The first statement partitions i into 2-bit groups and replaces each 2
bit group with the number of 1-bits set in that group. The second
statement partitions the number into 4-bit groups and adds together
the two 2-bit groups in each to get the number of 1-bits originally
set in that 4-bit group. The last statement partitions the number into
8-bit groups and adds the two 4-bit groups in each to get the number
of 1-bits in the 8-bit group. Then the multiplication adds the 8-bit
groups, leaving the sum in the upper 8 bits of the number. The shift
right-justifies the sum in the word.

Dave

On Jan 18, 10:54 am, Manisha <[email protected]> wrote:
> This is the very efficient code for counting set bits. I am not able
> to understand whats the algo/logic behind it. I tried running this
> code, it gives correct output but don't know why and how?
>
> Could anybody explain the logic with generic example?
> int NumberOfSetBits(int i)
> {
>     i = i - ((i >> 1) & 0x55555555);
>     i = (i & 0x33333333) + ((i >> 2) & 0x33333333);
>     return ((i + (i >> 4) & 0xF0F0F0F) * 0x1010101) >> 24;
>
> }
>
> I am sorry for posting this C code here and many of you might not be
> in touch with C. I would be happy to explain if any of the C related
> stuff is troubling you in this question.
-- 
You received this message because you are subscribed to the Google Groups 
"Algorithm Geeks" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/algogeeks?hl=en.


Reply via email to