Avoid single-chunk grouped accumulator states - #9597
2 benchmarks regressed
⚠️ 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.
⚡ 15 improved benchmarks
❌ 2 regressed benchmarks
✅ 1964 untouched benchmarks
⏩ 54 skipped benchmarks1
Warning
Please fix the performance issues or acknowledge them on CodSpeed.
Performance Changes
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Simulation | compress_fsst[(500, 64, 8)] |
530.5 µs | 598.4 µs | -11.35% |
| ❌ | Simulation | compact[(1024, 90)] |
1.6 µs | 1.8 µs | -10.87% |
| ⚡ | Simulation | fsl_sum_small |
134.3 µs | 94.5 µs | +42.2% |
| ⚡ | Simulation | fsl_sum_medium |
155.2 µs | 113.6 µs | +36.64% |
| ⚡ | Simulation | fsl_sum_large |
190.3 µs | 145 µs | +31.24% |
| ⚡ | Simulation | listview_sum_small |
173.1 µs | 137.2 µs | +26.11% |
| ⚡ | Simulation | listview_sum_medium |
201.4 µs | 164.2 µs | +22.66% |
| ⚡ | Simulation | listview_sum_large |
249 µs | 213.6 µs | +16.57% |
| ⚡ | Simulation | list_sum_small |
261.4 µs | 227.1 µs | +15.1% |
| ⚡ | Simulation | take[small_m/shuffled/primitive/nonnull/chunks=256/indices=64] |
608.1 µs | 533.5 µs | +14% |
| ⚡ | Simulation | take[small_m/shuffled/primitive/nonnull/chunks=2048/indices=64] |
801.4 µs | 705.7 µs | +13.56% |
| ⚡ | WallTime | words_gather_scalar_avx2[65536] |
9.4 µs | 8.3 µs | +13.37% |
| ⚡ | Simulation | list_sum_medium |
294.4 µs | 260.5 µs | +13.02% |
| ⚡ | Simulation | list_sum_large |
347.5 µs | 312.2 µs | +11.32% |
| ⚡ | Simulation | sum_i32_nullable_all_valid |
93.8 µs | 84.8 µs | +10.64% |
| ⚡ | Simulation | sum_f64_all_valid |
85.9 µs | 78 µs | +10.04% |
| ⚡ | Simulation | take[small_m/shuffled/primitive/nonnull/chunks=256/indices=16] |
256.9 µs | 233.4 µs | +10.04% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing mk/grouped-single-partial-flush (2ff2a6a) with develop (e4b3421)
Footnotes
-
54 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. ↩