first posted: 14 August 2006 (applies to ALL releases of cm1)
CM1 does not work properly for 2d runs with MPI (i.e., distributed memory). MPI should not be used for 2D simulations until further notice. [There are no problems with single-processor runs or OpenMP (shared memory) runs. There are no problems with 3D runs with MPI.]
This problem also applies to axisymmetric simulations. Do not run axisymmetric simulations with MPI.