[Beta Support]: random onnx decoder spikes on Nvidia GPU #22075

a10kiloham · 2026-02-21T18:45:56Z

a10kiloham
Feb 21, 2026

Describe the problem you are having

Once a day or every other day, the onnx utilization spikes up and demands a restart. I've had this going on for about a month now using the betas and RCs. Photo below of UI

Beta Version

0.17-rc2

Issue Category

Object Detection / Detectors

Frigate config file

mqtt:
  enabled: true
  host: 192.168.1.2
  port: 1883
  topic_prefix: frigate
  client_id: frigate
  stats_interval: 60

auth:
  enabled: false
  failed_login_rate_limit: 1/second;5/minute;20/hour
  trusted_proxies:
    - 172.18.0.0/16
    - 172.21.0.0/16
    - 192.168.1.0/24

  roles:
    frigate_user:
      - g5_patio
      - front_door
      - patio_doorbell
      - lightwell_rotate
      - doorbell
live:
  quality: 1

proxy:
  default_role: admin
  header_map:
    user: Remote-User
    role: Remote-Groups

detectors:
  onnx:
    type: onnx

model:
  path: plus://853ba3e3ab2f6e022a515846c3c84e26

go2rtc:
  retry_interval: 15
  streams:
    patio_doorbell:
      - rtsp://xxx:xxx@192.168.1.52:554/Preview_01_main
      - ffmpeg:http://192.168.1.52/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=xx#video=copy#audio=copy#audio=opus
    patio_doorbell_sub:
      - ffmpeg:http://192.168.1.52/flv?port=1935&app=bcs&stream=channel0_sub.bcs&user=admin&password=xxx
    front_door:
      - rtsp://xxx:xxx@192.168.1.146:554/Preview_01_main
    front_door_sub:
      - ffmpeg:http://192.168.1.146/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&passwod=xxx
    doorbell:
      - rtsp://xxx:xxx@192.168.1.78/Preview_01_sub?backchannel=0
      - ffmpeg:http://192.168.1.78/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=xxx#video=copy#audio=copy#audio=opus
    doorbell_sub:
      - ffmpeg:http://192.168.1.78/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=xxx
    g5_patio_high:
      - rtspx://192.168.1.1:7441/5vICp1bQfbYPWxxxx
    g5_patio:
      - rtspx://192.168.1.1:7441/xxxxx
    lightwell:
      - rtsp://xxx:xxx@192.168.1.51:8554/Preview_01_main #?video=h264#audio=copy#hardware
      - ffmpeg:kh_lightwell#audio=opus#video=h264#rotate=90#hardware
    lightwell_rotated:
      #- ffmpeg:http://192.168.1.51/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=xxx#video=copy#audio=copy#audio=opus#rotat
e=90
      - ffmpeg:lightwell#audio=opus#video=h264#rotate=-90#hardware
ffmpeg:
  hwaccel_args: preset-nvidia
  output_args:
    record: preset-record-generic-audio-aac

audio:
  enabled: false
  max_not_heard: 30
  min_volume: 500
  listen:
    - bark
    - fire_alarm
    - scream
    - speech
    - yell

detect:
  enabled: true

review:
  alerts:
    enabled: true
    # Optional: labels that qualify as an alert (default: shown below)
    labels:
      - person
      - face
      - package
      - royal_mail
      - ups
      - fedex
      - fox
      - license_plate
      - dpd
      - amazon
    # required_zones:
  detections:
    enabled: true

objects:
  track:
    - person
    - face
    - motorcycle
    - bicycle
    - school_bus
    - rodent
    - hedgehog
    - bus
    - garbage_truck
    - license_plate
    - dog
    - dpd
    - cat
    - car
    - amazon
    - fedex
    - dhl
    - royal_mail
    - fox
    - ups
    - bird
    - package
  filters:
    person:
      # Optional: minimum width*height of the bounding box for the detected object (default: 0)
      # decimal representing the percentage of the frame (0.000001 to 0.99).
      #min_area: 5000
      # Optional: maximum width*height of the bounding box for the detected object (default: 24000000)
      #max_area: 1000000
      # Optional: minimum width/height of the bounding box for the detected object (default: 0)
      #min_ratio: 0.5
      # Optional: maximum width/height of the bounding box for the detected object (default: 24000000)
      #max_ratio: 2.0
      # Optional: minimum score for the object to initiate tracking (default: shown below)
      min_score: 0.7
      # Optional: minimum decimal percentage for tracked object's computed score to be considered a true positive (default: shown below)
      threshold: 0.75

  genai:
    enabled: true
    prompt: Analyze the {label} in these images from the {camera} security
      camera at the front door of a home on a busy UK street. Focus on the
      actions, behavior, and potential intent of the {label}, rather than just
      describing its appearance. Look for package deliveries especially.
    object_prompts:
      person: Examine the main person in these images. What are they doing and
        what might their actions suggest about their intent (e.g., making a
        delivery to the parcel box, approaching a door, leaving an area,
        standing still)? Do not describe the surroundings or static details.

