Skip to content

Add io-uring based ObjectStore for local file I/O#21673

Draft
Dandandan wants to merge 5 commits intoapache:mainfrom
Dandandan:worktree-io-uring-object-store
Draft

Add io-uring based ObjectStore for local file I/O#21673
Dandandan wants to merge 5 commits intoapache:mainfrom
Dandandan:worktree-io-uring-object-store

Conversation

@Dandandan
Copy link
Copy Markdown
Contributor

@Dandandan Dandandan commented Apr 16, 2026

Which issue does this PR close?

  • N/A (new feature)

Rationale for this change

Local file reads in DataFusion use object_store::local::LocalFileSystem, which issues one pread() per byte range. For Parquet column chunk reads this means many individual syscalls. Linux's io_uring allows batching these into a single io_uring_enter().

What changes are included in this PR?

New crate datafusion-object-store-iouring providing IoUringObjectStore:

  • Dedicated io_uring worker thread with a 256-entry submission queue
  • Channel-based dispatch: unbounded mpsc for requests, oneshot for responses
  • Batched reads: get_ranges() submits all byte ranges as SQEs in one io_uring_enter()
  • Graceful fallback: probes io_uring at startup; falls back to LocalFileSystem when unavailable (EPERM in Docker/seccomp, non-Linux platforms)
  • Feature flag: io-uring on datafusion-execution and datafusion, enabled by default

Architecture:

Tokio tasks                          io_uring thread
───────────                          ───────────────
                                     IoUring (256 entries)
  send(ReadRanges) ──► mpsc ──►     submit SQEs (batch)
                                     submit_and_wait()
  await oneshot    ◄── oneshot ◄──   collect CQEs → Bytes

To actually enable io_uring in benchmark environments, a companion PR adds a custom seccomp profile: adriangb/datafusion-benchmarking#4.

Are these changes tested?

6 unit tests covering put/get round-trip, single and multi-range reads, head, list, and empty-range edge cases. The io_uring code path requires a Linux host with io_uring support; on macOS CI the fallback path is exercised.

Are there any user-facing changes?

New optional crate and default feature flag. No behavioral changes when io_uring is unavailable — transparent fallback to LocalFileSystem.

Introduces `datafusion-object-store-iouring`, a new crate that provides
an `IoUringObjectStore` using Linux's io_uring interface for
high-performance local file reads. A dedicated thread runs an io_uring
event loop, and read requests (`get_opts`, `get_ranges`) are dispatched
via channels — enabling batched syscalls where multiple byte-range reads
(e.g., Parquet column chunks) are submitted in a single `io_uring_enter()`
call instead of individual `pread()` calls.

Key design:
- Dedicated io_uring worker thread with a 256-entry submission queue
- Unbounded mpsc channel for requests, oneshot channels for responses
- Range reads batched per-request; chunked if exceeding ring capacity
- Write/list/copy/delete operations delegated to LocalFileSystem
- On non-Linux platforms, all operations fall back to LocalFileSystem
- Feature flag `io-uring` on `datafusion-execution` to opt in

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@github-actions github-actions bot added the execution Related to the execution crate label Apr 16, 2026
- Resolve merge conflicts with upstream (version 53.0.0, object_store 0.13.2)
- Update ObjectStore trait impl for 0.13.2 API changes:
  - copy/copy_if_not_exists → copy_opts(CopyOptions)
  - delete → delete_stream (required method)
  - PutMultipartOpts → PutMultipartOptions
  - Import ObjectStoreExt for head() convenience method
- Enable io-uring feature by default in datafusion-execution

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@Dandandan
Copy link
Copy Markdown
Contributor Author

