My general understanding is that when nVidia bought the company that originally developed PhysX, they basically killed the silicon product that was being offered and reimplemented the API to use the programmable shader language.
But, assuming they did keep the silicon chip around, the basic process would be that the driver would probe the card and figure out whether or not it had this PPU. If it did, then anything using the PhysX API would be directed to the PPU. Otherwise it would be directed to the GPU or CPU. The game wouldn't really know or care if you had a PPU or not. If you had one, then physics related calculations would be done by the PPU.
What you might want to do, is look into the history of 3D graphics cards, since they too started out as secondary add-in cards before 2D and 3D functions were all folded into a single chip. It's the same basic idea with the PPU, and you can also look into MPEG-2 video decoding cards that used to be sold. The basic process will be the same in both cases, so you should be able to apply the things you learn from one to the other.
Hello,
I'm working on some "articles" regarding physics in video games and I came to read a lot about NVIDIA physX card and also about several physics engines.
But so far I couldn't find anywhere something about how it really works. I know it is similar to a GPU, but still a lot of things are a little unclear for me. So if anyone can help me figure out these questions, please do.
So here are the questions:
1. What happens if you have a game with a physics engine and you also have a PPU (physics processing unit), like PhysX?
2. What happens if you have a game with a physics engine, but you don't have a PPU, only a normal GPU?
3. How will your game know that you have a PPU or not; and how is it solved that all the calculations will be done by the PPU?
That's it for now. Thanks in advance for the help.
Chowhound
Comic Vine
GameFAQs
GameSpot
Giant Bomb
TechRepublic