Read more.Quote:
LLVM compiler opened up in hopes of extending reach to other architectures.
Printable View
Read more.Quote:
LLVM compiler opened up in hopes of extending reach to other architectures.
When I read the title I was hoping CUDA would become platform agnostic... oh well, I guess I was being overly optimistic to think nVidia would allow competitors to benefit from their code to that extent.
I guess OpenCL will have to do.
Well, in the future who knows maybe they will open it up to AMD so that CUDA made by NVIDIA becomes well known and they'd no doubt say "Home of CUDA" or something =PQuote:
Originally Posted by ExHail
Alternatively AMD could open up its own architecture details and have a 3rd party work on the support.
Hmmmmmmmm
So long as everybody is making their own special-favour-my-hardware stuff it's largely irrelevant to end users - CUDA is one example but the other from NVIDIA is Physx of course.
I'd like it rolled into DirectX personally given the choice (not that I have anything against another third party doing it but it seems like a obvious move and echoes the past pain of glide et al).
Agree with Dangel....it should be in DX, although I would like to see support for off-loading retained.
DirectCompute would be more relevant, or better still OpenGL...