3 ms·
Because you are working in the cache. Also, you should use SIMD.
by dist-epoch 5mo ago
Because you are working in the cache.
Also, you should use SIMD.
- lacedeconstruct 5mo ago> Also, you should use SIMD. ironically no clang is better at auto vectorizing
- spider-mario 5mo 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?