pde-solvers-cuda

Requirements

Currently, only UNIX based operating systems are supported. Moreover, you need to have the following installed on your machine:

• C++ compiler (e.g. gcc, clang)
• CMake (at least VERSION 3.18)
• CUDA (not strictly needed)
• git (not strictly needed, you could also download the repo as .zip file)
• hdf5
• netcdf
• python
• python: netCDF4

Getting started

You can build pde-solvers-cuda as follows

git clone https://github.com/LouisHurschler/pde-solvers-cuda.git  
cd pde-solvers-cuda
mkdir build
cmake -B build -DENABLE_CUDA={ON, OFF}

Note that ENABLE_CUDA is set OFF by default

Running the application

The easiest way to run the application is by following these steps:

Create a NetCDF Input File

Use the Python function in scripts/create_netcdf_input.py to create a NetCDF input file. You can refer to the example provided in scripts/brusselator_script.py.

Run the Solver

After you have generated the input file, run the solver using:

./build/run_from_netcdf <filename> [1]

where <filename> denotes the path of the file generated by your script relative to the directory you're running the application.

This will run on the GPU if you have a GPU available, you have compiled the application with -DENABLE_CUDA=ON and if you run it with the optional 1 added at the end.

Output

After the executable finishes, the results are stored in a NetCDF file stored at the location specified in your input script.

You can use the Python function in scripts/plot_output_file.py to plot the output or use the NetCDF output otherwise.