Avoiding floating point is only needed when performance matters. When you do floating point calculations for, say a custom calculator, it doesn't matter, and you should use floating point (unless you really want to know the deviation). But for heavy 3D applications, fixed point should be used!