Not An Issue, looking for advice.

[marcosvanella]

Dear Sherry Li,

I'm shopping for a direct solver for one or several Poisson equations, which can be used in a single MPI process or in parallel. These need to be solved within the time-stepping of a CFD code (https://github.com/firemodels/fds), so potentially thousands of times in total per simulation.
We have been using in our solver the Intel MKL PARDISO and SPARSE_CLUSTER_SOLVERs which have proved efficient but we are currently having issues are they are not provided for ARM systems (Macs and servers).

I was wondering if you could provide us with some guidance as to, if superlu_dist could be used to replace these. Our code is written in fortran, it is parallel through mpi and openmp. We can have several poisson problems per MPI process or several mpi-processes, and the discretization of each leads to a poisitive difinite or indefinite symmetric matrix.

I'm looking forward to hear your thoughts whenever you have time.
Best Regards,
Marcos