Yesterday I couldn't reach the [email protected]
forum, now I read some threads there.
It seems the difference in architecture between ATI and NVidia, and (mainly) the drivers needed for them, are the reason that ATI cards are able to do the job, while NVidia cards aren't (yet).
Updating the NVidia drivers seem to delay things down (Vista, DX10).
Using CUDA would mean at least 2 different GPU clients and the additional work for Stanford to program it, not an option for them (they will invest programming time into optimizing the [email protected]
client(s), which is still their priority).
On the up hand, NVidia is working closely with Standford to find the bug(s), but as with all companies, they have to prioritize their work.
Sorry, I couldn't find the answer here but this is in short what I found about this subject.
Also I didn't know that, while GPU calculations are fast, they are limited in the type of calculations they can do. CPU calculations may be (a bit) slower, but they can run more complex calculations then GPU's can.
So, my CPU is folding without a minority complex.