It looks like Prescott will roll in February, one way or the other, and roll out big. Intel expects to make 70 million of them in 2004, which is more than half of the CPUs Intel would be expected to make next year.
However, it’s not good for overclockers (or Intel) when the biggest benefit a new generation of Intel processors is likely to give you is somewhat lower Hammer prices.
It doesn’t look like they’ve been able to do much about the heat problem. XBitLabs reports that socket 478 Prescotts will chew up over 100 watts, while “higher-speed” socket T Prescotts will chew up almost 120 watts.
Please note what that statement implies. It seems to say that socket 478 Prescotts will only come in lower speeds (perhaps only up to 3.4GHz), and higher speeds will only be available with the socket T form factor.
Of more immediate concern is what is likely to face Intel overclockers come next February.
The good news is that it probably will be easy to find a relatively low-speed Prescott at a reasonable price by late February, probably around $175.
The bad news is two-fold, one temporary, one permanent.
The temporary problem may be that you might be able to get a 2.8GHz Prescott pretty easily, but you may not be able to get a recently made 2.8GHz Prescott at all.
Intel has apparently gotten up to a C stepping in its efforts to get a product out the door. That doesn’t necessarily mean the Prescotts that show up in February will all be C-steppings.
Intel’s been mumbling about having lots of low-speed Celeron Prescotts around for quite some time now. Those clearly aren’t going to be C stepping chips. It may be quite possible that early, low-speed Prescotts will also be sold come February.
When the time comes, this little detail will be easy enough to check by matching sspecs to Intel documentation in a few months, and it may well turn out not to be a problem.
We’re just noting for those who might want to overclock a Prescott right away that it could be a problem, and likely a big one, adding to the other new requirements we’ve previously addressed.