From ScientificComputing
Revision as of 09:25, 5 July 2021 by Sfux (talk | contribs)

Jump to: navigation, search

If you would like to run orca in parallel, then you need to load the corresponding open_mpi module. For instance, when using orca 3.0.3, then you would need to load the following modules:

module load orca/3.0.3 open_mpi/1.6.5

Mapping of Orca and OpenMPI versions:

Old software stack:

Orca version OpenMPI version Module load command
3.0.1 1.6.5 module load orca/3.0.1 open_mpi/1.6.5
3.0.3 1.6.5 module load orca/3.0.3 open_mpi/1.6.5
4.0.0 2.0.2 module load new orca/4.0.0 open_mpi/2.0.2 2.0.2 module load new orca/ open_mpi/2.0.2
4.0.1 2.0.2 module load new orca/4.0.1 open_mpi/2.0.2
4.1.0 2.1.1 module load new orca/4.1.0 open_mpi/2.1.1
4.1.1 2.1.1 module load new orca/4.1.1 open_mpi/2.1.1
4.2.0 2.1.1 module load new orca/4.2.0 open_mpi/2.1.1

New software stack:

Orca version OpenMPI version Module load command
4.2.1 4.0.2 module load new orca/4.2.1 openmpi/4.0.2
5.0.0 4.0.2 module load new orca/5.0.0 openmpi/4.0.2

Further more you also need to add a line to the orca input file to specify the number of cores that should be used. For a 4 core job, you would add

%pal nprocs 4

to your orca input file.

The example input file provided on this wiki page would then change to

# Updated input file for the goldchloride example:
! BP ECP{def2-TZVP,def2-TZVP/J} def2-tzvp def2-tzvp/j tightscf opt
%pal nprocs 4
* xyz -2 2
Au  0   0   0
Cl  2.5 0   0
Cl -2.5 0   0
Cl  0   2.5 0
Cl  0  -2.5 0

and the submission command would be

bsub -n 4 -W 1:00 -R "rusage[mem=2048]" "orca test1.inp > test1.out"

Please note that you also need to request 4 cores from the batch system, when submitting the job. It is not sufficient to just change the orca input file.