Enabling Hardware Encoding in Jellyfin with NVIDIA GPU

Problem

• My Docker Jellyfin running on an i3-9100F CPU is too slow in encoding movies.
• The server has an NVIDIA 1600 GPU, so I must utilize the power of this beast.

This quick journal summarizes the key points and steps to be done to use an NVIDIA GPU on a Dockerized Jellyfin server.

Solution

Install NVIDI Toolkit

• Follow this guide: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
• Update NVIDIA driver and follow the above guide

Modify Jellyfin docker-compose.yml file

  jellyfin:
    image: linuxserver/jellyfin:10.8.11
    container_name: jellyfin
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=UTC
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=compute,utility,video
    volumes:
      - jellyfin-config:/config
      - torrent-downloads:/data
    
    devices:
      - /dev/dri:/dev/dri
      - /dev/nvidiactl:/dev/nvidiactl
      - /dev/nvidia0:/dev/nvidia0
      - /dev/nvidia-modeset:/dev/nvidia-modeset
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              capabilities: [gpu]
    ports:
      - 2284:8096
      - 7359:7359/udp
      - 8920:8920
    restart: always
    networks:
      - media-stack-network

• Run command: ck$ sudo usermod -aG video $USER
• Restart the docker service
• Confirm if GPU had been detected in the docker environment by below command: $ docker exec -it jellyfin nvidia-smi

Enable Hardware Acceleration in Jellyfin Dashboard

That’s the key ingredient! Enjoy the new performance of Jellyfin