motion:
  threshold: 30
  lightning_threshold: 0.8
  # 10 most sensitive, 30 medium, 50 least
  contour_area: 10
  improve_contrast: true

record:
  enabled: true
  expire_interval: 120
  sync_recordings: true
  preview:
    quality: high
  alerts:
    retain:
      days: 7
      mode: motion #active_objects
    #pre_capture: 5
    #post_capture: 5
  detections:
    retain:
      days: 7
      mode: motion
    #pre_capture: 5
    #post_capture: 5

  continuous:
    days: 0
  motion:
    days: 14
snapshots:
  enabled: true
  clean_copy: true
  timestamp: false
  bounding_box: false
  crop: true
  #height: 480
  required_zones: []
  retain:
    default: 14
    objects:
      person: 14
  quality: 75

cameras:
#########################################################################################################
  g5_patio:
    enabled: true
    best_image_timeout: 10
    ui:
      order: 5
    detect:
      enabled: true
      #fps: 5 # increase to 10 if vehicles move quickly across your frame. Higher than 10 is unnecessary and is not recommended.
      #min_initialized: 2
      #width: 1280
      #height: 720
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/g5_patio
          roles:
            - record
            - detect
      output_args:
        record: preset-record-ubiquiti
    motion:
      mask:
        - 0,0,0.477,0,0.463,0.677,0.399,0.79,0.412,1,0.19,1,0.196,0.259,0,0.074
        - 0.732,0,0.739,0.074,0.722,0.624,0.809,0.985,0.994,1,1,0.024
      threshold: 80
      contour_area: 95
      improve_contrast: false
    review:
      alerts:
        required_zones: patio_entry_gate_zone
    objects:
      mask:
        - 0.466,0,0.469,0.35,0.716,0.369,0.719,0.423,0.737,0.727,0.994,1,1,0.006
        - 0.171,0,0.48,0.011,0.48,1,0.164,1,0.172,0.102,0.176,0.18
      genai:
        enabled: false
    zones:
      patio_entry_gate_zone:
        genai:
          enabled: true
        coordinates:
          0.715,0.428,0.719,0.596,0.655,0.596,0.625,0.619,0.616,0.657,0.59,0.748,0.525,0.747,0.509,0.45,0.542,0.401,0.681,0.381
        inertia: 3
        loitering_time: 3
        objects:
          - person
          - face
          - package
          - cat
          - dog
          - fox
          - bird
          - royal_mail
        friendly_name: Patio_Entry_Gate_Zone
      patio_ground_zone:
        coordinates:
          0.452,0.68,0.488,0.601,0.678,0.601,0.732,0.708,0.83,0.925,0.822,0.996,0.593,0.995,0.116,0.996,0.191,0.853,0.262,0.963,0.369,0.969,0.399,0.80
5
        inertia: 3
        loitering_time: 0
        objects:
          - cat
          - dog
          - fox
          - person
          - bird
          - package
          - royal_mail
        friendly_name: Patio_Ground_Zone
    notifications:
      enabled: true
#########################################################################################################
  front_door:
    face_recognition:
      enabled: false
    enabled: true
    best_image_timeout: 20
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/front_door_sub
          roles:
            - detect
          #input_args: preset-http-reolink
        - path: rtsp://127.0.0.1:8554/front_door
          roles:
            - record
            - audio
    detect:
      enabled: true
      #fps: 5
    ui:
      order: 3
      dashboard: true
    zones:
      front_door_main_zone:
        coordinates:
          0.281,0.11,0.178,0.308,0.276,0.976,0.57,1,0.581,0.651,0.522,0.454,0.478,0.21,0.477,0.198
        inertia: 3
        loitering_time: 0
        friendly_name: Front_Door_Main_Zone
      speed_zone:
        coordinates: 0.449,0.478,0.505,0.557,0.597,0.247,0.562,0.228
        inertia: 1
        loitering_time: 0
        speed_threshold: 2
        distances: 1.2,6,1.2,6
        friendly_name: New_End_Speed_Zone
        objects: car
      front_door_right_zone:
        coordinates:
          0.58,0.981,0.594,0.208,0.834,0.314,0.99,0.556,0.912,0.953,0.75,0.959,0.686,0.962
        inertia: 3
        loitering_time: 0
        objects:
          - person
          - royal_mail
          - dpd
          - package
          - ups
          - face
        friendly_name: Front_Door_Right_Zone
    review:
      alerts:
        required_zones:
          - front_door_main_zone
          - speed_zone
          - front_door_right_zone
        enabled: true
      detections:
        enabled: true
    objects:
      genai:
        enabled: false
    motion:
      threshold: 53
      contour_area: 26
    notifications:
      enabled: true
#########################################################################################################
  patio_doorbell: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/patio_doorbell_sub
