| Home | About | ![]() |
Articles | ![]() |
News | ![]() |
Folding | ![]() |
Forums | ![]() |
Login | ![]() |
Register |
|
|
|
Page 1 of 1 pages for this article This Changes Everything: Intel Cancels Next-Generation P4 by Article Admin
![]()
Published: 05/06/2004
It?s been an open secret in the industry for the past few months that the Pentium-M design was likely to be the future of Intel?s CPU efforts rather than the Pentium 4. Ever since Centrino debuted, its been seen as having considerable potential on the desktop, so it wasn?t surprising to see those discussions resurrected after Prescott?s disappointing debut. Even Intel, however, didn?t seem to be expecting to make the change quite this soon. Rumors about the next generation P4?s dual core design (codenamed Tejas) had already begun to circulate on the web several months ago, with the general expectation being that the move to an enhanced version of Pentium-M technology would occur sometime afterwards. The Tejas design, in fact, is reportedly almost finished and ready for tape out, with full production originally scheduled for the end of the year. That won?t be happening. In what amounts to a major bombshell across the entire market, Intel has cancelled the next-generation P4 design because of its excessive thermal dissipation. While this isn?t entirely surprising considering single-core Prescott radiates over 100W in some cases, it?s still a tremendous shift in focus for Santa Clara that?s going to reshape the entire dynamic of the CPU market. It?s not clear what this means for Prescott or for the introduction of that CPU?s 64-bit capabilities, which was originally slated for the end of this year. Although Intel?s already strongly hinting that its next-generation desktop processors will be based on Dothan rather than Prescott, the switch may not be an easy one. Pentium-M processors currently use various flavors of the Intel 855 mobile chipset, which is specced for only a 400 MHz FSB, AGP 1.5v, and DDR333, along with a mobile version of the older ICH-4. While not bad at all for a mobile chipset, this obviously isn?t a platform that can compete on the desktop. The product churn in the channel as a result of this announcement is going to be tremendous. While it?s doubtful that Intel could actually switch gears and have a desktop version of the Pentium-M ready for widespread deployment by Christmas, this is an announcement that?s going to make Prescott look like a lame duck no matter what date Intel targets for the transition. Motherboard manufacturers who?ve spent months building new boards to conform to Intel?s Socket T (LGA775) aren?t going to be too happy either, though its possible Intel could design its desktop Pentium-M to run on the same boards Prescott used. Intel hasn?t given much guidance on when these new CPU?s will ship, but Q1-Q2 of 2005 is probably an optimistic target. The new Pentium-M derived core will, according to Intel, be a dual-core design, but how that core will be positioned remains unclear. If P4 moves to dual cores, however, expect Xeon to follow; Intel won?t want a dual core P4 stealing sales from single-core Xeon in the workstation or server market. Having dual cores could make differentiation between the Celeron and the Pentium quite easy; simply keep the Celeron as the single-chip derivative. As for AMD, they?re likely to benefit in the short-to-mid-term from the further confusion. It?s much harder to predict the long-term effect. The rumor mill reports that AMD?s .09 micron die shrink will meet more traditional expectations for a die shrink, which is to say that AMD?s .09 micron chip will run cooler, draw less power, and scale well. Against hot-running Prescott, that?s a significant advantage. Against the Pentium-M, things are less clear. Then, of course, there?s the issue of dual core vs. single core designs. AMD?s already announced its plans to move Opteron to a dual core design within 12 months, but what about Athlon 64? Given the advantages that a true multi-core design could introduce, AMD might have to scale that particular feature lower into the product line faster than they originally planned. There are a million other questions we don?t have answers to yet. How fast will it run? (Original estimates of Dothan mobile performance by Q4 04 were targeting a 2.13 GHz clockspeed on a 533 MHz FSB). How quickly can Intel design new chipsets? Will it support SSE3 and AMD64? What about the set of Tejas New Instructions that were under development? Will Intel now keep Prescott in the market longer and scale it higher? Early indications are that there will now be one to two new flavors of Prescott, but we have no concrete data. All good questions that we don?t have the answers to yet, so we?ll have to wait and see what develops. For AMD, this could turn out to be a very good thing in the short term as Intel revamps its product lines and redeploys resources, but the Pentium-M may prove a tougher competitor than the Pentium 4, especially if Intel deploys a multi-core version at the desktop level. We?ll keep you informed. Page 1 of 1 pages for this article Search
|