you don't need the first instruction if you're gonna mask it anyway

How do you do?
It is a dynamic way to check if AVX512 or AVX2 is present and then choose the fastest available implementation))

why 💀

I mean it's still O(n) but it's pretty slow compared to other solutions, about 10x slower (for each character).

Damn it! I knew that some math guy would solve it in O(1), good job...

a branchless solution is the biggest flex.

I hate that this is the optimal solution

Lol what? He used sort which is a built in function, also his solution is O(nlogn) when the optimal solution is O(n)

im guessing people just blindly click the upvotes just because its at the top

I thought of almost the same solution

smart

I mean you are not wrong...

I think he noticed that but he just sacrificed performance for compactness