CUDA 10.2 Released With VMM APIs, libcu++ As Parallel Standard C++ Library For GPUs

NVIDIA has released CUDA 10.2 for SuperComputing 19 week. CUDA 10.2 comes with some interesting changes, including to be the last release that will support Apple's macOS and the introduction of a standard C++ library for GPUs.

Highlights of CUDA 10.2 include:

- CUDA 10.2 now ships libcu++ as a parallel standard C++ library for graphics processors.

- CUDA Virtual Memory Management (VMM) APIs are added.

- This is the last CUDA release supporting Apple's macOS for developing and running applications. The next CUDA release will be completely dropping macOS support.

- New interoperability APIs for buffer allocation / synchronization / streaming. For 10.2 these new interop APIs are in beta.

- Increased texture size limit for Maxwell+ GPUs.

Downloads and more details on CUDA 10.2 via developer.nvidia.com.