cutil_math.h fmaxf() bug

Hello there,

cutil_math.h, as in CUDA 2.1 SDK:

inline float fmaxf(float a, float b)

{

  return a < b ? a : b;

}

This obviously should be “return a > b ? a : b;”

Or am I wrong?

atimb

So the devs aren’t interested in bugs ? :o

I suppose that is a bug. However, absolutely nobody should be using cutil for anything outside of the SDK samples. The same goes for the SDK makefile. They’re messy, inflexible, and subject to change from version to version.