MPI across multiple nodes using a container

Dear all,

I found my parallel FreeFem code could not scale across multiple nodes (40 processors per node) on the university’s ARC system, although it scale great on one node (less than 40 processors).
Can I ask:

  1. Does the MPI implementation of FreeFem within a container understand how to work with the SGE environment variables?
  2. What’s the right command to use in order to run a FreeFem script across multiple nodes?

When I say container in the above, I mean a singularity container freefem.sif using the following command

singularity build freefem.sif docker-archive://freefem.tar.gz