Go to a Macworld Expo keynote speech, and you’re bound to see it–a demo that shows a Mac blowing away its Wintel rivals. Apple executives run an Adobe Photoshop file on machines with the fastest available PowerPC and Pentium processors: the Mac sprints to the finish line first, and the Wintel PC staggers down the home stretch like a weekend jogger running a first marathon.
Apple’s point is really quite simple: megahertz numbers aren’t everything. (See ”
Does MHz Matter?
” elsewhere in this issue.) Chip architecture is just as important–that’s what helps a Power PC CPU outperform a faster Pentium chip. For example, at the January 2001 Macworld Expo, a 733MHz Power Mac G4 finished a Photoshop test 33 percent faster than a PC with a 1.5GHz Pentium 4 processor did.
Does Apple’s example hold up in a real-world road race? To find out, we ran six Photoshop 6.0.1 operations on the fastest PowerPC and Pentium chips we could find: a 733MHz Power Mac G4 and a Gateway Performance 1500XL with a 1.5GHz Pentium 4 processor (with added memory and an Nvidia GeForce2 MX graphics card, for more-comparable performance).
Going up against a Pentium 4 processor roughly twice its speed, the G4 more than held its own. The Mac and the PC performed most of the tasks in about the same time, with the Mac the decisive winner in two tests.
How did a Mac with a 733MHz processor outperform a 1.5GHz machine? Credit AltiVec, the subprocessor built into a G4–it allows the chip to render graphics and perform calculations faster than an ordinary PowerPC CPU. Adobe designed Photoshop to take advantage of AltiVec; hence the Mac’s speedy performance in Photoshop tests. (In April, Adobe announced a plug-in that optimizes Photoshop 6 for the Pentium 4.)
But try other applications, and the speed advantage that the Mac enjoys over a Pentium PC quickly evaporates. Operations in Microsoft Office took longer on the G4–more than three times as much time in one case. Only line-by-line scrolling tests in Word and Excel were faster on the Mac.
The MP3-encoding test is a perfect example of the PowerPC G4 chip’s worst-case scenario. Since the Mac version of the MVP player and encoder hasn’t been optimized for AltiVec, the G4 lags behind the PC version. (When running AltiVec-friendly SoundJam, the PowerPC edges out the Pentium. Our test also involved reading audio-CD data from the G4’s sluggish SuperDrive, which slowed things down further.)
Architectural differences between the Mac and Windows platforms account for the PowerPC’s lower Quake III frame rate. Most PCs have a sound card to take the load off the host CPU during game play; the extra processing demands on Apple Sound Manager sent the Mac’s score even lower. With Sound Extension turned off, the PowerPC’s scores improved.
It’s not just marketing spin when Apple says processor speed isn’t the only way to determine whether a computer will zip through a task. But if the other factors involved don’t favor the Mac, performance will lag behind that of a Pentium machine, megahertz gap or no.