nicolasaunai
commented
3 weeks ago Apparently it is something that occurs when using so many (too many) MPI processes and very small patches are created. Not sure what happens yet but this ran on 192 cores, fails whatever branch I'm on, and all goes well with 40 cores. TBC
Two different runs on branch nu
1)
2)
appear to crash for no reason. It's probably not too low hyper-resistivity because it looks diffusive enough. however in both cases the divergence appears on a L2 corner.