ORCA

ORCA is a modern electronic structure program package written by F. Neese, with contributions from many current and former coworkers and several collaborating groups. The binaries of ORCA are available free of charge for academic users for a variety of platforms.

Installed version

  • 3.0.2 sequential version
  • 3.0.2 parallel version with MPI (Open MPI)

Sequential version

Example of SGE script to use:

orca_seq.sge
#!/bin/bash
 
#$ -q all.q
#$ -N orca
#$ -o $JOB_NAME.$JOB_ID.out
#$ -l h_vmem=4G
 
module load orca/3.0.2
 
orca MyMol.inp > MyMol.out

Parallel version

Running ORCA in parallel

Note that when running ORCA in parallel, ORCA should NOT be started with mpirun: e.g. mpirun -np 4 orca etc. ORCA takes care of communicating with the OpenMPI interface on its own when needed (as long as the OpenMPI binaries and libraries are made available).

Use the !PalX keyword in the inputfile to tell ORCA to start multiple processes. E.g. to start a 4-process job, the input file might look like this: 

! BP86 def2-SVP def2-SVP/J Opt PAL4

or using block input: 

! BP86 def2-SVP def2-SVP/J Opt
%pal
nprocs 4
end

The inputfile can then be run, calling ORCA by a full path (Important!)

Example of SGE script

  • When using parallel.q queue the allowed number of processors to use are: 16,32 or 64
  • Make sure that requested slots is equal to the number of processors in the input file

orca_MPI.sge
#!/bin/bash
 
#$ -q parallel.q
#$ -pe mpi 16
 
#$ -N orca_mpi
#$ -o $JOB_NAME.$JOB_ID.out
#$ -l h_vmem=4G
 
 
module load orca/3.0.2
 
## save the cruuent directory
CURRENT_DIR=`pwd`
 
## cp input to TMPDIR
cp input.txt $TMPDIR
 
## go to TMPDIR
cd $TMPDIR
 
## Run it
${ORCA_HOME}/orca input.txt > $CURRENT_DIR/output.out
 
## Copy back results to CURRENT_DIR
cp input.gbw $CURRENT_DIR

Links