Motivation for this change

The motivation for this change was:

• Shared packaging code between docker, podman, and nvidia container runtime related packages
• Allow multiple container runtimes to have GPU support enabled

Things done

• Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS linux)
• Built on platform(s)
  • NixOS
  • macOS
  • other Linux distributions
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all pkgs that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• Determined the impact on package closure size (by running nix path-info -S before and after)
• Ensured that relevant documentation is up to date
• Fits CONTRIBUTING.md.

Additionally, I've tested both docker and podman with the nvidia container runtime in the following ways on a machine with an NVidia RTX 2080 Super:

1. docker: nvidia-smi using the --runtime nvidia flag

   docker run --runtime nvidia --rm nvidia/cuda nvidia-smi --list-gpus
   GPU 0: GeForce RTX 2080 SUPER (UUID: GPU-16efc9cd-6886-5c12-52f8-2bef887747d9)
   

2. docker: nvidia-smi using the --gpus all flag

   docker run --gpus all --rm nvidia/cuda nvidia-smi --list-gpus
   GPU 0: GeForce RTX 2080 SUPER (UUID: GPU-16efc9cd-6886-5c12-52f8-2bef887747d9)
   

3. docker: TensorFlow list_physical_devices using the --runtime nvidia flag

   docker run -e TF_CPP_MIN_LOG_LEVEL=1 -e LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu --runtime nvidia --rm tensorflow/tensorflow:latest-gpu python -c 'import tensorflow as tf; print(tf.config.list_physical_devices("GPU"))'
   [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]
   

4. docker: TensorFlow list_physical_devices using the --gpus all flag

   docker run -e TF_CPP_MIN_LOG_LEVEL=1 -e LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu --gpus all --rm tensorflow/tensorflow:latest-gpu python -c 'import tensorflow as tf; print(tf.config.list_physical_devices("GPU"))'
   [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]
   

5. podman: nvidia-smi using the --runtime nvidia flag

   podman run --runtime nvidia --rm nvidia/cuda nvidia-smi --list-gpus
   GPU 0: GeForce RTX 2080 SUPER (UUID: GPU-16efc9cd-6886-5c12-52f8-2bef887747d9)
   

6. podman: TensorFlow list_physical_devices using the --runtime nvidia flag

   podman run -e TF_CPP_MIN_LOG_LEVEL=1 -e LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu --runtime nvidia --rm tensorflow/tensorflow:latest-gpu python -c 'import tensorflow as tf; print(tf.config.list_physical_devices("GPU"))'
   [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]