I seem to see this argument a lot, about how we "shouldn't care about our 3DMark scores", because we "don't play 3DMark". I will agree with you that the acid test is how well our system performs in the games we play, but benchmarks EXIST so we have some way of judging how well our setups relate to others, and also it is a common way to measure performance (hence, "benchmark"). It is a way of giving us real numbers so we can compare our setup to someone else's. Obviously in the end all that matters is how our games perform, but at least 3DMark gives us something share with the group. It's an easy way to give objective performance results for varying systems.
I agree with you, and I'm not refuting the fact that benchmarks are a good, popular and standard way to compare performance between different systems. However, my point was that up to the point when I first posted in this thread, nobody had tested the new drivers with any games at all, or at least they weren't posting anything about it, and if you look back, there were tens of posts before I posted in this thread. Everyone was just raving about how well the new drivers are by just running 3DMark2001. Fast forward to Rand's post, and you'll see that according to his figures run on his system, the new drivers aren't really that good at all in real games, compared to the existing drivers. So, I guess I'm just trying to say that it isn't wise to jump to conclusions with just 1 benchmark, especially when this benchmark is a synthetic one rather than one run from within a real application/game that we use/play.
btw thanks for the detailed benchmarks Rand, it was really helpful to us all