Starting LAMMPS-Jobs

Lammps may be used in combination with different MPI versions but we recommend on our system the usage of Intel MPI. It proved to be the version running in shortest time when compared with others.

The submission of jobs for Lammps is embedded in sbatch-lammps which should contain all possible options. Please inform us in case something is missing.

To submit a job:

sbatch-lammps [--verbose] [--arch <XEON_SP_6126,..>] [--mpi <intel,openmpi,platform>] \ 
[--node <n>] [--project <p>] [--queue <q>] [--jobname <j>] \
[--walltime <t>]  [--output <o>] [--version <v>] \
-- Parameter/Options for Lammps

Optional options are enclosed in []. Some options require additional parameters indicated by a < > pair. The \ at the end of a line indicates a continuation in next line. Parallelisation is purely based on MPI and the recommended combination --mpi intel is default. Only complete compute nodes are selected, thus --node 2 means 2 compute nodes of type XEON_SP_6126 using 24 cores each.