Intel-AMD is a Cold War; there are no decisive battles. — Ed
“If it’s the ultimate game, then why are they playing it again next year?”
— Dallas running back Duane Thomas referring to the Super Bowl (Super Bowl VI)
Gee, Intel nudges ahead by a little bit by releasing the PIV, and some people are treating this like this was the second Dresden firestorm.
Having a $500 chip do a little bit better than a $250 chip using an older process technology is only going to appeal to those with more money than sense.
If AMD had nothing in the works, yes, this would be a matter of concern for AMD fans, but sheesh, Thoroughbred is coming, and while it’s really just a die-shrunk XP, that’s all AMD needs to stay in the ballgame for most of the year.
If AMD just had the Athlon design, that would be a longer-range matter of concern, too, but Clawhammer will be coming, and even if you take the early estimates on performance improvement with a grain of salt, the initial numbers look impressive.
Why Is Intel So Coy?
If Intel really did have AMD with their pants down, they seem remarkably mellow about pressing their advantage with hardware (as opposed to harangues).
The Intel roadmap has shown and continues to show very little movement until towards the end of the year, then it shows a rush. There will be more ramping in Northwood in the last three months of the year than the first nine.
This looks especially odd given Intel announcing that they’ll have .09 micron processors come the middle of 2003. You’d think Intel would spread out the ramping a bit more.
Why is that?
You could say Intel is just not going to be pushed by AMD very much for most of the year, and that the push upward seems to coincide with Clawhammer release. You could say that, and in the past, I’ve said that.
There’s another possible reason, though, and that’s been one of the major reason why I’ve been reluctant to talk a whole lot about buying Northwood systems.
Those who have been overclocking current PIVs towards 3GHz have been finding that it doesn’t do a whole lot of good with DDR-based systems. The initial Willamette was designed with dual-channel RDRAM in mind. Current PIV DDR systems match faster processors with less memory bandwidth.
This should wave a red flag, especially when Intel decides all on its lonesome that dual-channel DDR would really be a good idea for performance systems, oh, around the end of the year, just at the time when they decide to ramp up big time.
I don’t think this is pure coincidence.
This Is Not A Decision Point
Most people reading this are looking at upgrade prospects rather than buying a new computer for a new use.
The next major change will occur very late this year/the first half of next year. It will be Clawhammer vs. late model Northwood/early Prescott. Both will have (or better have) dual-channel DDR (or possibly some RDRAM equivalent) to support that (actual or equivalent) 3GHz+ speed.
For everyone with a computer today, that choice is going to involve a platform change. Even if you buy a PIV system today, your mobo will still be obsoleted by Granite Bay and its competitors.
Most people reading this probably have AMD systems. Looks like they’ll have a reasonable chance to be able to just pop in a Thoroughbred if they’ve upgraded recently. If their equipment is older and they can’t, no matter what they pick, it’s not going to have a long lifespan on the cutting or even very sharp edge.
What most of this year is going to be are preliminaries. Expect the guy on “top” to change a couple times during the year. If you change platforms every time the “lead” changes, you’ll be giving the local squirrels excess computers for Christmas.
There is and will be no compelling reason to prefer one platform over another in 2002. If you have Athlons, stick with AMD. If you have PIVs, stick with Intel. Sometimes, one will be on top, sometimes the other, at all times, it will at least be fairly close.
And don’t let anybody else persuade you otherwise.