │
│ QQuery 16 │ 13.31 / 13.47 ±0.11 / 13.61 ms │             13.39 / 13.52 ±0.11 / 13.66 ms │ no change │
│ QQuery 17 │ 69.22 / 69.89 ±0.60 / 70.88 ms │             68.98 / 69.80 ±0.70 / 70.69 ms │ no change │
│ QQuery 18 │ 64.54 / 67.72 ±1.96 / 70.67 ms │             64.24 / 66.51 ±2.14 / 69.99 ms │ no change │
│ QQuery 19 │ 29.15 / 29.34 ±0.26 / 29.85 ms │             28.90 / 29.13 ±0.15 / 29.35 ms │ no change │
│ QQuery 20 │ 33.75 / 34.20 ±0.27 / 34.49 ms │             33.84 / 34.15 ±0.17 / 34.33 ms │ no change │
│ QQuery 21 │ 53.91 / 54.50 ±0.79 / 56.06 ms │             53.42 / 54.00 ±0.37 / 54.45 ms │ no change │
│ QQuery 22 │ 15.72 / 15.79 ±0.10 / 15.98 ms │             15.59 / 15.92 ±0.38 / 16.65 ms │ no change │
└───────────┴────────────────────────────────┴────────────────────────────────────────────┴───────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 775.04ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 772.84ms │
│ Average Time (HEAD)                                       │  35.23ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │  35.13ms │
│ Queries Faster                                            │        0 │
│ Queries Slower                                            │        0 │
│ Queries with No Change                                    │       22 │
│ Queries with Failure                                      │        0 │
└───────────────────────────────────────────────────────────┴──────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 5191bcc (merge-base) | Changed: rich-T-kid/aggregate-dictionary-hash-cache

tpch — tpch_sf1

Query	Base	Changed	Change
Query 1	30.1 MiB	30.1 MiB	-0.0%
Query 2	23.1 MiB	23.1 MiB	+0.0%
Query 3	11.5 MiB	11.5 MiB	+0.0%
Query 4	22.3 MiB	32.2 MiB	+44.5%
Query 5	32.8 MiB	32.8 MiB	-0.0%
Query 6	832 B	832 B	+0.0%
Query 7	107.6 MiB	107.6 MiB	+0.0%
Query 8	39.8 MiB	39.8 MiB	+0.0%
Query 9	155.6 MiB	155.6 MiB	+0.0%
Query 10	53.3 MiB	53.3 MiB	+0.0%
Query 11	41.5 MiB	61.7 MiB	+48.6%
Query 12	12.6 MiB	22.5 MiB	+78.0%
Query 13	62.5 MiB	62.3 MiB	-0.2%
Query 14	7.7 MiB	7.7 MiB	+0.0%
Query 15	12.7 MiB	13.0 MiB	+2.4%
Query 16	33.5 MiB	43.2 MiB	+28.9%
Query 17	150.3 MiB	150.3 MiB	+0.0%
Query 18	127.4 MiB	127.3 MiB	-0.0%
Query 19	228.6 KiB	228.6 KiB	+0.0%
Query 20	63.6 MiB	58.7 MiB	-7.7%
Query 21	37.7 MiB	37.7 MiB	+0.0%
Query 22	32.0 MiB	31.8 MiB	-0.6%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpch	base (`5191bcc (merge-base)`)	155.6 MiB	1.2 GiB	1.0 GiB	7.6×
tpch	changed (`rich-T-kid/aggregate-dictionary-hash-cache`)	155.6 MiB	1.1 GiB	942.2 MiB	7.1×

Resource Usage

tpch — base (merge-base)

Metric	Value
Wall time	5.0s
Peak memory	1.2 GiB
Avg memory	487.1 MiB
CPU user	20.2s
CPU sys	1.7s
Peak spill	0 B

