<* Front end may already be .18. By contrast, Intel's shift to .18 was a total process revision of the front end that resulted in a massive improvement of transistor characteristics vs. .25. If they do in fact implement an effective, total process revision of the front end over what they are doing now, it may well be a different story. Going by some data, some gut on this one, that the front end they introduce on their ".18" is going to be an incremental improvement at best, as opposed to a massive process change.>
OK, I am familiar with this argument.
<* Going to .18 means new interconnects. Even though the paths are shorter, unless they really luck out, they will still be chasing down speed paths / RC delayed circuits. This is situation is not atypical. >
This is very interesting. I haven't heard this one before. Are you saying that gains in path delays and RC issues can come to naught! Have you seen this happen?
<* The main reason I said one speed grade is that those are the official signals coming out of AMD. Could be 2, I suppose.>
OK, you are taking AMD's official statements at face value then. I think they are sandbagging on MHz but what do I know. You may be right. Also, what is the "could be 2" deal? I don't understand.
<Example: MOT and G4. That was a really big pop, wasn't it?>
Don't even get there. Mot's microprocessors group is so poorly run - they can't deliver on a lot of stuff. You just need to look at the past decade to see how pathetic their execution was. Here is a company that has no focus and will probably never shine in any area consistantly because of that.
Chuck
P.S.: During the days of 88xxx and early PowerPC, I used to have a few friends there but now I don't have a single contact on the Mot microprocessor division. I don't know if the current PPC group has any resemblence to the team that was there 5 years back - the execution shows. |