          roles:
            - detect
          #input_args: preset-http-reolink
        - path: rtsp://127.0.0.1:8554/patio_doorbell
          roles:
            - record
            - audio
    detect:
      enabled: true
      #fps: 5
      #width: 1280
      #height: 720
    best_image_timeout: 5
    mqtt:
      enabled: true
      timestamp: false
      bounding_box: false
      crop: true
      quality: 75
      height: 480
    ui:
      order: 2
      dashboard: true
    zones:
      patio_entryway_zone:
        coordinates:
          0.496,0.221,0.612,0.25,0.624,0.333,0.703,0.563,0.691,1,0.113,1,0.119,0.913,0.174,0.367,0.215,0.262
        inertia: 6
        loitering_time: 2
        objects:
          - person
          - face
          - dog
          - cat
          - package
          - fox
          - royal_mail
        friendly_name: Patio_Entryway_Zone
    review:
      detections: {}
      alerts:
        required_zones: patio_entryway_zone
    motion:
      threshold: 80
      contour_area: 100
      improve_contrast: false
      mask:
        0.353,0.285,0.353,0.004,1,0,0.931,0.143,0.813,0.973,0.712,1,0.793,0.284,0.461,0.183,0.458,0.273,0.412,0.276
    notifications:
      enabled: true
#########################################################################################################
    objects:
      genai:
        enabled: true
  lightwell_rotate:
    face_recognition:
      enabled: false
    enabled: true
    best_image_timeout: 10
    ffmpeg:
      #output_args:
      # record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/lightwell_rotated #otated
          roles:
            - record
            - audio
            - detect
    detect:
      enabled: false
    ui:
      order: 9
      dashboard: true
    notifications:
      enabled: true
    zones:
      lightwell_entry_warning_zone:
        coordinates:
          0.535,0.497,0.178,0.494,0.133,0.476,0,0.454,0.027,0.993,0.964,0.996,0.987,0.122,0.564,0.399
        inertia: 3
        loitering_time: 0
        objects:
          - bird
          - cat
          - dog
          - fox
          - person
          - package
        friendly_name: Lightwell_Entry_Warning_Zone
      lightwell_entry_door_side_zone:
        coordinates:
          0.159,0.493,0.187,0.336,0.124,0.233,0.541,0.236,0.593,0.282,0.513,0.391,0.53,0.493
        inertia: 3
        loitering_time: 0
        objects:
          - cat
          - dog
          - fox
          - package
          - person
          - bird
        friendly_name: Lightwell_Entry_Door_Side_Zone
    motion:
      threshold: 21
      contour_area: 10
      improve_contrast: true
    review:
      alerts:
        required_zones: lightwell_entry_warning_zone
    objects:
      genai:
        enabled: true
  doorbell:
    audio:
      # Optional: Enable audio events (default: shown below)
      enabled: true
      # Optional: Configure the amount of seconds without detected audio to end the event (default: shown below)
      max_not_heard: 30
      # Optional: Configure the min rms volume required to run audio detection (default: shown below)
      # As a rule of thumb:
      #  - 200 - high sensitivity
      #  - 500 - medium sensitivity
      #  - 1000 - low sensitivity
      min_volume: 500
      # Optional: Types of audio to listen for (default: shown below)
      listen:
        - bark
        - fire_alarm
        - scream
        - speech
        - yell
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/doorbell
          roles:
            - record
            - audio
          input_args: preset-rtsp-restream
        - path: rtsp://127.0.0.1:8554/doorbell_sub
          roles:
            - detect
          #input_args: preset-http-reolink
    detect:
      enabled: true
      #fps: 8
      #width: 1280
      #height: 720
    best_image_timeout: 10
    mqtt:
      enabled: true
      timestamp: false
      bounding_box: false
      crop: true
      quality: 75
      #height: 1280
    ui:
      order: 1
      dashboard: true
    zones:
      front_door_near_zone:
        coordinates:
          0,0.216,0.288,0.324,0.432,0.392,0.655,0.482,0.7,0.59,0.695,0.643,0.67,1,0.478,0.984,0.044,0.984,0.033,0.753
        inertia: 3
        loitering_time: 3
        friendly_name: Front_Door_Near_Zone
    review:
      detections:
        enabled: true
      alerts:
        required_zones: front_door_near_zone
    motion:
      improve_contrast: true
      threshold: 56
      contour_area: 48
    objects:
      filters:
        person:
          mask:
            0.094,0.284,0.154,0,0.975,0.007,0.988,0.474,0.634,0.44,0.452,0.382,0.362,0.41
      genai:
        enabled: true
    notifications:
      enabled: true
version: 0.17-0