tpch — branch

Metric	Value
Wall time	5.0s
Peak memory	1.1 GiB
Avg memory	469.2 MiB
CPU user	20.0s
CPU sys	1.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-11-09T18:59:07Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark tpcds
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │    6.08 ms │                                    6.16 ms │     no change │
│ QQuery 2  │   79.43 ms │                                   78.94 ms │     no change │
│ QQuery 3  │   30.53 ms │                                   30.47 ms │     no change │
│ QQuery 4  │  714.55 ms │                                  638.43 ms │ +1.12x faster │
│ QQuery 5  │   54.03 ms │                                   54.21 ms │     no change │
│ QQuery 6  │   35.68 ms │                                   36.34 ms │     no change │
│ QQuery 7  │   74.18 ms │                                   73.74 ms │     no change │
│ QQuery 8  │   40.26 ms │                                   39.69 ms │     no change │
│ QQuery 9  │   53.28 ms │                                   53.10 ms │     no change │
│ QQuery 10 │   61.15 ms │                                   60.06 ms │     no change │
│ QQuery 11 │  453.51 ms │                                  400.16 ms │ +1.13x faster │
│ QQuery 12 │   27.22 ms │                                   28.53 ms │     no change │
│ QQuery 13 │  117.95 ms │                                  118.86 ms │     no change │
│ QQuery 14 │  398.69 ms │                                  394.92 ms │     no change │
│ QQuery 15 │   59.49 ms │                                   58.74 ms │     no change │
│ QQuery 16 │    6.76 ms │                                    6.80 ms │     no change │
│ QQuery 17 │   80.81 ms │                                   81.12 ms │     no change │
│ QQuery 18 │  111.03 ms │                                  114.41 ms │     no change │
│ QQuery 19 │   44.06 ms │                                   44.23 ms │     no change │
│ QQuery 20 │   33.79 ms │                                   33.34 ms │     no change │
│ QQuery 21 │   18.29 ms │                                   18.13 ms │     no change │
│ QQuery 22 │   78.54 ms │                                   67.09 ms │ +1.17x faster │
│ QQuery 23 │  334.91 ms │                                  330.15 ms │     no change │
│ QQuery 24 │  214.79 ms │                                  212.66 ms │     no change │
│ QQuery 25 │  111.72 ms │                                  109.32 ms │     no change │
│ QQuery 26 │   46.72 ms │                                   46.96 ms │     no change │
│ QQuery 27 │    6.93 ms │                                    6.88 ms │     no change │
│ QQuery 28 │   62.21 ms │                                   59.60 ms │     no change │
│ QQuery 29 │   98.88 ms │                                   98.79 ms │     no change │
│ QQuery 30 │   39.44 ms │                                   39.08 ms │     no change │
│ QQuery 31 │  117.45 ms │                                  115.54 ms │     no change │
│ QQuery 32 │   21.83 ms │                                   21.68 ms │     no change │
│ QQuery 33 │   39.02 ms │                                   38.47 ms │     no change │
│ QQuery 34 │   10.53 ms │                                   10.38 ms │     no change │
│ QQuery 35 │   78.60 ms │                                   75.48 ms │     no change │
│ QQuery 36 │    7.06 ms │                                    6.58 ms │ +1.07x faster │
│ QQuery 37 │    8.41 ms │                                    8.18 ms │     no change │
│ QQuery 38 │   71.23 ms │                                   70.52 ms │     no change │
│ QQuery 39 │   79.90 ms │                                   77.61 ms │     no change │
│ QQuery 40 │   25.76 ms │                                   24.73 ms │     no change │
│ QQuery 41 │   15.07 ms │                                   14.67 ms │     no change │
│ QQuery 42 │   25.62 ms │                                   25.20 ms │     no change │
│ QQuery 43 │    5.54 ms │                                    5.43 ms │     no change │
│ QQuery 44 │    9.71 ms │                                    9.35 ms │     no change │
│ QQuery 45 │   44.87 ms │                                   40.95 ms │ +1.10x faster │
│ QQuery 46 │   12.42 ms │                                   12.01 ms │     no change │
│ QQuery 47 │  239.44 ms │                                  229.56 ms │     no change │
│ QQuery 48 │   96.47 ms │                                   93.76 ms │     no change │
│ QQuery 49 │   73.89 ms │                                   73.88 ms │     no change │
│ QQuery 50 │   61.13 ms │                                   60.56 ms │     no change │
│ QQuery 51 │   98.36 ms │                                   98.08 ms │     no change │
│ QQuery 52 │   25.68 ms │                                   25.87 ms │     no change │
│ QQuery 53 │   30.63 ms │                                   31.20 ms │     no change │
│ QQuery 54 │   57.02 ms │                                   57.36 ms │     no change │
│ QQuery 55 │   25.96 ms │                                   25.62 ms │     no change │
│ QQuery 56 │   41.31 ms │                                   40.24 ms │     no change │
│ QQuery 57 │  163.50 ms │                                  164.61 ms │     no change │
│ QQuery 58 │  112.78 ms │                                  108.09 ms │     no change │
│ QQuery 59 │  112.90 ms │                                  112.38 ms │     no change │
│ QQuery 60 │   41.14 ms │                                   40.89 ms │     no change │
│ QQuery 61 │   11.47 ms │                                   11.51 ms │     no change │
│ QQuery 62 │   40.53 ms │                                   40.12 ms │     no change │
│ QQuery 63 │   31.11 ms │                                   30.87 ms │     no change │
│ QQuery 64 │  389.61 ms │                                  381.99 ms │     no change │
│ QQuery 65 │   85.98 ms │                                   82.51 ms │     no change │
│ QQuery 66 │   77.07 ms │                                   77.97 ms │     no change │
│ QQuery 67 │  248.72 ms │                                  249.96 ms │     no change │
│ QQuery 68 │   11.97 ms │                                   12.83 ms │  1.07x slower │
│ QQuery 69 │   55.51 ms │                                   56.22 ms │     no change │
│ QQuery 70 │   99.68 ms │                                  101.49 ms │     no change │
│ QQuery 71 │   36.86 ms │                                   37.68 ms │     no change │
│ QQuery 72 │ 1570.07 ms │                                 1785.90 ms │  1.14x slower │
│ QQuery 73 │   10.07 ms │                                   10.46 ms │     no change │
│ QQuery 74 │  302.51 ms │                                  308.32 ms │     no change │
│ QQuery 75 │  143.43 ms │                                  150.06 ms │     no change │
│ QQuery 76 │   35.68 ms │                                   37.60 ms │  1.05x slower │
│ QQuery 77 │   62.32 ms │                                   64.01 ms │     no change │
│ QQuery 78 │  289.87 ms │                                  309.86 ms │  1.07x slower │
│ QQuery 79 │   69.48 ms │                                   73.41 ms │  1.06x slower │
