Skip to content

README.md

Latest commit

 

History

History
65 lines (43 loc) · 1.82 KB

README.md

File metadata and controls

65 lines (43 loc) · 1.82 KB

DigiForests Docker Environment

This README provides instructions for setting up and using the Docker environment for the DigiForests project, which includes GPU support and a multi-stage build process.

Note: Host machine should support CUDA 11.8!

Prerequisites

  1. Install Docker on your system.

  2. Install NVIDIA Container Runtime: Follow the instructions in the NVIDIA Documentation.

  3. Verify GPU support in Docker:

    docker run --rm --gpus all nvidia/cuda:11.8.0-base-ubuntu22.04 nvidia-smi

    This should display your GPU information if everything is set up correctly.

Building the DigiForests Docker Image

The Dockerfile uses a 3-stage build process:

  1. Base CUDA and PyTorch setup
  2. MinkowskiEngine compilation
  3. DigiForests package installation

Note: The build process can take 30+ minutes, especially the MinkowskiEngine compilation.

Important Considerations

  • Ensure your host machine supports CUDA 11.8, as this is the base image which we build upon.
  • In the Dockerfile, adjust the TORCH_CUDA_ARCH_LIST environment variable to match your GPU's compute capability.

Build Command

docker build -t digiforests_devkit -f docker/Dockerfile .

Running the Container

Using Docker Run

docker run -it --rm --gpus all digiforests_devkit

Using Docker Compose

  1. Ensure your user ID and group ID are set in the environment:

    export UID=$(id -u)
export GID=$(id -g)

  2. Run the container:

    docker compose -f docker/compose.yaml run devkit

Customization

  • The compose.yaml file includes volume mounts for development. Adjust these as needed.
  • Uncomment the data volume mount in compose.yaml to access external data.