tree 4b4bac26ae4871cee034b3bd9c2fd4d067caf010
parent e6bed9c959779623413fcd55ebe3c7a01b7c2d95
author Lun Dong <lundong@google.com> 1651739204 -0700
committer Lun Dong <lundong@google.com> 1654545579 -0700

Add the option to optimize RISP4ML using RVV

This is the comprehensive code change for adding the option of using vector instructions to reduce # of instructions for the RISP4ML toolchain.

Summary:
- All 6 blocks (BLC, DG, WBG, Demosaic, Gamma and Downscaler) have been optimized.
- Numerical correctness has been verified. All unit tests passed (after
  necessary rewriting).
- Significant saving: 40% -> 1%

Change-Id: I6abc8f253f5a74044985ab64797e5de5a76364a2