With processor speeds these days, is it necessary to use fixed point math instead of floating point math for 3D calcs (transformations, etc.)?

Anything under a pentium class processor I would suggest. Since pentiums the FPU has been moved onto the CPU and the speed difference was reduced a lot.

But I''m sure some people would disagree?

n!

quote: Original post by TheViper

With processor speeds these days, is it necessary to use fixed point math instead of floating point math for 3D calcs (transformations, etc.)?

in my opinion it is not worth it to impliment fixed point math. A lot of people when using fixed point only used it for software rendering, not transformations and other 3d math. But the only real speed problem is float to int conversion, which can be rewritten with two lines of inline asm. In general its easier and cleaner to use floating point, and I have yet to see proof that fixed point is faster (considering you rewrite ftoint).