How is virtualflow parallelized? Does it use MPI or another communication protocol?
It uses a perfect parallization, also called delightful or pleasing parallelization: https://en.wikipedia.org/wiki/Embarrassingly_parallel
More details you can find in our publication: https://t.co/cB0yasUO0y
Meaning there is no need for OpenMP, MPI, and similar (since there is no communication at all), and one can thus obtain a perfectly linear scaling behavior