refactor: make file-statistics cache keys schema-aware

[Phoenix500526]

Which issue does this PR close?

• ClosesMake file-statistics cache keys schema-aware #23072.

Rationale for this change

File statistics are computed against a specific file_schema (their
column_statistics are positional, one per column), but the file-statistics
cache was keyed only by table and path. Reading the same path under a different
schema could therefore reuse statistics whose columns no longer line up,
panicking during statistics projection.

#22950 worked around this by bypassing the file-statistics cache entirely for
anonymous explicit-schema reads — correct, but it gave up cache reuse for them
(every such read recomputes statistics). #23072 asks to make the cache itself
schema-aware so those reads can reuse the cache safely instead of skipping it.

What changes are included in this PR?

• Add SchemaFingerprint — the per-column (name, data_type, nullable) of a
  file_schema, in order — and FileStatisticsCacheKey { table, path, schema },
  and key the file-statistics cache on it (FileStatisticsCache is now
  dyn Cache<FileStatisticsCacheKey, CachedFileMetadata>).
• ListingTable::do_collect_statistics_and_ordering builds the key with the
  file_schema fingerprint and uses the shared cache directly. The fix: isolate anonymous file statistics cache #22950
  bypass (statistics_cache helper / schema_source-based skip) is removed:
  different schemas now land in distinct entries (no stale cross-schema reuse),
  while a repeated read of the same schema reuses its entry.
• The fingerprint deliberately excludes field/schema metadata (it cannot
  affect statistics, and including it would needlessly fragment the cache) and
  partition columns (partition statistics are computed separately, outside this
  cache).
• Table-drop invalidation is unchanged: drop_table_entries matches on
  CacheKey::table_ref(), which still returns the table, so all schema variants
  for a dropped table are removed together.
• The list-files cache continues to key on TableScopedPath.

Are these changes tested?

Yes.

• Updated the fix: isolate anonymous file statistics cache #22950 regression test
  (anonymous_parquet_stats_cache_with_explicit_wider_schema): the wider
  explicit-schema read now lands in its own cache entry (2 entries, was 1 under
  the bypass) with correct statistics and no panic, and a repeated read of that
  schema is served from the cache (a cache hit, no new entry).
• Added unit tests for SchemaFingerprint: it distinguishes nullability and
  field order, and ignores field/schema metadata.
• cargo test for the file_statistics integration module and the
  datafusion-execution cache tests (including drop_table_entries) pass, along
  with cargo fmt --all and cargo clippy --all-targets --all-features -- -D warnings for the touched crates.

Are there any user-facing changes?

No change to query results, physical plans, or the serialized (proto) wire
format; file statistics are computed exactly as before.

One public API change (please add the api change label): the
FileStatisticsCache type alias now uses FileStatisticsCacheKey instead of
TableScopedPath as its key. Code that constructed keys for this cache directly
must switch to FileStatisticsCacheKey. SchemaFingerprint and
FileStatisticsCacheKey are newly public; TableScopedPath remains (still used
by the list-files cache). cargo-semver-checks will flag the key-type change,
which is expected.

[mkleen]

Thank you for working on this. I left a few comments.

[adriangb]

I am going to run the wide_schema benchmarks here. I am afraid that any change touching schemas is susceptible to introduce O(num_columns^X) operations.

[adriangb]

run bechmark wide_schema

env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

[adriangb]

run bechmark wide_schema

[adriangb]

run bechmark wide_schema

baseline:
env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

[mkleen]

run bechmark wide_schema

baseline:
env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

This is cool, i did not know this.

[adriangb]

run benchmark wide_schema

env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

[adriangb]

run benchmark wide_schema

baseline:
env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

[adriangb]

run benchmark wide_schema

[adriangb]

run bechmark wide_schema

baseline:
env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

This is cool, i did not know this.

Yep the idea here is to run baseline w/o cache and this branch w/ cache. Orthogonal to this PR but I want to see how it looks like.

Unfortunately I had a typo in benchmark, sorry for the noise.

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4808656966-702-b4kn8 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (a13e269) to ff677c4 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4808667999-703-92qlv 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (a13e269) to ff677c4 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4808669794-704-whlqh 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (a13e269) to ff677c4 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.00 79.8±0.27ms ? ?/sec 1.00 79.8±0.18ms ? ?/sec
wide_schema/Q01_wide 1.00 1025.8±5.54ms ? ?/sec 1.06 1087.0±3.82ms ? ?/sec
wide_schema/Q02_narrow 1.00 5.9±0.09ms ? ?/sec 1.05 6.2±0.05ms ? ?/sec
wide_schema/Q02_wide 1.00 899.4±3.46ms ? ?/sec 1.08 975.5±2.01ms ? ?/sec
wide_schema/Q03_narrow 1.00 14.8±0.23ms ? ?/sec 1.02 15.0±0.24ms ? ?/sec
wide_schema/Q03_wide 1.00 912.9±6.88ms ? ?/sec 1.07 977.0±3.51ms ? ?/sec
wide_schema/Q04_narrow 1.00 37.2±0.24ms ? ?/sec 1.01 37.6±0.20ms ? ?/sec
wide_schema/Q04_wide 1.00 990.8±6.47ms ? ?/sec 1.08 1068.2±6.72ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	980.2s
Peak memory	1.2 GiB
Avg memory	98.4 MiB
CPU user	399.6s
CPU sys	58.7s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	975.2s
Peak memory	1.2 GiB
Avg memory	108.3 MiB
CPU user	386.8s
CPU sys	50.4s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.00 80.3±0.44ms ? ?/sec 1.00 80.5±0.49ms ? ?/sec
wide_schema/Q01_wide 1.00 1016.6±5.73ms ? ?/sec 1.07 1085.9±8.10ms ? ?/sec
wide_schema/Q02_narrow 1.00 5.8±0.05ms ? ?/sec 1.05 6.1±0.06ms ? ?/sec
wide_schema/Q02_wide 1.00 893.2±5.93ms ? ?/sec 1.10 984.1±7.66ms ? ?/sec
wide_schema/Q03_narrow 1.00 14.5±0.26ms ? ?/sec 1.02 14.8±0.23ms ? ?/sec
wide_schema/Q03_wide 1.00 900.3±5.99ms ? ?/sec 1.11 997.2±11.39ms ? ?/sec
wide_schema/Q04_narrow 1.00 37.0±0.20ms ? ?/sec 1.02 37.8±0.23ms ? ?/sec
wide_schema/Q04_wide 1.00 983.9±3.82ms ? ?/sec 1.08 1060.0±7.98ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	980.2s
Peak memory	1.1 GiB
Avg memory	107.6 MiB
CPU user	398.4s
CPU sys	58.8s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	980.2s
Peak memory	1.2 GiB
Avg memory	104.2 MiB
CPU user	383.6s
CPU sys	51.3s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.03 84.0±1.81ms ? ?/sec 1.00 81.3±1.46ms ? ?/sec
wide_schema/Q01_wide 1.00 1041.9±15.17ms ? ?/sec 1.07 1112.7±14.96ms ? ?/sec
wide_schema/Q02_narrow 1.00 6.2±0.11ms ? ?/sec 1.03 6.4±0.14ms ? ?/sec
wide_schema/Q02_wide 1.00 921.0±21.09ms ? ?/sec 1.09 1007.2±20.47ms ? ?/sec
wide_schema/Q03_narrow 1.00 14.8±0.35ms ? ?/sec 1.05 15.6±0.17ms ? ?/sec
wide_schema/Q03_wide 1.00 938.0±23.35ms ? ?/sec 1.17 1097.5±26.47ms ? ?/sec
wide_schema/Q04_narrow 1.00 37.8±0.62ms ? ?/sec 1.04 39.1±0.90ms ? ?/sec
wide_schema/Q04_wide 1.00 994.7±8.18ms ? ?/sec 1.09 1080.9±23.55ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	970.2s
Peak memory	1.1 GiB
Avg memory	97.1 MiB
CPU user	381.4s
CPU sys	52.1s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	995.2s
Peak memory	1.1 GiB
Avg memory	109.4 MiB
CPU user	384.8s
CPU sys	51.6s
Peak spill	0 B

