In binary representation, hamming weight is the number of 1's. I came across web and found an O(1) answer to it:

v = v - ((v>>1) & 0x55555555);
v = (v & 0x33333333) + ((v>>2) & 0x33333333);
int count = ((v + (v>>4) & 0xF0F0F0F) * 0x1010101) >> 24;

However I don't quite understand the algorithm and cannot find a description of it anywhere. Can someone please explain it a little bit especially the last line (what the heck does *0x1010101 and then >> 24 mean)?

algorithm bit-manipulation
Edit
Report