Quote Originally Posted by jdarling
Well using my software based solution (ASM/SSE2/MMX) I can easily pull 60 fps (thats where I cap due to VSync) and have plenty of processor space left over. So on 2nd thought maby I'll keep going just because so many people think it can't be done .
It can be done, if optimized correctly, but you still won't be able to beat hardware even in simple things like alpha-blending. However, doing it in software can have its own advantages: since you don't stall the pipeline of using the texture you just rendered to, you can use this to create all kinds of motion blur, gaussian blur, fractal, distortion and other effects.

P.S. Can I ask what are you using SSE/SSE2 for?