---

File an issue against this benchmark runner

[mkleen]

Unfortunately we have regressions in the benchmarks:

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.00 80.3±0.44ms ? ?/sec 1.00 80.5±0.49ms ? ?/sec
wide_schema/Q01_wide 1.00 1016.6±5.73ms ? ?/sec 1.07 1085.9±8.10ms ? ?/sec
wide_schema/Q02_narrow 1.00 5.8±0.05ms ? ?/sec 1.05 6.1±0.06ms ? ?/sec
wide_schema/Q02_wide 1.00 893.2±5.93ms ? ?/sec 1.10 984.1±7.66ms ? ?/sec
wide_schema/Q03_narrow 1.00 14.5±0.26ms ? ?/sec 1.02 14.8±0.23ms ? ?/sec
wide_schema/Q03_wide 1.00 900.3±5.99ms ? ?/sec 1.11 997.2±11.39ms ? ?/sec
wide_schema/Q04_narrow 1.00 37.0±0.20ms ? ?/sec 1.02 37.8±0.23ms ? ?/sec
wide_schema/Q04_wide 1.00 983.9±3.82ms ? ?/sec 1.08 1060.0±7.98ms ? ?/sec

[mkleen]

run benchmark wide_schema

baseline:
env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

[mkleen]

run benchmark wide_schema

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4810951539-710-d6x86 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (9618f88) to ff677c4 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4810954479-711-k7mss 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (9618f88) to ff677c4 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.00 79.8±0.36ms ? ?/sec 1.02 81.4±0.53ms ? ?/sec
wide_schema/Q01_wide 1.00 1017.6±3.63ms ? ?/sec 1.02 1042.3±4.84ms ? ?/sec
wide_schema/Q02_narrow 1.00 6.2±0.10ms ? ?/sec 1.00 6.2±0.08ms ? ?/sec
wide_schema/Q02_wide 1.00 901.5±3.36ms ? ?/sec 1.02 916.6±6.25ms ? ?/sec
wide_schema/Q03_narrow 1.00 15.5±0.24ms ? ?/sec 1.00 15.5±0.26ms ? ?/sec
wide_schema/Q03_wide 1.00 912.1±5.79ms ? ?/sec 1.02 933.7±4.82ms ? ?/sec
wide_schema/Q04_narrow 1.00 37.1±0.12ms ? ?/sec 1.00 37.1±0.13ms ? ?/sec
wide_schema/Q04_wide 1.00 1008.7±5.74ms ? ?/sec 1.02 1029.2±9.65ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	640.1s
Peak memory	1.2 GiB
Avg memory	152.6 MiB
CPU user	382.9s
CPU sys	52.9s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	975.2s
Peak memory	1.1 GiB
Avg memory	104.0 MiB
CPU user	381.7s
CPU sys	52.0s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.01 81.3±0.44ms ? ?/sec 1.00 80.1±0.38ms ? ?/sec
wide_schema/Q01_wide 1.00 1033.7±7.03ms ? ?/sec 1.02 1054.8±6.46ms ? ?/sec
wide_schema/Q02_narrow 1.01 6.0±0.07ms ? ?/sec 1.00 6.0±0.08ms ? ?/sec
wide_schema/Q02_wide 1.00 911.3±4.97ms ? ?/sec 1.02 933.8±4.85ms ? ?/sec
wide_schema/Q03_narrow 1.00 15.0±0.30ms ? ?/sec 1.00 15.1±0.09ms ? ?/sec
wide_schema/Q03_wide 1.00 923.9±5.68ms ? ?/sec 1.02 942.4±5.85ms ? ?/sec
wide_schema/Q04_narrow 1.01 38.0±0.23ms ? ?/sec 1.00 37.5±0.28ms ? ?/sec
wide_schema/Q04_wide 1.00 1004.1±5.02ms ? ?/sec 1.01 1018.3±6.65ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	645.1s
Peak memory	1.1 GiB
Avg memory	147.8 MiB
CPU user	379.2s
CPU sys	53.5s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	960.2s
Peak memory	1.1 GiB
Avg memory	100.0 MiB
CPU user	384.2s
CPU sys	52.7s
Peak spill	0 B

---

File an issue against this benchmark runner

[github-actions]

Thank you for opening this pull request!

