SU2 _ MPI Problem
Hello All,
I was wondering if there is any control over how MPI divides the mesh when running on a HPC cluster. Sometimes when I run a ~7million cell 2D mesh It works with 200 cores but when I go any value above (lets say 360) it either gets stuck at the “Communicationg....halo layers) step or throws a MPI FATAL ERROR. Any comments or suggestions are appreciated. Mehdi |
All times are GMT -4. The time now is 04:09. |