NVIDIA Makes Available CUDA 6 Release Candidate
NVIDIA released on Wednesday the release candidate for CUDA 6, the latest major breakthrough in the company's parallel programming environment.
CUDA 6 features support for unified memory between the CPU and GPU, drop-in libraries for items like BLAS and FFTW, improved multi-GPU scaling for up to six GPUs, and many other improvements.
Details on the CUDA 6 release candidate can be found via the NVIDIA blog.
CUDA 6 RC can be downloaded from the NVIDIA Developer Zone. When it comes to the Linux binaries there's spins for RHEL/CentOS 5 and 6, Fedora 19, openSUSE 12.3, SUSE Linux Enterprise Server 11, Ubuntu 13.04, and Ubuntu 12.04 LTS.
CUDA 6 features support for unified memory between the CPU and GPU, drop-in libraries for items like BLAS and FFTW, improved multi-GPU scaling for up to six GPUs, and many other improvements.
Details on the CUDA 6 release candidate can be found via the NVIDIA blog.
CUDA 6 RC can be downloaded from the NVIDIA Developer Zone. When it comes to the Linux binaries there's spins for RHEL/CentOS 5 and 6, Fedora 19, openSUSE 12.3, SUSE Linux Enterprise Server 11, Ubuntu 13.04, and Ubuntu 12.04 LTS.
4 Comments