[HW Accel Support]: I have a 2060 nvidia installed and recognized but not being leveraged

[vectorr66]

Describe the problem you are having

My GPU usage is around 1 percent. My CPU usage is being pinned. I have frigate running in a docker from the unraid apps.

Version

version: 0.15-1

Frigate config file

mqtt:
  enabled: true
  host: 10.0.0.5
  user: mqtt_user
  password:
  port: 1883
  topic_prefix: frigate
  client_id: frigate
  stats_interval: 30
  
#detectors: 
#   tensorrt:
#     type: tensorrt
#     device: 0 #This is the default, select the first GPU
  
detectors:
  coral:
    type: edgetpu
    device: usb
    
ffmpeg:
  hwaccel_args: preset-nvidia-h264
  output_args:
    record: preset-record-generic-audio-aac

tls:
  enabled: false

# Include all cameras by default in Birdseye view
#birdseye:
#  enabled: true
#  mode: continuous

audio:
  enabled: true
  max_not_heard: 30
  min_volume: 500

#genai:
#  enabled: true
#  provider: gemini
#  api_key: AIzaSyDU4T5oMG9CKv8pqbIYaYR1xgZTdZrcNoQ
#  model: gemini-1.5-flash

go2rtc:
  streams:
    porch:
      - rtsp://@10.0.0.45:554/cam/realmonitor?channel=1&subtype=2
    office:
      - rtsp://@10.0.0.49:554/cam/realmonitor?channel=1&subtype=2
    office2:
      - rtsp://@10.0.0.23:554/cam/realmonitor?channel=1&subtype=2

cameras:
  porch:
    ffmpeg:
      inputs:
        - path: rtsp://@10.0.0.45:554/stream1
          input_args: preset-rtsp-restream
          roles:
            - record
            - audio
        - path: rtsp://@10.0.0.45:554/stream2
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
    motion:
      threshold: 30
    objects:
      track:
        - person
        - car
        - dog
    record:
      enabled: true
      retain:
        days: 3
      alerts:
        retain:
          days: 30
      detections:
        retain:
          days: 30
    snapshots:
      enabled: true
      retain:
        default: 2
    audio:
      enabled: true
      listen:
        - bark
        - fire_alarm
        - scream
        - speech
        - yell

  office:
    ffmpeg:
      inputs:
        - path: rtsp://@10.0.0.49:554/stream1
          input_args: preset-rtsp-restream
          roles:
            - record
            - audio
        - path: rtsp://@10.0.0.49:554/stream2
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
    motion:
      threshold: 30
    objects:
      track:
        - person
    record:
      enabled: true
      retain:
        days: 3
      alerts:
        retain:
          days: 30
      detections:
        retain:
          days: 30
    snapshots:
      enabled: true
      retain:
        default: 2
    audio:
      enabled: true
      listen:
        - bark
        - fire_alarm
        - scream
        - speech
        - yell
        
  office2:
    ffmpeg:
      inputs:
        - path: rtsp://@10.0.0.23:554/stream1
          input_args: preset-rtsp-restream
          roles:
            - record
            - audio
        - path: rtsp://@10.0.0.23:554/stream2
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
    motion:
      threshold: 30
    objects:
      track:
        - person
    record:
      enabled: true
      retain:
        days: 3
      alerts:
        retain:
          days: 30
      detections:
        retain:
          days: 30
    snapshots:
      enabled: true
      retain:
        default: 2
    audio:
      enabled: true
      listen:
        - bark
        - fire_alarm
        - scream
        - speech
        - yell
version: 0.15-1

docker-compose file or Docker CLI command

none

Relevant Frigate log output

none

Relevant go2rtc log output

none

FFprobe output from your camera

none

Install method

Docker Compose

Object Detector

Coral

Network connection

Wired

Camera make and model

tapoc120

Screenshots of the Frigate UI's System metrics pages

Any other information that may be helpful

No response

[NickM-27]

where is your CPU being pinned? the screenshot shows minimal CPU usage

[vectorr66]

I first starting looking today when my videos/recording were skipping like crazy. When I monitor the actual docker I get this.

[NickM-27]

what CPU do you have? Docker shows CPU % as a total of each core. So 100% CPU usage means a total of one core is being used, which may be very minimal depending on how many CPU cores there are.

[vectorr66]

I am not limiting what it can use. My whole question though is why is my GPU not given more. BUT if resources are not the issue then why are my videos all of a sudden studdering. https://we.tl/t-7wdoIk5Cxo

[NickM-27]

Right, so the CPU is also very low.

decoding video is a very easy task for a GPU, and regardless of having a GPU there are still tasks that only a CPU can do like moving the video data between the system and GPU memory

regarding the stuttering, one reason may be that you are using a restream preset but not actually restreaming

[vectorr66]

OK, I think that is it, not sure if I have the wrong preset or what not. Before my video was fine and not studdering and now it is. I am also not recording audio which is strange. I have what I would think a quite simple configuration. Am I using the wrong presets? I had this before for my ffmpeg preset-nvidia

[NickM-27]

for input_args you should delete those lines so it uses the default