run benchmarks

@adriangbot
Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4259337795-1353-g2nzm 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 worktree-io-uring-object-store (0bf32dd) to 7bfa3fb (merge-base) diff using: tpcds
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4259337795-1354-6rzbf 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 worktree-io-uring-object-store (0bf32dd) to 7bfa3fb (merge-base) diff using: tpch
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4259337795-1352-277vz 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 worktree-io-uring-object-store (0bf32dd) to 7bfa3fb (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 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 worktree-io-uring-object-store
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃           worktree-io-uring-object-store ┃       Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.91 / 7.31 ±0.66 / 8.61 ms │              7.09 / 7.55 ±0.67 / 8.88 ms │    no change │
│ QQuery 2  │        148.64 / 149.45 ±0.94 / 150.60 ms │        147.71 / 148.21 ±0.52 / 149.20 ms │    no change │
│ QQuery 3  │        114.73 / 115.31 ±0.45 / 116.05 ms │        115.24 / 115.66 ±0.32 / 116.14 ms │    no change │
│ QQuery 4  │    1455.75 / 1489.62 ±25.84 / 1515.29 ms │    1432.48 / 1460.41 ±22.91 / 1486.05 ms │    no change │
│ QQuery 5  │        176.56 / 178.51 ±1.67 / 181.33 ms │        176.48 / 177.67 ±1.72 / 181.02 ms │    no change │
│ QQuery 6  │       881.65 / 911.35 ±24.26 / 950.71 ms │       852.20 / 894.81 ±31.25 / 947.28 ms │    no change │
│ QQuery 7  │        352.58 / 357.54 ±2.74 / 360.44 ms │        348.43 / 351.94 ±1.79 / 353.34 ms │    no change │
│ QQuery 8  │        116.96 / 119.44 ±1.37 / 121.12 ms │        118.97 / 120.06 ±0.90 / 121.53 ms │    no change │
│ QQuery 9  │        103.35 / 106.09 ±2.87 / 110.01 ms │        108.11 / 110.62 ±3.51 / 117.56 ms │    no change │
│ QQuery 10 │        109.22 / 111.63 ±1.90 / 114.41 ms │        108.74 / 109.26 ±0.72 / 110.66 ms │    no change │
│ QQuery 11 │    1065.95 / 1082.20 ±10.29 / 1094.07 ms │     1032.53 / 1043.31 ±7.81 / 1054.16 ms │    no change │
│ QQuery 12 │           47.36 / 48.92 ±2.00 / 52.89 ms │           45.75 / 46.66 ±0.86 / 48.08 ms │    no change │
│ QQuery 13 │        419.30 / 423.51 ±3.77 / 430.56 ms │        412.48 / 417.19 ±3.18 / 421.99 ms │    no change │
│ QQuery 14 │     1020.42 / 1028.25 ±6.51 / 1040.01 ms │     1014.22 / 1027.39 ±9.17 / 1038.89 ms │    no change │
│ QQuery 15 │           16.32 / 17.87 ±1.24 / 19.69 ms │           16.87 / 18.26 ±1.05 / 19.72 ms │    no change │
│ QQuery 16 │              7.92 / 8.36 ±0.37 / 9.03 ms │             7.54 / 8.42 ±1.26 / 10.92 ms │    no change │
│ QQuery 17 │        237.83 / 239.76 ±1.83 / 243.14 ms │        234.69 / 238.48 ±2.30 / 241.36 ms │    no change │
│ QQuery 18 │        132.53 / 133.01 ±0.54 / 133.73 ms │        131.57 / 132.39 ±0.98 / 134.29 ms │    no change │
│ QQuery 19 │        161.25 / 163.01 ±1.39 / 165.14 ms │        160.78 / 162.43 ±1.22 / 164.14 ms │    no change │
│ QQuery 20 │           14.27 / 14.98 ±0.61 / 15.89 ms │           14.73 / 15.21 ±0.32 / 15.67 ms │    no change │
│ QQuery 21 │           19.87 / 20.71 ±0.57 / 21.63 ms │           20.31 / 21.00 ±0.65 / 22.15 ms │    no change │
│ QQuery 22 │        491.31 / 498.61 ±4.92 / 505.56 ms │        497.29 / 501.97 ±4.28 / 509.11 ms │    no change │
│ QQuery 23 │        943.07 / 950.03 ±5.47 / 957.08 ms │        916.45 / 930.73 ±8.40 / 942.85 ms │    no change │
│ QQuery 24 │        395.57 / 399.26 ±2.21 / 402.11 ms │        393.77 / 399.62 ±3.85 / 405.18 ms │    no change │
│ QQuery 25 │        348.91 / 350.92 ±1.86 / 354.13 ms │        349.77 / 353.28 ±2.60 / 356.72 ms │    no change │
│ QQuery 26 │           83.63 / 85.32 ±1.86 / 88.91 ms │           81.74 / 83.54 ±1.08 / 84.61 ms │    no change │
│ QQuery 27 │              7.55 / 7.84 ±0.35 / 8.52 ms │              7.16 / 7.78 ±0.65 / 8.64 ms │    no change │
│ QQuery 28 │        152.21 / 153.12 ±0.85 / 154.20 ms │        151.77 / 153.05 ±0.78 / 153.96 ms │    no change │
│ QQuery 29 │        288.51 / 293.53 ±3.00 / 296.67 ms │        290.21 / 292.91 ±1.92 / 295.84 ms │    no change │
│ QQuery 30 │           43.67 / 44.81 ±0.80 / 45.74 ms │           43.49 / 46.10 ±1.52 / 47.84 ms │    no change │
│ QQuery 31 │        177.23 / 177.86 ±0.43 / 178.52 ms │        175.32 / 177.34 ±1.08 / 178.33 ms │    no change │
│ QQuery 32 │           58.83 / 59.89 ±0.85 / 61.43 ms │           58.31 / 58.97 ±0.71 / 59.94 ms │    no change │
│ QQuery 33 │        142.89 / 144.30 ±1.52 / 147.05 ms │        143.66 / 145.19 ±1.00 / 146.67 ms │    no change │
│ QQuery 34 │              7.18 / 7.83 ±0.65 / 9.06 ms │              7.48 / 7.65 ±0.20 / 8.01 ms │    no change │
│ QQuery 35 │        111.24 / 112.21 ±0.90 / 113.62 ms │        108.87 / 112.54 ±1.88 / 114.19 ms │    no change │
│ QQuery 36 │              6.72 / 6.92 ±0.15 / 7.10 ms │              6.86 / 7.22 ±0.49 / 8.17 ms │    no change │
│ QQuery 37 │              8.43 / 9.00 ±0.36 / 9.38 ms │             8.64 / 9.25 ±0.65 / 10.47 ms │    no change │
│ QQuery 38 │          86.40 / 91.75 ±4.79 / 100.65 ms │          88.30 / 92.73 ±6.60 / 105.74 ms │    no change │
│ QQuery 39 │        131.00 / 132.81 ±1.36 / 134.69 ms │        127.52 / 133.74 ±3.77 / 137.09 ms │    no change │
│ QQuery 40 │        113.58 / 119.82 ±6.26 / 131.79 ms │        111.66 / 120.01 ±8.53 / 135.54 ms │    no change │
│ QQuery 41 │           15.08 / 16.49 ±1.51 / 18.71 ms │           15.04 / 16.13 ±1.11 / 18.22 ms │    no change │
│ QQuery 42 │        108.30 / 110.44 ±1.71 / 112.59 ms │        109.08 / 112.20 ±1.63 / 113.77 ms │    no change │
│ QQuery 43 │              6.40 / 6.50 ±0.06 / 6.58 ms │              6.44 / 7.03 ±0.68 / 7.89 ms │ 1.08x slower │
│ QQuery 44 │           12.48 / 12.79 ±0.22 / 13.08 ms │           12.32 / 12.60 ±0.23 / 12.91 ms │    no change │
│ QQuery 45 │           52.61 / 53.26 ±0.75 / 54.74 ms │           52.56 / 53.52 ±0.88 / 55.19 ms │    no change │
│ QQuery 46 │             8.92 / 9.42 ±0.56 / 10.37 ms │              8.89 / 9.27 ±0.33 / 9.90 ms │    no change │
│ QQuery 47 │       806.71 / 818.86 ±10.77 / 833.20 ms │        783.78 / 796.33 ±7.95 / 807.42 ms │    no change │
│ QQuery 48 │        303.78 / 308.89 ±4.61 / 314.70 ms │        295.72 / 300.97 ±3.89 / 306.71 ms │    no change │
│ QQuery 49 │        259.54 / 261.08 ±1.30 / 263.35 ms │        258.31 / 260.19 ±2.24 / 264.04 ms │    no change │
