vickybat
I am the night...I am...
Now this is kind of interesting news. Nvidia has surprisingly released the source code of its LLVM ( low level virtual machine) CUDA compiler. Developers and software tool vendors will greatly benefit from this as they can enable GPU support in software apps running on various processor and gpu architectures including intel's x86 and amd gpu's as well.
CUDA was well received being a powerful computing architecture that can be used with well know programming languages like c,c++ to assist gpu computing. Its similar to opencl but was proprietary until nvidia made the source code available. Previously it needed nvidia gpu's to aid gpu computing but now it has a chance to take advantage of other processor architectures.
Considering heterogeneous processor architectures as the future of computing, this is a great move by nvidia making cuda open considering its modular architecture and support for a vast number of programming languages. Developers now have a far greater headroom to utilize cuda in their apps and support a greater number of hardware. Open-cl now has a strong competitor.
Source
CUDA was well received being a powerful computing architecture that can be used with well know programming languages like c,c++ to assist gpu computing. Its similar to opencl but was proprietary until nvidia made the source code available. Previously it needed nvidia gpu's to aid gpu computing but now it has a chance to take advantage of other processor architectures.
Considering heterogeneous processor architectures as the future of computing, this is a great move by nvidia making cuda open considering its modular architecture and support for a vast number of programming languages. Developers now have a far greater headroom to utilize cuda in their apps and support a greater number of hardware. Open-cl now has a strong competitor.
Source