VarBinViewBuilder: support decoding Dict and Sparse arrays directly into the builder - #9236
3 benchmarks regressed
⚠️ Different runtime environments detected
Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.
⚡ 21 improved benchmarks
❌ 3 regressed benchmarks
✅ 1910 untouched benchmarks
⏩ 51 skipped benchmarks1
Warning
Please fix the performance issues or acknowledge them on CodSpeed.
Performance Changes
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Simulation | fsst_decompress_string |
7.2 ms | 8.3 ms | -12.88% |
| ❌ | Simulation | decompress_fsst[(10000, 4, 4)] |
793.9 µs | 899.8 µs | -11.76% |
| ❌ | Simulation | decompress_fsst[(10000, 4, 8)] |
815.4 µs | 920.9 µs | -11.46% |
| ⚡ | Simulation | decompress[u64, (1000, 16)] |
134.7 µs | 73.9 µs | +82.1% |
| ⚡ | Simulation | chunked_varbin_to_varbinview_builder[(10, 100)] |
516.5 µs | 331.5 µs | +55.8% |
| ⚡ | Simulation | chunked_varbin_to_varbinview_builder[(500, 2)] |
167.6 µs | 116.8 µs | +43.53% |
| ⚡ | Simulation | chunked_varbin_into_canonical[(10, 100)] |
696.3 µs | 489.5 µs | +42.25% |
| ⚡ | Simulation | chunked_varbin_opt_to_varbinview_builder[(10, 100)] |
790.9 µs | 568.7 µs | +39.07% |
| ⚡ | Simulation | chunked_varbin_into_canonical[(500, 2)] |
178.4 µs | 128.6 µs | +38.71% |
| ⚡ | Simulation | compact_sliced[(16384, 10)] |
2 ms | 1.4 ms | +37.15% |
| ⚡ | Simulation | chunked_varbin_opt_to_varbinview_builder[(500, 2)] |
166.2 µs | 121.4 µs | +36.83% |
| ⚡ | Simulation | compact_sliced[(4096, 10)] |
517.9 µs | 385.7 µs | +34.29% |
| ⚡ | Simulation | chunked_canonicalize_into[(1000, 50, 8, 4)] |
15.4 ms | 12.8 ms | +20.33% |
| ⚡ | Simulation | chunked_canonicalize_into[(1000, 50, 8, 16)] |
16.5 ms | 13.9 ms | +18.51% |
| ⚡ | Simulation | chunked_into_canonical[(1000, 50, 8, 4)] |
17.7 ms | 15.1 ms | +16.86% |
| ⚡ | Simulation | chunked_into_canonical[(1000, 50, 8, 16)] |
18.8 ms | 16.2 ms | +16.18% |
| ⚡ | Simulation | chunked_into_canonical[(1000, 50, 8, 64)] |
20.4 ms | 17.7 ms | +15.15% |
| ⚡ | Simulation | chunked_canonicalize_into[(1000, 50, 8, 64)] |
17.8 ms | 15.5 ms | +14.73% |
| ⚡ | Simulation | take[small_m/shuffled/primitive/nonnull/chunks=16384/indices=16] |
1.3 ms | 1.1 ms | +13% |
| ⚡ | Simulation | take[small_m/shuffled/primitive/nonnull/chunks=16384/indices=256] |
3.7 ms | 3.4 ms | +11.54% |
| ... | ... | ... | ... | ... | ... |
ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing rk/varbinview-gather-scatter (f442d48) with develop (ad5de22)2
Footnotes
-
51 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. ↩
-
No successful run was found on
develop(39fde8c) during the generation of this report, so ad5de22 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