│ QQuery 80 │   98.62 ms │                                  103.79 ms │  1.05x slower │
│ QQuery 81 │   29.28 ms │                                   30.29 ms │     no change │
│ QQuery 82 │   17.76 ms │                                   18.55 ms │     no change │
│ QQuery 83 │   34.01 ms │                                   34.34 ms │     no change │
│ QQuery 84 │   31.94 ms │                                   32.23 ms │     no change │
│ QQuery 85 │   97.80 ms │                                  101.22 ms │     no change │
│ QQuery 86 │   25.86 ms │                                   26.16 ms │     no change │
│ QQuery 87 │   69.27 ms │                                   69.48 ms │     no change │
│ QQuery 88 │   60.46 ms │                                   61.55 ms │     no change │
│ QQuery 89 │   36.78 ms │                                   39.13 ms │  1.06x slower │
│ QQuery 90 │   16.90 ms │                                   17.45 ms │     no change │
│ QQuery 91 │   41.00 ms │                                   41.06 ms │     no change │
│ QQuery 92 │   30.97 ms │                                   31.05 ms │     no change │
│ QQuery 93 │   50.41 ms │                                   51.82 ms │     no change │
│ QQuery 94 │   39.35 ms │                                   41.24 ms │     no change │
│ QQuery 95 │   81.71 ms │                                   85.60 ms │     no change │
│ QQuery 96 │   24.32 ms │                                   24.75 ms │     no change │
│ QQuery 97 │   52.00 ms │                                   53.52 ms │     no change │
│ QQuery 98 │   39.18 ms │                                   40.01 ms │     no change │
│ QQuery 99 │   57.35 ms │                                   59.33 ms │     no change │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 9755.06ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 9845.07ms │
│ Average Time (HEAD)                                       │   98.54ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   99.45ms │
│ Queries Faster                                            │         5 │
│ Queries Slower                                            │         7 │
│ Queries with No Change                                    │        87 │
│ Queries with Failure                                      │         0 │
└───────────────────────────────────────────────────────────┴───────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │           6.08 / 6.78 ±1.20 / 9.18 ms │                6.16 / 6.77 ±1.08 / 8.93 ms │     no change │
│ QQuery 2  │        79.43 / 80.48 ±0.65 / 81.19 ms │             78.94 / 79.68 ±0.68 / 80.88 ms │     no change │
│ QQuery 3  │        30.53 / 30.92 ±0.27 / 31.32 ms │             30.47 / 30.64 ±0.15 / 30.92 ms │     no change │
│ QQuery 4  │     714.55 / 718.88 ±4.43 / 726.15 ms │          638.43 / 646.32 ±7.43 / 658.78 ms │ +1.11x faster │
│ QQuery 5  │        54.03 / 54.75 ±0.72 / 56.10 ms │             54.21 / 54.81 ±0.72 / 56.12 ms │     no change │
│ QQuery 6  │        35.68 / 36.57 ±0.55 / 37.14 ms │             36.34 / 37.37 ±0.54 / 37.83 ms │     no change │
│ QQuery 7  │        74.18 / 75.73 ±1.85 / 79.26 ms │             73.74 / 74.66 ±0.53 / 75.31 ms │     no change │
│ QQuery 8  │        40.26 / 40.89 ±0.66 / 42.13 ms │             39.69 / 40.14 ±0.43 / 40.77 ms │     no change │
│ QQuery 9  │        53.28 / 57.00 ±3.15 / 62.26 ms │             53.10 / 55.77 ±2.40 / 59.76 ms │     no change │
│ QQuery 10 │        61.15 / 61.43 ±0.20 / 61.71 ms │             60.06 / 60.59 ±0.30 / 60.87 ms │     no change │
│ QQuery 11 │     453.51 / 467.91 ±8.77 / 477.73 ms │         400.16 / 415.99 ±11.17 / 429.23 ms │ +1.12x faster │
│ QQuery 12 │        27.22 / 28.18 ±0.63 / 29.14 ms │             28.53 / 29.46 ±0.64 / 30.44 ms │     no change │
│ QQuery 13 │     117.95 / 118.59 ±0.59 / 119.50 ms │          118.86 / 121.57 ±2.88 / 126.61 ms │     no change │
│ QQuery 14 │     398.69 / 400.43 ±1.92 / 404.14 ms │          394.92 / 398.59 ±2.60 / 402.79 ms │     no change │
│ QQuery 15 │        59.49 / 63.86 ±3.07 / 68.87 ms │             58.74 / 61.84 ±1.70 / 63.76 ms │     no change │
│ QQuery 16 │           6.76 / 7.07 ±0.20 / 7.26 ms │                6.80 / 7.01 ±0.11 / 7.15 ms │     no change │
│ QQuery 17 │        80.81 / 81.97 ±1.14 / 84.04 ms │             81.12 / 81.59 ±0.34 / 81.99 ms │     no change │
│ QQuery 18 │     111.03 / 115.34 ±2.52 / 117.96 ms │          114.41 / 118.00 ±5.83 / 129.61 ms │     no change │
│ QQuery 19 │        44.06 / 44.63 ±0.33 / 44.97 ms │             44.23 / 44.52 ±0.26 / 44.85 ms │     no change │
│ QQuery 20 │        33.79 / 34.19 ±0.30 / 34.70 ms │             33.34 / 33.85 ±0.37 / 34.31 ms │     no change │
│ QQuery 21 │        18.29 / 18.98 ±0.80 / 20.52 ms │             18.13 / 18.92 ±1.01 / 20.92 ms │     no change │
│ QQuery 22 │        78.54 / 81.17 ±2.99 / 85.99 ms │             67.09 / 69.17 ±1.07 / 70.07 ms │ +1.17x faster │
│ QQuery 23 │     334.91 / 341.02 ±6.63 / 353.58 ms │          330.15 / 340.42 ±7.07 / 348.01 ms │     no change │
│ QQuery 24 │     214.79 / 221.65 ±6.72 / 233.06 ms │          212.66 / 220.29 ±5.74 / 226.74 ms │     no change │
│ QQuery 25 │     111.72 / 114.64 ±2.94 / 120.07 ms │          109.32 / 114.95 ±3.63 / 120.46 ms │     no change │
│ QQuery 26 │        46.72 / 47.38 ±0.53 / 48.01 ms │             46.96 / 48.37 ±1.83 / 52.00 ms │     no change │
│ QQuery 27 │           6.93 / 7.20 ±0.14 / 7.30 ms │                6.88 / 7.44 ±0.29 / 7.69 ms │     no change │
│ QQuery 28 │        62.21 / 63.02 ±0.68 / 63.95 ms │             59.60 / 63.38 ±2.00 / 65.37 ms │     no change │
│ QQuery 29 │      98.88 / 100.73 ±1.83 / 104.08 ms │           98.79 / 100.15 ±1.47 / 102.48 ms │     no change │
│ QQuery 30 │        39.44 / 40.15 ±0.64 / 41.06 ms │             39.08 / 40.38 ±1.66 / 43.59 ms │     no change │
│ QQuery 31 │     117.45 / 118.60 ±1.22 / 120.94 ms │          115.54 / 120.21 ±4.29 / 128.32 ms │     no change │
│ QQuery 32 │        21.83 / 22.42 ±0.55 / 23.22 ms │             21.68 / 22.01 ±0.29 / 22.48 ms │     no change │
│ QQuery 33 │        39.02 / 40.08 ±0.63 / 40.95 ms │             38.47 / 38.81 ±0.20 / 39.01 ms │     no change │
│ QQuery 34 │        10.53 / 10.83 ±0.21 / 11.19 ms │             10.38 / 10.99 ±1.01 / 13.01 ms │     no change │
│ QQuery 35 │        78.60 / 79.83 ±0.73 / 80.72 ms │             75.48 / 77.92 ±1.34 / 79.55 ms │     no change │
│ QQuery 36 │           7.06 / 7.24 ±0.16 / 7.53 ms │                6.58 / 6.97 ±0.30 / 7.38 ms │     no change │
│ QQuery 37 │           8.41 / 8.58 ±0.14 / 8.75 ms │                8.18 / 8.37 ±0.19 / 8.71 ms │     no change │
│ QQuery 38 │        71.23 / 72.26 ±1.17 / 74.51 ms │             70.52 / 71.89 ±1.31 / 74.29 ms │     no change │
│ QQuery 39 │        79.90 / 82.73 ±2.55 / 86.00 ms │             77.61 / 79.31 ±1.31 / 81.57 ms │     no change │
│ QQuery 40 │        25.76 / 26.73 ±1.21 / 28.90 ms │             24.73 / 25.43 ±0.59 / 26.27 ms │     no change │
│ QQuery 41 │        15.07 / 15.16 ±0.09 / 15.33 ms │             14.67 / 14.91 ±0.20 / 15.27 ms │     no change │
│ QQuery 42 │        25.62 / 26.36 ±0.60 / 27.43 ms │             25.20 / 26.49 ±2.36 / 31.20 ms │     no change │
