There seems to be alot of controversy right now as to the validity of 3DMark2000 scores. I would propose we switch to the Evolva rolling demo for benchmarking that gives you scores with bumpmapping on and off, Real world type benchmarks are always a better idea in my book then synthetic benchmarks that can easily be manipulated through (certain) driver optimisations.
[This message has been edited by cpld005 (edited 08-24-2000).]