│ QQuery 50 │        236.37 / 240.62 ±3.23 / 245.27 ms │        233.37 / 238.52 ±3.52 / 242.77 ms │    no change │
│ QQuery 51 │        185.86 / 188.10 ±2.47 / 192.42 ms │        183.45 / 187.58 ±3.15 / 191.76 ms │    no change │
│ QQuery 52 │        109.19 / 110.74 ±1.14 / 112.15 ms │        111.59 / 113.35 ±2.11 / 116.69 ms │    no change │
│ QQuery 53 │        104.79 / 106.95 ±1.51 / 109.05 ms │        106.14 / 107.96 ±1.10 / 108.99 ms │    no change │
│ QQuery 54 │        150.00 / 153.78 ±2.25 / 157.02 ms │        152.33 / 153.01 ±0.72 / 154.38 ms │    no change │
│ QQuery 55 │        109.19 / 110.95 ±1.44 / 113.33 ms │        110.65 / 112.35 ±1.60 / 115.10 ms │    no change │
│ QQuery 56 │        146.09 / 147.25 ±1.43 / 149.85 ms │        144.06 / 147.32 ±2.17 / 150.26 ms │    no change │
│ QQuery 57 │        180.04 / 183.19 ±1.85 / 185.76 ms │        180.58 / 183.18 ±1.51 / 185.13 ms │    no change │
│ QQuery 58 │        297.77 / 307.10 ±9.29 / 319.70 ms │        305.26 / 311.55 ±4.65 / 317.72 ms │    no change │
│ QQuery 59 │        207.33 / 210.41 ±1.70 / 212.55 ms │        208.87 / 211.45 ±2.54 / 215.87 ms │    no change │
│ QQuery 60 │        148.71 / 150.03 ±0.85 / 151.27 ms │        146.50 / 150.56 ±2.20 / 153.12 ms │    no change │
│ QQuery 61 │           13.71 / 14.01 ±0.21 / 14.25 ms │           14.01 / 14.11 ±0.10 / 14.28 ms │    no change │
│ QQuery 62 │      961.76 / 986.80 ±22.11 / 1015.32 ms │      930.98 / 977.49 ±27.28 / 1013.58 ms │    no change │
│ QQuery 63 │        108.72 / 110.05 ±2.14 / 114.32 ms │        105.41 / 108.31 ±1.50 / 109.77 ms │    no change │
│ QQuery 64 │        718.69 / 724.64 ±3.86 / 730.53 ms │        711.68 / 719.34 ±4.18 / 722.75 ms │    no change │
│ QQuery 65 │        276.22 / 281.05 ±4.95 / 290.36 ms │        271.05 / 274.05 ±1.90 / 275.60 ms │    no change │
│ QQuery 66 │        249.08 / 265.68 ±9.90 / 278.40 ms │       248.31 / 267.12 ±11.21 / 282.89 ms │    no change │
│ QQuery 67 │        331.30 / 339.10 ±6.65 / 350.13 ms │        329.70 / 339.71 ±5.35 / 345.46 ms │    no change │
│ QQuery 68 │            9.46 / 11.00 ±0.96 / 12.45 ms │            9.78 / 11.38 ±1.14 / 13.06 ms │    no change │
│ QQuery 69 │        107.14 / 109.50 ±2.01 / 112.88 ms │        105.64 / 107.92 ±2.11 / 111.24 ms │    no change │
│ QQuery 70 │       344.40 / 357.65 ±12.75 / 381.87 ms │       340.32 / 356.73 ±13.83 / 380.95 ms │    no change │
│ QQuery 71 │        139.99 / 144.38 ±2.68 / 147.02 ms │        138.10 / 139.03 ±0.83 / 140.14 ms │    no change │
│ QQuery 72 │       651.30 / 665.73 ±10.44 / 681.70 ms │        630.47 / 650.08 ±9.99 / 658.55 ms │    no change │
│ QQuery 73 │              7.47 / 8.16 ±0.45 / 8.71 ms │              8.03 / 8.74 ±0.40 / 9.24 ms │ 1.07x slower │
│ QQuery 74 │        678.87 / 688.56 ±6.58 / 698.64 ms │        656.49 / 663.54 ±5.15 / 669.85 ms │    no change │
│ QQuery 75 │        283.66 / 286.91 ±2.69 / 291.70 ms │        279.43 / 282.09 ±1.97 / 285.38 ms │    no change │
│ QQuery 76 │        136.99 / 137.96 ±0.90 / 139.22 ms │        137.07 / 138.45 ±1.68 / 141.69 ms │    no change │
│ QQuery 77 │        192.21 / 194.15 ±1.49 / 196.45 ms │        191.71 / 193.08 ±1.44 / 195.26 ms │    no change │
│ QQuery 78 │        359.27 / 363.38 ±2.87 / 366.92 ms │        348.91 / 354.94 ±4.05 / 361.48 ms │    no change │
│ QQuery 79 │        252.86 / 256.07 ±3.30 / 261.06 ms │        249.39 / 252.40 ±2.83 / 257.24 ms │    no change │
│ QQuery 80 │        327.24 / 330.12 ±3.02 / 335.60 ms │        325.45 / 327.14 ±1.48 / 329.67 ms │    no change │
│ QQuery 81 │           26.95 / 28.38 ±1.53 / 31.25 ms │           27.04 / 28.11 ±1.19 / 29.70 ms │    no change │
│ QQuery 82 │        204.26 / 206.43 ±2.13 / 209.97 ms │        203.54 / 206.51 ±2.75 / 211.73 ms │    no change │
│ QQuery 83 │           41.25 / 42.02 ±0.70 / 43.11 ms │           39.50 / 39.92 ±0.45 / 40.72 ms │    no change │
│ QQuery 84 │           49.64 / 50.82 ±0.87 / 51.67 ms │           49.76 / 50.29 ±0.50 / 51.17 ms │    no change │
│ QQuery 85 │        154.16 / 155.54 ±1.09 / 156.68 ms │        153.45 / 154.74 ±1.13 / 156.36 ms │    no change │
│ QQuery 86 │           40.18 / 40.95 ±0.66 / 41.87 ms │           41.17 / 42.08 ±0.74 / 43.29 ms │    no change │
│ QQuery 87 │           90.09 / 93.31 ±2.80 / 98.08 ms │           88.43 / 92.25 ±3.36 / 97.73 ms │    no change │
│ QQuery 88 │        104.06 / 104.87 ±0.73 / 105.79 ms │        102.81 / 103.59 ±0.69 / 104.62 ms │    no change │
│ QQuery 89 │        121.55 / 122.34 ±0.69 / 123.53 ms │        121.28 / 122.85 ±1.33 / 124.84 ms │    no change │
│ QQuery 90 │           24.20 / 25.45 ±0.95 / 27.01 ms │           24.37 / 24.97 ±0.80 / 26.55 ms │    no change │
│ QQuery 91 │           66.01 / 66.99 ±0.96 / 68.67 ms │           64.11 / 66.73 ±1.34 / 67.78 ms │    no change │
│ QQuery 92 │           58.73 / 61.23 ±1.43 / 62.45 ms │           59.66 / 60.24 ±0.66 / 61.33 ms │    no change │
│ QQuery 93 │        195.13 / 198.75 ±2.28 / 201.56 ms │        195.30 / 195.84 ±0.40 / 196.52 ms │    no change │
│ QQuery 94 │           63.37 / 63.90 ±0.42 / 64.48 ms │           63.08 / 63.79 ±0.71 / 65.10 ms │    no change │
│ QQuery 95 │        131.65 / 133.86 ±1.27 / 135.08 ms │        131.40 / 133.03 ±0.90 / 133.87 ms │    no change │
│ QQuery 96 │           75.48 / 76.63 ±0.84 / 77.76 ms │           71.35 / 75.22 ±2.06 / 76.76 ms │    no change │
│ QQuery 97 │        130.55 / 132.51 ±1.73 / 135.18 ms │        129.67 / 131.51 ±1.54 / 134.09 ms │    no change │
│ QQuery 98 │        155.25 / 159.18 ±2.43 / 161.85 ms │        157.74 / 161.11 ±2.33 / 164.42 ms │    no change │
│ QQuery 99 │ 10888.28 / 10938.14 ±32.39 / 10981.56 ms │ 10881.34 / 10920.80 ±25.00 / 10944.22 ms │    no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴──────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                             ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                             │ 32855.78ms │
│ Total Time (worktree-io-uring-object-store)   │ 32604.87ms │
│ Average Time (HEAD)                           │   331.88ms │
│ Average Time (worktree-io-uring-object-store) │   329.34ms │
│ Queries Faster                                │          0 │
│ Queries Slower                                │          2 │
│ Queries with No Change                        │         97 │
│ Queries with Failure                          │          0 │
└───────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 164.6s
Peak memory 5.7 GiB
Avg memory 4.7 GiB
CPU user 273.4s
CPU sys 18.0s
Peak spill 0 B

