NOTICE: This is a work in progress, subject to change.

Description

GPU-accelerated particle tracking for OpenFOAM. You can use this repository without compilation. Jump to Running with Docker for more details, or follow the instructions below for a native installation.

Credits

This repository is built upon the following repositories:

• RTXAdvect
• tetMeshQueries
• OpenFOAM

Kudos to the authors!

Requirements

• NVIDIA CUDA Toolkit 10.1
• OptiX 7.0
• OpenFOAM v2106

Installing dependencies

Check the NVIDIA CUDA Installation Guide for Linux for installing the NVIDIA CUDA Toolkit 10.1. Note that the maximum supported GCC version is 8.

Download OptiX 7.0 for Linux. You must be a member of the NVIDIA Developer Program to download OptiX.

The following command will install OptiX in a local folder:

./NVIDIA-OptiX-SDK-7.0.0-linux64.sh --include-subdir --skip-license

Only OpenFOAM v2106 is currently supported. Compile and install OpenFOAM v2106 environment before continuing.

Building

Set the environment variables in etc/bashrc pointing to your installation paths, for example:

export RTX=false
export OptiX_INSTALL_DIR=${HOME}/cudaParticlesFoam/NVIDIA-OptiX-SDK-7.0.0-linux64
export CUDA_HOME=/usr/local/cuda-10.1

If your graphics card has ray tracing cores set RTX=true for additional hardware acceleration [1].

Set the environment for this repository with:

source etc/bashrc

Load OpenFOAM's environment:

source ${HOME}/OpenFOAM/OpenFOAM-v2106/etc/bashrc

Run the following commnad for building the cudaParticleAdvection library:

make lib

And finally, build the OpenFOAM solvers:

make applications

If you change the variables on etc/bashrc do a make clean and repeat the process before building lib and applications again.

Running

Running with native installation

Go to one of the tutorials:

cd tutorials/incompressible/cudaParticlesUncoupledFoam/pitzDaily

Finally, run the tutorial with:

./Allrun

Running with Docker

You need to first configure your machine for using GPUs within Docker containers. Follow this link for instructions.

Set up the environment with:

source etc/bashrc

Go to one of the tutorials:

cd tutorials/incompressible/cudaParticlesUncoupledFoam/pitzDaily

Finally, run the tutorial with:

runWithDocker ./Allrun

You will see the results and logs. Running containers can be checked with docker ps. Containers can be killed with docker kill CONTAINER_ID.

References

• [1] Wang, Bin, et al. "An GPU-accelerated particle tracking method for Eulerian–Lagrangian simulations using hardware ray tracing cores." Computer Physics Communications 271 (2022): 108221.

Licenses

The whole project is licensed under the GNU General Public License v3.0 except the code inside the third_party directory which is licensed under Apache Licence 2.0.

OPENFOAM® is a registered trade mark of OpenCFD Limited, producer and distributor of the OpenFOAM software via www.openfoam.com.