Until now, testing DVD performance and video quality has been a very subjective exercise with far too much room for error. Scaling performance is hard to judge, and calculating CPU utilization over a set period of time isn't as accurate as one would hope for a precise measurement. For this reason, MadOnion.com (in conjunction with Faroudja Laboratories) has developed Video 2000.

Although not a common household name, Faroudja is renowned for their line of picture enhancing products used in broadcast and high-end home theatre equipment. When these guys say they know how to test video capabilities, they really mean it.

In order to understand how our various cards will be tested, we'll first look at the tests that Video 2000 performs:

1) Quality - As we said earlier, without quality, performance is a moot point. Video 2000 tests the up/down scaling of both the blitter and overlay. DVD Playback and Color Space Conversion are also tested for accuracy.

Blitter and Overlay are both methods for displaying video data on your screen. By testing the quality of these two display techniques we are able to interactively judge how resizing affects their output.

In testing DVD playback, Video 2000 examines line-flicker, feathering, double imaging and tearing. The Color Space Conversion test is designed to reveal any loss of precision in converting YUV data (optimal for storing information) into RGB data (the kind that can be displayed), resulting in color banding.

Quality tests constitute 40% of the final score.

2) Performance - Once an adapter's quality has been established, performance is tested using another small barrage of trials.

Once again the blitter is tested, only this time for speed. From this test, a number in the form of MB/s can be calculated. Data transfer is then tested, and also recorded in a MB/s format.