Ars Technica logo. Serving the PC enthusiast for over 5x10-2 centuries  

Subscribe to Ars Technica!

Have news? Send it in.

 
Ars Guides.
  Buyer's Guide
  How-To's & Tweaks
  Product Reviews
  Ars Shopping Engine

Technopaedia.
  Technical Blackpapers
  CPU Theory & Praxis
  Ars OpenForum
  Search Ars

Columnar Edifice.
  Wankerdesk
  AskArs!
  Diary of a Geek
  Game.Ars Report   Mac.Ars takes on...
  Linux.Ars

Site Info.
  Subscribe to Ars
  Ars Merchandise
  Who We Ars
  Advertising
  Links



Intel's Processor Number Schema

   by Ken "Caesar" Fisher

 

The purpose of this article is twofold. First, we've received numerous requests to explicate the reasoning behind Intel's new processor number schema. So, we'll give you the lowdown on that, knowing that it may be old hat for our friends working in the channel.

Second, we've been working hard on our clairvoyance, and at the end of this article we'll provide you with a glimpse of Intel's plans for its desktop and mobile lines up through early 2005. It's based on real information and real sources; we're not rumor mongering here.

The good old days

Maybe the old days were good, and maybe they weren't. Industry analysts, especially those that study different CPU architectures, have long called for an end to the megahertz myth, that is, the very basic and pervasive notion that a CPU's internal clockspeed is an indicator of its performance relative to any other CPU.

Whatever the merits of clockspeed are (and we've documented our opinion on numerous occasions), we learned in March what we had long expected: Intel is ready to drop the megahertz from their product names because, well, it megahurtz, if you know what I mean.

The problem was our little pal named Prescott. Originally meant to be the greatest thing since laser-sliced bread, after several delays the ugly facts began peeking their heads around the corner: the move to a 90 nanometer fab process proved that the Pentium 4's daring NetBurst architecture was indeed less extensible than previously imagined.

We hinted at this through the end of 2003, and in the same month that Intel announced the move to model numbers, they let another bit of news slip, too: the Pentium M architecture is the future of Intel's Desktop offerings. NetBurst is dead, and while 2004 will see more speed improvements in the Pentium 4, the plan is to create time and space to bring the Pentium 3, I mean, uh (did I just say that?), the Pentium M, Back to the Future.

You can see the quandary: imagine late-life Pentium 4s at 4.0 GHz, when the Pentium M will easily trail by more than a gigahertz (probably 2GHz). But this Pentium M is the wave not only of the mobile future, but of the desktop future as well. Quite a quandary, indeed. Hence the move to processor numbers. As sources have revealed to us, Intel states it plainly:

Sum of all processor features > GHz alone

Indeed. Intel's position is that the payoff for the move to processor numbers will come in the form of greater end-user awareness of the complexities of performance. Let's look at the philosophy of the new naming scheme, as Intel sees it.

Intel's naming philosophy

Processor numbers, according to Intel, represent the "sum of innovative features and technologies" in a CPU. These features are listed by Intel as:

  • Architecture (Pentium M, Pentium 4, etc.). Not only will this relate to different base architectures, but it will also reflect things such as process technology (90nm fab).
  • Clockspeed (3.4 GHz, etc.).
  • Cache (2MB L2 cache, etc.)
  • Frontside Bus (FSB, e.g., 533MHz, 800MHz, 1066MHz)
  • "Future Intel Technologies"

Only Mulder and Scully can tell you about the last one. Intel has said from on high that the new processor numbers should be used stating in May 4 in all advertising locations where GHz would have been previously used. Partners who do not have this new system in place by July 1st will not be eligible for advertising kickbacks from the company, Intel has purportedly said privately.

 

Next: the generic outlay

 


Dual 2.5GHz Power Mac G5 review

The Sims 2 review

Pipelining: an overview (Part II)

System Guide: September edition

Pipelining: an overview (Part I)

Chris Sawyer's Locomotion review

Multicore, dual-core, and the future of Intel

System Guide: gaming boxes

TrackIR3 Pro review

Doom 3: the review

PowerPC on Apple: An Architectural History, Part I

Virtual machine shootout: Virtual PC vs. VMware

The Pentium: An Architectural History � Part II

Joint Operations: Typhoon Rising game review

AirPort Express review

The Pentium: An Architectural History � Part I

The Ars guide to PCI Express

Beyond Divinity game review

The future of Prescott

Interview with Mozilla.org's Scott Collins

Thief: Deadly Shadows game review

USB 2.0 Hi-Speed Flash drive review

A closer look at Intel's processor numbers and 2004 road map

Far Cry game review

Dell Latitude D800 laptop review

HP Compaq nc6000 laptop review

Hitman: Contracts game review

Deploying a small business Windows 2003 network

Alternative AIM clients for Windows

Inside GNOME 2.6

/etc

OpenForum

Distributed Computing

Take the Poll Technica

FAQ: Celeron overclocking

 

Copyright © 1998-2004 Ars Technica, LLC