Please register or login. There are 0 registered and 1136 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s November 23 - 04:00pm EST 
Hardware Analysis
      
Forums Product Prices
  Contents 
 
 

  Latest Topics 
 

More >>
 

    
 
 

  The Megahertz Myth, Apple, AMD and Intel. 
  Jan 10, 2002, 07:30am EST 
 

Introduction


By: Sander Sassen

Megahertz, the good old, ever effective, marketing tool has been under fire from a number of manufactures as of late. Apple has been touting their computer's supremacy and clock cycle efficiency ever since the early days and AMD has been educating their prospective customers ever since they came up with their new naming scheme. But although both Apple and AMD claim that clockspeed doesn't always equal performance they both still use clockspeed as a marketing tool, and so does Intel.

Granted, ever since Intel and AMD designed new and different x86 architectures for their CPUs, the clockspeed-equals-performance argument has become almost academic and isn't as clear-cut as we'd like, we're simply not comparing apples to apples. But there's more, clock cycle efficiency is also often used to determine how efficient a CPU handles instructions compared to another. Although comparing clock cycle efficiency looks like a valid comparison there's a few things that should be taken into account.

First off, stating that clock cycle efficiency determines CPU performance is just as bad as saying clockspeed equals performance. Clock cycle efficiency simply determines how many instructions a CPU handles per clock cycle under ideal conditions, if there's a cache miss or a branch prediction miss we'll loose valuable clock cycles making up for it, thus severely cutting into a CPU's clockspeed efficiency. So there's no such thing as a raw and clear-cut clockspeed efficiency comparison, it is an indication how efficient a CPU executes but not an indication of real world performance.

A good example is Intel's Pentium 4, with its deep 20-stage pipeline. Suppose the branch prediction unit makes a bad prediction and the CPU was just about to process the data in the last stage of the pipeline, it now has to flush the entire pipeline and start over again, losing valuable clock cycles in the process. The longer the pipeline the more performance is lost. Same applies to the branch prediction unit, if it mispredicts too often that'll also cut into the CPUs performance, so both the pipeline and the branch prediction units need to be designed and optimized to offer the best performance. The real world performance of the Pentium 4 however shows that a good design can negate most of the drawbacks of a deep pipeline and yield excellent results.

That all of this technical jargon is often used to convince the consumer to favor one particular CPU or platform over another is no secret. However it is rather disappointing to see that all of the manufacturers, despite their efforts to educate their prospective customers, still resort to using clockspeed as a marketing tool. In the following pages we'll take a closer look at how clockspeed is still being used to sell their products, especially by Apple and to a lesser extend by AMD and Intel.



1. Introduction
2. Apple, Megahertz myth?
3. AMD, Performance index?
4. Intel, Clockspeed rules?

Discuss This Article (8 Comments) - If you have any questions, comments or suggestions about the article and/or its contents please leave your comments here and we'll do our best to address any concerns.

 

    
 
 

  Related Articles 
 
 

  Newsletter 
 
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.