tpcds — branch

Metric Value
Wall time 163.3s
Peak memory 5.5 GiB
Avg memory 4.6 GiB
CPU user 270.4s
CPU sys 17.9s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 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 worktree-io-uring-object-store
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃        worktree-io-uring-object-store ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.24 / 4.54 ±6.46 / 17.45 ms │          1.19 / 4.49 ±6.44 / 17.36 ms │     no change │
│ QQuery 1  │        15.33 / 15.62 ±0.15 / 15.75 ms │        15.09 / 15.55 ±0.32 / 16.10 ms │     no change │
│ QQuery 2  │        44.99 / 45.31 ±0.23 / 45.70 ms │        44.71 / 44.93 ±0.25 / 45.40 ms │     no change │
│ QQuery 3  │        39.83 / 40.90 ±1.36 / 43.54 ms │        39.82 / 40.93 ±0.62 / 41.62 ms │     no change │
│ QQuery 4  │     290.75 / 294.95 ±3.71 / 301.88 ms │     292.10 / 297.03 ±4.15 / 304.01 ms │     no change │
│ QQuery 5  │     346.28 / 349.29 ±2.95 / 354.20 ms │     341.95 / 351.12 ±7.02 / 360.18 ms │     no change │
│ QQuery 6  │           5.06 / 7.07 ±1.38 / 9.36 ms │           5.17 / 6.53 ±1.42 / 9.26 ms │ +1.08x faster │
│ QQuery 7  │        17.89 / 18.14 ±0.23 / 18.48 ms │        17.59 / 17.79 ±0.12 / 17.93 ms │     no change │
│ QQuery 8  │     413.63 / 422.79 ±5.71 / 430.70 ms │     413.00 / 430.06 ±8.64 / 436.07 ms │     no change │
│ QQuery 9  │    679.17 / 692.82 ±11.89 / 711.63 ms │     650.32 / 662.51 ±7.28 / 672.45 ms │     no change │
│ QQuery 10 │       95.46 / 99.41 ±4.86 / 108.58 ms │       94.84 / 99.32 ±3.10 / 104.53 ms │     no change │
│ QQuery 11 │     110.59 / 112.32 ±1.00 / 113.51 ms │     108.42 / 110.11 ±1.15 / 111.64 ms │     no change │
│ QQuery 12 │     349.35 / 355.92 ±3.54 / 359.44 ms │     341.30 / 348.22 ±3.67 / 351.97 ms │     no change │
│ QQuery 13 │    467.63 / 477.01 ±10.17 / 494.39 ms │     461.98 / 473.19 ±9.32 / 483.94 ms │     no change │
│ QQuery 14 │     361.04 / 366.48 ±3.23 / 370.40 ms │     347.62 / 351.68 ±3.95 / 358.16 ms │     no change │
│ QQuery 15 │    367.65 / 385.56 ±15.31 / 406.78 ms │    352.20 / 373.19 ±13.88 / 390.42 ms │     no change │
│ QQuery 16 │     734.54 / 741.42 ±7.63 / 754.09 ms │     723.68 / 731.32 ±4.66 / 737.12 ms │     no change │
│ QQuery 17 │     728.52 / 740.80 ±7.47 / 749.94 ms │     714.10 / 720.69 ±5.95 / 731.06 ms │     no change │
│ QQuery 18 │ 1469.22 / 1493.30 ±17.11 / 1516.36 ms │ 1476.34 / 1491.61 ±19.05 / 1529.14 ms │     no change │
│ QQuery 19 │       37.74 / 45.17 ±12.59 / 70.21 ms │       39.26 / 48.49 ±17.09 / 82.66 ms │  1.07x slower │
│ QQuery 20 │    724.68 / 745.27 ±16.99 / 771.78 ms │    716.46 / 734.35 ±18.40 / 757.27 ms │     no change │
│ QQuery 21 │     767.80 / 775.97 ±8.22 / 789.71 ms │    763.85 / 784.28 ±22.63 / 827.45 ms │     no change │
│ QQuery 22 │ 1140.49 / 1152.81 ±10.86 / 1166.59 ms │  1131.07 / 1135.08 ±2.81 / 1139.18 ms │     no change │
│ QQuery 23 │ 3121.24 / 3153.06 ±30.63 / 3207.13 ms │ 3113.15 / 3132.45 ±16.38 / 3161.26 ms │     no change │
│ QQuery 24 │     102.21 / 104.68 ±3.01 / 110.48 ms │      99.43 / 107.39 ±5.99 / 114.67 ms │     no change │
│ QQuery 25 │     140.02 / 142.33 ±2.11 / 145.59 ms │     141.58 / 143.50 ±1.31 / 145.31 ms │     no change │
│ QQuery 26 │     101.92 / 104.10 ±1.85 / 107.40 ms │      98.56 / 101.80 ±2.45 / 104.53 ms │     no change │
│ QQuery 27 │     852.81 / 859.98 ±7.55 / 874.23 ms │    850.12 / 862.61 ±11.36 / 877.32 ms │     no change │
│ QQuery 28 │ 3244.04 / 3271.69 ±17.80 / 3291.78 ms │ 3231.96 / 3268.81 ±21.66 / 3291.95 ms │     no change │
│ QQuery 29 │        50.92 / 57.19 ±6.23 / 68.95 ms │        50.56 / 55.04 ±5.02 / 63.91 ms │     no change │
│ QQuery 30 │     370.72 / 378.83 ±7.16 / 390.72 ms │     367.90 / 374.82 ±5.94 / 384.85 ms │     no change │
│ QQuery 31 │    362.65 / 380.25 ±11.15 / 397.76 ms │     377.60 / 395.70 ±9.42 / 403.10 ms │     no change │
│ QQuery 32 │ 1041.10 / 1065.12 ±19.15 / 1096.14 ms │ 1213.93 / 1250.69 ±35.84 / 1309.99 ms │  1.17x slower │
│ QQuery 33 │ 1459.29 / 1480.44 ±11.42 / 1492.81 ms │ 1485.09 / 1552.30 ±44.90 / 1603.99 ms │     no change │
│ QQuery 34 │ 1526.29 / 1539.82 ±13.72 / 1559.55 ms │ 1526.12 / 1617.63 ±88.30 / 1777.08 ms │  1.05x slower │
│ QQuery 35 │    393.98 / 406.61 ±10.74 / 425.43 ms │     384.56 / 399.57 ±9.81 / 413.58 ms │     no change │
│ QQuery 36 │     113.94 / 122.30 ±4.32 / 126.39 ms │     117.56 / 122.10 ±3.43 / 126.85 ms │     no change │
│ QQuery 37 │        48.79 / 49.74 ±0.88 / 51.24 ms │        49.44 / 52.41 ±1.55 / 53.58 ms │  1.05x slower │
│ QQuery 38 │        76.04 / 77.75 ±1.16 / 78.98 ms │        76.59 / 79.62 ±1.90 / 82.57 ms │     no change │
│ QQuery 39 │     209.77 / 221.85 ±6.25 / 227.60 ms │     214.23 / 221.81 ±8.58 / 237.65 ms │     no change │
│ QQuery 40 │        24.63 / 26.77 ±1.47 / 28.55 ms │        24.70 / 26.68 ±1.31 / 28.49 ms │     no change │
│ QQuery 41 │        19.97 / 21.24 ±0.71 / 22.05 ms │        20.58 / 22.07 ±1.47 / 24.85 ms │     no change │
│ QQuery 42 │        19.94 / 21.31 ±1.20 / 23.49 ms │        19.52 / 20.57 ±0.62 / 21.42 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                             ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                             │ 22867.95ms │
│ Total Time (worktree-io-uring-object-store)   │ 23080.06ms │
│ Average Time (HEAD)                           │   531.81ms │
│ Average Time (worktree-io-uring-object-store) │   536.75ms │
│ Queries Faster                                │          1 │
│ Queries Slower                                │          4 │
│ Queries with No Change                        │         38 │
│ Queries with Failure                          │          0 │
└───────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 115.3s
Peak memory 37.5 GiB
Avg memory 26.6 GiB
CPU user 1090.8s
CPU sys 87.4s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 116.3s
Peak memory 34.1 GiB
Avg memory 26.0 GiB
CPU user 1082.9s
CPU sys 100.6s
Peak spill 0 B

