[Feature] Add SVE optimization for copyBits function in BitUtil.h #58
Description
Feature Category
Performance Optimization
Problem / Use Case
The readBits function is one of the hotspots in Spark-Bolt, and its internal implementation is divided into two functions: fillBits and copyBits. Compared to fillBits, which is only used for filling 0s and 1s, we chose the more meaningful copyBits to implement SVE.
Proposed Solution
We have completed the SVE optimization for copyBits within the readBits function. The function is scalar loop implementations in Velox, not SVE. We hope to submit them to Bolt soon.
References / Prior Art
No response
Importance
Medium (Nice to have)
Willingness to Contribute
Yes, I can submit a PR