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

Details

 Cloning apache/main
Building datafusion v54.0.0 (current)
Built [ 100.718s] (current)
Parsing datafusion v54.0.0 (current)
Parsed [ 0.033s] (current)
Building datafusion v54.0.0 (baseline)
Built [ 100.974s] (baseline)
Parsing datafusion v54.0.0 (baseline)
Parsed [ 0.034s] (baseline)
Checking datafusion v54.0.0 -> v54.0.0 (no change; assume patch)
Checked [ 0.859s] 223 checks: 223 pass, 30 skip
Summary no semver update required
Finished [ 204.560s] datafusion
Building datafusion-catalog-listing v54.0.0 (current)
Built [ 43.842s] (current)
Parsing datafusion-catalog-listing v54.0.0 (current)
Parsed [ 0.012s] (current)
Building datafusion-catalog-listing v54.0.0 (baseline)
Built [ 44.703s] (baseline)
Parsing datafusion-catalog-listing v54.0.0 (baseline)
Parsed [ 0.012s] (baseline)
Checking datafusion-catalog-listing v54.0.0 -> v54.0.0 (no change; assume patch)
Checked [ 0.109s] 223 checks: 223 pass, 30 skip
Summary no semver update required
Finished [ 90.066s] datafusion-catalog-listing
Building datafusion-common v54.0.0 (current)
Built [ 32.556s] (current)
Parsing datafusion-common v54.0.0 (current)
Parsed [ 0.061s] (current)
Building datafusion-common v54.0.0 (baseline)
Built [ 32.423s] (baseline)
Parsing datafusion-common v54.0.0 (baseline)
Parsed [ 0.061s] (baseline)
Checking datafusion-common v54.0.0 -> v54.0.0 (no change; assume patch)
Checked [ 0.951s] 223 checks: 223 pass, 30 skip
Summary no semver update required
Finished [ 67.912s] datafusion-common
Building datafusion-execution v54.0.0 (current)
Built [ 29.196s] (current)
Parsing datafusion-execution v54.0.0 (current)
Parsed [ 0.023s] (current)
Building datafusion-execution v54.0.0 (baseline)
Built [ 29.187s] (baseline)
Parsing datafusion-execution v54.0.0 (baseline)
Parsed [ 0.024s] (baseline)
Checking datafusion-execution v54.0.0 -> v54.0.0 (no change; assume patch)
Checked [ 0.332s] 223 checks: 221 pass, 2 fail, 0 warn, 30 skip
--- failure constructible_struct_adds_field: externally-constructible struct adds field ---
Description:
A pub struct constructible with a struct literal has a new pub field. Existing struct literals must be updated to include the new field.
ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.48.0/src/lints/constructible_struct_adds_field.ron
Failed in:
field CachedFileMetadata.schema_fingerprint in /home/runner/work/datafusion/datafusion/datafusion/execution/src/cache/cache_manager.rs:101
--- failure method_parameter_count_changed: pub method parameter count changed ---
Description:
A publicly-visible method now takes a different number of parameters, not counting the receiver (self) parameter.
ref: https://doc.rust-lang.org/cargo/reference/semver.html#fn-change-arity
impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.48.0/src/lints/method_parameter_count_changed.ron
Failed in:
datafusion_execution::cache::cache_manager::CachedFileMetadata::new takes 3 parameters in /home/runner/work/datafusion/datafusion/target/semver-checks/git-apache_main/29eb218ca8a229e0f3215fb041a6f8becb4e6db0/datafusion/execution/src/cache/cache_manager.rs:107, but now takes 4 parameters in /home/runner/work/datafusion/datafusion/datafusion/execution/src/cache/cache_manager.rs:110
datafusion_execution::cache::cache_manager::CachedFileMetadata::is_valid_for takes 1 parameters in /home/runner/work/datafusion/datafusion/target/semver-checks/git-apache_main/29eb218ca8a229e0f3215fb041a6f8becb4e6db0/datafusion/execution/src/cache/cache_manager.rs:122, but now takes 2 parameters in /home/runner/work/datafusion/datafusion/datafusion/execution/src/cache/cache_manager.rs:127
Summary semver requires new major version: 2 major and 0 minor checks failed
Finished [ 60.185s] datafusion-execution

[mkleen]

run benchmark clickbench_partitioned

[mkleen]