File an issue against this benchmark runner

- Fix missing `StreamExt` import for `.boxed()` on Linux code path
- Fix `GetResult.range` type: `Range<u64>` in object_store 0.13.2
- Add `io-uring` feature to datafusion core, forwarding to execution
- Add to core's default features so benchmarks get it automatically

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@github-actions github-actions bot added the core Core DataFusion crate label Apr 16, 2026
CI runners and Docker containers often block io_uring_setup via
seccomp filters (EPERM). Instead of failing hard, probe availability
at construction time and gracefully fall back to LocalFileSystem
for all read operations when io_uring cannot be initialized.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@Dandandan
Copy link
Copy Markdown
Contributor Author

run benchmarks

@adriangbot
Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4260127134-1366-226ds 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 worktree-io-uring-object-store (87539f6) to 7bfa3fb (merge-base) diff using: tpch
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4260127134-1365-xbxvj 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 worktree-io-uring-object-store (87539f6) to 7bfa3fb (merge-base) diff using: tpcds
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4260127134-1364-5gzvs 6.12.55+ #1 SMP Sun Feb 1 08:59:41 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 worktree-io-uring-object-store (87539f6) to 7bfa3fb (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete


File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 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 worktree-io-uring-object-store
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃           worktree-io-uring-object-store ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.76 / 7.18 ±0.72 / 8.61 ms │              6.67 / 7.12 ±0.70 / 8.51 ms │     no change │
│ QQuery 2  │        144.65 / 146.10 ±1.01 / 147.70 ms │        146.89 / 148.13 ±0.63 / 148.66 ms │     no change │
│ QQuery 3  │        113.26 / 114.31 ±0.80 / 115.41 ms │        113.97 / 114.76 ±0.77 / 116.17 ms │     no change │
│ QQuery 4  │    1316.03 / 1344.16 ±18.57 / 1359.75 ms │    1296.60 / 1329.65 ±18.68 / 1349.14 ms │     no change │
│ QQuery 5  │        173.71 / 175.79 ±2.34 / 179.74 ms │        173.77 / 174.54 ±0.73 / 175.53 ms │     no change │
│ QQuery 6  │       825.35 / 867.65 ±21.82 / 888.25 ms │       802.40 / 833.75 ±23.05 / 871.14 ms │     no change │
│ QQuery 7  │        345.00 / 346.28 ±2.08 / 350.38 ms │        341.20 / 346.29 ±4.04 / 352.67 ms │     no change │
│ QQuery 8  │        114.09 / 117.70 ±1.84 / 118.97 ms │        116.17 / 117.57 ±1.12 / 119.57 ms │     no change │
│ QQuery 9  │        101.28 / 110.39 ±9.49 / 128.77 ms │        101.39 / 105.38 ±2.91 / 108.78 ms │     no change │
│ QQuery 10 │        106.89 / 107.98 ±0.65 / 108.78 ms │        107.12 / 108.70 ±0.99 / 109.86 ms │     no change │
│ QQuery 11 │       903.33 / 918.24 ±10.12 / 932.52 ms │        896.44 / 905.51 ±6.45 / 916.15 ms │     no change │
│ QQuery 12 │           45.86 / 47.26 ±0.79 / 48.22 ms │           44.65 / 46.72 ±1.33 / 48.08 ms │     no change │
│ QQuery 13 │        399.06 / 401.01 ±1.44 / 402.90 ms │        398.25 / 403.08 ±3.23 / 407.83 ms │     no change │
│ QQuery 14 │     1007.67 / 1013.22 ±4.66 / 1020.97 ms │     1001.39 / 1011.09 ±9.61 / 1025.15 ms │     no change │
│ QQuery 15 │           15.45 / 16.30 ±0.87 / 17.97 ms │           15.46 / 16.49 ±1.54 / 19.53 ms │     no change │
│ QQuery 16 │              7.17 / 7.54 ±0.48 / 8.42 ms │              7.28 / 8.30 ±0.98 / 9.79 ms │  1.10x slower │
│ QQuery 17 │        230.66 / 232.40 ±1.05 / 233.52 ms │        227.80 / 230.27 ±2.08 / 234.06 ms │     no change │
│ QQuery 18 │        128.37 / 129.60 ±1.06 / 131.22 ms │        128.32 / 129.54 ±1.40 / 132.23 ms │     no change │
│ QQuery 19 │        153.56 / 155.13 ±1.30 / 156.64 ms │        156.04 / 156.83 ±0.77 / 158.26 ms │     no change │
│ QQuery 20 │           13.45 / 13.90 ±0.37 / 14.45 ms │           13.72 / 14.14 ±0.36 / 14.63 ms │     no change │
│ QQuery 21 │           19.38 / 20.08 ±0.56 / 20.73 ms │           19.47 / 20.07 ±0.66 / 21.33 ms │     no change │
│ QQuery 22 │        482.14 / 487.32 ±2.94 / 491.09 ms │        489.13 / 493.34 ±4.00 / 498.41 ms │     no change │
│ QQuery 23 │        866.45 / 875.57 ±6.38 / 883.98 ms │       865.53 / 882.66 ±14.83 / 909.03 ms │     no change │
│ QQuery 24 │        381.29 / 384.17 ±2.74 / 388.91 ms │        387.34 / 390.45 ±1.61 / 391.86 ms │     no change │
│ QQuery 25 │        342.80 / 344.58 ±1.40 / 346.56 ms │        346.11 / 346.65 ±0.69 / 348.00 ms │     no change │
