Computational Science Community Wiki

Running Fluent on Mace01

Overview

Prerequisites

Fluent uses SSH to start MPI-related processes. Therefore you will need to ensure you have promptless, passwordless SSH access across the Mace01 cluster to run parallel Fluent jobs. This is done using an SSH key and an appropriate known_hosts file.

New users will have the required SSH configuration set up for them automatically. Established users may not have the required configuration.

Fluent and SGE

Fluent works well with SGE, once suitable queues (and parallel environments) are configured. Furthermore, it is not necessary to use a qsub script; instead, simply give Fluent some SGE-related command-line arguments — see the examples given below.

Running Serial Jobs

To run serial Fluent batch jobs on Mace01, use the serial.q queue: at the command-line, simply enter

    /software/Fluent.Inc/bin/fluent 3d -g -pnet -ssh -sge -sgeq serial.q -i <journal-file>

where journal-file is your Fluent journal file, and Fluent will then handle the SGE queue submission. As output from the above command shows, this is actually equivalent to using the qsub script

    #!/bin/bash

    #$ -cwd 
    #$ -q serial.q 
    #$ -S /bin/bash

    export FLUENT_INC=/software/Fluent.Inc 

    /software/Fluent.Inc/bin/fluent 3d -g -sge -sgeq serial.q -i input.jou

You can check the progress of your job by using the qstat utility, in the usual way.

Running Parallel Jobs

To run parallel Fluent jobs on Mace01, use the parallel.q queue; use the fluent.pe SGE parallel environment. At the command-line, simply enter

  /software/Fluent.Inc/bin/fluent 3d -g -t8 -pnet -ssh -sge -sgeq parallel-R2.q -sgepe fluent.pe 8 -i input.jou
      # ...or use the "parallel-R5.q" queue...

adjusting the number of processors requested as you require, e.g., changing both -t4 and -sgepe fluent.pe 4 to -t8 and -sgepe fluent.pe 8, respectively.