Running in parallel

From FDTD++ wiki
Jump to: navigation, search

The following describes how to run FDTD++ in parallel.

Note that there is a known issue with FDTD++: The simulation parameters nxprocs, nyprocs, and nzprocs must be set in the parameters file for all parallel simulations.

Instructions

In your work directory, open a terminal. Then, the precise command to execute FDTD++ in parallel depends on the version of MPI use to compile it. Typically though, it will be similar to[1]:

openmpirun -n 8 FDTDxx param mat geom.unv

where param, mat, and geom.unv were defined and discussed in the Running in serial section, and the simulation, in this example, is to be executed in 8 processes. For versions of MPI other than OpenMPI, openmpirun -n may need to be replaced by mpirun -np, etc.

Notes and references

  1. This assumes that the FDTD++ directory has been set in an environmental variable, as described here.

Notes and references