│ QQuery 43 │           5.54 / 5.76 ±0.15 / 6.00 ms │                5.43 / 5.59 ±0.14 / 5.81 ms │     no change │
│ QQuery 44 │          9.71 / 9.87 ±0.13 / 10.09 ms │                9.35 / 9.72 ±0.25 / 9.99 ms │     no change │
│ QQuery 45 │        44.87 / 46.60 ±1.15 / 47.94 ms │             40.95 / 42.79 ±1.72 / 45.51 ms │ +1.09x faster │
│ QQuery 46 │        12.42 / 12.61 ±0.17 / 12.80 ms │             12.01 / 12.36 ±0.28 / 12.61 ms │     no change │
│ QQuery 47 │     239.44 / 245.81 ±6.09 / 256.59 ms │          229.56 / 235.25 ±3.86 / 239.33 ms │     no change │
│ QQuery 48 │       96.47 / 97.52 ±1.59 / 100.68 ms │             93.76 / 95.99 ±2.02 / 99.41 ms │     no change │
│ QQuery 49 │        73.89 / 74.70 ±0.71 / 75.58 ms │             73.88 / 74.45 ±0.53 / 75.13 ms │     no change │
│ QQuery 50 │        61.13 / 64.42 ±4.01 / 72.30 ms │             60.56 / 62.55 ±2.22 / 66.85 ms │     no change │
│ QQuery 51 │      98.36 / 100.72 ±2.56 / 105.69 ms │           98.08 / 100.52 ±3.15 / 106.75 ms │     no change │
│ QQuery 52 │        25.68 / 26.14 ±0.23 / 26.30 ms │             25.87 / 26.29 ±0.28 / 26.60 ms │     no change │
│ QQuery 53 │        30.63 / 33.01 ±4.23 / 41.44 ms │             31.20 / 31.47 ±0.17 / 31.73 ms │     no change │
│ QQuery 54 │        57.02 / 57.76 ±0.69 / 58.77 ms │             57.36 / 57.83 ±0.40 / 58.45 ms │     no change │
│ QQuery 55 │        25.96 / 26.52 ±0.44 / 27.05 ms │             25.62 / 26.23 ±0.79 / 27.75 ms │     no change │
│ QQuery 56 │        41.31 / 41.80 ±0.47 / 42.60 ms │             40.24 / 41.63 ±0.90 / 43.00 ms │     no change │
│ QQuery 57 │     163.50 / 167.68 ±4.50 / 176.44 ms │          164.61 / 166.53 ±1.09 / 167.76 ms │     no change │
│ QQuery 58 │     112.78 / 114.93 ±2.51 / 119.41 ms │          108.09 / 111.39 ±3.66 / 118.49 ms │     no change │
│ QQuery 59 │     112.90 / 113.92 ±1.05 / 115.64 ms │          112.38 / 115.12 ±3.71 / 122.42 ms │     no change │
│ QQuery 60 │        41.14 / 41.75 ±0.50 / 42.65 ms │             40.89 / 41.84 ±0.68 / 42.80 ms │     no change │
│ QQuery 61 │        11.47 / 11.73 ±0.15 / 11.90 ms │             11.51 / 11.81 ±0.28 / 12.31 ms │     no change │
│ QQuery 62 │        40.53 / 42.55 ±1.89 / 46.05 ms │             40.12 / 41.84 ±2.32 / 46.41 ms │     no change │
│ QQuery 63 │        31.11 / 31.82 ±0.59 / 32.78 ms │             30.87 / 31.38 ±0.41 / 31.94 ms │     no change │
│ QQuery 64 │     389.61 / 393.27 ±4.28 / 401.39 ms │          381.99 / 389.35 ±5.91 / 398.47 ms │     no change │
│ QQuery 65 │        85.98 / 87.49 ±1.52 / 90.17 ms │             82.51 / 86.44 ±5.12 / 96.24 ms │     no change │
│ QQuery 66 │        77.07 / 79.30 ±1.36 / 81.02 ms │             77.97 / 79.86 ±1.39 / 81.70 ms │     no change │
│ QQuery 67 │     248.72 / 262.27 ±9.96 / 274.92 ms │          249.96 / 254.41 ±4.37 / 261.96 ms │     no change │
│ QQuery 68 │        11.97 / 12.08 ±0.10 / 12.26 ms │             12.83 / 12.89 ±0.04 / 12.95 ms │  1.07x slower │
│ QQuery 69 │        55.51 / 55.92 ±0.30 / 56.34 ms │             56.22 / 59.38 ±4.26 / 67.76 ms │  1.06x slower │
│ QQuery 70 │      99.68 / 102.78 ±3.01 / 107.70 ms │          101.49 / 103.28 ±1.87 / 106.37 ms │     no change │
│ QQuery 71 │        36.86 / 37.33 ±0.47 / 38.10 ms │             37.68 / 38.24 ±0.49 / 39.05 ms │     no change │
│ QQuery 72 │ 1570.07 / 1657.65 ±53.86 / 1717.06 ms │      1785.90 / 1863.33 ±62.09 / 1957.90 ms │  1.12x slower │
│ QQuery 73 │        10.07 / 10.41 ±0.36 / 11.01 ms │             10.46 / 10.75 ±0.29 / 11.11 ms │     no change │
│ QQuery 74 │     302.51 / 309.53 ±5.64 / 319.11 ms │          308.32 / 312.46 ±3.39 / 318.10 ms │     no change │
│ QQuery 75 │     143.43 / 146.95 ±3.30 / 153.03 ms │          150.06 / 154.81 ±6.63 / 167.97 ms │  1.05x slower │
│ QQuery 76 │        35.68 / 38.79 ±4.85 / 48.43 ms │             37.60 / 38.09 ±0.31 / 38.45 ms │     no change │
│ QQuery 77 │        62.32 / 63.32 ±0.79 / 64.72 ms │             64.01 / 64.96 ±1.04 / 66.92 ms │     no change │
│ QQuery 78 │    289.87 / 299.37 ±11.41 / 321.52 ms │          309.86 / 322.20 ±8.37 / 335.85 ms │  1.08x slower │
│ QQuery 79 │        69.48 / 72.60 ±4.61 / 81.78 ms │             73.41 / 74.64 ±0.66 / 75.35 ms │     no change │
│ QQuery 80 │       98.62 / 99.75 ±0.69 / 100.65 ms │          103.79 / 105.13 ±1.41 / 107.60 ms │  1.05x slower │
│ QQuery 81 │        29.28 / 31.64 ±2.04 / 35.42 ms │             30.29 / 31.01 ±0.75 / 32.30 ms │     no change │
│ QQuery 82 │        17.76 / 19.60 ±1.92 / 23.16 ms │             18.55 / 19.04 ±0.41 / 19.74 ms │     no change │
│ QQuery 83 │        34.01 / 35.30 ±1.58 / 38.37 ms │             34.34 / 34.81 ±0.36 / 35.41 ms │     no change │
│ QQuery 84 │        31.94 / 32.24 ±0.18 / 32.51 ms │             32.23 / 32.84 ±0.59 / 33.94 ms │     no change │
│ QQuery 85 │       97.80 / 98.79 ±1.12 / 100.81 ms │          101.22 / 104.76 ±5.05 / 114.73 ms │  1.06x slower │
│ QQuery 86 │        25.86 / 27.39 ±1.97 / 31.20 ms │             26.16 / 26.66 ±0.29 / 27.06 ms │     no change │
│ QQuery 87 │        69.27 / 69.95 ±0.44 / 70.57 ms │             69.48 / 70.19 ±0.70 / 71.13 ms │     no change │
│ QQuery 88 │        60.46 / 61.39 ±0.47 / 61.71 ms │             61.55 / 61.90 ±0.23 / 62.23 ms │     no change │
│ QQuery 89 │        36.78 / 37.16 ±0.24 / 37.48 ms │             39.13 / 39.73 ±0.85 / 41.40 ms │  1.07x slower │
│ QQuery 90 │        16.90 / 17.18 ±0.21 / 17.52 ms │             17.45 / 17.68 ±0.17 / 17.94 ms │     no change │
│ QQuery 91 │        41.00 / 42.60 ±2.22 / 46.96 ms │             41.06 / 41.65 ±0.41 / 42.29 ms │     no change │
│ QQuery 92 │        30.97 / 32.25 ±1.56 / 34.99 ms │             31.05 / 31.98 ±0.69 / 32.92 ms │     no change │
│ QQuery 93 │        50.41 / 51.60 ±1.12 / 53.32 ms │             51.82 / 55.01 ±4.04 / 62.95 ms │  1.07x slower │
│ QQuery 94 │        39.35 / 39.54 ±0.14 / 39.72 ms │             41.24 / 42.66 ±1.67 / 45.42 ms │  1.08x slower │
│ QQuery 95 │        81.71 / 84.41 ±1.88 / 87.04 ms │             85.60 / 87.35 ±1.25 / 88.98 ms │     no change │
│ QQuery 96 │        24.32 / 24.57 ±0.21 / 24.85 ms │             24.75 / 25.28 ±0.46 / 25.90 ms │     no change │
│ QQuery 97 │        52.00 / 52.37 ±0.38 / 52.97 ms │             53.52 / 56.72 ±3.81 / 64.12 ms │  1.08x slower │
│ QQuery 98 │        39.18 / 39.58 ±0.24 / 39.86 ms │             40.01 / 41.18 ±0.95 / 42.69 ms │     no change │
│ QQuery 99 │        57.35 / 58.92 ±2.49 / 63.88 ms │             59.33 / 60.25 ±0.82 / 61.74 ms │     no change │
└───────────┴───────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 10025.29ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 10119.42ms │
│ Average Time (HEAD)                                       │   101.27ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   102.22ms │
│ Queries Faster                                            │          4 │
│ Queries Slower                                            │         11 │
│ Queries with No Change                                    │         84 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 5191bcc (merge-base) | Changed: rich-T-kid/aggregate-dictionary-hash-cache

