"I thought the issue was can intel execute flawlessly in the future, and can what it has in the pipeline make money."
The issue is whether or not Intel can execute, not whether it can do it flawless or not. Yes, a P4 at 0.13 micron, possibly with some of the axed stuff put back in, could be one heck of a processor. Sorry about the qualifications, I am so aghast at the reality of the P4 that I tend to waffle around when the future is mentioned. Sure, if the P4 can scale to 2.5GHz or 3GHz in 130nm, then it could be a good chip. But we are talking about something that can only be speculated on. It is just as reasonable to speculate on a pony chip running at about the same frequencies at 130nm, it very well may happen. Sorry, I was less than impressed with the P4 running briefly at 2GHz on a 180nm process, a Tbird might have been able to do the same.
I freely admit that I am very impressed with the bandwidth that the i850 can deliver.
I look at x86-64 and I see an 80386 equivalent, I look at the Itanium and I see the iAPX432. When Intel says that the first generation was a proof of concept, just wait for the second version which will be the "real" Itanium, I hear an echo of a past era. The iAPX432 only had two generations...
And yes, I am very dubious about VLIW-type architectures. I think they are fundamentally flawed and only have a future in niche applications.
Thank you for the history of the "botched transistor" process, I was unaware of it's origins. It makes me wonder though, with Intel's deserved reputation for process control, why did it get released to mass manufacturing? Was Intel worried that far back about the Athlon?
I've been following "EE Times" about the new Intel process, it looks really good. It will be interesting to see if Intel can move it to mass production on the timetable it says, and whether or not it's public dissing of SOI proves to be another mistake.
Intel has great engineers, they have proven it time and again. The question is whether Barrett should be shot or hanged, though. Ropes are re-useable, but not as sure... |
