OpenMP is a compilation option. If you look at the makefile in the FDS
Repository under FDS_Compilations, you see it is an option in the
Intel Fortran build. Christian Rogsch is leading this effort, and he's
currently "thread checking", that is, making sure the parallel version
gives the same answer as serial. If the OpenMP version is used, then
automatically all cores or processors on a given machine are used in
the calc.
We hope eventually to use OpenMP and MPI in the same calculation.
OpenMP will speed the processing of a given mesh, while MPI still
allows us to split the job over multiple machines.