The in-game benchmark was used with a few different configurations. First, the 8800 GTS by itself, then the GTX 275 by itself, and finally the GTX 275 used for graphics and the 8800 GTS dedicated to physics. Tests were run with the following settings:
1680x1050 resolution
Very High detail
Everything else enabled except for V-Sync, MSAA, and 3D Vision
Batman PhysX Performance
1680x1050 resolution, 0xAA, Very High Graphics Settings
Card
PhysX Off
Normal PhysX Setting
High PhysX Setting
GeForce 8800 GTS 512
91
42
35
GeForce GTX 275
119
55
46
GeForce GTX 275 + 8800 GTS 512
120
66
59
Notes
With PhysX off, you get only the most basic physics, equal to what is present in the console versions of the game. As expected, the GTX 275 is substantially faster than the 8800 GTS, and the dual setup netted no performance gain without PhysX effects enabled.
The Normal setting brings volumetric smoke/fog and dynamic debris like papers and leaves, but sees each number reduced by about half. The same margin between the 8800 GTS and GTX 275 exists as with PhysX turned off. A very respectable 20% is gained with one GPU dedicated to physics calculations.
Framerates are reduced further when the PhysX are turned to High, as realistic cloth effects beyond the cape are added and smoke/debris are further refined. The individual scores maintain the identical proportion seen in previous tests, but the really interesting part is with the pair’s result. The increase in performance brought by the dedicated 8800 GTS is even greater on the High setting than on Normal, coming to about 28%.
Comparing these results to the official recommendations (GTX 260 for Normal and GTX 260 + 9800 GTX for High), it becomes apparent that they are on the conservative side. Depending on what you think is a playable framerate, you could pull off High PhysX for the most part with a mid-range 8-series video card.
How do the PhysX effects look?
Digg
Del.icio.us
Reddit
SU
furl
