[ The PC Guide | Systems and Components Reference Guide | The Processor | Processor Architecture and Operation | Processor Performance ] Processor Performance Factors Since the performance of a processor is (in most cases) based on how many instructions it can execute in a given time, it has become common to use the words "performance" and "speed" somewhat interchangeably. Unfortunately, the word "speed" has too many meanings when it comes to processors, and as a result frequently one person will use it to mean one thing and another person to mean quite another. There are two major factors that determine the performance level of a processor:
Since clock speed is easy to see and understand--it's just a number--and architecture is both complex and difficult to understand, it's not surprising that the former receives much more attention than the latter. This is unfortunate, because looking at just the clock speed of a processor is very deceptive, because it tells only one part of the picture. In fact, this is moreso today than ever before due to the much greater variety in processor designs and technologies. The "P" rating scheme was invented by Intel competitors AMD and Cyrix to provide what they feel is an assessment of the value of their processors that is more fair than using just clock speed. One great example showing the importance of architecture is AMD's K5 series of processors. Some of these processors actually have very different performance ratings despite running at the same clock speed! For example, the K5-PR100 and the K5-PR133 both run at the same clock speed, 100 MHz, but the PR133 has approximately 33% better performance, due to changes that were made to the processor's internal architecture. Clock speed can be used to compare processors only if they are identical internally. (Obviously, the K5-PR100 and K5-PR133 aren't the same internally if they have the same clock speed and different performance. I think AMD should have given them more clearly different designations to avoid this confusion.) This means you can only use clock speed to compare the performance of otherwise identical processors. A Pentium 200 is in fact 20% faster than a Pentium 166. But it isn't 20% faster than a Pentium with MMX 166, because of the latter's architectural improvements, such as a larger internal cache.
|