"one source, run everywhere"

So, I snapped this of the khronos group twitter feed.

http://www.par4all.org/

it’s supposed to be able to take C and convert the code to run on CUDA. But more importantly it started work to be able to convert C to OpenCL.

maybe it’s interesting for developers.

if you could just convert the C code for the blender render and nodes to OpenCL and have it executed on the much faster GPU.