float fmaxf(float x, float y);
fmaxf determines the maximum of x and y.
fmaxf (NaN, y) is y. fmaxf(x, NaN) is x.