23rd August 2005, 9:31 AM
A PPU is an interesting idea, however LL has pointed out a lot here. I'll also say that the graphics processing really has to know pretty much every single thing a PPU might be doing. A PPU on a seperate card (as I've heard rumored before) seems similar to the idea of a card that only does 3D and needs a 2D graphics card in there (as they did some time ago).
I'm thinking that a GPU would actually still be doing a lot of the work but maybe a PPU would have a lot of shortcuts and predesigned algorythms and such that the graphics chip could access every now and then by sending the data and some calls there. (By "every now and then" of course I mean "every single frame*). The communication between the two would need to be as fast as possible. In fact, it may do well several years from now to just put all those algorythms straight into the GPU.
That Icarus image, if real, looks like some sort of rock album cover. What a big headed serious version of Pit that is.
I'm thinking that a GPU would actually still be doing a lot of the work but maybe a PPU would have a lot of shortcuts and predesigned algorythms and such that the graphics chip could access every now and then by sending the data and some calls there. (By "every now and then" of course I mean "every single frame*). The communication between the two would need to be as fast as possible. In fact, it may do well several years from now to just put all those algorythms straight into the GPU.
That Icarus image, if real, looks like some sort of rock album cover. What a big headed serious version of Pit that is.
"On two occasions, I have been asked [by members of Parliament], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able to rightly apprehend the kind of confusion of ideas that could provoke such a question." ~ Charles Babbage (1791-1871)