The X800XT is not 15% faster than the 6800U in almost any real world game. Again, these idiots miss the point, a good gaming benchmark needs to predict overall system performance, not just the differences in a sinlge compinent of the system. For example, the X800XT overall score is 2.5 times higher than the 9800 pro's overall score, but in aquamark the difference would only be 1.5 times higher at most (I know this based on my two systems A64 3200+/6800U and AXP 3200+/9700pro), and that is closer to the real difference in fps rates you get in most games. Maybe that's what they intended, so you could take it for what it is, but it's not a good linear gaming performance predictor.
It's kind of like if we compare the memory badwidth of a P4 2.4C to A64 3400+, the pentoium will still have about double the bandwidth because of it's dual channel support and 800MHz FSB, but we all know the 3400+ will dominate in any real world gaming test by a wide margin. Bottom line, performance in real world games depends on many other factors, and that overall performance which is quantified by the fps rate, is what people care about.