Fluent Job Submission:
- For fluent there is no graphical user interface provided
- You must submit your job in batch mode using qsub
- Linux script is provided and you may change source file name and output file name
- You can change number of processors and number of nodes according to your need
your script begins here
#!/bin/sh
#name the program, your default output error file are Test_fluent.oJobid
#and Test_fluent.eJobid
#PBS -N Test_fluent
#follwing 2 lines ensures that you'll be notified by email when your job is done
#PBS -M your_user_id@tigermail.auburn.edu
#PBS -m e
#you are asking for 4 node 8 processor each, 32 processor as a total for 50hrs
#after 50 hours your job will be killed
#PBS -l nodes=4:ppn=8,walltime=50:00:00
#your directory path can be obtained by pwd
#PBS -d /home/your_dirctory_path/
#loading variables do not change this
export PATH=/export/apps/ansys_inc/v130/fluent/bin:$PATH
export FLUENT_ARCH=lnamd64
export NO_LOCAL=1
#writing mpd_nodes to boot mpd in these nodes and conf_file to select processors
`sort -u $PBS_NODEFILE > mpd_nodes`
#assignin nhosts variable to number of nodes
nhosts=`cat mpd_nodes | wc -l`
#generating conf_file to select processors
`sort $PBS_NODEFILE > conf_file`
proc=`cat conf_file | wc -l`
#printing initial timestamp
date > fluent.out
#printing which host performed computation
/bin/hostname >> fluent.out
#booting mpd on the selected nodes
#mpdboot -n $nhosts -v -f mpd_nodes
#if you want to use intel as communication port the use -mpi=intel, also uncomment mpdboot and mpdallexit, by default you can use hp and leave it unchanged
#executing fluent with 64 processor providing mpich communication connector
#and conf_file as config file and i=source_file >> writing_to_output_file
fluent 3ddp -g -t$proc -mpi=hp -cnf=conf_file -i fluent.jou >> fluent.out
#end time stamp
date >> fluent.out
#stop mpd
#mpdallexit
end of your script
- Create a directory using "mkdir testing"
- Go to that directory using "cd testing"
- Write a script file in vi editor similar to these
- Make sure it is executable by "chmod 744 file_name.sh"
- Submit the job using "qsub ./file_name.sh"
- Give a "showq" to see if your job is running
- You can terminate the session using "exit" but your job will be running and you can check back the results after 1-2 days whenever your job is done (you'll also be notified by email)
Running two or more fluent job simultaneously (if you are using -mpi=intel):
- Comment following 2 lines (in your script)using # sign in front of them
`sort -u $PBS_NODEFILE > mpd_nodes`
nhosts=`cat mpd_nodes | wc -l`
- Comment
mpdboot -n $nhosts -v -f mpd_nodes
- Comment last line
mpdallexit
- Using vi editor write a mpd_nodes file that has following 4 lines and save it
compute-1
compute-2
compute-3
compute-4
- ssh to compute-1 using
ssh compute-1
- To start mpd manually execute the following command
mpdboot -n 4 -v -f mpd_nodes
- Type "exit" to exit from compute-1
- Submit your jobs using qsub ./script.sh
- When all of your job is done, ssh compute-1 and type "mpdallexit" to stop mpd manually
For details you may want to visit
http://www.ansys.com/products/fluid-dynamics/fluent/\
FIT websit
http://my.fit.edu/itresources/manuals/fluent6.3/help/index.htm