walnut-REE / RTXAdvect

Particle tracking code accelerated by GPU RTX cores

rtxAdvect - RTX-accelerated Tet-Mesh Particle Advection

Building

==================================================================

Dependencies

Optix 7, CUDA 10, and a recent NVIDIA driver.

Checkout w/ Submodule

Mind this project uses a git submodule, so idealy clone with --recursive flag:

git clone --recursive [e-mail]:ingowald/rtxTetAdvect

If you did close without this flag, you can afterwards also do a

git submodule init
git submodule update

Building then via cmake and your favorite compiler toolchain.

Building on Linux

For linux, I assume that nvcc is in the default path, and that there is environment variable names 'OptiX_INSTALL_DIR` that points to the optix SDK install directory

cd rtxTetAdvect
mkdir bin
cd bin
cmake ..
make

If you have any missing dependencies ccmake will tell you... just fix and re-run cmake

Building on Windows

For windows, install Visual Studio (2017+), CUDA (10.1+) and Optix 7 first. Then run cmake-gui and click configure with following paramters: CUDA_HOST_COMPILER = C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.26.28801/bin/Hostx64/x64/cl.exe

OptiX_INCLUDE = C:/ProgramData/NVIDIA Corporation/OptiX SDK 7.0.0/include

Usage

==================================================================

cudaParticleAdvection.exe --num-particles 5000 --num-steps 50000 --input_mesh verts.dat cells.dat --input_tet_velocity_field solutions_cell.dat -dt 2e-3 --seeding-box  -1.5 -12.0 -1.5 50.5 -8.5 50.5 --save-streamline-to-vtk Streamline.vtk

License

==================================================================

Apache Licence 2.0 - pretty much do with it what you like; no warranties.

Collaborators

==================================================================

UofU & NVidia

• Ingo Wald
• Will Usher
• Nate Morrical

LSU:

• Bin Wang