Docker(CUDA) Support For SHARK

[karthikkurella]

Introduction of Dockerfile with CUDA Environment

• CUDA Base Image: Utilizes nvidia/cuda:12.3.1-runtime-ubuntu22.04 as the base image to ensure compatibility with NVIDIA's CUDA.

• Package Updates and Installations:

  • Runs apt update and apt upgrade for the latest packages.

• Vulkan SDK Installation:

  • Adds the Vulkan SDK repository and installs the Vulkan SDK.

• Python 3.11 Setup:

  • Installs Python 3.11 and sets it as the default Python version.

• Virtual Environment for SHARK:

  • Sets up and ensures automatic activation of a virtual environment for SHARK, enhancing isolation and dependency management.

• Interactive Shell Configuration:

  • Configures Docker to start with an interactive bash shell, allowing for manual intervention and debugging within the container.

Running Instructions

To build and run the Docker container based on this Dockerfile, use the following commands:

1. Build the Docker Image:

   docker build -t your-image-name .
   

2. Run the image:

   docker run --gpus all -it  --rm -p 8080:8080 your-image-name

[dan-garvey]

Hey thanks for contributing. We're planning a pretty significant overhaul, so this may stop working in the near future, but I'm happy to commit it if anyone would find it useful.

I'm not very familiar with docker, so I'm unclear as to the reason to use a cuda docker as the base if you're installing vk?

[karthikkurella]

Thank you @dan-garvey, The reason is that docker:cuda image contains CUDA driver to interact with Nvidia GPU.