Run Cuda HPC applications from prepared docker containers.
https://ngc.nvidia.com/registry
For example LAMMPS.
