Apple: World's Fastest?

Not very likely, and here's why.

Brad Grimes

Brad Grimes is a former PC World executive editor.

Let's kick this off with a little full disclosure: I own an IBM-compatible PC. It's my fifth IBM-compatible PC. I've never owned an Apple computer, although I've used many of them, including the 64-bit Power Mac G5.

Still, I can tell you, without a hint of bias, that the Apple Power Mac G5 is not "the world's fastest personal computer," as Apple has claimed in its advertising.

Or maybe it is. Or maybe it's not.

Frankly, I don't know for sure how fast the Power Mac G5 is. I only know from experience that it's wicked fast.

But people who know a lot better than I jumped on Apple's back when it began claiming to sell "the world's fastest personal computer" last year. Their gripe? That the tests on which Apple based its claims weren't fair. And frankly, those experts were right, if only because Apple tested its Power Mac G5 against Dell PCs (more full disclosure--I own two Dells), without regard for the fact that the world's fastest personal computer might actually be a Gateway, or an IBM, or even an Alienware Aurora Extreme custom-built gaming PC.

But critics had another issue with Apple's claim, aside from the fact the company clearly didn't test every personal computer in the world. Experts said the Standard Performance Evaluation Corporation benchmark test, which was administered by a third party, did not test the Dell machine in its optimal configuration.

The testing firm pitted the Power Mac G5 against a pair of Intel-based Dell systems running Linux and an open-source software compiler called the GNU Compiler Collection. To oversimplify, the compiler acts like a translator, taking source code and turning it into executable object code. Compilers can be fast or slow, and can be more suited to one computer than another.

Apple said this was the fairest way to compare, ahem, apples to apples. But for as long as there have been Apple Macs and IBM-compatible PCs, computer users have known that we've got a case of apples and oranges on our hands, and to try and compare the two based on speed is folly.

Not only is it hard to come up with good speed tests for both Macs and IBM-compatible PCs, but few people would care if someone did. For the large part, folks like me don't buy an IBM-compatible PC because it's faster than an Apple Mac. I bought mine because it was cheaper and ran more software. By the same token, Apple users don't buy Macs because they're faster than IBM-compatible PCs. They buy them because they're easy to use, because they don't crash all the time, and (when it comes to every graphic designer I've ever met) because graphical applications run better on a Mac.

Benchmarks, Schmenchmarks

When Apple laid its claim to the title of "world's fastest," tech analysts were quick to point out that if you ran the same Dell PCs with Windows XP and an Intel-optimized compiler, the Dell PCs would have scored much higher on the tests.

As a result of the controversy, British TV regulators banned Apple's "world's fastest" ads, and recently the U.S. Better Business Bureau asked Apple to stop using the claim in its advertising because it was misleading.

So what is the world's fastest personal computer? Who knows. I doubt very highly that any IBM-compatible PC can lay claim to the title.

Benchmark testing is an inexact science. A system that breaks speed records on one test might struggle on another. In my days reviewing computers, I've seen PC makers deconstruct benchmark tests and then optimize their systems to earn higher scores.

Don't trust benchmark scores, especially when a computer manufacturer is spouting them. If, for whatever reason, your buying decision comes down to a single score, at least look for application benchmark scores, like those used by PC World and other reputable publications. It's one thing to know how fast all the components in a system run; it's a whole other thing to know, comparatively, how a system will run the kind of software you use every day.

I know this: The Power Mac G5 is very fast. So are a few Intel Pentium 4-based systems I've tested and several AMD Athlon 64-bit PCs. I also know that when PC World pitted a Power Mac G5 against an Athlon-based PC, in what the magazine would grant was an imperfect test, the G5 wasn't faster.

If every personal computer in the world ran the same operating system, the same software, the same hardware, the same compilers, and so on, we might be able to find the world's fastest. But I don't think anyone wants that. If all the computers were the same, how would any computer get better?