Pentium 4 Foster may sink the Itanic
Willamette team seems to win the day
When Intel set up the IA-64 project down there in Satan Clara, it sparked off quite an internal fight at the chip firm, as we reported at the time, passim.
Intel poured a heap of resources into the Merced (soon to be the Itanic) platform, and that caused quite a few hackles to be raised amongst the Willamette-Foster team, who wondered just why their own company was setting about the IA-64 project in just such a fashion.
Old-time readers will recall the multiple hirings Intel made of what we then termed "babes in the wood" to work on the Merced project, and how the whole IA-64 project seemed dogged by delays, delays and yet more delays.
That came to a head at the last IDF when Intel turned to the pilot project, which, as we revealed here some days back, will now apply to the McKinley project too.
IA-64 all seems to have gone a little low-profile at Intel, but whether this is because of pressure from its customers or for internal reasons, it's hard at a distance to say.
The fact is that Compaq has never been that keen on implementing Intel's plans for the Itanium and the McKinley chips, its server division plainly telling us some time ago that it wanted to take the eight-way SMP route first with Foster and then with McKinley, when the time was right.
HP, on the other hand, being Intel's partner in the Merced project, has always been very gung-ho about everything, but an interview with Compaq's now retired Jesse Lipcon, which you can find here, in PDF form, has some interesting other views on the whole subject.
Speaking on the whole IA-64 project, Lipcon, who now has little to lose and few axes to grind, said the following: "We think that Intel and HP have chosen the more difficult path, and ultimately the wrong path."
While Intel has always said that IA-64 processors will not come into their own until 2003 and 2004, its emphasis on the Pentium 4 next year now seems to cut down the Itanic and the McKinley to second fiddles in the chip orchestra.
Intel's roadmap shows it is pretty confident that it can clock Fosters and Pentium 4s to 2GHz and above, and that as early as Q3 next year -- Itanics however, still do not clock as high as Intel wanted them to earlier this year.
Will they ever reach the giggle hurtz speeds that Foster and the Pentium 4 seem capable of now?
The powers at Intel insist that you cannot compare their IA-64 family with their IA-32 family and we suppose there is some merit to this argument, but it's all looking a trifle embarrassing, at least from the outside.
If Intel is still as committed to the IA-64 platform, on which they've spent oodles of money in the past, we would expect them to announce some major breakthrough and real soon now. Instead, as the roadmap we saw last week showed, it seems to be pilots, pilots all the way, at least for the whole of 2001. The OEMs are keeping pretty quiet about it too.
Tracking the progress of the Merced-Itanium will cause many an attack of the collywobbles for commentators. Linley Gwennap, for example, then publisher of the Microprocessor Report newsletter, said in 1998: "If you look at this [Merced vs. competition] as a chess match, then Intel has already won the game." ®
Intel's Itanic, McKinley chips go misty-eyed
Intel's cunning server plans
Intel-Compaq deal maketh eight way servers
AMD may trash Intel's Merced on price performance
AMD to take on Intel in SMP market