Sure, over 3.2Ghz is optimistic...
but there is a difference between the "average" expectation and the possibilities. I do agree that I expected prescott well over 5Ghz, and opteron near 3Ghz on 90nm. This is just historical scaling, and I assume that whatever malaise is currently ailing the industry will be magically fixed. I do suppose the physical walls they are hitting is really slowing them down, but the current progress INTC has made on scaling frequency for netburst is nearly zero for the past two years. Athlon 64 has been well below historical levels.
My "middle of the road" expectations for Conroe are 3Ghz at introduction with perhaps a bump to 3.33Ghz a quarter to six months after introduction. Note that these two numbers are only off by 10%, which in the large scheme of things is a really small percentage. Prescott ended up off by almost 100% from my expectations!
With regards to AMD, they would be crazy to be optimizing for performance rather than cost right now. Later in 2006 this equation may change and they may need to tune for more performance at a higher cost / lower yield. I suspect there is some upside in performance to be had between this and the transition to 65nm. How quickly they realize performance and power improvements from 65nm remains to be seen.
--Alan
AI
Market Data 
Markets 
