Quote Originally Posted by Epsilon84 View Post
Let me just say this: 15 - 20% IPC gains from a die shrink would be absolutely incredible and unheard of, but many here seem to think its a realistic figure? 15 - 20% improvements are more akin to an architectural overhaul, for example K8 -> K10 gained about that much.

Just as a point of comparison, Penryn gained on average 5 - 6% per clock through a larger cache and several minor architectural improvements. AMD would have to work miracles to get 15 - 20% gains from Deneb, frankly I very much doubt it but I would be more than happy to eat humble pie if proven wrong.
You've a point, but it's not an unknown thing Deneb does consist out of more than just a die shrink and will have some improvements left and right.

But your example of Intel's move does show improvement/clock is possible, but it also shows something else. Penryn is/was a very good architecture. Nobody can seriously deny that. Now, they say 'Dont fix what's not broken', but also you cant fix what's not broken. Now Agena, with the start of B2, the quick fix of B3 (not argumenting whether TLB is gone or not), K10 is simply not a perfect architecture, there are some flaws. Now if those flaws are more obvious than Intel's, if AMD managed to find a lot of flaws, high bottlenecking flaws etc, they can all fix that with Deneb.

Therefor your argument 'Intel only gained 5~6%, so 20% for AMD is impossible' isnt completely valid either. Now I know AMD made some weird claims before the launch of Barcelone, then again, I'd never take such claims, either from AMD or Intel, serious. I do take it with me in my arguments, but not as being completely convinced, it can only help you in your arguments.