│ QQuery 26 │           80.68 / 82.92 ±1.47 / 85.26 ms │           81.88 / 83.90 ±1.20 / 85.10 ms │     no change │
│ QQuery 27 │              6.92 / 7.27 ±0.21 / 7.46 ms │              7.21 / 7.55 ±0.47 / 8.44 ms │     no change │
│ QQuery 28 │        148.39 / 149.12 ±0.39 / 149.46 ms │        150.02 / 151.88 ±1.38 / 153.78 ms │     no change │
│ QQuery 29 │        282.79 / 284.22 ±1.39 / 286.74 ms │        281.21 / 284.78 ±2.34 / 288.35 ms │     no change │
│ QQuery 30 │           42.60 / 45.24 ±1.80 / 47.26 ms │           44.24 / 45.47 ±0.97 / 46.86 ms │     no change │
│ QQuery 31 │        168.13 / 172.29 ±3.39 / 178.36 ms │        172.95 / 174.80 ±1.62 / 177.36 ms │     no change │
│ QQuery 32 │           56.77 / 58.31 ±1.13 / 60.11 ms │           57.56 / 58.91 ±1.51 / 61.82 ms │     no change │
│ QQuery 33 │        140.50 / 142.37 ±1.23 / 143.64 ms │        142.47 / 143.49 ±0.92 / 144.93 ms │     no change │
│ QQuery 34 │              6.97 / 7.25 ±0.38 / 8.01 ms │              7.10 / 7.38 ±0.28 / 7.82 ms │     no change │
│ QQuery 35 │        105.62 / 108.81 ±1.61 / 109.89 ms │        107.27 / 109.38 ±1.18 / 110.59 ms │     no change │
│ QQuery 36 │              6.52 / 6.89 ±0.22 / 7.20 ms │              6.81 / 7.01 ±0.16 / 7.30 ms │     no change │
│ QQuery 37 │             8.17 / 9.29 ±0.86 / 10.67 ms │              8.47 / 8.85 ±0.24 / 9.14 ms │     no change │
│ QQuery 38 │           82.98 / 86.54 ±3.83 / 93.47 ms │           85.82 / 89.74 ±3.88 / 97.06 ms │     no change │
│ QQuery 39 │        124.73 / 128.21 ±2.37 / 131.53 ms │        123.40 / 129.18 ±3.61 / 134.77 ms │     no change │
│ QQuery 40 │        111.70 / 116.59 ±6.47 / 129.17 ms │        108.93 / 115.47 ±7.92 / 131.06 ms │     no change │
│ QQuery 41 │           14.44 / 16.20 ±1.33 / 17.77 ms │           15.09 / 15.86 ±0.80 / 17.29 ms │     no change │
│ QQuery 42 │        107.16 / 108.96 ±1.61 / 111.27 ms │        109.04 / 110.27 ±1.12 / 112.14 ms │     no change │
│ QQuery 43 │              6.01 / 6.11 ±0.12 / 6.35 ms │              6.15 / 6.26 ±0.13 / 6.51 ms │     no change │
│ QQuery 44 │           11.64 / 12.11 ±0.45 / 12.88 ms │           11.56 / 12.00 ±0.33 / 12.49 ms │     no change │
│ QQuery 45 │           51.17 / 51.99 ±1.05 / 53.99 ms │           50.65 / 52.53 ±1.38 / 54.02 ms │     no change │
│ QQuery 46 │              8.48 / 8.70 ±0.18 / 8.99 ms │              8.69 / 9.08 ±0.30 / 9.54 ms │     no change │
│ QQuery 47 │       705.72 / 720.13 ±13.53 / 743.63 ms │        734.46 / 741.62 ±8.98 / 758.37 ms │     no change │
│ QQuery 48 │        291.61 / 294.39 ±1.79 / 296.88 ms │        293.57 / 295.17 ±1.38 / 297.67 ms │     no change │
│ QQuery 49 │        253.53 / 255.96 ±2.18 / 259.35 ms │        253.80 / 255.12 ±0.74 / 255.69 ms │     no change │
│ QQuery 50 │        222.46 / 226.63 ±3.10 / 230.42 ms │        216.31 / 223.70 ±5.45 / 230.56 ms │     no change │
│ QQuery 51 │        178.61 / 182.65 ±2.22 / 185.24 ms │        182.10 / 184.08 ±1.58 / 185.96 ms │     no change │
│ QQuery 52 │        107.57 / 108.62 ±1.19 / 110.79 ms │        107.98 / 109.58 ±1.52 / 111.58 ms │     no change │
│ QQuery 53 │        102.80 / 103.72 ±0.82 / 105.24 ms │        102.78 / 103.57 ±0.80 / 105.01 ms │     no change │
│ QQuery 54 │        146.95 / 148.07 ±0.61 / 148.68 ms │        146.50 / 147.56 ±0.99 / 149.21 ms │     no change │
│ QQuery 55 │        107.02 / 108.34 ±0.83 / 109.65 ms │        107.12 / 108.72 ±1.09 / 110.12 ms │     no change │
│ QQuery 56 │        140.98 / 141.44 ±0.31 / 141.77 ms │        140.71 / 143.27 ±2.89 / 148.58 ms │     no change │
│ QQuery 57 │        175.07 / 176.52 ±1.06 / 177.73 ms │        175.40 / 176.55 ±0.78 / 177.49 ms │     no change │
│ QQuery 58 │        297.46 / 302.16 ±3.81 / 307.38 ms │        292.43 / 298.90 ±5.78 / 307.48 ms │     no change │
│ QQuery 59 │        197.24 / 199.86 ±2.08 / 203.24 ms │        201.36 / 202.98 ±1.44 / 204.96 ms │     no change │
│ QQuery 60 │        146.23 / 146.65 ±0.51 / 147.57 ms │        143.49 / 145.80 ±1.26 / 147.17 ms │     no change │
│ QQuery 61 │           13.06 / 13.31 ±0.24 / 13.77 ms │           13.20 / 13.46 ±0.20 / 13.75 ms │     no change │
│ QQuery 62 │       907.56 / 934.75 ±24.63 / 973.32 ms │       895.13 / 930.66 ±34.51 / 987.66 ms │     no change │
│ QQuery 63 │        104.97 / 106.79 ±1.04 / 107.67 ms │        105.20 / 107.11 ±1.62 / 109.27 ms │     no change │
│ QQuery 64 │        683.42 / 691.19 ±5.03 / 697.56 ms │        676.58 / 688.40 ±8.45 / 701.68 ms │     no change │
│ QQuery 65 │        247.96 / 252.25 ±4.22 / 260.15 ms │        249.69 / 253.35 ±2.37 / 256.52 ms │     no change │