tpcds — tpcds_sf1

Query	Base	Changed	Change
Query 1	198.0 KiB	199.6 KiB	+0.8%
Query 2	97.4 MiB	78.4 MiB	-19.5%
Query 3	2.3 MiB	2.1 MiB	-6.8%
Query 4	87.8 MiB	87.8 MiB	-0.1%
Query 5	12.5 MiB	12.5 MiB	+0.0%
Query 6	4.7 MiB	4.7 MiB	+0.2%
Query 7	3.6 MiB	3.8 MiB	+4.9%
Query 8	2.1 MiB	2.0 MiB	-3.8%
Query 9	32.1 KiB	32.1 KiB	+0.0%
Query 10	2.0 MiB	2.0 MiB	+0.0%
Query 11	58.9 MiB	58.9 MiB	+0.0%
Query 12	114.4 MiB	114.4 MiB	-0.0%
Query 13	13.0 MiB	13.7 MiB	+5.1%
Query 14	170.9 MiB	171.2 MiB	+0.2%
Query 15	4.1 MiB	4.2 MiB	+2.3%
Query 16	118.6 KiB	118.6 KiB	+0.0%
Query 17	41.0 MiB	41.0 MiB	-0.0%
Query 18	21.2 MiB	21.2 MiB	+0.0%
Query 19	2.4 MiB	2.2 MiB	-8.2%
Query 20	114.6 MiB	114.6 MiB	+0.0%
Query 21	4.3 MiB	3.2 MiB	-25.3%
Query 22	61.8 MiB	61.8 MiB	+0.0%
Query 23	213.2 MiB	212.3 MiB	-0.4%
Query 24	38.7 MiB	38.7 MiB	+0.0%
Query 25	99.9 MiB	99.9 MiB	+0.0%
Query 26	3.6 MiB	3.9 MiB	+6.9%
Query 27	1.6 MiB	1.7 MiB	+8.0%
Query 28	1.5 MiB	1.3 MiB	-9.1%
Query 29	104.7 MiB	104.8 MiB	+0.0%
Query 30	3.7 MiB	3.8 MiB	+3.2%
Query 31	13.4 MiB	13.4 MiB	+0.0%
Query 32	2.9 MiB	2.7 MiB	-7.5%
Query 33	5.8 MiB	4.2 MiB	-27.1%
Query 34	6.0 MiB	6.0 MiB	+0.0%
Query 35	17.5 MiB	17.5 MiB	+0.0%
Query 36	10.3 MiB	10.3 MiB	+0.0%
Query 37	3.3 MiB	3.3 MiB	+0.0%
Query 38	10.8 MiB	10.8 MiB	-0.0%
Query 39	53.5 MiB	53.5 MiB	+0.0%
Query 40	12.2 MiB	12.2 MiB	+0.0%
Query 41	2.1 MiB	2.1 MiB	+0.0%
Query 42	1.5 MiB	1.3 MiB	-12.8%
Query 43	275.0 KiB	309.8 KiB	+12.6%
Query 44	233.2 KiB	233.2 KiB	+0.0%
Query 45	4.6 MiB	4.6 MiB	-1.3%
Query 46	6.2 MiB	6.2 MiB	+0.0%
Query 47	132.6 MiB	128.5 MiB	-3.0%
Query 48	11.4 MiB	11.8 MiB	+3.3%
Query 49	52.9 MiB	52.4 MiB	-1.1%
Query 50	17.3 MiB	17.3 MiB	+0.1%
Query 51	178.0 MiB	177.7 MiB	-0.2%
Query 52	1.8 MiB	1.7 MiB	-0.6%
Query 53	20.6 MiB	30.6 MiB	+48.6%
Query 54	2.1 MiB	2.2 MiB	+4.4%
Query 55	1.7 MiB	1.9 MiB	+17.2%
Query 56	3.2 MiB	2.9 MiB	-7.6%
Query 57	120.8 MiB	120.8 MiB	-0.0%
Query 58	4.4 MiB	4.4 MiB	+0.0%
Query 59	11.7 MiB	11.7 MiB	+0.0%
Query 60	5.2 MiB	5.5 MiB	+5.4%
Query 61	1.2 MiB	1.1 MiB	-6.4%
Query 62	4.0 MiB	4.0 MiB	+0.0%
Query 63	30.6 MiB	30.6 MiB	+0.0%
Query 64	115.3 MiB	105.3 MiB	-8.7%
Query 65	29.7 MiB	29.7 MiB	-0.0%
Query 66	14.9 MiB	15.0 MiB	+0.0%
Query 67	140.8 MiB	141.5 MiB	+0.5%
Query 68	6.2 MiB	6.2 MiB	+0.0%
Query 69	3.9 MiB	3.6 MiB	-6.9%
Query 70	44.0 MiB	44.0 MiB	+0.0%
Query 71	50.4 MiB	40.4 MiB	-19.7%
Query 72	97.2 MiB	97.2 MiB	+0.0%
Query 73	6.0 MiB	6.0 MiB	+0.0%
Query 74	36.0 MiB	36.0 MiB	+0.0%
Query 75	22.6 MiB	23.1 MiB	+2.6%
Query 76	4.3 MiB	4.3 MiB	+0.0%
Query 77	4.1 MiB	4.1 MiB	+0.0%
Query 78	335.7 MiB	335.7 MiB	+0.0%
Query 79	9.3 MiB	9.3 MiB	+0.0%
Query 80	41.2 MiB	41.7 MiB	+1.3%
Query 81	4.0 MiB	4.0 MiB	-0.1%
Query 82	4.9 MiB	4.9 MiB	+0.0%
Query 83	9.0 MiB	9.7 MiB	+7.5%
Query 84	11.9 MiB	11.9 MiB	+0.0%
Query 85	17.0 MiB	17.0 MiB	+0.0%
Query 86	82.1 MiB	91.9 MiB	+12.0%
Query 87	12.9 MiB	12.9 MiB	+0.0%
Query 88	1.7 MiB	1.6 MiB	-0.8%
Query 89	112.2 MiB	112.2 MiB	-0.0%
Query 90	351.7 KiB	351.7 KiB	+0.0%
Query 91	31.2 MiB	31.5 MiB	+0.8%
Query 92	2.7 MiB	2.7 MiB	+0.0%
Query 93	15.4 MiB	15.1 MiB	-1.6%
Query 94	3.4 MiB	3.4 MiB	-0.0%
Query 95	23.0 MiB	23.5 MiB	+2.2%
Query 96	224.7 KiB	224.7 KiB	+0.0%
Query 97	33.1 MiB	33.1 MiB	-0.0%
Query 98	114.7 MiB	114.7 MiB	-0.0%
Query 99	1.7 MiB	1.9 MiB	+13.7%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
tpcds	base (`5191bcc (merge-base)`)	335.7 MiB	2.2 GiB	1.9 GiB	6.8×
tpcds	changed (`rich-T-kid/aggregate-dictionary-hash-cache`)	335.7 MiB	2.0 GiB	1.6 GiB	6.0×

