Please register or login. There are 0 registered and 164 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s November 26 - 03:21am EST 
Hardware Analysis
Forums Product Prices

  Latest Topics 

More >>


  Intel's Prescott, a Pentium 4 on Prozac? 
  Feb 02, 2004, 08:30am EST 
By: Sander Sassen

It has been a long wait, Intel was supposed to introduce the successor to the Northwood core months ago and just introduced it today, on February 2nd, in the first quarter of 2004. Our initial observations of the 3.2GHz Prescott Pentium 4 at least show us that Intel has managed to get the kinks out of its 90nm manufacturing process. Whilst the 90nm manufacturing process means Intel can shave a few dollars off of the manufacturing costs the same can probably be said about the performance of the Prescott Pentium 4. Our benchmarks show that a 3.2GHz Prescott performs just about as well as a 3.06GHz Northwood, which is running off of a mere 533MHz FSB. So basically were not impressed, especially not since weve been following the development of the Prescott since its conception about two years ago. If we factor in all the improvements it was said to have wed expect it to perform a whole lot better.

Pentium III processor

Fig 1. The Pentium III processor that was replaced by the Pentium 4 about three years ago.

But wait, isnt this a similar situation like weve seen with the introduction of the Pentium 4 a few years ago? At that time the 1.5GHz Pentium 4 had a hard time keeping ahead of the fastest Pentium IIIs and many people raised the question why the Pentium 4s IPC, instructions per clockcycle, was that much lower. Today weve come to see that a reduction in IPC allowed the Pentium 4 to scale much higher in frequency than the Pentium III would ever have been able to. Do we see a similar approach in the Prescott? Not really, Intel took drastic measures this time as they increased both the latency of the processor caches and the length of the branch prediction pipeline to create more headroom for higher clockspeeds. Strange, as traditionally a die-shrink meant faster silicon and a reduction in overall latency and signal propagation through the processor core. But apparently the Prescott has other problems that prevent Intel from running it at full steam. Well keep a close eye on newer revisions of the Prescott core that might perform better and rid the new Pentium of its current Prozac subscription.

Sander Sassen


 Last Post 


  Voice Your Opinion 
Start New Discussion Topic


  Related Articles 

A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.