Intel says it should be able to hit 5 GHz “easily” with the CPU, which is about what you’d expect from a .09 micron CPU.
Based on the projected manufacturing numbers, we’ll probably see introduction next April or May, with serious production by the fall.
Since we don’t want fast CPUs, but cheap fast CPUs, we need to estimate when we can estimate that on the Intel side.
Based on prior experience, expect the initial Prescotts to be able to overclock to a bit more than 4GHz, say 4.2-4.3 with high-end air. As time and steppings go by, that speed will drift up to 5GHz or a little better (again, using high-end air).
It probably won’t be a lot more than 5GHz since Intel said it was going to introduce a successor to Prescott, Tejas, in 2004 to get to 6GHz.
A year from today, we’ll probably be in the same position we’re in now. Now, we’re waiting for a cheap easy 3GHz; next year, we’ll be waiting for a cheap easy 5GHz.
Will we see the early introduction of the Prescott version of the 1.6A? A good argument can be made either way on this.
On the one hand, Intel had an interest in phasing out big orders for big fat Willies as soon as possible, which won’t be such a big factor this time around.
On the other hand, OEMs very much liked the low-speed Northwoods, and no doubt would like a cool running Prescott, too.
So I think we’ll see them, the question is whether it will be sooner or later.
The answer to that lies in how quickly the new 300 mm fabs get up to speed. These fabs are meant to crank out a lot of chips very cheaply compared to earlier processes, and their production looks targeted towards .09 micron CPUs.
If the fabs get up to speed quickly, we’ll probably see them sooner. If not, later.
Again, going by prior history, the Prescott 1.6A will probably come in at about 2.67GHz.
Don’t Let This Stop You Now
We’ve advised for some time now that the ideal upgrade points are a PIV C1 stepping combined with a dual DDR board on the Intel side, and a Barton CPU upgrade on the AMD side. This news doesn’t change that at all.
On the Intel side, the next step after dual DDR will be dual DDR-II, and that will probably not be around for at least a year. Any new Intel motherboards between Granite Bay and that will only be evolutionary changes.
That year-from-now board will include serial ATA (maybe serial ATA 2.0), and likely include DDR-II and PCI Express (though early 2004 may be more like it). If you can’t upgrade a lot (indeed, especially if you plan on skipping this upgrade point), you might as well wait for a board with these features so as not to get obsoleted quickly.
What About AMD?
Given AMD’s frequent, if not spasmodic changes of plan recently, it seems foolish to try to predict very much.
The existence of Barton largely reduces the likelihood of a successful launch for a .13 micron Clawhammer.
Intel’s announced moves will put enormous pressure on a financially strapped AMD to convert to .09 micron, and given their troubles with .13 micron/SOI. . . .