Originally posted by jimmyjames123
Then both NVIDIA and ATI are cheating. Like it or not, both of these companies "optimize" for this benchmark. This makes a perfectly level comparison using a synthetic benchmark very difficult, no doubt about that.
no... not quite...
first off when the nv30 initially gained a huge performance boost they were not rendering @ the required precision level specified in the dx9 spec... ie... fp24 and higher... this has been debated on this and other forums to a degree where most people AGREE such a thing was happening..
secondly... there is no indication from any research done that shows that ati is indeed hacking in this benchmark... they have maintained pretty much the same performance from day one... which means 1 of 2 things to me... either they do not have the capability to optimise like nvidia and gain 2 times the performance from 1 driver release to the next (bear in mind there are some who are harping on the presence of ATi as a beta member of futuremarks 3dmark03 suite) or secondly... ati are staying within spec and hence cannot indeed achieve a higher score in 3dmark03 without more efficient drivers... which is why their incremental increases are so small or non-existent in this benchmark compared to the 100% increase nvidia's drivers have netted it and the rather large % increases we have seen with these drivers..
if the playing field were equal both the companies would benchmark using drivers that followed the dx9 spec and rendered the scenes per the specifications of the benchmark makers..