Skip to content

[WIP] Use Kahan (Neumaier) summation for float sums#8373

Closed
dimitarvdimitrov wants to merge 3 commits into
developfrom
mitko/kahan-float-aggregations
Closed

[WIP] Use Kahan (Neumaier) summation for float sums#8373
dimitarvdimitrov wants to merge 3 commits into
developfrom
mitko/kahan-float-aggregations

Carry the Kahan compensation term in float sum partials

86623b1
Select commit
Loading
Failed to load commit list.
CodSpeed HQ / CodSpeed Performance Analysis failed Jun 11, 2026 in 0s

Performance Regression: -20.43%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚡ 2 improved benchmarks
❌ 7 regressed benchmarks
✅ 1527 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

Mode Benchmark BASE HEAD Efficiency
Simulation sum_f64_all_valid 294.2 µs 509.8 µs -42.28%
Simulation sum_f64_nulls_clustered 629.1 µs 1,033.5 µs -39.13%
Simulation sum_f64 711.9 µs 1,105.9 µs -35.63%
Simulation sum_f64_clustered_nulls 553.3 µs 768.5 µs -28%
Simulation decompress_rd[f64, (100000, 0.0)] 845.5 µs 1,024.6 µs -17.48%
Simulation decompress_rd[f32, (100000, 0.0)] 499.1 µs 587.2 µs -15%
Simulation bitwise_not_vortex_buffer_mut[128] 244.4 ns 273.6 ns -10.66%
Simulation chunked_varbinview_canonical_into[(100, 100)] 309 µs 272.8 µs +13.24%
Simulation chunked_varbinview_into_canonical[(100, 100)] 361 µs 326.3 µs +10.63%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.


Comparing mitko/kahan-float-aggregations (86623b1) with mitko/sum/f64-bench (4c8059c)

Open in CodSpeed