You are here: Home Help/User Support Parallel Computing in the CAE Lab Cluster Job Example: Fluent

Cluster Job Example: Fluent

Fluent

Fluent is one of the only commercial programs licensed at TTU that can really take advantage of parallel environments. Unlike ANSYS and Abaqus, it can divide the solution work up among several systems and combine their individual solutions into one overall solution. Speed increases of multiple systems versus a single Pentium 4 Xeon CPU are shown below.

Fluent Speed Increase with Parallel Solutions

Sample input file (pipe_sf.txt)

rc pipe_sf.cas
rd pipe_sf.dat
file/autosave/data-frequency 2000
solve/it/10000
wc pipe_sf.cas
y
wd pipe_sf.dat
y
exit

This input file will read an already-saved case and data file, automatically save intermediate results every 2000 iterations, and solve up to 10000 iterations before saving the updated case and data files.

Sample qsub command file (fluent_demo.sh)

#!/bin/bash
# Simple multi-node Fluent job

# Request 4 CPUs, with >=1 CPU used per system
#PBS -l nodes=4:ppn=1

# Reserve 24 hours on selected CPUs
#PBS -l walltime=24:00:00

# Give the job a descriptive name for emails (name must start with a letter)
#PBS -N Pipe_steady_flow

# Email the address given below when the job starts, ends normally, or aborts
#PBS -m bea
#PBS -M myusername@tntech.edu

# Change to directory from which job was submitted
cd $PBS_O_WORKDIR

# set number of processors to run on
# (list of node names is in file $PBS_NODEFILE)
nprocs=`wc -l $PBS_NODEFILE | awk '{ print $1 }'`

# Run default version of Fluent in parallel across nodes in $PBS_NODEFILE
fluent 3ddp -t$nprocs -pnmpi -cnf=$PBS_NODEFILE -g -i pipe_sf.txt >& pipe_sf.out

This shell script will run a Fluent job on 4 CPUs selected by the queue scheduler, and will redirect everything that would have been printed on the screen to a .out file that can be monitored for convergence criteria.