Intel Lays It On
Intel has been futzing around with its EE and Xeons. They’re adding two additional layers of metal to their processors.
Hmmm. The last time somebody did something like that was when AMD added a layer which turned a Thoroughbred A into a Thoroughbred B. A few of you out there might attest that that was a good thing to do. 🙂
What’s important to note is that Intel isn’t doing this to PressHots, but to the old reliable Northwoods.
If this change means getting some hundreds more MHz out of the old Northwood design, we’ll no doubt hear about 3.6GHz Northwoods in a month or two.
It seems like Intel is taking a two-track approach to CPUs these days, a form of parallel processing, so to speak. They’re going along with the initial plans and hoping somehow, someway they can fix their bleeding, if not broken, 90nm process technology, while running a parallel contingency plan in case they don’t.
In the short run, that means relatively quick fixes like this one. In the longer run, that’s the reason for all the talk of fitting mobile technology into desktops. That would look to be the contingency plan if they can’t fix things in time for Tejas.
Even if Tejas does get fixed, it probably wouldn’t hurt Intel to have a cool and quiet line of desktop processors around, anyway.
More Parallel Processing
A few months ago, we heard weird stories about Intel coming out with a modular processor. What this technology really appears to have been is what Intel is now calling Twin Castle.
Succintly, Twin Castle is a chipset technology that will let mobo makers be very, very flexible in their designs, and make upgrades very easily. So flexible, in fact, that a future mobo could go both ways: IA64 and IA32 (with or without the E).
This doesn’t mean that Intel will do something like that, it just makes it easier for them to take that route down the road.
Again, it’s a form of parallel processing. There are plenty of other advantages to justify this technology without Intel going hybrid, but this would help Intel to take that road if it so chooses.
It may not make a lot of sense today, but two or three years from now, when we’re down to 65nm technology and multicore are becoming mainstream, such a transitional processor could make a lot of sense.
For those who have already kissed Itanium off, remember that the guys who count haven’t. Also remember that they weren’t planning to bring Itanium to the desktop until 2007 or so anyway.
It’s not dead yet. It may well die a desktop death a few years from now, but it’s not dead yet. It’s just lurking in the realm of high-end serverdom (where it’s finally starting to get some traction), waiting for mainstream technology to catch up to it.