run benchmark wide_schema

baseline:
env:
DATAFUSION_RUNTIME_FILE_STATISTICS_CACHE_LIMIT: 0

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4851059287-764-hvcks 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (985b8f3) to d58e0c6 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4851061669-765-llklf 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (985b8f3) to d58e0c6 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and issue_23072
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ issue_23072 ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0 │ 1.23 / 3.99 ±5.47 / 14.93 ms │ 1.27 / 4.02 ±5.40 / 14.81 ms │ no change │
│ QQuery 1 │ 12.98 / 13.24 ±0.25 / 13.62 ms │ 12.67 / 12.84 ±0.11 / 12.97 ms │ no change │
│ QQuery 2 │ 36.03 / 36.41 ±0.37 / 36.93 ms │ 35.78 / 36.15 ±0.28 / 36.42 ms │ no change │
│ QQuery 3 │ 31.08 / 32.12 ±0.55 / 32.61 ms │ 30.42 / 30.67 ±0.18 / 30.88 ms │ no change │
│ QQuery 4 │ 1649.65 / 1750.78 ±82.62 / 1868.33 ms │ 1689.81 / 1751.40 ±65.21 / 1869.59 ms │ no change │
│ QQuery 5 │ 1643.37 / 1704.54 ±69.58 / 1837.83 ms │ 1812.34 / 1899.68 ±71.10 / 1992.65 ms │ 1.11x slower │
│ QQuery 6 │ 1.27 / 1.42 ±0.23 / 1.87 ms │ 1.31 / 1.48 ±0.25 / 1.97 ms │ no change │
│ QQuery 7 │ 14.21 / 20.53 ±12.46 / 45.45 ms │ 13.62 / 13.80 ±0.11 / 13.92 ms │ +1.49x faster │
│ QQuery 8 │ 1986.44 / 2082.42 ±69.70 / 2193.24 ms │ 2032.28 / 2116.79 ±89.00 / 2263.80 ms │ no change │
│ QQuery 9 │ 467.70 / 490.94 ±18.57 / 510.33 ms │ 475.58 / 505.07 ±18.89 / 527.22 ms │ no change │
│ QQuery 10 │ 80.82 / 82.44 ±1.14 / 83.72 ms │ 75.07 / 77.33 ±1.15 / 78.29 ms │ +1.07x faster │
│ QQuery 11 │ 97.49 / 105.19 ±13.42 / 131.97 ms │ 90.71 / 96.31 ±8.32 / 112.77 ms │ +1.09x faster │
│ QQuery 12 │ 1617.66 / 1737.18 ±154.19 / 2011.59 ms │ 1795.44 / 1969.00 ±99.89 / 2075.27 ms │ 1.13x slower │
│ QQuery 13 │ 548.42 / 700.26 ±141.55 / 887.09 ms │ 550.90 / 699.54 ±92.55 / 809.54 ms │ no change │
│ QQuery 14 │ 544.85 / 564.35 ±19.31 / 600.82 ms │ 541.21 / 557.75 ±18.81 / 594.43 ms │ no change │
│ QQuery 15 │ 1987.63 / 2025.47 ±43.19 / 2087.29 ms │ 1971.44 / 2034.33 ±42.19 / 2099.57 ms │ no change │
│ QQuery 16 │ 4208.49 / 4337.37 ±104.44 / 4509.71 ms │ 4203.57 / 4379.49 ±150.62 / 4637.17 ms │ no change │
│ QQuery 17 │ 4112.09 / 4368.72 ±151.60 / 4550.03 ms │ 4308.23 / 4386.25 ±73.01 / 4515.52 ms │ no change │
│ QQuery 18 │ 18343.21 / 18535.48 ±200.52 / 18807.01 ms │ 18357.24 / 18682.53 ±211.25 / 18923.66 ms │ no change │
│ QQuery 19 │ 28.16 / 28.70 ±0.55 / 29.47 ms │ 27.88 / 38.51 ±17.20 / 72.32 ms │ 1.34x slower │
│ QQuery 20 │ 514.74 / 530.52 ±22.52 / 573.29 ms │ 511.83 / 521.09 ±8.42 / 533.56 ms │ no change │
│ QQuery 21 │ 511.48 / 519.35 ±4.56 / 524.75 ms │ 524.33 / 534.55 ±5.66 / 541.80 ms │ no change │
│ QQuery 22 │ 978.59 / 1022.83 ±30.81 / 1066.09 ms │ 978.29 / 988.10 ±7.63 / 997.18 ms │ no change │
│ QQuery 23 │ 3014.40 / 3084.86 ±69.91 / 3216.53 ms │ 3046.88 / 3090.33 ±45.37 / 3168.19 ms │ no change │
│ QQuery 24 │ 41.24 / 53.47 ±15.14 / 79.30 ms │ 40.96 / 49.47 ±16.67 / 82.81 ms │ +1.08x faster │
│ QQuery 25 │ 110.13 / 111.33 ±0.90 / 112.89 ms │ 111.35 / 113.64 ±2.78 / 119.09 ms │ no change │
│ QQuery 26 │ 41.53 / 43.94 ±2.71 / 47.76 ms │ 41.74 / 42.14 ±0.44 / 42.91 ms │ no change │
│ QQuery 27 │ 664.85 / 676.06 ±6.15 / 682.42 ms │ 676.60 / 681.25 ±3.57 / 685.41 ms │ no change │
│ QQuery 28 │ 3523.00 / 3739.36 ±188.92 / 4002.10 ms │ 3398.36 / 3644.95 ±226.16 / 4069.56 ms │ no change │
│ QQuery 29 │ 42.23 / 48.93 ±12.35 / 73.62 ms │ 40.43 / 42.52 ±3.08 / 48.55 ms │ +1.15x faster │
│ QQuery 30 │ 578.29 / 594.14 ±14.34 / 612.57 ms │ 544.70 / 573.50 ±22.98 / 614.67 ms │ no change │
│ QQuery 31 │ 286.91 / 296.86 ±10.65 / 316.36 ms │ 313.53 / 330.50 ±18.56 / 363.91 ms │ 1.11x slower │
│ QQuery 32 │ 951.49 / 978.33 ±19.82 / 1011.15 ms │ 986.32 / 1060.31 ±53.48 / 1135.57 ms │ 1.08x slower │
│ QQuery 33 │ 26654.01 / 28910.61 ±1812.18 / 31558.80 ms │ 28632.23 / 29368.01 ±666.23 / 30339.71 ms │ no change │
│ QQuery 34 │ 27736.18 / 28298.48 ±406.38 / 28973.62 ms │ 26864.53 / 29085.06 ±1742.33 / 30766.08 ms │ no change │
│ QQuery 35 │ 1096.23 / 1188.58 ±108.59 / 1400.28 ms │ 1058.74 / 1210.51 ±198.67 / 1588.31 ms │ no change │
│ QQuery 36 │ 165.48 / 184.08 ±21.11 / 224.58 ms │ 154.10 / 175.40 ±15.99 / 200.95 ms │ no change │
│ QQuery 37 │ 38.22 / 44.15 ±10.86 / 65.83 ms │ 38.47 / 50.17 ±13.73 / 69.33 ms │ 1.14x slower │
│ QQuery 38 │ 42.43 / 43.09 ±0.55 / 44.04 ms │ 42.51 / 44.22 ±1.37 / 45.65 ms │ no change │
│ QQuery 39 │ 178.90 / 193.67 ±13.67 / 217.47 ms │ 186.95 / 194.46 ±5.19 / 199.53 ms │ no change │
│ QQuery 40 │ 14.25 / 14.63 ±0.30 / 14.92 ms │ 14.31 / 14.88 ±0.47 / 15.50 ms │ no change │
│ QQuery 41 │ 13.75 / 14.02 ±0.26 / 14.43 ms │ 13.51 / 13.82 ±0.27 / 14.16 ms │ no change │
│ QQuery 42 │ 12.90 / 13.26 ±0.27 / 13.70 ms │ 12.88 / 13.07 ±0.12 / 13.21 ms │ no change │
└───────────┴────────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 109226.12ms │
│ Total Time (issue_23072) │ 111134.87ms │
│ Average Time (HEAD) │ 2540.14ms │
│ Average Time (issue_23072) │ 2584.53ms │
│ Queries Faster │ 5 │
│ Queries Slower │ 6 │
│ Queries with No Change │ 32 │
│ Queries with Failure │ 0 │
└────────────────────────────┴─────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	550.1s
Peak memory	12.0 GiB
Avg memory	6.5 GiB
CPU user	4918.3s
CPU sys	322.5s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	560.1s
Peak memory	12.1 GiB
Avg memory	6.5 GiB
CPU user	4956.7s
CPU sys	319.6s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.00 79.5±0.18ms ? ?/sec 1.00 79.6±0.19ms ? ?/sec
wide_schema/Q01_wide 1.00 1015.3±3.72ms ? ?/sec 1.00 1013.6±5.80ms ? ?/sec
wide_schema/Q02_narrow 1.01 5.8±0.08ms ? ?/sec 1.00 5.8±0.08ms ? ?/sec
wide_schema/Q02_wide 1.00 891.4±3.52ms ? ?/sec 1.00 890.1±1.80ms ? ?/sec
wide_schema/Q03_narrow 1.01 14.4±0.14ms ? ?/sec 1.00 14.3±0.20ms ? ?/sec
wide_schema/Q03_wide 1.00 895.7±4.49ms ? ?/sec 1.01 903.1±2.87ms ? ?/sec
wide_schema/Q04_narrow 1.01 37.3±0.21ms ? ?/sec 1.00 37.0±0.13ms ? ?/sec
wide_schema/Q04_wide 1.01 994.3±4.70ms ? ?/sec 1.00 985.4±2.62ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	640.1s
Peak memory	1.1 GiB
Avg memory	160.4 MiB
CPU user	405.6s
CPU sys	59.4s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	920.2s
Peak memory	1.1 GiB
Avg memory	106.0 MiB
CPU user	400.0s
CPU sys	58.9s
Peak spill	0 B

---

File an issue against this benchmark runner

[mkleen]

