Originally Posted by Necrofire
What makes this process different than the ones used for making CPUs?
If there isn't any difference, then why aren't they making CPUs on the current 40nm process?
A number of likely reasons. They may have gone to TMSC for the 40nm GPU fab for production capacity and cost. Apparently, NV did the same thing with not so good results.
With 28 nm processing, they could theoretically put almost 1.5x+ the transistors in the same die size, and run it even faster.