Originally Posted by leopold1985
here you go, this time I have used the exact same setting as yours. the score has dropped as expected but still well over 50.
Fluidmark benchmark is non-illustrative in terms of the question at hand.
There's virtually no rendering load in Fluidmark, unlike the situation with actual games. The main card basically has nothing else to do besides physX, so it'll score much better alone than it will with a less powerful dedicated card. Literally, ALL you're comparing when you look at Fluidmark in this way is: 'which of these two is a better physX card'?
Games are different, esp. if you crank the settings and give your main card a lot of 'other work' to do beside physX.
I wouldn't go out and buy a physX card, but if you have one laying around it's worth throwing in ... in actual games, a 560ti for rendering + dedicated 9600gt for physX will usually perform better than a 560ti doing both ... but again it does depend somewhat on how hard to run you make the game via the settings.
Mafia 2 is the best example because having a dedicated physX card is the only way to force the GPU to handle cloth physX instead of the CPU.
If you do searches on this site for 'brettjv physx' you'll see lots threads w/charts and graphs I've done re: this subject.
EDIT: Ooops ... this was a necro-bump. I'm locking this thread ...Edited by brettjv - 9/27/11 at 5:08pm