Skip to content

Use for_each_mask_word in run-end bool run count and decimal split - #11498

Merged
robert3005 merged 1 commit into
developfrom
rk/mask-word-runend-decimal
Nov 16, 2026
Merged

robert3005 merged 1 commit into
developfrom
rk/mask-word-runend-decimal

Conversation

@robert3005

@robert3005 robert3005 commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor

Follow-up to #10362, which added vortex_compute::lane_kernels::for_each_mask_word.

Changes

  • vortex-array: for_each_transition_word in runend/compress.rs now uses for_each_mask_word and low_bits_mask. The separate remainder branch is gone.
  • vortex-array: add a count_bool_runs benchmark to run_end_compress. It counts the runs of a bitmap that starts mid-byte, as a sliced boolean array does.
  • vortex-decimal-byte-parts: split_wide now uses for_each_mask_word in place of the hand-written 64-value chunk walk. Add the vortex-compute dependency.

Replace the hand-written 64-bit chunk walks in the run-end boolean
transition count and in the decimal byte parts wide split with
`vortex_compute::lane_kernels::for_each_mask_word`. Add a benchmark that
counts the runs of a bitmap that starts mid-byte.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Robert Kruszewski <github@robertk.io>
@robert3005 robert3005 added the changelog/chore A trivial change label Nov 16, 2026
@robert3005
robert3005 enabled auto-merge (squash) November 16, 2026 13:17
@robert3005
robert3005 merged commit 15eb6c1 into develop Nov 16, 2026
108 of 109 checks passed
@robert3005
robert3005 deleted the rk/mask-word-runend-decimal branch November 16, 2026 13:49
@codspeed

codspeed Bot commented Nov 16, 2026

Copy link
Copy Markdown

Merging this PR will improve performance by 19.67%

2 warnings

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚠️ 1 benchmark is dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

⚡ 16 improved benchmarks
✅ 1927 untouched benchmarks
🆕 5 new benchmarks
⏩ 53 skipped benchmarks1

Performance Changes

Mode Benchmark BASE HEAD Efficiency
⚡ Walltime compare_u8_neon 3 µs 1.6 µs +87.45%
⚡ Walltime dbp_split_kernel_mixed_null_neon[(I128, 32768)] 29.8 µs 23.8 µs +24.84%
⚡ Walltime compare_int_eq_neon 4.8 µs 3.9 µs +24.39%
⚡ Walltime compare_int_constant_neon 3.9 µs 3.2 µs +22.18%
⚡ Walltime dbp_split_kernel_mixed_null_neon[(I128, 8192)] 8 µs 6.6 µs +21.33%
⚡ Walltime add_constant_shapes_neon[(32768, ConstantPerRow)] 16.7 µs 14.3 µs +16.5%
⚡ Walltime add_constant_shapes_neon[(32768, PerRowNullableConstant)] 17.7 µs 15.2 µs +16.44%
⚡ Walltime compare_f32_neon 4.9 µs 4.2 µs +16.4%
⚡ Walltime add_constant_shapes_neon[(32768, PerRowConstant)] 16.8 µs 14.4 µs +16.1%
⚡ Walltime scalar_subtract_neon 14.8 µs 12.8 µs +15.53%
⚡ Walltime dbp_split_kernel_mixed_null_avx512[(I128, 8192)] 8.6 µs 7.5 µs +13.79%
⚡ Walltime dbp_split_kernel_mixed_null_neon[(I256, 16384)] 23.2 µs 20.6 µs +12.8%
⚡ Walltime compare_int_neon 4.3 µs 3.9 µs +11.44%
⚡ Walltime add_u32_nonnull_neon 14.5 µs 13.1 µs +10.83%
⚡ Walltime compare_int_nullable_neon 4.8 µs 4.3 µs +10.79%
⚡ CPU Simulation ArrowExportVTable_to_arrow_field 20.2 µs 18.4 µs +10.16%
🆕 CPU Simulation count_bool_runs[(10000, 1024)] N/A 3.9 µs N/A
🆕 CPU Simulation count_bool_runs[(10000, 16)] N/A 3.9 µs N/A
🆕 CPU Simulation count_bool_runs[(10000, 256)] N/A 3.9 µs N/A
🆕 CPU Simulation count_bool_runs[(10000, 4)] N/A 3.9 µs N/A
... ... ... ... ... ...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.


Comparing rk/mask-word-runend-decimal (da52d35) with develop (8c02880)2

Open in CodSpeed

Footnotes

  1. 53 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

  2. No successful run was found on develop (c68edb3) during the generation of this report, so 8c02880 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

changelog/chore A trivial change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants