GPU Dependency in Docker Image Run Command Prevents Execution on Non-GPU Systems

[sudo-deep]

Environment 🖥️

• OS Version: Ubuntu 20.04

• Do you have a dedicated GPU ? [No]

• Are you running this project natively or using Docker ? [Docker]

• Is the issue present in the main branch or any of the development branches ? [Development branch]

Description 📖

• Expected behavior: Docker image runs successfully
• Actual behavior: Docker daemon throws an error

Steps to reproduce 👀

1. Run run_devel.sh
2. Select Option 2 (Ubuntu 22.04)
3. Make sure you either test it on a non-gpu machine or uninstall nvidia container toolkit

Output 💥

docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]]

Additional Note 🚧

To address the docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]] issue when running without GPU, a conditional statement needs to be added in the run_devel.sh file.

Suggested solution:

# Add conditional check for GPU availability before using --gpus flag
if [ -z "$(command -v nvidia-smi)" ]; then
  # Handle case when no GPU is present
  echo "No GPU detected. Running without --gpus flag."
  # Add your additional commands here for non-GPU execution
else
  # GPU is present, use --gpus all
  echo "GPU detected. Running with --gpus all."
  # Add your original run command here with --gpus all
fi