Just a single instruction avx512 loop outperforms movsb by 10x on my computer.

I’m on ryzen 7600x. Just an example that it doesn’t need to trash instruction cache or have 10 loops behind conditionals