Return the maximum value of two single-precision floating-point numbers, ignoring NaN. - View it on GitHub