Archive notice: This article was originally published on April 15, 2016. Links and embedded videos are preserved as part of the historical record.
This is intended to allow CUDA programs to run on non-Nvidia GPUs as well. Programs would simply have to be recompiled; no changes would have to be made to the code.
If this is true, it could herald the end of OpenCL. AMD has tried, but has so far been unable to penetrate the market for CUDA programs. Nvidia will not like this development either, because its special CUDA applications ensure solid sales of expensive GPUs in the scientific sector.
Manufacturers of video applications could be pleased: they would no longer have to maintain different code branches for different manufacturers. Instead, code can simply be written in one CUDA version for Nvidia/AMD and Intel.
