I want to know if Precision can support FSI computing for 50 million or even 80 million grids, and how stable is it?
The biggest meshes I have tested contain 20 and 50 million nodes on single cores.
The main issue are memory (for longer implicitly-coupled simulations using acceleration) and RBF system matrices.
Runtime is really only an issue for RBF solvers. In general, preCICE aims to be insignificant in comparison to the runtime of solvers.