MPI Run

The code can be run in Message Passing Interface (MPI). The code is made in such a way that it can run:

in the REPL, selecting SegregatedVMSSolver.solve(simcase,with_debug). It is useful to debug the code and visualize errors.
in MPI, selecting SegregatedVMSSolver.solve(simcase,with_mpi)

When running in MPI the code cannot be easily executed in the REPL. Instead, one has to run them from a terminal using the mpiexecjl script as provided by MPI.jl.

For example with the command:

mpiexecjl -n 4 julia --project=. run_simulation.jl

It is possible to launch the tests changing the backend.

Run in a Cluster

It is possible to run a simulation in a cluster, creating a suitable bash file, example the following run_sim.sh file can be used. The user has to specify where julia is installed and to activate a suitable project where there is SegregatedVMSSolver in the dependancies. The run_simulation.jl is the julia file of the simulation itself. Be sure that the number of processors specified in simulation is corresponding to the number of processors used for running the simulation in the cluster.

#!/bin/sh
export PATH=$HOME/julia-1.11.1/bin/:$PATH

srun julia --project=../../../ -O3 --check-bounds=no -L run_simulation.jl

This is a command example to launch the simulation on a single node (-N1) a cluster using slurm, using 80 cores -n80, specifing the partition of 80 cores -p80CORE and a total time of 400 hours -t 400:0:0.

sbatch -t 400:0:0 -N 1 -n80 -p80CORE run_sim.sh