Floating = Point Hardware

I notice that some graphic programs (like Quake) rely upon the floating =point hardware. Does anybody know why?As far as I can see, graphic programs use two kinds of data:1: A value describing the color of a pixel. Normally up to 24 bits, which= may be a length suitable for being moved by a floating point processor, but will= not be subject to FP computations.2: Two or three integers indicating a point's coordinates in space. = Normally 16 bits should suffice, but in extreme cases 32 bits may be used.So why use a FP processor? For integer computations on coordinates?Or does anybody know about any real FP data?