No one noticed that the demonstration shows the videos running in slightly different modes – they don’t mention whether or not this may be an oversight or deliberate malfeasance on NVidia’s part – Nvidia has been known in the past to cheat on the driver level to improve benchmark scores. There is no mention of the fact that the Nvidia scene is a low-motion scene therefor requiring less decoding of the MPEG2 stream, while the supposedly slower computer is showing an action-intensive scene which required more decoding (the more that’s changing on screen and the faster it’s changing, the more data is required to describe it). They also don’t mention what the sources of the two videos are, i.e. DVD, Blu-Ray or from disk, or even if the two video sources are the same type of media. For all we know, the Nvidia screen on the left could be paused video.
In other words, the demonstration is completely worthless in the context of this article. And the really sad thing is the photo is credited to the author, who supposedly saw the demonstration (and took the photo with his cellphone camera, looks like).