3 Comments

Summary:

I normally don’t write about Intel, or the processor business, but today, I am making an exception. Last three days I have watched and waited for someone to write about Intel and what’s going on inside the Chipzilla, also known as Intel. The company decides to […]

I normally don’t write about Intel, or the processor business, but today, I am making an exception. Last three days I have watched and waited for someone to write about Intel and what’s going on inside the Chipzilla, also known as Intel. The company decides to cancel Tejas desktop and Jayhawk server processors, and the whole world decides to focus on Dothan chip. And then today, Michael Fister decides to quit the company and joins Cadence Design Systems as the CEO. In isolation, these two events should not be related but they are.

Exhibit A: Why? because Fister was the guy who was spearheading the Itanium project, which all agree is living up to its hack-given name, Itanic. Fister was “responsible for the design, development, and marketing of IA-32 processors, including the Pentium Pro, Pentium II and III, Celeron, and Xeon processors,” writes ESM. I think if one looks underneath the hood, Fister’s exit is a clear sign that there is a bigger problem. Was Fister punished for faliure, or was it something he knows which we don’t?

Exhibit B: Now lets focus on the cancellations of the two chips. Instead of Tejas and Jawhawk (future Xeon) intel says it will focus on releasing a dual-core desktop processor in 2005 based on the Pentium M microarchitecture used in the Banias and upcoming Dothan laptop chips, followed in 2006 by a server-oriented derivative of the same design. What that means there will be no new desktop processor from intel till late 2004/2005. That is when Tejas makes its debut. In that sense, this could be a golden opportunity for AMD to take market share, especially now that it has a leading edge product in Opteron.

bq. The company is already trying to position the decision as a Ïroadmap adjustment,Ó and other analysts have said the change is a Ïnet positiveÓ for IntelÌs future. Investors can be glad that Intel is no longer throwing good money after bad, but thatÌs a small silver lining in a big black cloud.

Kumar slams a fist in the solar plexus when he writes: Intel is losing its ability to develop and deliver market-leading products? In just a few months, Intel has terminated development of its most advanced x86 architecture and marginalized the Itanium family, which was once the anointed successor to x86.

DonÌt expect to see a dual-core desktop processor from Intel until mid-2005, and this first effort will probably fall short of IntelÌs traditional performance roadmap for the same timeframe, Kumar writes. The Pentium 4, in fact, has improved only marginally since the 3.06GHz version arrived at the end of 2002. It is ironic that Intel which has always made fun of Apple’s approach to higher performance – that multiprocessing – and now is resorting to the same approach.

Excellent guide to all the Chipzilla chips.

  1. Maybe Intel’s focus is shifting more to WiFi and WiMAX chips as the CPU market slowly shifts toward commodity? Just a thought. Nobody gets excited about GHz except gamers anymore, and the consoles are starting to market to them with network gaming…

    Share
  2. that might be so, but Intel’s revenue stream from WiFi is miniscule and they are not really kicking butt there. WiMAX, as I see it is nothing more than “slideware” right now. where is the silicon. I think this company is not focusing on the very basics – the processors which throw up billions for it to experiment with other crap. anyway that’s my take.

    Share
  3. This is true — this “roadmap adjustment” is huge. Intel processors usually take 18 months for “moderate” change iterations to materialize. (cache bump, FSB bump, etc.) That is, all the previous generations have validated the architectural viability & are getting “tweaks”. To pull in the schedule of dual core chips by 1 year is H-U-G-E. Even assuming that large-scale changes like this were assumed to take 4 years of development, that’s a 25% pull-in. That’s not trivial.

    And Yes, Fister’s head rolled. Intel execs as a whole don’t leave the company unless there’s a very good reason. Someone of Fister’s stature gets offers all the time, but leaving Intel? That’s only done when they hit a glass ceiling.

    Share

Comments have been disabled for this post