posted 17 years ago
IEEEremainder() and a deep knowledge of the IEEE floating-point standard are important if, and probably only if, you're doing relatively serious numeric computing (Monte Carlo simulation, quantum chemistry, large-scale modeling or statistics). There are complex numerical algorithms in these areas whose results cannot be easily checked except by comparing their results to a "gold standard", some previous well-known good implementation. In this case, you need to ensure that your results are identical to the last bit to the gold standard, or your code is suspect.
Also, there are situations -- for example, adding very small and very large numbers -- where a relatively deep knowledge of floating-point math is necessary to ensure that you get the right results. But for computer graphics and other more common floating-point applications, none of this matters, at least not most of the time.