> I believe at AMD's CC they confirmed that Mustang would have no architectural changes, just larger cache.
If they said that, then they are mistaken.
> The Microprocessor Watch, a free publication from MPR, expects P4 to hit 2GHz on it's current process, P858, before
> moving to .13u and god only knows how fast it will run from there. I'm as anxious as you to see the real scores......
I am skeptical of MDR (the guys who make Microprocessor Watch). They will post any company's prelim report as their own expectations. MDR was the group that posted that Willamette and Merced would be released in 1998. While they are not a fundamentally bad group, you should be careful in what you believe.
Thank you for clarifying that P858 is the 180nm process. Do they have a subdesignation for this current version with the notched poly gate things? I mean, there must have been some noting for the possibly process-related tweak that brought the Coppermine down to 90sqmm, no?
Anyway, I do not believe that AMD stated that Mustang has no architectural changes. c't did report at one point that the changes in Mustang were cache-only, but ... well, I really can't comment on the situation save for that the information c't has was tenuous at best. But don't hesitate in believing that c't kicks ass regardless of this. :)
The reason MDR reported that they expect P4 to hit 2GHz in its current process is because Intel's roadmaps (the one they've been distributing) specifically state that P4 will be at 2GHz in Q2 of 2001, and it is fairly commonly accepted that the 130nm ramp will start to pick up steam in terms of product exiting the production pipeline in Q3, so the logical conclusion is indeed that 2GHz will be at 180nm (though it is possible that it could be a hybrid or a further tweaked process, just like Intel's "0.22µn;m" process -- labeled as a quarter micron process -- that debuted with the Katmai/PIII).
-JC |
