The image is build on the official CuPy image, which comes with Python 3.10.12, numpy 1.26.4, scipy 1.13.0. Additionally, the following repos & packages are installed:
- git
- astropy
- matplotlib
- tomlkit
- jupyter
- https://github.com/spacetelescope/poppy
- https://github.com/jmeyers314/batoid
- https://docs.hcipy.org/0.4.0/
- https://docs.ray.io/en/latest/ray-overview/installation.html#installation
Can be built for multiple architectures with buildx. Add a version number tag and, if it's backwards compatible, the latest tag as well. For example:
docker buildx build --platform linux/amd64,linux/arm64 --push -t pearlhub/analysis:vx -t pearlhub/analysis:latest .
- open the Command Palette (Shift+Cmd+P in Mac or Shift+Ctrl+P in Windows)
- type 'Dev Containers: Reopen in Container'
- wait (it could take a few minutes)
docker run --gpus all -it cupy/cupy