[HW Accel Support]: How decode video with two same model Nvidia P2000 card?

[rasko1974]

Describe the problem you are having

Could you explain configuration file when we need decode one stream from one gpu another stream from second GPU? All GPU cards the same Nvidia cuda P2000 model.

Version

13.2

Frigate config file

go2rtc:
  ffmpeg:
  streams:
    cam1:
    - rtsp://admin:[email protected]:554/profile1#video=copy
    cam2:
    - rtsp://admin:[email protected]:8087/profile1#video=copy
#

cameras:
  cam1:
    ffmpeg:
      hwaccel_args: preset-nvidia-h265
      inputs:
      - path: rtsp://127.0.0.1:8554/cam1
        input_args: preset-rtsp-restream
        roles:
        - detect
        - record
      output_args:
        record: preset-record-generic-audio-aac
    detect:
      width: 2592
      height: 1520
      fps: 5
      max_disappeared: 25
    objects:
      filters:
        person:
          mask: 2592,0,2592,1327,1265,258,1075,278,1075,0
    snapshots:
      enabled: true
      clean_copy: true
      timestamp: false
      bounding_box: false
      crop: true
      height: 112
      retain:
        default: 3
    record:
      enabled: false
      retain:
        days: 0
      events:
        retain:
          default: 3
          mode: motion
        pre_capture: 3
        post_capture: 10
        #

docker-compose file or Docker CLI command

services:
  frigate:
    container_name: frigate
    runtime: nvidia
    environment:
      - NVIDIA_VISIBLE_DEVICES=1,5
      - CUDA_VISIBLE_DEVICES=1,0
      - NVIDIA_DRIVER_CAPABILITIES=compute,utility,video
    privileged: true # this may not be necessary for all setups
    restart: unless-stopped
    image: ghcr.io/blakeblackshear/frigate:stable
    shm_size: "4024mb" # update for your cameras based on calculation above
    devices:
      - /dev/dri:/dev/dri
      - /dev/apex_0    # Passes a PCIe Coral, follow driver instructions here https://coral.ai/docs/m2/get-started/#2a-on-linux
      - /dev/apex_1
      - /dev/apex_2
      - /dev/apex_3
      - /dev/apex_4
      - /dev/apex_5
      - /dev/apex_6
      - /dev/apex_7
      - /dev/dri/renderD128:/dev/dri/renderD128
      - /dev/dri/renderD129:/dev/dri/renderD129
#      - /dev/dri/card0
#      - /dev/dri/card1
#      - /dev/dri/card2
      - /dev/nvidia0:/dev/nvidia0
      - /dev/nvidia1:/dev/nvidia1
      - /dev/nvidia-caps:/dev/nvidia-caps
      - /dev/nvidiactl:/dev/nvidiactl
      - /dev/nvidia-modeset:/dev/nvidia-modeset
      - /dev/nvidia-uvm:/dev/nvidia-uvm
      - /dev/nvidia-uvm-tools:/dev/nvidia-uvm-tools
    deploy:
     resources:
       reservations:
         devices:
         - driver: nvidia
           device_ids: ['0','1']
#           count: 2  # number of GPUs
           capabilities: [gpu]
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /root/frigate/config:/config
      - /root/frigate/database/:/database
      - /root/frigate/media:/media/frigate
      - /root/frigate/config/a2z.png:/opt/frigate/frigate/images/birdseye.png:ro
      - type: tmpfs # Optional: 1GB of memory, reduces SSD/SD Card wear
        target: /tmp/cache
        tmpfs:
          size: 2000000000
    ports:
      - "5000:5000"
      - "8554:8554" # RTSP feeds
      - "8555:8555/tcp" # WebRTC over tcp
      - "8555:8555/udp" # WebRTC over udp
    environment:
      FRIGATE_RTSP_PASSWORD: "yildiz123"
    labels:
      traefik.enable: "true"

Relevant log output

no errors

FFprobe output from your camera

no errors

Operating system

Other Linux

Install method

Docker Compose

Network connection

Wired

Camera make and model

Dahua

Any other information that may be helpful

Thanks for a support

[NickM-27]

You can add a custom arg to global_args being -gpu 0 or -gpu 1 and that should choose the GPU to use

[rasko1974]

Thanks. Works