CUDA Raycasting Demo

I started this as a little learner project in order to teach myself some skills that pushed a little bit past what was available in the beginner tutorials:

• Programming using CUDA/C++.
• Configure a CMake Project to create both libraries and executables.
• Create Python bindings for an existing library.

Building C++

In order to build this library yourself, you will have to download and unzip libtorch from https://pytorch.org/get-started/locally/. Once unzipped, move the libtorch/ directory to a new folder called external, which should be at the top level of this repo. From there you should be ready to go using CMake

1. mkdir build
2. cd build
3. cmake ..
4. cmake --build . --config <Debug or Release>

Running C++ Test

Simply search the generated build/test/<Debug or Release>/ and run cuda_test.exe from the command line.

Building Python Bindings

For this you will need to have the python torch library installed in your environment in addition to libtorch mentioned in the last section Building C++

1. Go to cuda_raycast/python/setup.py and edit the directories under include_dirs and library_dirs to match your installation of torch.
2. On the command line, navigate to cuda_raycast/python
3. Run python setup.py install.

Running Python Test

On the commandline run python cuda_raycast/python/ray.py.