| Wednesday, June 15, 2005 |
15:27 - Ours not to reason why
http://www.xbitlabs.com/news/cpu/display/20050609121500.html
|
(top)  |
Steven Den Beste sends this latest bit of analyst speculation (this time from those ca-razy Russians at Xbit Labs) upon the Apple/Intel deal. The gist is that IBM unceremoniously dumped Apple after making it clear that they just didn't want to make any more PowerPC chips. Like, at all.
“IBM no longer sells standard products – Apple’s processor was the last product. IBM’s exit from semiconductor products started several years ago and is now fully completed,” Mr. Petrov he added.
While this is a big deal for Apple to throw out IBM’s chips from its machines, significance of this business for IBM is not really evident from the revenue standpoint. Last quarter, according to some reports, Apple sold 1.07 million computers, which is not really a lot. Still, it could be important for IBM to provide Apple with PowerPC chips in order to leverage the influence of this architecture. For instance, all three next-generation game consoles from Nintendo, Microsoft and Sony will be powered by chips that use PowerPC architecture. Furthermore, IBM offers consumer electronics designers to design PowerPC derivative processors for their needs.
The Petrov Group analyst thinks that the Cell processor is important, but that is now done and over. However, Power architecture and Linux continue to be critical fuel to IBM’s corporate revenue growth – from $100 billion to $200 billion in the next ten years, according to Petrov Group.
So IBM is happy to continue making chips it's already completed designing—but as far as future chips go, it's not just that IBM's roadmap doesn't coincide with Apple's, it's that IBM does't have a roadmap. So all that fooferaw about multicore 970s and low-power 970s and country-fried 970s with a light dusting of powdered sugar were all just so much wishful thinking. This certainly goes along well with the theory that IBM wasn't prepared to make any mobile G5 processors because Apple would be the only buyer, which wouldn't justify the business case for the development of said chip; but it certainly seems to speak ill of IBM's ability to communicate its intentions to its partners. I mean, if IBM had no intention of developing any 970-based chips beyond the Cell and its game-console siblings, and instead wanted to focus solely on Power chips for Linux mainframes, shouldn't they maybe have told Apple about this at some point?
Hell, maybe they did, even as far back as the G5's introduction. You'd think, in that case, Steve would have refrained from promising that "Believe me, this architecture has legs" and that we'd have reached 3 GHz by summer 2004. (But then, people have been rumoring that phantom G5s "have legs" since before the 1GHz barrier had been breached.)
Maybe the G5, with its backward compatibility and its initial demo-friendly muscularity advantage, was always meant as a stop-gap... but that sure assumes a lot of scheming genius on Steve's part. Not that that makes it untrue, of course.
Though I'm not sure why Mr. Petrov is talking about IBM's "exit from semiconductor products" and then going on to confirm that they're dedicated to the Power architecture going forward. What, is Power a quantum-computing chip or something?
UPDATE: Boy, switchers still keep cropping up—in the darnedest places, too. This can't be a good sign for the Linux faithful. (Via J Greely.)
|
|
Brian Tiemann
