Mac vs. PC: Which Is Faster?

Welcome Guest. | Log In| Register | Membership Benefits

August 21, 1998 (5:08 PM EDT)

By John Gartner,

Lately, Apple's been on a roll, with iMacs flying off the shelves and the company's stock price more than double what it was at the beginning of this year. In addition, the company has loudly claimed that its G3 processor is superior to Intel's Pentium II chips.

In NSTL testing, Apple's G3 processor was faster overall, but trailed Pentium IIs in some application tests. Is speed more important than usability? Share your opinion in TechWeb threads.

But an independent test lab's findings confirm that while Motorola's PowerPC G3 chips are indeed faster than their Intel counterparts in CPU-intensive applications and synthetic benchmarks, real-world tests show Intel-based PCs provide better overall performance.

NSTL, an international IT testing organization and subsidiary of CMP Media, which publishes TechWeb, recently came to that conclusion when comparing G3s and Pentium IIs. According to Stephen Platt, director of technical services at NSTL, preliminary testing at their labs, in fact, mirrors the BYTEmark CPU results posted on the Apple website.

Apple says that a PowerMac G3/300 scores 10.2 on the BYTEmark test, more than twice as fast as the 4.3 rating the Pentium II got, and 45 percent higher than a 400-MHz Pentium II. "The G3 is faster than the Pentium II when running very processor-intensive activities," said Platt.

However, data from Apple and NSTL indicate that because CPU performance is only one of many factors that contribute to system performance, users won't experience nearly the same increase in speed. In tests using Adobe Photoshop, which also stress a computer's memory architecture, disk, and video, Apple claims a 33 percent improvement over a similar 266-MHz Intel PC. NSTL found similar results using its Photoshop test when comparing the G3 to 266-MHz computers, but a 300-MHz G3 finished only about 5 percent faster than a PC using an AMD K6-2 processor.

"For activities requiring large amounts of processor-intensive memory manipulation, a Power Mac G3 is around 10 percent to 30 percent faster than a PC running at the same basic processor clock rate," Platt said.

NSTL's office-productivity application testing showed PCs are much faster in graphics-intensive functions where screen redraws are necessary. In a suite of tests using FileMaker Pro, Microsoft Word, and Excel, PCs consistently outperformed PowerMacs. Micron and IBM PCs performed the tasks, which included scrolling and repainting of text as well as table and chart generation, in less than half the time of PowerMacs.

Platt attributed some of the disparity in results to the differing graphics subsystems of Macs and PCs. "The graphics component may be equally or more important than the CPU, and to a lesser extent, disk performance also affects your daily operation," he said. For example, the faster 266-MHz PC was tested with a Diamond Viper V330, which has a 128-bit graphics processing chip as opposed to the PowerMac's 64-bit ATI Rage II+.

The NSTL technicians also performed their low-level Intermark Video Test which confirmed the latest Macs trail PCs in video performance. "When replaying sequences of graphics commands, the Mac is much slower than the PC," said Platt.

'G3-based PowerMacs' performance, from the user's perspective, is not fantastically superior to that of comparable Pentium II-based systems.'
-- Stephen Platt
NSTL

While Apple's claims of superior process prove accurate, Platt said they should not be misconstrued as resulting in more powerful computers. "G3-based PowerMacs' performance, from the user's perspective, is not fantastically superior to that of comparable Pentium II-based systems," he said. "While its core [processor] performance is faster than that of the corresponding Intel Pentium II-based system, this is more than offset by its slower graphics performance, resulting in lesser real-world performance. "

Test Results And Notes (Compiled by NSTL)

The most recent versions of each application were used -- on the Mac, we used Office 98, on the PC, Office 97, and FileMaker Pro 4.0 and Photoshop 5.0 were used on both platforms.

The application tests -- using Excel, Word, FileMaker Pro, and Photoshop -- emulate sequences of common activities, demonstrating how the system would respond during typical use. One sequence, for example, consists of opening a file, inserting text, performing a spell-check, and writing and closing the file. Other sequences include chart preparation, database querying and displaying, and sets of image operations.

The bars show time to complete the operation in seconds for each test. Smaller times indicate better performance.

The Excel and Word tests blend disk, processor, and display activity. For the Excel test on the Power Mac G3 266, the time taken is unusually high. Engineers were unable to determine the cause of this reproducible slow score.

Mac vs. PC: Word And Excel

The FileMaker test mixes data manipulation and disk operation. The Photoshop test performs global image manipulation, including filtering applications and resizing, to stress the CPU and memory subsystem.

Mac vs. PC Benchmarks: Photoshop And
Filemaker

Test Configurations

NSTL tested the PCs and Macs as configured and submitted by the manufacturer using the default software drivers, operating system settings and adapters. Testing was done to emulate a typical user machine as received; additional peripherals were not installed in an effort to balance configurations.

The Macintosh tests were run in the default graphics mode. Virtual memory was turned off for all tests.

Test Configurations