run benchmark clickbench_partitioned

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4852067698-768-p4z46 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (985b8f3) to d58e0c6 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and issue_23072
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ issue_23072 ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0 │ 1.24 / 4.12 ±5.61 / 15.33 ms │ 1.28 / 4.14 ±5.58 / 15.30 ms │ no change │
│ QQuery 1 │ 12.62 / 13.20 ±0.31 / 13.53 ms │ 13.29 / 13.35 ±0.05 / 13.44 ms │ no change │
│ QQuery 2 │ 37.69 / 38.40 ±0.81 / 39.90 ms │ 36.17 / 36.55 ±0.31 / 37.11 ms │ no change │
│ QQuery 3 │ 30.96 / 31.64 ±0.70 / 32.98 ms │ 31.06 / 31.68 ±0.34 / 32.05 ms │ no change │
│ QQuery 4 │ 1714.29 / 1746.04 ±27.34 / 1796.12 ms │ 1763.36 / 1788.54 ±21.49 / 1822.28 ms │ no change │
│ QQuery 5 │ 1571.87 / 1757.61 ±106.95 / 1892.75 ms │ 1642.84 / 1751.52 ±119.46 / 1983.81 ms │ no change │
│ QQuery 6 │ 1.30 / 1.45 ±0.22 / 1.89 ms │ 1.30 / 1.47 ±0.24 / 1.94 ms │ no change │
│ QQuery 7 │ 14.34 / 14.56 ±0.14 / 14.74 ms │ 14.73 / 14.82 ±0.10 / 14.96 ms │ no change │
│ QQuery 8 │ 2023.91 / 2065.80 ±34.15 / 2116.79 ms │ 2086.04 / 2122.21 ±56.16 / 2233.84 ms │ no change │
│ QQuery 9 │ 507.54 / 531.76 ±20.45 / 569.51 ms │ 462.50 / 510.36 ±29.44 / 552.37 ms │ no change │
│ QQuery 10 │ 79.04 / 81.44 ±3.34 / 88.05 ms │ 77.51 / 88.21 ±17.21 / 122.53 ms │ 1.08x slower │
│ QQuery 11 │ 91.33 / 99.59 ±12.99 / 125.39 ms │ 87.35 / 89.84 ±1.68 / 91.88 ms │ +1.11x faster │
│ QQuery 12 │ 1605.60 / 1724.41 ±79.85 / 1845.09 ms │ 1690.92 / 1818.20 ±65.61 / 1867.39 ms │ 1.05x slower │
│ QQuery 13 │ 463.15 / 635.32 ±149.26 / 862.59 ms │ 472.16 / 740.54 ±165.96 / 931.91 ms │ 1.17x slower │
│ QQuery 14 │ 565.21 / 569.11 ±2.74 / 572.82 ms │ 535.72 / 569.04 ±23.95 / 608.06 ms │ no change │
│ QQuery 15 │ 1892.03 / 1952.74 ±72.36 / 2074.14 ms │ 1886.81 / 1936.03 ±43.16 / 2006.95 ms │ no change │
│ QQuery 16 │ 4144.74 / 4469.12 ±195.52 / 4725.84 ms │ 4196.60 / 4416.77 ±165.98 / 4642.64 ms │ no change │
│ QQuery 17 │ 4099.81 / 4408.11 ±163.33 / 4588.98 ms │ 4446.56 / 4486.46 ±32.47 / 4530.24 ms │ no change │
│ QQuery 18 │ 18250.79 / 18588.21 ±503.27 / 19590.18 ms │ 17909.75 / 18302.75 ±316.06 / 18671.73 ms │ no change │
│ QQuery 19 │ 28.36 / 29.51 ±0.98 / 31.29 ms │ 28.65 / 28.89 ±0.31 / 29.50 ms │ no change │
│ QQuery 20 │ 517.25 / 527.69 ±8.17 / 537.80 ms │ 515.52 / 528.51 ±8.54 / 539.08 ms │ no change │
│ QQuery 21 │ 521.92 / 525.58 ±2.86 / 530.28 ms │ 514.22 / 518.56 ±4.86 / 527.88 ms │ no change │
│ QQuery 22 │ 991.13 / 1001.67 ±7.44 / 1014.11 ms │ 990.42 / 995.78 ±3.16 / 1000.12 ms │ no change │
│ QQuery 23 │ 3079.88 / 3106.31 ±23.01 / 3138.58 ms │ 3080.04 / 3118.87 ±29.33 / 3151.51 ms │ no change │
│ QQuery 24 │ 41.54 / 66.36 ±28.28 / 113.60 ms │ 41.63 / 48.46 ±7.20 / 57.68 ms │ +1.37x faster │
│ QQuery 25 │ 111.94 / 112.64 ±0.51 / 113.23 ms │ 111.28 / 112.76 ±1.27 / 114.98 ms │ no change │
│ QQuery 26 │ 42.17 / 43.32 ±1.79 / 46.87 ms │ 41.90 / 42.44 ±0.72 / 43.85 ms │ no change │
│ QQuery 27 │ 679.69 / 684.32 ±4.39 / 692.62 ms │ 675.50 / 679.50 ±3.15 / 683.73 ms │ no change │
│ QQuery 28 │ 3521.77 / 3641.27 ±105.93 / 3799.39 ms │ 3559.15 / 3739.09 ±186.79 / 4016.28 ms │ no change │
│ QQuery 29 │ 41.34 / 41.67 ±0.25 / 42.10 ms │ 41.07 / 41.67 ±0.52 / 42.58 ms │ no change │
│ QQuery 30 │ 562.66 / 585.14 ±16.47 / 609.21 ms │ 568.44 / 593.37 ±19.65 / 617.62 ms │ no change │
│ QQuery 31 │ 294.28 / 317.30 ±14.13 / 336.49 ms │ 298.91 / 310.29 ±10.61 / 328.08 ms │ no change │
│ QQuery 32 │ 1030.14 / 1082.63 ±44.01 / 1149.74 ms │ 1002.67 / 1022.84 ±19.61 / 1055.80 ms │ +1.06x faster │
│ QQuery 33 │ 26497.16 / 28747.99 ±1432.36 / 30924.53 ms │ 25559.35 / 28390.95 ±1728.23 / 31009.38 ms │ no change │
│ QQuery 34 │ 27366.22 / 30830.49 ±2170.57 / 33956.56 ms │ 25878.77 / 28682.12 ±2195.43 / 30751.40 ms │ +1.07x faster │
│ QQuery 35 │ 1072.28 / 1157.77 ±118.20 / 1387.90 ms │ 1103.06 / 1182.07 ±129.62 / 1440.82 ms │ no change │
│ QQuery 36 │ 158.54 / 172.50 ±9.32 / 187.29 ms │ 159.06 / 168.26 ±4.74 / 172.12 ms │ no change │
│ QQuery 37 │ 37.92 / 56.08 ±33.02 / 122.05 ms │ 37.85 / 54.23 ±23.87 / 101.70 ms │ no change │
│ QQuery 38 │ 42.15 / 45.88 ±2.66 / 50.30 ms │ 42.51 / 46.19 ±1.96 / 48.26 ms │ no change │
│ QQuery 39 │ 189.78 / 208.48 ±25.10 / 258.11 ms │ 186.28 / 202.80 ±20.83 / 243.17 ms │ no change │
│ QQuery 40 │ 14.59 / 15.12 ±0.37 / 15.72 ms │ 14.91 / 15.44 ±0.41 / 16.07 ms │ no change │
│ QQuery 41 │ 14.33 / 14.66 ±0.21 / 14.83 ms │ 14.25 / 14.55 ±0.28 / 15.09 ms │ no change │
│ QQuery 42 │ 13.77 / 13.91 ±0.14 / 14.12 ms │ 13.76 / 13.89 ±0.09 / 14.02 ms │ no change │
└───────────┴────────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 111760.93ms │
│ Total Time (issue_23072) │ 109274.03ms │
│ Average Time (HEAD) │ 2599.09ms │
│ Average Time (issue_23072) │ 2541.26ms │
│ Queries Faster │ 4 │
│ Queries Slower │ 3 │
│ Queries with No Change │ 36 │
│ Queries with Failure │ 0 │
└────────────────────────────┴─────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	560.1s
Peak memory	11.6 GiB
Avg memory	6.6 GiB
CPU user	4945.0s
CPU sys	320.9s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	550.1s
Peak memory	11.7 GiB
Avg memory	6.5 GiB
CPU user	4960.2s
CPU sys	329.8s
Peak spill	0 B

---

File an issue against this benchmark runner

[kosiew]

@Phoenix500526

Thanks for working through the earlier feedback. I think the overall direction is much better, especially keeping TableScopedPath as the cache key and moving schema validation into the cached value. The remaining concern is performance.

Although the schema fingerprint hash is now precomputed, successful cache hits still perform an O(schema width) comparison. Given the reported wide_schema and clickbench_partitioned results, I think it would be good to make the common cache hit path O(1) before merging.

[kosiew]

This is much closer to what I had in mind, but I think there is still one remaining performance issue.

