feat(cuda): add a FastLanes delta decode kernel - #9535
Performance Regression: -12.05%
⚠️ 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
❌ 3 regressed benchmarks
✅ 2019 untouched benchmarks
⏩ 12 skipped benchmarks1
🗄️ 1 archived benchmark run2
Warning
Please fix the performance issues or acknowledge them on CodSpeed.
Performance Changes
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | WallTime | words_gather_dispatch_avx512[1024] |
9 ns | 13 ns | -30.77% |
| ❌ | WallTime | words_gather_scalar_avx2[65536] |
8.2 µs | 9.4 µs | -12.03% |
| ❌ | Simulation | cold_misaligned[(16, 64)] |
382.7 µs | 429.1 µs | -10.82% |
| ⚡ | WallTime | cuda/bitpacked_u16/unpack/5bw[100M] |
613.2 µs | 556.8 µs | +10.14% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing ji/cuda-delta-kernel (0b47892) with develop (62648ef)
Footnotes
-
12 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. ↩
-
1 benchmark was run, but is now archived. If it was deleted in another branch, consider rebasing to remove it from the report. Instead if it was added back, click here to restore it. ↩