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	55.0s
Peak memory	2.2 GiB
Avg memory	1.5 GiB
CPU user	202.1s
CPU sys	6.4s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	55.0s
Peak memory	2.0 GiB
Avg memory	1.3 GiB
CPU user	210.3s
CPU sys	6.4s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-11-09T19:08:43Z

🤖 Benchmark completed (GKE) | trigger

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

Comparing rich-T-kid/aggregate-dictionary-hash-cache (8caaa35) to 5191bcc (merge-base) diff

Run configuration

run benchmark clickbench_partitioned
env:
  DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
  DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"

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 rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃       HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │    1.13 ms │                                    1.15 ms │     no change │
│ QQuery 1  │   11.27 ms │                                   11.55 ms │     no change │
│ QQuery 2  │   35.17 ms │                                   36.02 ms │     no change │
│ QQuery 3  │   31.64 ms │                                   31.56 ms │     no change │
│ QQuery 4  │  247.49 ms │                                  261.37 ms │  1.06x slower │
│ QQuery 5  │  855.86 ms │                                  779.79 ms │ +1.10x faster │
│ QQuery 6  │    1.21 ms │                                    1.20 ms │     no change │
│ QQuery 7  │   12.91 ms │                                   13.10 ms │     no change │
│ QQuery 8  │  356.86 ms │                                  359.92 ms │     no change │
│ QQuery 9  │  496.23 ms │                                  509.88 ms │     no change │
│ QQuery 10 │  203.20 ms │                                  204.31 ms │     no change │
│ QQuery 11 │  215.47 ms │                                  216.06 ms │     no change │
│ QQuery 12 │  848.78 ms │                                  793.01 ms │ +1.07x faster │
│ QQuery 13 │ 1146.92 ms │                                 1118.70 ms │     no change │
│ QQuery 14 │  884.24 ms │                                  904.78 ms │     no change │
│ QQuery 15 │  307.49 ms │                                  314.66 ms │     no change │
│ QQuery 16 │ 1340.77 ms │                                 1360.94 ms │     no change │
│ QQuery 17 │ 1324.28 ms │                                 1361.40 ms │     no change │
│ QQuery 18 │ 2078.41 ms │                                 2136.94 ms │     no change │
│ QQuery 19 │   28.04 ms │                                   28.28 ms │     no change │
│ QQuery 20 │ 1071.12 ms │                                 1091.25 ms │     no change │
│ QQuery 21 │ 1401.56 ms │                                 1414.22 ms │     no change │
│ QQuery 22 │ 3301.99 ms │                                 3322.16 ms │     no change │
│ QQuery 23 │ 9473.28 ms │                                 9508.75 ms │     no change │
│ QQuery 24 │   96.75 ms │                                   97.07 ms │     no change │
│ QQuery 25 │  357.58 ms │                                  357.11 ms │     no change │
│ QQuery 26 │   98.15 ms │                                   98.29 ms │     no change │
│ QQuery 27 │ 1168.29 ms │                                 1172.25 ms │     no change │
│ QQuery 28 │ 2620.90 ms │                                 2538.62 ms │     no change │
│ QQuery 29 │   38.67 ms │                                   39.34 ms │     no change │
│ QQuery 30 │  567.12 ms │                                  566.58 ms │     no change │
│ QQuery 31 │  555.92 ms │                                  567.08 ms │     no change │
│ QQuery 32 │ 1116.56 ms │                                 1163.18 ms │     no change │
│ QQuery 33 │ 3905.94 ms │                                 3432.90 ms │ +1.14x faster │
│ QQuery 34 │ 3986.18 ms │                                 3543.18 ms │ +1.13x faster │
│ QQuery 35 │  337.42 ms │                                  324.16 ms │     no change │
│ QQuery 36 │  157.29 ms │                                  135.36 ms │ +1.16x faster │
│ QQuery 37 │   98.93 ms │                                   98.90 ms │     no change │
│ QQuery 38 │   80.95 ms │                                   83.11 ms │     no change │
│ QQuery 39 │  339.28 ms │                                  322.56 ms │     no change │
│ QQuery 40 │   15.76 ms │                                   15.18 ms │     no change │
│ QQuery 41 │   14.31 ms │                                   14.18 ms │     no change │
│ QQuery 42 │   13.65 ms │                                   13.75 ms │     no change │
└───────────┴────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 41244.98ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 40363.77ms │
│ Average Time (HEAD)                                       │   959.19ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   938.69ms │
│ Queries Faster                                            │          5 │
│ Queries Slower                                            │          1 │
│ Queries with No Change                                    │         37 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

