Feeds

Intel: next-gen dual-cores not NetBurst

Going back to Pentium III, sort of

  • alert
  • submit to reddit

Intel has confirmed its second-generation dual-core desktop, notebook and server processors will not be based on the Pentium 4's NetBurst architecture, but something more akin to the Pentium III.

We knew this already, of course, but it's nice to see Intel going public at long last.

The revelation comes from the chip giant's former CTO and now Digital Enterprise Group head, Pat Gelsinger, in an interview with German pro computing site Golem.de.

Asked if next year will see Intel waving goodbye to NetBurst, Gelsinger replies: "Yes, with our products, which are developed under the code names Conroe, Woodcrest and Merom."

'Merom' is the next generation of Pentium M, which has never been NetBurst-based, deriving instead from the Pentium III, which because of its shorter instruction pipeline, was more suitable for power-sensitive systems such as notebooks. Witness the very different power consumption specifications of 'Banias', the 90nm Pentium M, and 'Prescott', the 90nm P4.

With dual- and eventually multi-core processors on the way, there's much less need to focus on enabling ever greater clock frequencies, so the P4's long pipeline is less relevant. It's also less efficient, from an electrical power perspective and "arithmetic performance", says Gelsinger.

"We decided for a number of reasons to move to an architecture with a less deep pipeline. In this regard, it rather resembles the Pentium III.

"However, one of the advantages of NetBurst was that we could add very simply things like virtualisation and HyperThreading," he added. "We're not throwing those things away."

But it's clear from the tone of Gelsinger's comments that the Pentium M architecture is going through something of a complex redesign to incorporate these features, which is possibly why Intel is reserving it for second- rather than first-generation dual-core processors.

Intel confirmed the existence of 'Conroe', the desktop implementation of Merom, during the Tokyo-hosted Intel Developer Forum in April. The server incarnation, Woodcrest, was mentioned last week at the company's analyst conference.

Conroe has been the subject of speculation for more than a year now, but came into focus when Intel tore up its roadmap and cancelled Tejas, the successor to Prescott. In April 2004, it was clear the Pentium M-derived Conroe would succeed the NetBurst architecture.

Pentium M architecture's unsuitability for HyperThreading could be a potential spanner in the works: its pipeline is too short for HT to work effectively. But HT's performance boost is minor compared to what you get from adding a second core, so Intel may not be too bothered about this. The dual-core Pentium D doesn't support HT, though it can do so if Intel enables the technology. That it doesn't may be a way of distinguishing the D from the dual-core Extreme Edition, as well as an attempt to remove HT use from mainstream desktop CPUs, paving the way for Conroe.

Gelsigner's comments can be read either way - it's not clear from what he said whether Conroe will have HT or not. We shall see. ®

Related stories

Intel feels healthy and dual-core happy
Intel: 65nm output to surpass 90nm production Q3 06
Intel confirms 'Conroe'
Desktop Dothans 'will not replace Prescott'
Intel to 'ditch' Pentium 4 core after Prescott
Intel's deskbook CPU platform merger plan

Whitepapers

Best practices for enterprise data
Discussing how technology providers have innovated in order to solve new challenges, creating a new framework for enterprise data.
Implementing global e-invoicing with guaranteed legal certainty
Explaining the role local tax compliance plays in successful supply chain management and e-business and how leading global brands are addressing this.
Advanced data protection for your virtualized environments
Find a natural fit for optimizing protection for the often resource-constrained data protection process found in virtual environments.
How modern custom applications can spur business growth
Learn how to create, deploy and manage custom applications without consuming or expanding the need for scarce, expensive IT resources.
High Performance for All
While HPC is not new, it has traditionally been seen as a specialist area – is it now geared up to meet more mainstream requirements?