ORGANIZE_USER_update_boundary_particles_Version
version of USER_update_boundary_particles.f90 to be used
The default value is
ORGANIZE_USER_update_boundary_particles_Version = 3
If
COMP_SharedMemoryForBE = true, then ORGANIZE_USER_update_boundary_particles_Version=2 will not work.
-
- Version 1 is original.
- Version 2 tries to distribute the computations of the geometry nodes to several MPI-processes, and then broadcast the data by MPI_bcast.
- Version 3 implements MPI-shared-memory movement of the boundary. Aditionally, it does not touch boundary elements which are flagged by MOVE-1 (see ALIAS attributes)
Special feature for version 2:
This invokes version 2 (first integer digit), but also tells
MESHFREE to use no more than 64 broadcasting processes.
In this logic,
is the same, and 0 broadcasting processes means no broadcasting at all, rather each process computes all necessary movement data on its own.