Originally Posted by ShiningArcanine
I refuse to believe that Cypress is a superior product to Fermi. Fermi is a true general purpose GPU with C++ support and a IEEE754 compliant floating point implementation, which Cypress is not.
I think this situation parallels Betamax versus VHS, where Fermi is Betamax and Cypress is VHS, which is sad. I really wanted to get my hands on a Fermi processor to program it. If this is true and I am lucky, Nvidia will have Fermi II out before I graduate next spring.
If you're a developer or somebody that needs cuda support to run their applications then you really have no other choice but to wait for Fermi,and i'll happily concede that the GP-GPU infrestructure is a lot better developed for Nvidia products than it is for ATI,no question about that.
But most are just regular users running more common applications,for which Cuda means absolutely squat for them in real terms,and there's no point in waiting for Fermi in those conditions.