Can iMac toast the PC?
August 18, 1998
by David Essex and Paul Heltzel
(IDG) -- Apple Computer says its low-cost iMac, which debuted this weekend, is 40 percent faster than systems based on Intel's 400-MHz Pentium II processor. The hype-heavy iMac uses a 233-MHz Motorola PowerPC, or G3, processor, which Apple says outperforms the fastest PC available. But analysts and benchmark experts say that claim may not present a full picture.
Nathan Brookwood, microprocessor analyst at Dataquest, warns that Apple's benchmark numbers, based on BYTE magazine's BYTEMark tests of raw CPU speed, "seem a little out of line to me." Brookwood says results he obtained from IBM and Intel based on tests from the Standard Performance Evaluation Corporation do not support Apple's claim.
Extrapolating from the SPEC numbers (which measure raw processing power) for a faster 300-MHz G3 processor, Brookwood determined that the iMac's processor might at best achieve a 10.3 rating for CPU integer operations, and 6.6 on floating-point operations. In these tests Intel's Pentium II-400 shows better performance, rating 15.8 and 11.4, respectively. (Integer operations predominate in software that emphasizes data movement, such as databases and word processors; floating-point is important in math-intensive programs such as 3D games and graphics software.)
Since SPEC testing is typically used to measure the performance of UNIX workstations, Will Swearingen, strategic marketing manager for PowerPC at Motorola, suggests that those tests don't correlate to real-world performance. "If you're looking to gauge PC systems performance you would want to use the BYTEMark tests," argues Swearingen. "It's not just a random benchmark. The G3 outperforms the fastest PC in the Wintel space."
Truth in benchmarking?
What kicked off all this hoopla? This statement: "The iMac for $1299 toasts the fastest PC money can buy at any price--the Pentium II-400." That quote from Steve Jobs, Apple's interim CEO, was included in an August 13 Apple press release.
A similar assertion about Power Macs sparked bitter controversy last April that saw Intel discrediting Apple's claims. BYTE, which suspended publication in May, posted a FAQ on its Web site explaining that the Power Mac had superior performance on a particular type of integer test. The magazine noted that the Pentium II bested the Power Mac on some other tests that employ full-blown applications to test other subsystems such as storage. It cited differences in software compilers as another factor in the Power Mac's ability to run the test faster.
According to Intel spokesperson Seth Walker the company "doesn't comment on other people's product's directly." He did however debate the merits of using one measure to access the performance of a PC.
Bill Rinko-Gay, PC World's Network Test Center Director, points out that a more complete picture would be gained by also using application-based testing that shows how quickly common tasks are accomplished. "Benchmarks don't work your computer the way you work your computer," said Rinko-Gay. "There are some applications in which the G3 is a faster processor. But your PC isn't just a processor. It really comes down to this: How fast can you do your daily work?" That can be determined only by testing with applications.
Back to the top
© 2000 Cable News Network. All Rights Reserved.
Terms under which this service is provided to you.
Read our privacy guidelines.