In order to figure out the issues I’ve been experiencing trying to batch run MFiX using dmp, I’ve come up with several questions related to how MFiX relays information to MPIs.
Is it possible that if a timestep diverges, MFiX cant properly send messages to MPI, resulting in return PID 0?
What MPI should I load into my shell?
Which fortran compiler is recommended for use when using MPI? Are there certain compilers which don’t work at all with MPI?
I’m seeing success when batch solving serially, but my dmp scripts are giving me extremely vague errors about MPI. The following is what I see when batch solving with dmp and mpi:
The attached files are the serial, dmp scripts I’m submitting
particle-bed-campaign - serial.TXT (106 Bytes)
particle-bed-campaign - DMP.TXT (174 Bytes)