│ QQuery 66 │        241.39 / 251.48 ±6.17 / 258.81 ms │        242.05 / 254.97 ±8.10 / 267.01 ms │     no change │
│ QQuery 67 │        311.10 / 322.67 ±6.78 / 330.90 ms │        301.36 / 313.25 ±7.76 / 323.68 ms │     no change │
│ QQuery 68 │            9.48 / 10.41 ±0.58 / 11.12 ms │             8.50 / 9.79 ±1.13 / 11.77 ms │ +1.06x faster │
│ QQuery 69 │        102.27 / 104.00 ±1.18 / 105.65 ms │        102.93 / 104.57 ±1.34 / 106.25 ms │     no change │
│ QQuery 70 │       334.56 / 351.18 ±21.35 / 392.64 ms │       338.38 / 352.79 ±14.70 / 380.76 ms │     no change │
│ QQuery 71 │        134.87 / 137.32 ±2.43 / 141.62 ms │        134.50 / 136.77 ±1.25 / 138.36 ms │     no change │
│ QQuery 72 │       608.99 / 626.20 ±11.05 / 643.67 ms │        616.38 / 622.92 ±7.67 / 636.53 ms │     no change │
│ QQuery 73 │             7.16 / 8.33 ±1.30 / 10.78 ms │             7.08 / 8.39 ±1.41 / 10.53 ms │     no change │
│ QQuery 74 │       565.24 / 583.91 ±12.89 / 598.28 ms │        561.51 / 567.27 ±3.82 / 572.91 ms │     no change │
│ QQuery 75 │        277.52 / 279.70 ±1.61 / 282.26 ms │        275.28 / 279.00 ±2.29 / 281.59 ms │     no change │
│ QQuery 76 │        132.35 / 135.22 ±1.62 / 137.06 ms │        133.55 / 133.96 ±0.28 / 134.32 ms │     no change │
│ QQuery 77 │        189.03 / 190.72 ±1.19 / 192.49 ms │        188.13 / 189.94 ±1.78 / 192.51 ms │     no change │
│ QQuery 78 │        338.57 / 342.70 ±4.11 / 350.05 ms │        336.03 / 343.14 ±4.64 / 350.40 ms │     no change │
│ QQuery 79 │        233.46 / 237.02 ±3.49 / 243.41 ms │        229.89 / 233.94 ±3.21 / 237.76 ms │     no change │
│ QQuery 80 │        323.12 / 327.05 ±2.38 / 330.09 ms │        322.35 / 324.47 ±2.28 / 328.71 ms │     no change │
│ QQuery 81 │           26.85 / 27.72 ±0.57 / 28.58 ms │           26.63 / 27.61 ±0.97 / 29.21 ms │     no change │
│ QQuery 82 │        198.53 / 199.90 ±1.40 / 202.04 ms │        195.80 / 199.73 ±3.04 / 204.15 ms │     no change │
│ QQuery 83 │           38.30 / 39.65 ±1.25 / 41.87 ms │           38.80 / 39.65 ±0.85 / 41.27 ms │     no change │
│ QQuery 84 │           48.75 / 49.93 ±1.42 / 52.71 ms │           48.19 / 49.13 ±0.91 / 50.60 ms │     no change │
│ QQuery 85 │        148.86 / 151.67 ±2.55 / 156.13 ms │        148.45 / 151.59 ±1.82 / 153.77 ms │     no change │
│ QQuery 86 │           38.15 / 40.31 ±1.41 / 41.81 ms │           39.36 / 40.11 ±0.44 / 40.65 ms │     no change │
│ QQuery 87 │           85.70 / 88.67 ±2.58 / 93.28 ms │           83.38 / 87.31 ±4.30 / 95.58 ms │     no change │
│ QQuery 88 │        100.38 / 101.15 ±0.75 / 102.33 ms │        100.32 / 100.74 ±0.53 / 101.76 ms │     no change │
│ QQuery 89 │        117.05 / 120.71 ±1.92 / 122.38 ms │        119.68 / 120.85 ±0.78 / 122.12 ms │     no change │
│ QQuery 90 │           23.30 / 23.97 ±0.39 / 24.41 ms │           23.32 / 23.88 ±0.45 / 24.42 ms │     no change │
│ QQuery 91 │           63.61 / 64.51 ±0.75 / 65.52 ms │           63.34 / 64.62 ±0.73 / 65.60 ms │     no change │
│ QQuery 92 │           57.00 / 58.02 ±0.64 / 58.81 ms │           56.81 / 57.81 ±0.56 / 58.47 ms │     no change │
│ QQuery 93 │        184.01 / 188.22 ±2.40 / 190.60 ms │        185.01 / 187.90 ±1.55 / 189.17 ms │     no change │
│ QQuery 94 │           60.73 / 61.26 ±0.38 / 61.86 ms │           60.10 / 62.16 ±2.12 / 66.18 ms │     no change │
│ QQuery 95 │        127.57 / 129.02 ±0.96 / 130.29 ms │        127.76 / 128.43 ±0.61 / 129.49 ms │     no change │
│ QQuery 96 │           72.97 / 74.11 ±0.78 / 75.43 ms │           70.31 / 72.70 ±1.89 / 74.98 ms │     no change │
│ QQuery 97 │        125.37 / 128.43 ±2.58 / 132.44 ms │        122.76 / 126.40 ±2.10 / 128.61 ms │     no change │
│ QQuery 98 │        152.63 / 156.23 ±2.63 / 159.89 ms │        152.74 / 155.51 ±2.13 / 157.81 ms │     no change │
│ QQuery 99 │ 10790.78 / 10815.29 ±27.17 / 10867.50 ms │ 10717.67 / 10755.77 ±22.69 / 10786.04 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                             ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                             │ 31533.39ms │
│ Total Time (worktree-io-uring-object-store)   │ 31421.43ms │
│ Average Time (HEAD)                           │   318.52ms │
│ Average Time (worktree-io-uring-object-store) │   317.39ms │
│ Queries Faster                                │          1 │
│ Queries Slower                                │          1 │
│ Queries with No Change                        │         97 │
│ Queries with Failure                          │          0 │
└───────────────────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric Value
Wall time 158.0s
Peak memory 5.2 GiB
Avg memory 4.5 GiB
CPU user 260.5s
CPU sys 17.4s
Peak spill 0 B

