Are you doing i % n
, where n
is a power of two? There’s a neat alternative way to do that: i & (n-1)
. See how it works by choosing n == 8
, so i % 8
is the same as i & 7
. 7
is 0b111
, so i & 0b111
removes all but the three least significant bits. For instance, 14 % 8 == 6
. But 14 & 7 == 0b1110 & 0b0111 == 0b0110 == 0b110 = 6
.
For example, you might use this when implementing a ring buffer with power-of-two length.
I wrote this because I felt like it. This post is my own, and not associated with my employer.
Jim. Public speaking. Friends. Vidrio.