4 ms·
> Also, you should use SIMD. ironically no clang is better at auto vectorizing
by lacedeconstruct 4mo ago
> Also, you should use SIMD.
ironically no clang is better at auto vectorizing
- spider-mario 4mo agoBetter than what? And do you use `-mavx2` or do you let it target baseline x86_64 and miss out on 8-float vectors? How do you make sure its autovectorisation is successful?