wumeibanfa commented on PR #59426:
URL: https://github.com/apache/doris/pull/59426#issuecomment-3694536117
# BM_bytes32_mask_to_bits32_mask Performance Evaluation
## Benchmark Results
### BM_bytes32_mask_to_bits32_mask
| Size (bytes) | Origin (ns) | SVE (ns) | Improvement |
|--------------|-------------|----------|-------------|
| 32 | 3.92 | 1.09 | +72.2% |
| 64 | 7.86 | 2.18 | +72.3% |
| 256 | 31.4 | 8.67 | +72.4% |
| 1024 | 125 | 34.7 | +72.2% |
> Improvement = (Origin − SVE) / Origin
---
## Throughput Comparison
- **Origin implementation**
~7.6 GB/s (stable across all sizes)
- **SVE implementation**
~27.4–27.5 GB/s (stable across all sizes)
This represents an **approximately 3.6× throughput increase** when using SVE.
## Summary
SVE optimization delivers a **substantial and consistent performance
improvement** for `bytes32_mask_to_bits32_mask`:
- **~72% latency reduction**
- **~3.6× throughput increase**
---
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
