Speaker
Martijn Marsman
(University of Vienna, Austria)
Description
In this talk I will discuss the parallelisation strategy VASP employs to distribute its work and data, and the ways a user may influence these to optimise the performance of VASP on high-performance computing (HPC) resources, be it CPU-only or GPU-accelerated hardware. The main focus will be on aspects of the parallelisation using MPI, but I will briefly touch upon the use of MPI + OpenMP as well.
Primary author
Martijn Marsman
(University of Vienna, Austria)
