From: Jerry Coffin 

At 10:56 AM 1/22/2004, you wrote:
>I have to round up to a stepped list of values in an embedded system.  It
>occurs a few times, and I looking for something that will be small/fast.
>
>The steps are multiples of 3, and the subject number is a float and the
>result is the nearest step that is greater than the subject_number.
>
>What I'm doing now is next_step = ((int)(subject_number / 3) * step_size) +
>step size

I'm not really happy with my solution, but it may be at least a minor improvement:

int step(float in) {
         static int adds[] = {0, 2, 1};

         int i=(int)in;
         return i+adds[i%3];
}

At least if I understand your description correctly, this gives correct
results, and it shouldn't normally do anything in floating point (though
converting a float to an int may produce a distressing amount of code all
by itself).  Of course, this will fail utterly if your input is outside the
range that can be represented in an int.
         Later,
         Jerry.

--- BBBS/LiI v4.01 Flag-5