NVIDIA PhysX Quick Test

As you guys likely know NVIDIA issued a new PhysX supporting driver to the press, which (driver) actually already got leaked.

This driver will enable and allow a significant set of Physics functions to be ran over the GPU. Now I just put this quickly to the test, though in very early stages testing I must say it pushes out some pretty interesting results.

First off, there are some new included demos in the additional PhysX software you need to install, which really work out well. This trick is however to get some software going with this new feature. Currently I am testing a GeForce 9800 GTX Black Edition from XFX; tried out the new PhysX driver and installed a press version NVIDIA PhysX software build on there (not available just yet).

Though within games offloading PhysX to the GPU would take a toll on your framerate there seems to be one application benefiting from this new option greatly .. 3DMark Vantage.

Let me show you a test run with the new GeForce 177.39 driver and then with the NVIDIA PhysX (8.06.12) driver installed:

• PhysX driver on: Overall Performance score 7493 - GPU score 6029
• PhysX driver off: Overall Performance score 5823 - GPU score 6022

It causes a bit of a concern as what exactly is the new score now telling us ? Definitely not GPU performance, but it does boost the overall Vantage score significantly. Looking further ahead ... If you measure game performance with PhysX enabled in a game that actually supports it, the framerate will be slower opposed to a competitor who doesn't have that function enabled/supported.

My suggestion to NVIDIA, definitely make an on/off switch in the drivers so the press can measure objectively. It's a great feature though, and I love to see this well developed and become a standard in games.

If you like to discuss NVIDIA PhysX, we've opened up a thread right here.