Nvidia Cheats in 3Dmark2003?
 

You have to read this article:
http://www.extremetech.com/article2/...1086025,00.asp

Apparently Nvidia changed things in their drivers so they score better in 3Dmark2003.

Read the article for more information

did this not already happen with the 5800 ? When nVidia released 3dmark03 specific drivers?


if the quality stays as it should this time around, there is no problem.. otherwise :/

3dmark03 aint a very respresentable benchmark anyway when it comes down to measuring real gameplay speed...

I know,

but it is used by a lot of normal users who only look at this benchmark to make their decision of which card they are going to buy.

So in that way they are indeed cheating.

The rest of us, are also using other benchmarks, so for them it's not a big deal.

pretty much sums it up:

But is it possible to find a solution against cheating like nVidia did?
How to test/bench the real power of a GPU?

Moving the camera around @ random could prevent what nVidia did, but maybe those "@random" flights can give other results in the end..

just use game as a benchmark
you start up a round of Q3A / UT2003 , record a demo & then timedemo it on every card, should give you quite accurate results

But then again, when you want to use a standard demo/flyby (each time the same path) , you give the opportunity to tweak the drivers so they make the card look better.

In some way there has to be a standard bench (a game might be a good idea), but with a certain amount of "coincidence" so the manufacturer can't predict what his card will have to show (in comparison with 3Dmark200* )

"you start up a round of Q3A / UT2003 , record a demo "

this is random, always :)

I believe this is the source of the dispute; http://inquirerinside.com/?article=8498

yep it seems futuremark & nvidia are no best buds