tpcds — branch

Metric Value
Wall time 157.4s
Peak memory 5.5 GiB
Avg memory 4.6 GiB
CPU user 260.5s
CPU sys 16.5s
Peak spill 0 B

File an issue against this benchmark runner

@adriangbot
Copy link
Copy Markdown

🤖 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 worktree-io-uring-object-store
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃        worktree-io-uring-object-store ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.24 / 4.60 ±6.55 / 17.69 ms │          1.22 / 4.53 ±6.46 / 17.45 ms │     no change │
│ QQuery 1  │        15.02 / 15.41 ±0.20 / 15.58 ms │        14.56 / 14.96 ±0.33 / 15.56 ms │     no change │
│ QQuery 2  │        45.66 / 46.12 ±0.55 / 47.18 ms │        44.85 / 45.11 ±0.20 / 45.37 ms │     no change │
│ QQuery 3  │        40.62 / 41.44 ±0.99 / 43.38 ms │        40.86 / 41.18 ±0.34 / 41.80 ms │     no change │
│ QQuery 4  │     288.31 / 293.25 ±3.49 / 299.09 ms │     293.13 / 296.88 ±2.67 / 300.90 ms │     no change │
│ QQuery 5  │     343.01 / 353.71 ±5.92 / 359.18 ms │     345.68 / 355.65 ±5.98 / 364.00 ms │     no change │
│ QQuery 6  │           6.11 / 6.30 ±0.18 / 6.63 ms │           5.18 / 6.79 ±1.33 / 9.07 ms │  1.08x slower │
│ QQuery 7  │        17.80 / 18.10 ±0.24 / 18.31 ms │        17.12 / 18.20 ±0.95 / 19.61 ms │     no change │
│ QQuery 8  │    421.00 / 434.71 ±17.40 / 468.82 ms │     417.91 / 428.99 ±9.16 / 444.16 ms │     no change │
│ QQuery 9  │    675.29 / 689.04 ±12.36 / 711.49 ms │     675.61 / 690.22 ±8.13 / 699.56 ms │     no change │
│ QQuery 10 │       95.96 / 98.26 ±2.49 / 103.01 ms │       94.16 / 97.54 ±3.04 / 102.70 ms │     no change │
│ QQuery 11 │     110.40 / 111.88 ±0.85 / 112.88 ms │     107.59 / 108.67 ±0.76 / 109.71 ms │     no change │
│ QQuery 12 │     346.67 / 351.74 ±3.44 / 356.38 ms │     352.68 / 362.60 ±6.27 / 371.02 ms │     no change │
│ QQuery 13 │     472.31 / 484.26 ±9.48 / 501.30 ms │     471.94 / 486.20 ±8.61 / 493.40 ms │     no change │
│ QQuery 14 │     348.67 / 356.89 ±4.40 / 361.62 ms │     350.16 / 355.16 ±3.84 / 358.91 ms │     no change │
│ QQuery 15 │     360.53 / 367.60 ±5.69 / 374.94 ms │    351.54 / 371.86 ±12.29 / 388.00 ms │     no change │
│ QQuery 16 │     731.87 / 737.94 ±4.31 / 742.84 ms │     727.75 / 735.55 ±6.34 / 743.49 ms │     no change │
│ QQuery 17 │     714.80 / 730.24 ±8.65 / 737.85 ms │     729.40 / 735.99 ±3.76 / 740.82 ms │     no change │
│ QQuery 18 │ 1453.58 / 1498.85 ±37.37 / 1547.52 ms │ 1466.46 / 1522.15 ±43.87 / 1596.54 ms │     no change │
│ QQuery 19 │       36.61 / 44.86 ±13.46 / 71.69 ms │       37.21 / 52.05 ±19.78 / 89.33 ms │  1.16x slower │
│ QQuery 20 │    723.34 / 746.77 ±25.20 / 786.09 ms │    728.61 / 747.53 ±20.82 / 788.04 ms │     no change │
│ QQuery 21 │     765.57 / 772.61 ±7.02 / 785.05 ms │     771.42 / 778.43 ±5.40 / 784.66 ms │     no change │
│ QQuery 22 │  1141.56 / 1147.92 ±5.66 / 1156.94 ms │  1147.58 / 1152.69 ±2.69 / 1154.94 ms │     no change │
│ QQuery 23 │ 3138.55 / 3163.20 ±23.22 / 3205.42 ms │ 3170.24 / 3185.71 ±10.56 / 3201.24 ms │     no change │
│ QQuery 24 │     101.23 / 105.59 ±2.34 / 108.34 ms │     104.34 / 108.45 ±3.74 / 114.71 ms │     no change │
│ QQuery 25 │     141.74 / 143.18 ±0.81 / 144.01 ms │     141.12 / 144.48 ±3.60 / 150.98 ms │     no change │
│ QQuery 26 │     106.40 / 107.47 ±0.58 / 108.00 ms │     103.79 / 105.59 ±1.92 / 109.24 ms │     no change │
│ QQuery 27 │     856.65 / 861.23 ±3.28 / 866.40 ms │     859.37 / 862.40 ±2.02 / 864.62 ms │     no change │
│ QQuery 28 │ 3269.62 / 3302.54 ±27.66 / 3340.17 ms │ 3284.73 / 3303.13 ±12.33 / 3319.04 ms │     no change │
│ QQuery 29 │        50.59 / 54.67 ±5.03 / 63.99 ms │        50.96 / 56.96 ±7.06 / 68.56 ms │     no change │
│ QQuery 30 │     379.65 / 386.59 ±5.94 / 394.66 ms │     364.73 / 377.79 ±7.86 / 389.34 ms │     no change │
│ QQuery 31 │    387.37 / 403.94 ±13.49 / 424.81 ms │    381.82 / 397.66 ±12.33 / 418.20 ms │     no change │
│ QQuery 32 │ 1230.97 / 1266.40 ±41.09 / 1345.56 ms │ 1075.43 / 1095.06 ±29.76 / 1154.20 ms │ +1.16x faster │
│ QQuery 33 │ 1547.42 / 1638.72 ±49.25 / 1681.33 ms │  1512.54 / 1523.88 ±9.07 / 1537.78 ms │ +1.08x faster │
│ QQuery 34 │ 1585.50 / 1679.02 ±64.18 / 1779.76 ms │ 1532.35 / 1563.64 ±24.44 / 1594.58 ms │ +1.07x faster │
│ QQuery 35 │    402.13 / 434.34 ±30.36 / 488.62 ms │     406.18 / 413.51 ±6.88 / 424.75 ms │     no change │
│ QQuery 36 │     121.89 / 124.72 ±2.46 / 128.16 ms │     113.16 / 121.89 ±4.72 / 127.32 ms │     no change │
│ QQuery 37 │        50.04 / 51.23 ±1.07 / 52.66 ms │        47.92 / 49.53 ±1.09 / 51.25 ms │     no change │
│ QQuery 38 │        77.66 / 78.76 ±1.03 / 80.32 ms │        75.31 / 77.30 ±1.42 / 79.28 ms │     no change │
│ QQuery 39 │     217.26 / 226.06 ±5.38 / 234.05 ms │     204.26 / 216.86 ±8.35 / 228.27 ms │     no change │
│ QQuery 40 │        25.39 / 27.67 ±2.61 / 32.60 ms │        23.16 / 27.63 ±2.64 / 30.78 ms │     no change │
│ QQuery 41 │        21.20 / 22.45 ±1.11 / 24.49 ms │        20.67 / 21.92 ±1.41 / 24.41 ms │     no change │
│ QQuery 42 │        19.92 / 20.32 ±0.39 / 20.98 ms │        19.84 / 20.42 ±0.39 / 20.80 ms │     no change │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary                             ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)                             │ 23450.60ms │
│ Total Time (worktree-io-uring-object-store)   │ 23082.76ms │
│ Average Time (HEAD)                           │   545.36ms │
│ Average Time (worktree-io-uring-object-store) │   536.81ms │
│ Queries Faster                                │          3 │
│ Queries Slower                                │          2 │
│ Queries with No Change                        │         38 │
│ Queries with Failure                          │          0 │
└───────────────────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric Value
Wall time 117.8s
Peak memory 32.2 GiB
Avg memory 24.9 GiB
CPU user 1096.5s
CPU sys 106.4s
Peak spill 0 B

clickbench_partitioned — branch

Metric Value
Wall time 116.4s
Peak memory 37.3 GiB
Avg memory 27.2 GiB
CPU user 1100.1s
CPU sys 87.3s
Peak spill 0 B

File an issue against this benchmark runner

- Remove unnecessary `as u64` cast (meta.size is already u64 in 0.13.2)
- Allow clippy::result_large_err on execute_read_ranges (object_store::Error
  is large by design)
- Fix broken rustdoc links to LocalFileSystem (cfg-gated away when
  io-uring feature is enabled)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Dandandan added a commit to Dandandan/datafusion-benchmarking that referenced this pull request Apr 16, 2026
DataFusion now includes io_uring-based local file I/O
(apache/datafusion#21673), but the default container seccomp profile
blocks io_uring_setup.

Adds an init container that writes a seccomp profile (default allowlist
+ io_uring_setup/enter/register) to the node's kubelet seccomp dir.
The runner container then references it via Localhost seccomp type.

No DaemonSet or cluster-wide changes needed — it's self-contained
in each benchmark Job.
Dandandan added a commit to Dandandan/datafusion-benchmarking that referenced this pull request Apr 16, 2026
DataFusion now includes io_uring-based local file I/O
(apache/datafusion#21673), but the default container seccomp profile
blocks io_uring_setup.

Adds a checked-in seccomp profile (services/seccomp/io-uring-allowed.json)
that's the standard containerd default allowlist plus three syscalls:
io_uring_setup, io_uring_enter, io_uring_register.

Deployment:
- A Pulumi DaemonSet (services/seccomp.ts) copies the profile from a
  ConfigMap to /var/lib/kubelet/seccomp/profiles/ on every node
- The controller and benchmark-main workflow reference it via
  seccompProfile.type: Localhost

No init containers needed — the DaemonSet keeps the profile present
on all nodes continuously.
Dandandan added a commit to Dandandan/datafusion-benchmarking that referenced this pull request Apr 16, 2026
DataFusion now includes io_uring-based local file I/O
(apache/datafusion#21673), but the default container seccomp profile
blocks io_uring_setup.

Adds a checked-in seccomp profile (services/seccomp/io-uring-allowed.json)
that's the standard containerd default allowlist plus three syscalls:
io_uring_setup, io_uring_enter, io_uring_register.

Deployment:
- A Pulumi DaemonSet (services/seccomp.ts) copies the profile from a
  ConfigMap to /var/lib/kubelet/seccomp/profiles/ on every node
- The controller and benchmark-main workflow reference it via
  seccompProfile.type: Localhost

No init containers needed — the DaemonSet keeps the profile present
on all nodes continuously.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

core Core DataFusion crate execution Related to the execution crate

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants