...

• The main requirement is a graphics card (GPU) supporting OpenGL 4.3. Note that since we only have Nvidia hardware, we haven't been able to test the code on AMD and Intel GPUs.
• The graphics card needs to have enough video memory to accomodate all data that should be visualized in a single image (e.g., if the size of your simulation data is on the order of 500 MB per variable and time step, and you would like to visualize u and v wind components, temperature and cloud cover simultaneously, you will need at least 5x500MB = 2,5GB (Met.3D also requires pressure) of video memory - more memory will increase performance during animation).
• The main system memory needs to be at least as large as the video memory. Data is cached on system memory, hence, the more, the better the performance.
• Met.3D makes some use of CPU multithreading to parallelize computations. Multi-code CPUs will hence also increase performance.

Linux

We provide a pre-compiled binary distribution for Linux as a conda package, which is the simplest way to use Met.3D. Note that this distribution corresponds to the latest tagged version in the "master" branch in the git repository.

If you would like to compile the software yourself (e.g., to use a different version than the "master" branch or to make changes by yourself), detailed installation instructions can be found here.

Windows

As noted above, Met.3D should compile and run on Windows. If you would like to contribute, help would be much appreciated.

...