Running bamps on clusters
To schedule a bamps run on clusters using SLURM, use the script bashsbamps found in the exe directory.
Modify it to include your username, email and desired output location, as well as a job name and a launch script to execute when the job is launched.
A minimal launch script is bamps-single, also found in the exe directory.
<put an example invocation here>
Be aware that while commands in bashsbamps get executed immediately when your job is submitted,
commands in the launch script only get executed once the job actually starts.
This means that modifying the launch script while the job is in the queue will be reflected in the run.
ARA
To compile bamps on ARA, a few steps are needed,
since not all of the modules on ARA are configured correctly as of april 2020.
First, run
source /cluster/intel/parallel_studio_xe_2019/bin/psxevars.sh
This will load the necessary environment variables to enable MKL.
Once this has completed, you can load the modules
compiler/intel/2019-Update5 mpi/intel/2019-Update5
in that order.
Afterwards, to complete enabling Intel MPI, the following environment variables need to be set:
export MPI_ROOT='/cluster/intel/compilers_and_libraries_2019.5.281/linux/mpi' export MPI_HOME='/cluster/intel/compilers_and_libraries_2019.5.281/linux/mpi' export MPI_RUN='/cluster/intel/compilers_and_libraries_2019.5.281/linux/mpi/bin64/mpirun'
With this, bamps can be compiled.
To schedule a run, use bashsbamps as detailed above.
NOTE: You must also set up your environment before submitting a job, otherwise it will fail upon launch.