SU2 is an open source CFD solver developed by the School of Aeronautics and Astronautics at Stanford University. It is based on C++ and Python and is similar to OpenFOAM, but does not support polyhedral meshes. Compared with OpenFOAM, SU2 has more advantages in solving high-speed compressible flows.
Download SU2 source code:
mkdir $HOME/su2code && cd $HOME/su2code
# Only clone the latest commit version to speed up downloading
git clone https://github.com/su2code/SU2.git --depth=1
Define environment variables and create a new configuration file su2.env:
touch $HOME/su2code/su2.env
chmod +x $HOME/su2code/su2.env
vim $HOME/su2code/su2.env
Add the following content to the su2.env file, save and exit:
#!/bin/sh
# Define SU2 environment variables
export SU2_RUN=$HOME/su2code/bin # su2 installation path after compilation
export SU2_HOME=$HOME/su2code/SU2 # su2 source code folder path
export PATH=$PATH:$SU2_RUN
export PYTHONPATH=$SU2_RUN:$PYTHONPATH
The compiler configuration file meson_options.txt is located in the SU2 source code folder. Adjust the compilation options according to your needs:
vim $HOME/su2code/SU2/meson_options.txt
Here, enable mpi and blas support and modify the values of the following two lines:
option('with-mpi', type : 'feature', value : 'enabled', description: 'enable MPI support')
option('enable-openblas', type : 'boolean', value : true, description: 'enable BLAS and LAPACK support via OpenBLAS')
If it is an Intel machine, it is recommended to enable mkl support.
The default supported blas library is openblas. You need to download the openblas library first:
sudo apt install libopenblas-dev -y
Enter the downloaded source code directory and run the compiler
# Load environment variables
source $HOME/su2code/su2.env
# Enter the source code folder
cd $SU2_HOME
# Configure the compiler and generate ninja build files
# The configuration process will automatically download external dependencies from git
# It takes a lot of time...
./meson.py build --prefix=$SU2_RUN/..
# Start compiling and installing
./ninja -C build install
Verify that the installation was successful:
SU2_CFD --help
If the installation is successful, the software version number and help information will be output.
Write the slurm calculation script for SU2:
#!/bin/bash
#SBATCH --job-name=su2_test
#SBATCH --partition=debug
#SBATCH --output=%j.out
#SBATCH --error=%j.err
#SBATCH -N 1
#SBATCH --ntasks-per-node=32
cd $SLURM_SUBMIT_DIR
source ${HOME}/su2code/su2.env # The absolute path should be filled in
export CFG_FILE=`find . -name "*.cfg"`
export MACHINEFILE=$SLURM_JOBID.nodes
scontrol show hostnames $SLURM_JOB_NODELIST > $MACHINEFILE
mpiexec -np $SLURM_NPROCS --machinefile $MACHINEFILE SU2_CFD $CFG_FILE
Put the script file, SU2’s cfg file and mesh into the same folder, and submit the calculation task through the sbatch command.