SchemaFingerprint::eq first compares the precomputed hash and then falls back to comparing every column. On a successful cache hit, the hashes match, so we still end up walking the entire Vec<(String, DataType, bool)>. That means schema validation on cache hits is still O(schema width).

The precomputed hash helps reject different schemas cheaply, but it does not make the common successful validation path O(1). That also lines up with the benchmark results. wide_schema is still slightly slower and clickbench_partitioned Q6 still shows a noticeable regression.

Could we make successful validation constant time as well? For example, adding a pointer identity fast path for shared Arc<SchemaFingerprint> values would let the common case avoid the deep comparison while still preserving collision safety.

[Phoenix500526]

Good point. I added an Arc::ptr_eq fast path for the shared-fingerprint case, while keeping exact equality as a fallback for equivalent fingerprints from different Arc allocations. This keeps collision safety and existing cache reuse behavior, but makes the common successful validation path O(1).

[Phoenix500526]

Could you please help run a benchmark? @kosiew@mkleen

[mkleen]

run benchmark clickbench_partitioned

[mkleen]

run benchmark wide_schema

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4873836253-836-g9bp2 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (ec13f59) to d58e0c6 (merge-base) diff using: wide_schema
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4873833542-835-2pgcd 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (ec13f59) to d58e0c6 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and issue_23072
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ issue_23072 ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0 │ 1.23 / 3.96 ±5.40 / 14.76 ms │ 1.29 / 4.26 ±5.69 / 15.63 ms │ 1.08x slower │
│ QQuery 1 │ 12.56 / 12.74 ±0.12 / 12.87 ms │ 12.84 / 13.12 ±0.16 / 13.29 ms │ no change │
│ QQuery 2 │ 36.84 / 37.20 ±0.23 / 37.45 ms │ 36.23 / 36.55 ±0.32 / 37.11 ms │ no change │
│ QQuery 3 │ 30.69 / 31.56 ±0.76 / 32.42 ms │ 31.59 / 32.16 ±0.61 / 33.30 ms │ no change │
│ QQuery 4 │ 1693.17 / 1738.48 ±60.86 / 1855.33 ms │ 1673.21 / 1730.54 ±33.23 / 1759.82 ms │ no change │
│ QQuery 5 │ 1679.26 / 1767.71 ±80.89 / 1912.18 ms │ 1627.51 / 1745.53 ±121.32 / 1926.86 ms │ no change │
│ QQuery 6 │ 1.30 / 1.45 ±0.23 / 1.90 ms │ 1.27 / 1.43 ±0.27 / 1.97 ms │ no change │
│ QQuery 7 │ 13.95 / 14.34 ±0.22 / 14.61 ms │ 14.08 / 14.16 ±0.06 / 14.24 ms │ no change │
│ QQuery 8 │ 2024.08 / 2101.78 ±54.85 / 2176.52 ms │ 2043.14 / 2130.90 ±88.52 / 2272.94 ms │ no change │
│ QQuery 9 │ 459.93 / 519.22 ±47.41 / 587.25 ms │ 484.03 / 522.40 ±35.17 / 571.81 ms │ no change │
│ QQuery 10 │ 77.46 / 83.94 ±5.84 / 94.51 ms │ 76.67 / 78.42 ±1.30 / 80.70 ms │ +1.07x faster │
│ QQuery 11 │ 87.24 / 89.72 ±2.38 / 93.79 ms │ 91.24 / 94.95 ±5.43 / 105.59 ms │ 1.06x slower │
│ QQuery 12 │ 1768.87 / 1835.87 ±63.44 / 1953.82 ms │ 1744.14 / 1898.37 ±80.78 / 1979.68 ms │ no change │
│ QQuery 13 │ 463.21 / 664.40 ±146.65 / 900.16 ms │ 438.88 / 613.65 ±127.51 / 822.15 ms │ +1.08x faster │
│ QQuery 14 │ 528.14 / 565.30 ±23.23 / 601.07 ms │ 535.51 / 553.75 ±25.40 / 604.14 ms │ no change │
│ QQuery 15 │ 1896.25 / 2018.24 ±75.37 / 2121.25 ms │ 1877.90 / 1989.68 ±80.19 / 2094.41 ms │ no change │
│ QQuery 16 │ 4142.15 / 4311.62 ±152.29 / 4529.81 ms │ 4191.63 / 4407.83 ±145.10 / 4586.91 ms │ no change │
│ QQuery 17 │ 4216.50 / 4502.62 ±285.89 / 5050.35 ms │ 4292.73 / 4481.12 ±157.82 / 4753.71 ms │ no change │
│ QQuery 18 │ 18007.89 / 18268.35 ±283.02 / 18723.53 ms │ 18119.26 / 18734.43 ±389.08 / 19315.59 ms │ no change │
│ QQuery 19 │ 28.06 / 33.11 ±8.54 / 50.13 ms │ 28.39 / 35.73 ±11.63 / 58.49 ms │ 1.08x slower │
│ QQuery 20 │ 519.93 / 532.68 ±11.27 / 550.60 ms │ 514.37 / 519.07 ±4.77 / 527.11 ms │ no change │
│ QQuery 21 │ 514.65 / 526.24 ±9.88 / 542.21 ms │ 511.35 / 520.99 ±6.08 / 528.88 ms │ no change │
│ QQuery 22 │ 1000.76 / 1018.26 ±10.94 / 1033.62 ms │ 989.47 / 1005.27 ±11.16 / 1020.37 ms │ no change │
│ QQuery 23 │ 3049.90 / 3150.55 ±69.83 / 3230.20 ms │ 3137.72 / 3183.22 ±35.30 / 3217.83 ms │ no change │
│ QQuery 24 │ 42.66 / 46.41 ±4.65 / 54.75 ms │ 40.94 / 42.13 ±1.31 / 44.43 ms │ +1.10x faster │
│ QQuery 25 │ 110.49 / 116.45 ±7.70 / 130.12 ms │ 110.54 / 113.36 ±2.11 / 115.91 ms │ no change │
│ QQuery 26 │ 41.80 / 42.40 ±0.75 / 43.85 ms │ 42.12 / 45.89 ±4.27 / 54.10 ms │ 1.08x slower │
│ QQuery 27 │ 671.59 / 682.06 ±7.17 / 692.32 ms │ 668.85 / 671.56 ±3.69 / 678.77 ms │ no change │
│ QQuery 28 │ 3481.43 / 3644.74 ±188.55 / 4012.74 ms │ 3478.17 / 3736.11 ±150.12 / 3938.72 ms │ no change │
│ QQuery 29 │ 40.39 / 46.87 ±11.68 / 70.22 ms │ 40.11 / 78.39 ±63.91 / 204.72 ms │ 1.67x slower │
│ QQuery 30 │ 564.32 / 624.43 ±69.75 / 756.04 ms │ 559.37 / 593.40 ±21.22 / 624.45 ms │ no change │
│ QQuery 31 │ 289.38 / 307.37 ±11.43 / 324.05 ms │ 288.32 / 306.49 ±12.58 / 323.47 ms │ no change │
│ QQuery 32 │ 971.11 / 1019.83 ±34.48 / 1076.57 ms │ 946.73 / 1004.34 ±40.36 / 1067.42 ms │ no change │
│ QQuery 33 │ 27785.86 / 30548.27 ±2301.40 / 33016.98 ms │ 26730.68 / 28851.23 ±1639.95 / 31651.67 ms │ +1.06x faster │
│ QQuery 34 │ 28648.00 / 29597.23 ±893.84 / 31168.75 ms │ 27403.46 / 29205.92 ±1512.55 / 31833.37 ms │ no change │
│ QQuery 35 │ 991.36 / 1122.67 ±114.67 / 1335.91 ms │ 1100.43 / 1183.59 ±133.79 / 1450.53 ms │ 1.05x slower │
│ QQuery 36 │ 163.29 / 208.25 ±56.66 / 316.16 ms │ 162.38 / 174.82 ±9.46 / 188.91 ms │ +1.19x faster │
│ QQuery 37 │ 39.56 / 44.29 ±6.80 / 57.71 ms │ 37.05 / 52.97 ±25.27 / 103.28 ms │ 1.20x slower │
│ QQuery 38 │ 45.43 / 46.40 ±0.93 / 48.10 ms │ 42.75 / 45.27 ±2.07 / 48.62 ms │ no change │
│ QQuery 39 │ 188.92 / 198.99 ±6.95 / 209.21 ms │ 179.09 / 198.54 ±22.47 / 242.15 ms │ no change │
│ QQuery 40 │ 14.41 / 15.12 ±0.51 / 15.79 ms │ 14.39 / 14.51 ±0.19 / 14.89 ms │ no change │
│ QQuery 41 │ 14.61 / 14.91 ±0.25 / 15.26 ms │ 13.33 / 13.81 ±0.33 / 14.23 ms │ +1.08x faster │
│ QQuery 42 │ 14.06 / 14.19 ±0.09 / 14.29 ms │ 12.89 / 13.31 ±0.27 / 13.63 ms │ +1.07x faster │
└───────────┴────────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 112170.23ms │
│ Total Time (issue_23072) │ 110693.16ms │
│ Average Time (HEAD) │ 2608.61ms │
│ Average Time (issue_23072) │ 2574.26ms │
│ Queries Faster │ 7 │
│ Queries Slower │ 7 │
│ Queries with No Change │ 29 │
│ Queries with Failure │ 0 │
└────────────────────────────┴─────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	565.1s
Peak memory	12.1 GiB
Avg memory	6.4 GiB
CPU user	4969.7s
CPU sys	336.9s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	555.1s
Peak memory	11.9 GiB
Avg memory	6.6 GiB
CPU user	4959.8s
CPU sys	338.9s
Peak spill	0 B

