Well, there's only 1 point in favor of PhysX: they put the physics topic on the table for devs. They showed the world what could be done separating the physics engine from the games allowing this 3rd party API/engine to work it's magic. So, as a first stone (correct me here if I'm wrong, please), PhysX really worked out nice. Now, thanks to nVidia and ATI (according to an article saying ATI said "Hell no!" to support PhysX, favoring it's own and then open proyects), it came to a stop in our eyes.
I like what we have now in terms of "physics acceleration" and it's developing fast. And remember, the whole point behind having "better video cards" has always been about the EYECANDY, and we really can tell that physics is another eyecandy out there having the same importance than mostly every other form of eyecandy (AA, AF, types of lights, etc).
Now, the GTX480... That thing is supposed to have a LOT of horsepower to do both worlds fine: heavy OpenCL/CUDA/DC/PhysX calculations and DX11/OpenGL/DX10 eyecandy goodness. Down the road we're supposed to see the above I said put into practice, because my bet is that OpenCL and DirectCompute is little to none on current games and the GTX4xx will show it's fangs when we get there, but that's just my interpretation of things. Right now, the GTX480 is a lackluster gaming card for games (as obvious as it may sound, it ain't for a lot of folks).
Better drivers and more DX11 games using OpenCL/CD is what the GTX480 needs... Maybe... 😛
I'm curious as to what the 5xxx series do in terms of OpenCL and DirectCompute right now...
Cheers!
EDIT: Typo xD