How We Put Windows to the Test
We conducted three types of performance evaluations: automated WorldBench 6 tests, hand-timed tests for file management and waking from hibernation, and video frame-rate tests using popular games. We ran WorldBench 6 with Windows XP Professional SP2, Windows Vista Business, and Vista Business SP1 on a higher-end desktop and a low-end but Vista-ready laptop. (For each system's specs, see "WorldBench 6 Results: XP vs. Vista vs. Vista SP1.") We ran each test three times, and then averaged the scores. To learn how our WorldBench 6 tests work, consult "WorldBench 6 Beta 2 FAQ, Version 1.1."
To test video, we measured the frame rate (in frames per second) that each system could handle from two games, Doom 3 and Far Cry. We ran tests with antialiasing on and off, and at the 1024 by 768 and 1280 by 1024 resolutions.
Finally, we ran hand-timed tests to measure file copying, file compressing and decompressing, and recovery time from hibernation on both systems.
XP Wins Convincingly
As the composite scores indicate, XP outperformed both Vista and Vista SP1 on our automated WorldBench 6 tests by a significant margin, on the desktop and again on the notebook. Vista SP1 finished slightly ahead of plain Vista on the desktop test bed, but managed only a tie with it on the laptop.
In the areas where Service Pack 1 improved on pre-SP1 Vista's performance, the gains were far from dazzling.
In our desktop WorldBench 6 tests, XP, which dominated most of the re sults, finished last on a couple of measures: the Nero CD-burning test and the Photoshop image-effects test. The latter marked the only time Vista SP1 emerged as the clear winner.
In the Nero test for desktops, Vista ran 4 percent faster than Vista SP1 and 12 percent faster than XP.
On the laptop, XP had the best score on every test. The laptop results indicate that, as we expected, XP is the better choice for lower-power PCs.
On both test beds we used, the SP1 results were frequently the same or extremely close to the results for Vista. But the Photoshop and Firefox tests showed little consistency in scores between the desktop and laptop PCs running the two Vistas. Hardware differences definitely influenced the outcome.