Skip to content

Installation

Environment setup

We recommend installing flooder in a clean new Anaconda environment as described below:

1
2
3
conda create -n flooder-env python=3.9 -y
conda activate flooder-env
conda install pip git -y

flooder is available on PyPi (here) and can be installed via:

1
pip install flooder

Development installation

In case you want to contribute to flooder, clone the GitHub repo and run

1
2
3
git clone https://github.com/plus-rkwitt/flooder
cd flooder
pip install -e .

In case you do not want to install anything, you can also execute examples or tests from within the checked-out folder by specifying PYTHONPATHas

1
2
3
git clone https://github.com/plus-rkwitt/flooder
cd flooder
PYTHONPATH=. python examples/example_01_cheese_3d.py

GPU requirements

Our implementation relies heavily on custom Triton kernels (although we support CPU computation as well) for maximum performance. According to the official Triton compatibility page, you need a NVIDIA GPU with compute capabilty (check
here) of at least 8.0 (e.g., GTX 3080, etc.).