The rCUDA Team is proud to announce a new and revolutionary version of the rCUDA framework which will include many new functionalities as well as boosted performance. This new version, cooked for over a year, will incorporate pipelined transfers, full multi-thread and multi-node capabilities, CUDA 4.1 support, global scheduler integration, support for CUDA C extensions, and native InfiniBand support.

The capabilities of this new version were presented in the HPC Advisory Council Switzerland Conference 2012 http://www.hpcadvisorycouncil.com/events/2012/Switzerland-Workshop/Presentations/Day_2/13_rCUDA.pdf, where a live demonstration was played, letting the LAMMPS simulator accessing 12 GPUs spread across a cluster from a single node.

This version will be released under non-free license, although we will keep on maintaining the free version as usual. We estimate this new version to be available for free early testing on July 1st, althoug only selected institutions will be eligible for this purpose. If you are interested in early testing the new rCUDA, please proceed to the "Contact us" section and let us know.