Release of OpenCruncher – Version 7.2

Release

OpenCrunchers parallel calculation capabilities have been massively improved.

Release notes:

  • Input and output
    • Introduced a mesh import util and a native mesh format
    • Solver file IO works fully parallel now
    • Reduced result and restart file size
  • Parallelization & High performance computing (HPC)
    • Infiniband network support
    • Make use of the NVIDIA Collective Communications Library (NCCL)
  • Multiple bug fixes (e.g. device zero ghost calls)

OpenCruncher can now run on any GPU system of unlimited size and automatically uses the fastest network interconnect.