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

Comparing HEAD and rich-T-kid_aggregate-dictionary-hash-cache
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                   HEAD ┃ rich-T-kid_aggregate-dictionary-hash-cache ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │            1.13 / 1.77 ±1.14 / 4.04 ms │                1.15 / 1.82 ±1.19 / 4.19 ms │     no change │
│ QQuery 1  │         11.27 / 11.48 ±0.13 / 11.61 ms │             11.55 / 11.76 ±0.14 / 11.91 ms │     no change │
│ QQuery 2  │         35.17 / 35.65 ±0.29 / 35.96 ms │             36.02 / 36.35 ±0.44 / 37.20 ms │     no change │
│ QQuery 3  │         31.64 / 33.49 ±1.99 / 37.07 ms │             31.56 / 32.27 ±0.50 / 32.86 ms │     no change │
│ QQuery 4  │      247.49 / 253.93 ±4.30 / 258.74 ms │          261.37 / 264.03 ±2.61 / 268.75 ms │     no change │
│ QQuery 5  │     855.86 / 868.82 ±13.36 / 893.63 ms │         779.79 / 798.87 ±14.80 / 817.27 ms │ +1.09x faster │
│ QQuery 6  │            1.21 / 1.38 ±0.24 / 1.86 ms │                1.20 / 1.36 ±0.23 / 1.82 ms │     no change │
│ QQuery 7  │         12.91 / 13.03 ±0.09 / 13.17 ms │             13.10 / 13.26 ±0.09 / 13.34 ms │     no change │
│ QQuery 8  │      356.86 / 359.35 ±1.90 / 361.78 ms │         359.92 / 372.72 ±13.17 / 397.09 ms │     no change │
│ QQuery 9  │     496.23 / 514.44 ±15.42 / 535.85 ms │         509.88 / 528.61 ±11.97 / 544.18 ms │     no change │
│ QQuery 10 │      203.20 / 204.08 ±1.00 / 205.95 ms │          204.31 / 209.69 ±9.14 / 227.93 ms │     no change │
│ QQuery 11 │      215.47 / 216.60 ±0.94 / 218.08 ms │          216.06 / 217.47 ±0.83 / 218.37 ms │     no change │
│ QQuery 12 │     848.78 / 876.22 ±15.29 / 889.94 ms │         793.01 / 805.19 ±11.16 / 820.03 ms │ +1.09x faster │
│ QQuery 13 │  1146.92 / 1193.13 ±41.22 / 1249.96 ms │      1118.70 / 1162.75 ±38.93 / 1231.56 ms │     no change │
│ QQuery 14 │     884.24 / 907.70 ±14.22 / 923.44 ms │         904.78 / 922.70 ±16.66 / 949.69 ms │     no change │
│ QQuery 15 │      307.49 / 314.86 ±4.46 / 319.17 ms │          314.66 / 322.62 ±6.68 / 331.03 ms │     no change │
│ QQuery 16 │  1340.77 / 1376.05 ±20.40 / 1396.67 ms │      1360.94 / 1401.53 ±21.57 / 1419.80 ms │     no change │
│ QQuery 17 │  1324.28 / 1339.74 ±15.78 / 1368.16 ms │      1361.40 / 1380.30 ±22.97 / 1421.94 ms │     no change │
│ QQuery 18 │  2078.41 / 2121.44 ±25.05 / 2147.83 ms │      2136.94 / 2217.26 ±43.48 / 2257.23 ms │     no change │
│ QQuery 19 │        28.04 / 35.79 ±11.64 / 58.61 ms │             28.28 / 29.94 ±1.54 / 32.66 ms │ +1.20x faster │
│ QQuery 20 │   1071.12 / 1079.24 ±7.38 / 1092.74 ms │       1091.25 / 1099.02 ±8.57 / 1115.64 ms │     no change │
│ QQuery 21 │  1401.56 / 1416.44 ±12.36 / 1438.97 ms │      1414.22 / 1431.48 ±15.68 / 1460.75 ms │     no change │
│ QQuery 22 │  3301.99 / 3368.79 ±53.26 / 3461.62 ms │      3322.16 / 3356.63 ±18.20 / 3375.86 ms │     no change │
│ QQuery 23 │ 9473.28 / 9624.68 ±126.16 / 9810.63 ms │      9508.75 / 9624.58 ±73.84 / 9724.50 ms │     no change │
│ QQuery 24 │      96.75 / 104.78 ±11.76 / 127.28 ms │          97.07 / 151.37 ±96.94 / 345.00 ms │  1.44x slower │
│ QQuery 25 │      357.58 / 361.69 ±4.02 / 368.63 ms │          357.11 / 361.21 ±4.99 / 370.73 ms │     no change │
│ QQuery 26 │       98.15 / 100.10 ±1.42 / 101.83 ms │            98.29 / 99.18 ±1.37 / 101.91 ms │     no change │
│ QQuery 27 │  1168.29 / 1186.85 ±15.55 / 1214.74 ms │      1172.25 / 1186.18 ±10.97 / 1205.97 ms │     no change │
│ QQuery 28 │  2620.90 / 2673.22 ±45.57 / 2754.88 ms │      2538.62 / 2585.40 ±56.38 / 2691.04 ms │     no change │
│ QQuery 29 │         38.67 / 38.98 ±0.22 / 39.25 ms │             39.34 / 39.77 ±0.29 / 40.21 ms │     no change │
│ QQuery 30 │      567.12 / 575.87 ±9.94 / 595.25 ms │          566.58 / 573.26 ±5.40 / 582.94 ms │     no change │
│ QQuery 31 │      555.92 / 563.32 ±9.34 / 581.07 ms │          567.08 / 579.61 ±8.76 / 593.79 ms │     no change │
│ QQuery 32 │ 1116.56 / 1237.17 ±106.14 / 1435.66 ms │     1163.18 / 1283.58 ±143.12 / 1536.39 ms │     no change │
│ QQuery 33 │ 3905.94 / 4092.98 ±132.35 / 4257.53 ms │     3432.90 / 3674.79 ±284.01 / 4230.23 ms │ +1.11x faster │
│ QQuery 34 │ 3986.18 / 4243.20 ±164.03 / 4437.43 ms │      3543.18 / 3623.18 ±65.91 / 3715.53 ms │ +1.17x faster │
│ QQuery 35 │     337.42 / 355.44 ±15.22 / 382.40 ms │         324.16 / 345.07 ±28.54 / 401.60 ms │     no change │
│ QQuery 36 │      157.29 / 163.04 ±5.21 / 171.29 ms │         135.36 / 151.29 ±10.16 / 167.35 ms │ +1.08x faster │
│ QQuery 37 │       98.93 / 108.99 ±6.57 / 119.27 ms │           98.90 / 105.61 ±5.40 / 114.78 ms │     no change │
│ QQuery 38 │         80.95 / 88.58 ±4.99 / 95.73 ms │             83.11 / 87.95 ±3.69 / 92.77 ms │     no change │
│ QQuery 39 │     339.28 / 360.28 ±16.51 / 388.36 ms │         322.56 / 336.33 ±11.05 / 350.70 ms │ +1.07x faster │
│ QQuery 40 │         15.76 / 16.62 ±0.69 / 17.49 ms │             15.18 / 15.85 ±0.48 / 16.65 ms │     no change │
│ QQuery 41 │         14.31 / 15.61 ±2.17 / 19.94 ms │             14.18 / 14.38 ±0.16 / 14.66 ms │ +1.09x faster │
│ QQuery 42 │         13.65 / 13.81 ±0.13 / 14.01 ms │             13.75 / 13.87 ±0.09 / 13.98 ms │     no change │
└───────────┴────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                                         ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                                         │ 42468.61ms │
│ Total Time (rich-T-kid_aggregate-dictionary-hash-cache)   │ 41470.08ms │
│ Average Time (HEAD)                                       │   987.64ms │
│ Average Time (rich-T-kid_aggregate-dictionary-hash-cache) │   964.42ms │
│ Queries Faster                                            │          8 │
│ Queries Slower                                            │          1 │
│ Queries with No Change                                    │         34 │
│ Queries with Failure                                      │          0 │
└───────────────────────────────────────────────────────────┴────────────┘

Memory Pool Peaks

Peak MemoryPool reservation per query — what DataFusion's accounting believes it reserved. Recorded only for benchmarks that write a results JSON (the dfbench suites, and the suites bench.sh runs through the Criterion SQL harness since #25644), and only under a memory limit: DATAFUSION_RUNTIME_MEMORY_LIMIT, or one the suite sets itself.