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

group HEAD issue_23072
----- ---- -----------
wide_schema/Q01_narrow 1.01 80.6±0.38ms ? ?/sec 1.00 80.1±0.51ms ? ?/sec
wide_schema/Q01_wide 1.02 1040.5±5.02ms ? ?/sec 1.00 1023.8±5.48ms ? ?/sec
wide_schema/Q02_narrow 1.00 6.0±0.10ms ? ?/sec 1.00 6.0±0.11ms ? ?/sec
wide_schema/Q02_wide 1.00 912.4±7.61ms ? ?/sec 1.00 909.0±7.56ms ? ?/sec
wide_schema/Q03_narrow 1.02 15.4±0.17ms ? ?/sec 1.00 15.1±0.36ms ? ?/sec
wide_schema/Q03_wide 1.00 938.3±8.11ms ? ?/sec 1.00 939.4±10.22ms ? ?/sec
wide_schema/Q04_narrow 1.00 37.6±0.40ms ? ?/sec 1.00 37.5±0.41ms ? ?/sec
wide_schema/Q04_wide 1.00 996.9±6.75ms ? ?/sec 1.01 1001.9±9.94ms ? ?/sec

Resource Usage

wide_schema — base (merge-base)

Metric	Value
Wall time	640.1s
Peak memory	1.1 GiB
Avg memory	151.0 MiB
CPU user	371.2s
CPU sys	52.7s
Peak spill	0 B

wide_schema — branch

Metric	Value
Wall time	935.2s
Peak memory	1.1 GiB
Avg memory	101.7 MiB
CPU user	380.5s
CPU sys	53.1s
Peak spill	0 B

---

File an issue against this benchmark runner

[kosiew]

run benchmark clickbench_partitioned

[adriangbot]

🤖 Benchmark running (GKE) | trigger
Instance:c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4875007595-840-csw5t 6.12.85+ #1 SMP Mon May 11 08:17:35 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

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

Comparing issue/23072 (ec13f59) to d58e0c6 (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

---

File an issue against this benchmark runner

[adriangbot]

🤖 Benchmark completed (GKE) | trigger

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

CPU Details (lscpu)

