Installation Guide

Installation Guide#

The xsnow package is designed to install smoothly via pip.
It supports both standard CPU-based analysis and optional GPU acceleration (via CuPy) for performance-critical workloads.

Warning

The xsnow package is not yet published on PyPI.
To install it, you currently need to clone the repository from GitLab and install locally:

git clone https://gitlab.com/avacollabra/postprocessing/xsnow
cd xsnow
# pick an install command from below

Standard Installation (CPU Only)#

For most users, the standard installation is sufficient.
This installs the core package and essential dependencies such as xarray and pandas. Consider installing it into a dedicated virtual or conda environment.

pip install xsnow

Once installed, verify it works:

python -c "import xsnow; print(xsnow.__version__)"

Installation with GPU Support#

If you have a compatible NVIDIA GPU, xsnow can leverage CuPy for substantial speedups in heavy computations.

Prerequisites#

  • An NVIDIA GPU with a recent CUDA driver installed

  • You do not need the full CUDA toolkit; the driver is enough

  • Linux is the most widely tested platform for CuPy prebuilt wheels (Windows/macOS have limited GPU support)

Install#

pip install xsnow[gpu]

This command installs the base package plus a matching CuPy build. When available, xsnow methods will automatically detect and offload work to the GPU.

Tip: if no GPU is available, the package gracefully falls back to CPU execution.

Developer Installation#

If you want to contribute to xsnow, use an editable install. This way, any local changes to the source code are reflected immediately without reinstallation.

git clone https://gitlab.com/avacollabra/postprocessing/xsnow
cd xsnow
pip install -e .[gpu,testing,dev]

The dev extras include tooling such as test dependencies, jupyter, ipython, matplotlib. You can now run the test suite to confirm everything works:

pytest
Contents