docker-compose.yml

x-gpu-enabled: &gpu-enabled
  devices:
    - driver: nvidia
      count: all
      capabilities:
        - gpu # Enables GPU access for the container.

x-gpu-disabled: &gpu-disabled
  devices: [] # Disables GPU access (default for systems without an NVIDIA GPU).

services:
  ebook2audiobookxtts:
    image: athomasson2/ebook2audiobook
    platform: linux/amd64
    tty: true
    stdin_open: true
    ports:
      - 7860:7860 # Maps container's port 7860 to the host's port 7860.
    command: python app.py
    deploy:
      resources:
        reservations:
          <<: *gpu-disabled # Use *gpu-enabled if you have an NVIDIA GPU.
          # --- CPU Memory (RAM) Reservation ---
          # memory: 4g   # Uncomment to reserve 4GB of system RAM (minimum required by the container).
          # --- GPU VRAM (Indirect Control) ---
          # devices:     # Uncomment and configure to limit GPU VRAM (requires GPU-enabled above).
          #   - driver: nvidia
          #     count: 1 # Use fractional GPU count (e.g., 0.5) to indirectly limit VRAM.
          #     capabilities:
          #       - gpu
        limits:
          # --- CPU Memory (RAM) Limit ---
          # memory: 4g   # Uncomment to set a 4GB upper limit on system RAM usage by the container.
    volumes: []
    # - ./:/home/user/app  # Uncomment to map the local directory to the container.

# Additional Notes:
# - "CPU Memory (RAM)" refers to the system RAM used by the container.
# - "GPU VRAM" refers to the graphics memory allocated to the container's GPU tasks.
# - To enable GPU VRAM limits, ensure the NVIDIA Docker runtime is installed and active.
# - The `memory` options (RAM) use units like 'm' (megabytes) or 'g' (gigabytes).
# - The `count` parameter for GPU limits controls how much of the GPU (and indirectly VRAM) is accessible.