The fastest approach for 472G

Revision en1, by i_love_sqrt_decomp, 2025-08-01 07:02:42

Here is my best implementation for 472G: 331895855. It works the same way as 8014415, just with AVX2 intrinsics instead of SSE code. It's about 30% faster than popcnt (which is already fast) and 2-3x faster than the old SSE code. It's currently the fastest on CF.

Why was the top solutions using fast I/O libraries?

History

 
 
 
 
Revisions
 
 
  Rev. Lang. By When Δ Comment
en1 English i_love_sqrt_decomp 2025-08-01 07:02:42 476 Initial revision (published)