nvidia Cg language question

[acejj26]

i'm sure a lot of you have read about the new high-level Cg language from nvidia. for those of you who haven't, it basically allows for game programmers to code their game once in an easy language and then they can compile it in DX or OGL. my question is this, though: i've read that there will certainly be optimizations for nvidia's GPU's in the compiler, but down the road, there will also be optimizations for other GPU's (R300, Parhelia, etc); so will there be different game versions for different GPU's depending on how the game is optimized?? or is their some way that the compiler can optimize the game for all supported GPU's??

can anyone venture a guess about this?

 

[Thorstadt]

Hey, I just read this in an article and it should answer your question:

Seitz said Nvidia will make the basic Cg code available for other makers of graphics chips to write their own compilers. "We're going to keep our secret sauce, but we'll let other hardware vendors plug their stuff in," he said.

You can read the whole thing here.

 

[acejj26]

i completely understand that much

but what i wanted to know is...say a game developer writes a game in Cg and then compiles it. he can optimize it for the R300, the NV30, the Parhelia, etc. so, does he have to make different versions of the game, each one compiled to be optimized with a certain GPU in mind? or can he compile the game once with all the optimizations built in?? if he does that, how would that work when the game is ran? would the program run and query the system to figure out what GPU is being used and then it configures itself on the fly?

i'm just curious here about how games will be optimized using this new language.