I think now that the initial disappointments at the low scores have subsided and people have had time to reflect, I believe a very KEY conclusion has been made, that 3DMark03 is really a videocard benchmark and not a gaming benchmark.
I used to think 3DMark was a gauge on game performance and maybe older versions were. This version sure doesn't seem that way though.
As far as the difference in scores between various Nvidia boards being almost negligible, I agree this is a little alarming. We need some explanation from Futuremark about that one.
However, the difference in scores between a Ti4600 and 9700 Pro I fully expected and don't see why people got so alarmed. The 9700 Pro IS nearly twice as powerful as a Ti4600 so a Ti4600 scoring 1600 and a 9700 Pro scoring 4500 is to me totally reasonable.
As far as the generally abysmal frame rates running the tests (my P3 1 Ghz and Ti4600 sometimes registered ZERO FPS!), this too I think should be expected. 3DMark is a stress benchmark. It's SUPPOSED to drive your system to its knees. As PC technology continues to grow in power, eventually even 3DMark03 will be spitting out 18,000 marks on the systems of a couple years from now. This is the same situation for all prior versions of 3DMark. Our systems at the time really gagged but within a couple years we were eating it for lunch then, you guessed it, new version of 3DMark comes out and the cycle began anew.