In relation to what I wrote yesterday, I would like to mention my idea of what Nvidia could be working on with their purchase of AGEIA in February.

I call it the GPPPU: General-purpose computing on physics processing units.

In other words, a physics processing unit that can also handle most of the applications that are usually dedicated to GPUs. It will be akin to the GPGPU, which uses “a GPU, which typically handles computation only for computer graphics, to perform computation in applications traditionally handled by the CPU”.

In the GPGPU arena, Nvidia is already competing with ATI (now owned by AMD); it is also in a war of words with Intel over the fate of the traditional CPU.

But if Nvidia is working on making a GPU that can do a CPU’s job, then how will they have the time and money to make a PPU that can do a GPU’s job?