Abaqus software has options for running calculations on more than one CPU.

The following is a sample PBS script to achieve parallelism with abaqus.

For better performance it is recommend that local disk is used for all file I/O on the system (storm or cyclone) that is computing your job.

It is important that you do the following three things in your PBS script:

1. request multiple CPUs in the options provided to PBS (eg. ncpus=2)
   
2. request the same number of CPUs  when invoking abaqus (cpus=2)
3. select the thread model for parallelism (mp_mode=threads)

#!/bin/sh
#
########### Send email when job is terminated or aborted
#PBS -m ae
#
########### Email to the following address
#PBS -M SOMEONE@uq.edu.au
#
########### Job name
#PBS -N Job_001
#
########### Queue
#PBS -q ia64
#
########### Maximum amount of time it will run for (Hours:Minutes: Seconds)
#PBS -l walltime=2:00:00
#
########### ncpus = Number of cpu required MAKE SURE THIS MATCHES WHAT ABAQUS USES
########### cput = Cpu time require, simulation more than this will be terminated (Hours:Minutes:Seconds)
########### mem = Memory required
#PBS -l ncpus=2,cput=2:00:00,mem=2gb
#
########### The working directory
cd $PBS_O_WORKDIR  

SCRATCHDIR=/scratch/SOMEONE/Job_001
#Ensure that the scratch directory exists and create if necessary
if [ ! -d $SCRATCHDIR ]; then
  mkdir -p $SCRATCHDIR
fi
 

#Copy input file to the scratch space
cp $PBS_O_WORKDIR/test.inp $SCRATCHDIR

#Change into the scratch directory

cd $SCRATCHDIR

############ Execute command for abaqus
/HPC/apps/Abaqus/Commands/abq655 job=test interactive cpus=2 mp_mode=threads memory=2048mb scratch=$SCRATCHDIR

#Copy output files from the scratch space

cp $SCRATCHDIR/test.* $PBS_O_WORKDIR

#FOOTNOTES
#
#
# Execution procedure for ABAQUS/Standard and ABAQUS/Explicit
#
# abaqus job=job-name  [analysis | datacheck | parametercheck | continue |
#                        convert={select | odb | state | all} | recover |
#                        syntaxcheck | information={environment | local |
#                        memory | release | status | support | system | all}]
#                      [input=input-file]
#                      [user={source-file | object-file}]
#                      [oldjob=oldjob-name]
#                      [fil={append | new}]
#                      [globalmodel={results file-name |
#                        output database file-name}]
#                      [cpus=number-of-cpus]
#                      [parallel={domain | loop]
#                      [domains=number-of-domains]
#                      [mp_mode={mpi | threads}]
#                      [standard_parallel={all | solver}]
#                      [memory=memory-size]
#                      interactive | background | queue=[queue-name]
#                        [after=time]
#                      [double]
#                      [scratch=scratch-dir]
##                      [output_precision={single | full}]