Base: 5191bcc (merge-base) | Changed: rich-T-kid/aggregate-dictionary-hash-cache

clickbench_partitioned

Query	Base	Changed	Change
Query 0	0 B	0 B	0.0%
Query 1	104 B	104 B	+0.0%
Query 2	936 B	936 B	+0.0%
Query 3	312 B	312 B	+0.0%
Query 4	843.8 MiB	843.9 MiB	+0.0%
Query 5	1.4 GiB	1.3 GiB	-6.3%
Query 6	0 B	0 B	0.0%
Query 7	60.0 MiB	40.4 MiB	-32.7%
Query 8	947.2 MiB	967.6 MiB	+2.2%
Query 9	552.0 MiB	593.9 MiB	+7.6%
Query 10	72.1 MiB	73.7 MiB	+2.2%
Query 11	75.2 MiB	75.0 MiB	-0.3%
Query 12	1.6 GiB	1.5 GiB	-7.6%
Query 13	1.6 GiB	1.5 GiB	-4.9%
Query 14	1.6 GiB	1.6 GiB	+1.9%
Query 15	1.2 GiB	1.2 GiB	+1.9%
Query 16	3.3 GiB	3.5 GiB	+4.1%
Query 17	3.4 GiB	3.3 GiB	-5.0%
Query 18	7.1 GiB	6.9 GiB	-2.6%
Query 19	0 B	0 B	0.0%
Query 20	104 B	104 B	+0.0%
Query 21	1.6 MiB	1.1 MiB	-29.9%
Query 22	2.4 MiB	2.3 MiB	-2.5%
Query 23	12.2 MiB	13.0 MiB	+6.4%
Query 24	6.3 MiB	6.7 MiB	+7.4%
Query 25	6.9 MiB	8.1 MiB	+17.6%
Query 26	7.5 MiB	7.4 MiB	-1.6%
Query 27	2.5 MiB	2.5 MiB	+3.0%
Query 28	1.9 GiB	1.8 GiB	-5.1%
Query 29	624 B	624 B	+0.0%
Query 30	667.8 MiB	650.9 MiB	-2.5%
Query 31	1.5 GiB	1.5 GiB	-4.6%
Query 32	7.2 GiB	7.2 GiB	-0.5%
Query 33	9.9 GiB	9.6 GiB	-3.3%
Query 34	9.9 GiB	9.8 GiB	-1.4%
Query 35	660.8 MiB	664.8 MiB	+0.6%
Query 36	105.4 MiB	100.1 MiB	-5.0%
Query 37	8.1 MiB	8.4 MiB	+3.7%
Query 38	5.7 MiB	6.0 MiB	+5.1%
Query 39	213.4 MiB	213.9 MiB	+0.3%
Query 40	2.6 MiB	2.5 MiB	-2.7%
Query 41	3.1 MiB	3.3 MiB	+9.1%
Query 42	1.9 MiB	1.7 MiB	-10.5%

Pool accounting vs. process RSS

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

Benchmark	Side	Max pool peak	Peak RSS	Gap	RSS / pool
clickbench_partitioned	base (`5191bcc (merge-base)`)	9.9 GiB	12.4 GiB	2.5 GiB	1.3×
clickbench_partitioned	changed (`rich-T-kid/aggregate-dictionary-hash-cache`)	9.8 GiB	12.6 GiB	2.8 GiB	1.3×

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	215.0s
Peak memory	12.4 GiB
Avg memory	5.3 GiB
CPU user	2046.2s
CPU sys	146.4s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	210.0s
Peak memory	12.6 GiB
Avg memory	5.4 GiB
CPU user	1971.7s
CPU sys	145.0s
Peak spill	0 B

File an issue against this benchmark runner

jayzhan211

Thanks @Rich-T-kid !

github-actions Bot added common Related to common crate physical-plan Changes to the physical-plan crate labels Oct 27, 2026

Rich-T-kid force-pushed the rich-T-kid/aggregate-dictionary-hash-cache branch from e1aa5f4 to d49dd6d Compare October 27, 2026 20:22

Rich-T-kid commented Oct 27, 2026

View reviewed changes

Rich-T-kid force-pushed the rich-T-kid/aggregate-dictionary-hash-cache branch from d49dd6d to 0044519 Compare October 27, 2026 23:09

This was referenced Nov 8, 2026

fix(parquet): honor requested dictionary value type apache/arrow-rs#12126

Closed

perf(aggregate): specialize single dictionary groups #27262

Closed

jayzhan211 approved these changes Nov 14, 2026

View reviewed changes

Merge branch 'main' into rich-T-kid/aggregate-dictionary-hash-cache

64bfad2

jayzhan211 enabled auto-merge November 14, 2026 01:21

jayzhan211 added this pull request to the merge queue Nov 14, 2026

github-merge-queue Bot removed this pull request from the merge queue due to failed status checks Nov 14, 2026

jayzhan211 added this pull request to the merge queue Nov 14, 2026

Merged via the queue into apache:main with commit 8f279e1 Nov 14, 2026
43 checks passed

github-actions Bot added the v57.0.0 label Nov 14, 2026

Conversation

Rich-T-kid commented Oct 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

What is the testing strategy for this PR?

Are there any user-facing changes?

Uh oh!

Rich-T-kid commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

Rich-T-kid commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

Rich-T-kid commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

Rich-T-kid Oct 27, 2026

Choose a reason for hiding this comment

Uh oh!

Rich-T-kid commented Oct 27, 2026

Uh oh!

Rich-T-kid commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

adriangbot commented Oct 27, 2026

Uh oh!

Rich-T-kid commented Nov 8, 2026

Uh oh!

adriangbot commented Nov 8, 2026

Uh oh!

adriangbot commented Nov 8, 2026

Uh oh!

adriangbot commented Nov 8, 2026

Uh oh!

adriangbot commented Nov 8, 2026

Uh oh!

adriangbot commented Nov 8, 2026

Uh oh!

adriangbot commented Nov 8, 2026

Uh oh!

Rich-T-kid commented Nov 9, 2026

Uh oh!

adriangbot commented Nov 9, 2026

Uh oh!

Rich-T-kid commented Oct 27, 2026 •

edited

Loading