Now that I think about it, maybe I WILL write a way to run several benchmarks in succession and then create a formula to give an overall score.
Heh... might be more useful than 3DMark200x. Yes, I think 2001 is a POS. I think 2003 is a POS. I've had games give me minimum framerates 10 or 15 FPS higher after I did some goofy tweak or driver change, but my 3DMark score would go down 500 points. I'd say, "WTF!?" and go back to enjoying my now-even-niftier gaming.
I'd much rather have standard benchmark utilities based on real games rather than Synthetic Benchmark of the Week (VulpineMark, SharkMark, whatever the PowerVR one was, the list goes on and on) so then maybe driver teams would spend time optimizing for games instead of entirely for 3DMark.
Oh well, I can dream, can't I.