"We use floating point quite extensively in some of our stuff" (my emphasis).

Based on that experience, what are your criteria for using/avoiding floating-point?

Do you have specially optimised libraries, or just "out-of-the-box" C?

Just curious.