feat: support nullable Arrow Device array export #8104
+224
−68
CodSpeed HQ / CodSpeed Performance Analysis
failed
May 26, 2026 in 0s
Performance Regression: -4.36%
⚠️ 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.
⚠️ Different runtime environments detected
Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.
⚡ 1 improved benchmark
❌ 2 regressed benchmarks
✅ 1248 untouched benchmarks
Warning
Please fix the performance issues or acknowledge them on CodSpeed.
Performance Changes
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ⚡ | Simulation | chunked_varbinview_opt_into_canonical[(1000, 10)] |
239.6 µs | 203 µs | +17.99% |
| ❌ | Simulation | chunked_varbinview_opt_canonical_into[(1000, 10)] |
187.5 µs | 224.2 µs | -16.38% |
| ❌ | Simulation | baseline_lt[16, 65536] |
217.6 µs | 245.4 µs | -11.34% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing ad/cudf3-null-masks (2d89c69) with develop (6ddc4d5)
Loading