Computer benchmarks have always been surrounded by an aura of secrecy to non-computer people. They use completely foreign sounding units of measurement, and what do they really measure anyway? And most important of all, how do they affect an average computer user? Since it looks like theyíre around to stay, I thought Iíd uncover some of the mystery and educate myself, and hereís what I found.

One of the popular benchmarks is the Winstone, a test developed by Ziff Davis Labs to measure performance between PCs running widely used applications. The most important thing to remember about all benchmarks is that they should never be compared to each other because they all measure different things in different ways. Winstone itself has two versions: Winstone 32, which measures performance with 32-bit applications, and Winstone 96, which uses 16-bit applications. Although the applications they both use are similar, even these two benchmarks cannot be compared to each other unless, of course, youíre comparing your own computerís performance while running 16 versus 32-bit software.

Winstone works by starting with a base machine: A Dell^® PC with a 486XS processor running Windows^® 3.1 or Windows 95^®, depending on the version. The score of this machine is set to a 10.0, which is relative, having no meaning whatsoever until associated with another score. This accounts for the lack of measurement standards; Winstone just compares two numbers that would otherwise have no meaning. Therefore, a computer with a score of 50.0 is five times faster than the base machine. It is also important to remember that these scores are not representative of processor performance but the overall performance of your entire system configuration while running predefined tasks on the different software applications.

The Winstone benchmark is actually quite a good measurement compared to some of the other widely available tests because it measures graphical applications in addition to the standard office suite-type packages. The applications included in Winstone 32 are

Adobe^TM PageMaker^TM
Corel Corporation CorelDRAW!^®
Microsoft^® PowerPoint^®
Borland^® Paradox^®
Microsoft Access^®
Microsoft Excel
Lotus^® Word Pro^TM
Microsoft Word

Winstone 96 adds Borland dBASE^®, Lotus 1-2-3^®, Novell^® Quattro Pro^®, Lotus Ami Pro^®, Novell WordPerfect,^® and Microsoft Works. Keep in mind that the versions of the above software applications are different depending on which Winstone test is being used. Once the tests are run the computer is given an overall system score as well as a score for Desktop Publishing/Business Graphics, Database, Spreadsheet, and Word Processing (the Spreadsheet and Word Processing scores are combined in the Winstone 32 test).

See? Itís not so hard to understand. However, there is one final thing to keep in mind. If youíre testing the performance of two machines with similar or even identical hardware, donít worry if their scores are different. Subtle nuances in structure and minute details that you wonít think of, like temperature, fluctuate constantly, even with machines built from the same manufacturer. So consider the mystery unveiled. You can get your CD copy of Winstone from Ziff Davis for a $5.00 shipping fee ($6.00 in Canada) at http://www.zdnet.com/zdbop/reqfrm.html. Also, be sure to check out Gateway's current benchmarks, available for most of our systems.

Copyright © 1996 Gateway 2000 Inc. All rights reserved.
Please see our Legal Information. Please send feedback to Webmaster Central.