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

Details

Comparing HEAD and issue_23072
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query ┃ HEAD ┃ issue_23072 ┃ Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0 │ 1.25 / 3.97 ±5.40 / 14.77 ms │ 1.22 / 4.00 ±5.46 / 14.93 ms │ no change │
│ QQuery 1 │ 12.85 / 12.91 ±0.04 / 12.97 ms │ 12.49 / 12.86 ±0.30 / 13.34 ms │ no change │
│ QQuery 2 │ 36.07 / 36.23 ±0.11 / 36.41 ms │ 35.58 / 35.89 ±0.21 / 36.24 ms │ no change │
│ QQuery 3 │ 30.88 / 31.41 ±0.48 / 32.25 ms │ 30.69 / 30.95 ±0.17 / 31.13 ms │ no change │
│ QQuery 4 │ 1617.67 / 1675.04 ±43.23 / 1735.13 ms │ 1621.34 / 1685.19 ±63.10 / 1766.38 ms │ no change │
│ QQuery 5 │ 1616.69 / 1782.39 ±146.90 / 2028.89 ms │ 1708.44 / 1805.84 ±71.79 / 1905.96 ms │ no change │
│ QQuery 6 │ 1.23 / 1.40 ±0.24 / 1.87 ms │ 1.27 / 1.42 ±0.24 / 1.90 ms │ no change │
│ QQuery 7 │ 13.90 / 19.29 ±10.52 / 40.32 ms │ 13.99 / 14.04 ±0.07 / 14.17 ms │ +1.37x faster │
│ QQuery 8 │ 1969.97 / 2057.08 ±55.84 / 2113.04 ms │ 1960.47 / 2014.23 ±39.14 / 2061.41 ms │ no change │
│ QQuery 9 │ 476.11 / 494.61 ±19.70 / 532.60 ms │ 480.53 / 506.50 ±16.19 / 530.68 ms │ no change │
│ QQuery 10 │ 77.41 / 84.92 ±13.38 / 111.65 ms │ 74.96 / 76.96 ±1.06 / 77.80 ms │ +1.10x faster │
│ QQuery 11 │ 87.44 / 89.82 ±2.42 / 94.46 ms │ 86.74 / 90.96 ±3.01 / 95.48 ms │ no change │
│ QQuery 12 │ 1644.10 / 1802.57 ±118.81 / 1999.77 ms │ 1555.69 / 1676.78 ±86.22 / 1788.11 ms │ +1.08x faster │
│ QQuery 13 │ 410.89 / 544.81 ±144.37 / 806.90 ms │ 428.99 / 680.11 ±203.65 / 893.86 ms │ 1.25x slower │
│ QQuery 14 │ 540.94 / 557.31 ±14.74 / 581.53 ms │ 546.88 / 555.23 ±8.20 / 569.47 ms │ no change │
│ QQuery 15 │ 1883.56 / 1942.65 ±46.78 / 1993.64 ms │ 1863.43 / 1940.60 ±41.77 / 1985.10 ms │ no change │
│ QQuery 16 │ 4108.25 / 4353.76 ±206.58 / 4719.00 ms │ 4117.76 / 4219.34 ±88.81 / 4370.33 ms │ no change │
│ QQuery 17 │ 4080.25 / 4169.05 ±57.09 / 4258.54 ms │ 4196.19 / 4280.59 ±114.45 / 4506.55 ms │ no change │
│ QQuery 18 │ 17879.28 / 18425.92 ±327.50 / 18742.55 ms │ 17416.76 / 18098.20 ±563.44 / 18812.11 ms │ no change │
│ QQuery 19 │ 27.96 / 28.56 ±0.52 / 29.25 ms │ 28.13 / 29.08 ±1.47 / 32.01 ms │ no change │
│ QQuery 20 │ 518.89 / 523.94 ±6.46 / 536.53 ms │ 510.20 / 524.59 ±9.12 / 537.02 ms │ no change │
│ QQuery 21 │ 513.70 / 523.51 ±8.77 / 536.43 ms │ 521.87 / 530.44 ±7.60 / 542.09 ms │ no change │
│ QQuery 22 │ 983.89 / 1000.53 ±11.98 / 1017.86 ms │ 988.03 / 998.76 ±6.36 / 1006.29 ms │ no change │
│ QQuery 23 │ 3033.67 / 3067.76 ±21.66 / 3097.74 ms │ 3048.05 / 3082.81 ±21.25 / 3110.74 ms │ no change │
│ QQuery 24 │ 41.17 / 41.73 ±0.68 / 43.05 ms │ 42.56 / 45.91 ±3.34 / 50.61 ms │ 1.10x slower │
│ QQuery 25 │ 110.44 / 119.56 ±9.73 / 137.87 ms │ 111.20 / 112.29 ±1.08 / 114.28 ms │ +1.06x faster │
│ QQuery 26 │ 41.80 / 44.27 ±4.21 / 52.65 ms │ 41.49 / 47.01 ±8.75 / 64.22 ms │ 1.06x slower │
│ QQuery 27 │ 672.05 / 681.36 ±8.26 / 695.86 ms │ 670.77 / 679.45 ±12.21 / 703.29 ms │ no change │
│ QQuery 28 │ 3516.43 / 3605.49 ±156.93 / 3919.07 ms │ 3528.23 / 3661.67 ±114.57 / 3856.39 ms │ no change │
│ QQuery 29 │ 40.44 / 53.07 ±17.98 / 86.77 ms │ 40.00 / 40.48 ±0.44 / 41.17 ms │ +1.31x faster │
│ QQuery 30 │ 550.44 / 564.41 ±13.79 / 590.62 ms │ 551.41 / 574.21 ±21.29 / 611.24 ms │ no change │
│ QQuery 31 │ 299.99 / 306.05 ±5.37 / 315.69 ms │ 282.46 / 291.21 ±8.34 / 302.86 ms │ no change │
│ QQuery 32 │ 940.60 / 969.50 ±26.90 / 1012.44 ms │ 935.37 / 972.72 ±24.55 / 997.53 ms │ no change │
│ QQuery 33 │ 26260.53 / 27583.92 ±1460.56 / 29972.19 ms │ 25992.03 / 28000.01 ±1054.33 / 29048.38 ms │ no change │
│ QQuery 34 │ 27036.02 / 27733.53 ±542.61 / 28689.48 ms │ 24285.16 / 27818.28 ±1916.29 / 29988.25 ms │ no change │
│ QQuery 35 │ 961.07 / 1072.64 ±144.03 / 1355.53 ms │ 970.47 / 1018.75 ±24.85 / 1036.70 ms │ +1.05x faster │
│ QQuery 36 │ 155.37 / 175.26 ±13.61 / 195.36 ms │ 160.41 / 172.42 ±11.57 / 192.58 ms │ no change │
│ QQuery 37 │ 36.90 / 40.29 ±3.63 / 47.30 ms │ 36.78 / 39.82 ±4.03 / 47.01 ms │ no change │
│ QQuery 38 │ 43.28 / 54.40 ±19.84 / 94.04 ms │ 40.89 / 50.41 ±12.27 / 74.68 ms │ +1.08x faster │
│ QQuery 39 │ 180.09 / 193.27 ±11.56 / 214.03 ms │ 184.09 / 188.50 ±4.57 / 195.09 ms │ no change │
│ QQuery 40 │ 14.44 / 14.99 ±0.47 / 15.87 ms │ 14.21 / 15.63 ±2.27 / 20.15 ms │ no change │
│ QQuery 41 │ 13.78 / 13.84 ±0.06 / 13.92 ms │ 13.59 / 13.92 ±0.21 / 14.21 ms │ no change │
│ QQuery 42 │ 13.07 / 13.34 ±0.18 / 13.61 ms │ 13.25 / 13.56 ±0.27 / 14.03 ms │ no change │
└───────────┴────────────────────────────────────────────┴────────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 106510.35ms │
│ Total Time (issue_23072) │ 106651.59ms │
│ Average Time (HEAD) │ 2476.98ms │
│ Average Time (issue_23072) │ 2480.27ms │
│ Queries Faster │ 7 │
│ Queries Slower │ 3 │
│ Queries with No Change │ 33 │
│ Queries with Failure │ 0 │
└────────────────────────────┴─────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	535.1s
Peak memory	11.5 GiB
Avg memory	6.4 GiB
CPU user	4819.1s
CPU sys	305.8s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	535.1s
Peak memory	12.1 GiB
Avg memory	6.5 GiB
CPU user	4814.9s
CPU sys	315.0s
Peak spill	0 B

---

File an issue against this benchmark runner

[kosiew]

from:

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 112170.23ms │
│ Total Time (issue_23072) │ 110693.16ms │
│ Average Time (HEAD) │ 2608.61ms │
│ Average Time (issue_23072) │ 2574.26ms │
│ Queries Faster │ 7 │
│ Queries Slower │ 7 │
│ Queries with No Change │ 29 │
│ Queries with Failure │ 0 │
└────────────────────────────┴─────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ Benchmark Summary ┃ ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ Total Time (HEAD) │ 106510.35ms │
│ Total Time (issue_23072) │ 106651.59ms │
│ Average Time (HEAD) │ 2476.98ms │
│ Average Time (issue_23072) │ 2480.27ms │
│ Queries Faster │ 7 │
│ Queries Slower │ 3 │
│ Queries with No Change │ 33 │
│ Queries with Failure │ 0 │
└────────────────────────────┴─────────────┘

there is no consistent reproducible performance regression

[kosiew]

@Phoenix500526
Thanks for your persistence in this.

Looks 👍 to me

[mkleen]

Thanks @Phoenix500526 👍