Abaqus

How to Run Abaqus on Mesocentre Cluster.

  • Only DMA researchers are allowed to use Abaqus on the Mesocentre
  • The license server is hosted at Femto-ST. Make sure you have access to it.

Interactive/Graphical mode

Objective:

  • Run the graphical interface a.k.a CAO.
  • Post/pre processing
  • Small Jobs run

We will use mesoshared.univ-fcomte.fr. An interactive node shared by all users and not managed by SGE

  1. Connect to mesoshared.univ-fcomte.fr using x2go as explained here.
  2. Once connected, open terminal (Top menu: Applications→System Tools> Terminal
  3. Change directory to WORK: cd WORK (remember this is the Work space)
  4. Load abaqus module 
    $ module load abaqus
  5. Run CAE 
    $ abq2017 cae -mesa

    This will open a Grapical window of abaqus/CAE.

On can also run Abaqus batch mode (command line) just for testing (for long computation thanks to use SGE, see bellow). For example: 

$ module load intel # if we use ifort 
$ module load abaqus
$ abq2017 job=nom_job interactive # for short job and testing purpose

Batch mode

Objective:

  • Run Abaqus solver on bacth mode using SGE
  • Long run
  • Multiprocessing run (SMP)

Here is an example of SGE script to adapt: 

#!/bin/bash
 
#$ -o $JOB_NAME.o$JOB_ID
 
#$ -N job_abaqus_batch  # modify me
 
#$ -q all.q
 
#$ -pe openmp 8  # modify me MAX=16
 
#$ -l h_vmem=4G  # modify me
 
module load abaqus
module load intel # if we use ifort
job=myjob  # modify me
abq2017  cpus=$NSLOTS job=$job interactive

  • This script need to be saved in WORK directory
  • It run Abaqus in one node using 8 cores and 8x4G of memory.

Submit the script by: 

$ qsub script.sge

Check the execution by: 

$ qstat

Abaqus only has 52 tokens. If the job does not find any tokens it ends up with errors. We are working to optimise this: queuing jobs until the tokens are available.