[amitkamboj]

Hi!

I need to run shape_optimization.py on cluster using pbspro job scheduler.

I am interested in running the SU2 problem in parallel. For a flow solving problem using SU2_CFD, I am able to do using the following pbspro script file,

(# indicates the comments, #PBS are read by pbspro)

#! /bin/csh
#######PBS -l walltime=48:00:00
#PBS -N nca6
#PBS -q workq
#PBS -l select=1:ncpus=2:mpiprocs=2
#PBS -l place=scatter:excl
#PBS -V

# Go to the directory from which you submitted the job
# to split the domain
# mpirun -n 2 /local/bin/SU2_PRT inv_NACA0012.cfg

cd $PBS_O_WORKDIR

cat $PBS_NODEFILE > pbsnodes

mpirun --machinefile pbsnodes -n 02 /local/bin/SU2_CFD inv_NACA0012.cfg > /dev/null

# finally..
# scp restart_flow.dat solution_flow.dat
# SU2_SOL inv_NACA0012.cfg

Also, I am able to run the shape_optimization.py through,

shape_optimization.py -f rotating_NACA0012.cfg -n 4

The process works perfectly fine. But when the problem size increases I need to submit the job on more than 1 node using the scheduler, which I am not able to do.

So I want to create a similar script file so that I can submit shape_optimization.py on cluster. Which I am not able to do and need help with.

Would be grateful for any help you can provide.

Thanks in Advance!

[hlk]

Originally Posted by amitkamboj

Hi!

I need to run shape_optimization.py on cluster using pbspro job scheduler.

I am interested in running the SU2 problem in parallel. For a flow solving problem using SU2_CFD, I am able to do using the following pbspro script file,

(# indicates the comments, #PBS are read by pbspro)

#! /bin/csh
#######PBS -l walltime=48:00:00
#PBS -N nca6
#PBS -q workq
#PBS -l select=1:ncpus=2:mpiprocs=2
#PBS -l place=scatter:excl
#PBS -V

# Go to the directory from which you submitted the job
# to split the domain
# mpirun -n 2 /local/bin/SU2_PRT inv_NACA0012.cfg

cd $PBS_O_WORKDIR

cat $PBS_NODEFILE > pbsnodes

mpirun --machinefile pbsnodes -n 02 /local/bin/SU2_CFD inv_NACA0012.cfg > /dev/null

# finally..
# scp restart_flow.dat solution_flow.dat
# SU2_SOL inv_NACA0012.cfg

Also, I am able to run the shape_optimization.py through,

shape_optimization.py -f rotating_NACA0012.cfg -n 4

The process works perfectly fine. But when the problem size increases I need to submit the job on more than 1 node using the scheduler, which I am not able to do.

So I want to create a similar script file so that I can submit shape_optimization.py on cluster. Which I am not able to do and need help with.

Would be grateful for any help you can provide.

Thanks in Advance!

Thanks for your question.
I think this thread includes a lot of notes on using PBS with SU2
http://www.cfd-online.com/Forums/su2...using-pbs.html

[economon]

Hi,

I might also add that you can directly modify the command line call for parallel runs (to add machine files or change the mpiexec call directly) in the file interface.py that will be located in the $YOUR_SU2_INSTALL/SU2/run/ directory.

Hope this helps,
Tom