Closed oleavr closed 3 years ago
I gave this a shot and unfortunately it performs slower than #11 in all cases :sweat_smile:. Going to close this so I can focus on the other implementation as well as prefetching.
Yay, was hoping this one would be the slowest, as it's also the most memory-hungry approach 😄
Similar to #11 but inlining the implementation in every block. (Probably not a good idea, but worth trying.)