camera_groups: {}
tls:
  enabled: false

genai:
  provider: openai
  api_key:xx
  model: gpt-4o
semantic_search:
  enabled: true
  reindex: false
  model_size: large

notifications:
  enabled: true
  email: xxx@xxx.com

face_recognition:
  enabled: true
  model_size: large
  detection_threshold: 0.55
  min_area: 4000
  min_faces: 3
  unknown_score: 0.85
  recognition_threshold: 0.93
  blur_confidence_filter: true

lpr:
  enabled: true
  device: GPU
  model_size: large
  enhancement: 8

classification:
  bird:
    enabled: true

Relevant Frigate log output

many of these:
frigate  | 2026-02-21 18:34:16.842408577  [2026-02-21 18:34:16] frigate.record.maintainer      WARNING : Too many unprocessed recording segments in cache for front_door. This likely indicates an issue with the detect stream, keeping the 6 most recent segments out of 7 and discarding the rest...
frigate  | 2026-02-21 18:34:16.842408577  [2026-02-21 18:34:16] frigate.record.maintainer      WARNING : Too many unprocessed recording segments in cache for front_door. This likely indicates an issue with the detect stream, keeping the 6 most recent segments out of 7 and discarding the rest...
frigate  | 2026-02-21 18:34:16.842408577  [2026-02-21 18:34:16] frigate.record.maintainer      WARNING : Too many unprocessed recording segments in cache for front_door. This likely indicates an issue with the detect stream, keeping the 6 most recent segments out of 7 and discarding the rest...

frigate  | 2026-02-16 14:27:15.044488668  [2026-02-16 14:27:15] frigate.detectors.plugins.onnx INFO    : ONNX: loading /config/model_cache/853ba3e3ab2f6e022a515846c3c84e26
frigate  | 2026-02-16 14:27:15.528016629  [2026-02-16 14:27:15] frigate.detectors.plugins.onnx INFO    : ONNX: /config/model_cache/853ba3e3ab2f6e022a515846c3c84e26 loaded
frigate  | 2026-02-19 13:34:59.680063969  [2026-02-19 13:34:59] frigate.detectors.plugins.onnx INFO    : ONNX: loading /config/model_cache/853ba3e3ab2f6e022a515846c3c84e26
frigate  | 2026-02-19 13:35:00.487996424  [2026-02-19 13:35:00] frigate.detectors.plugins.onnx INFO    : ONNX: /config/model_cache/853ba3e3ab2f6e022a515846c3c84e26 loaded

Nvidia-SMI
Sat Feb 21 18:44:22 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 590.48.01              Driver Version: 590.48.01      CUDA Version: 13.1     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA RTX A2000 12GB          Off |   00000000:02:00.0 Off |                  Off |
| 48%   75C    P2             29W /   70W |    2590MiB /  12282MiB |      4%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A         2416830      C   frigate.detector:onnx                   138MiB |
|    0   N/A  N/A         3042371      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          136MiB |
|    0   N/A  N/A         3042378      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          125MiB |
|    0   N/A  N/A         3075135      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          348MiB |
|    0   N/A  N/A         3075140      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          454MiB |
|    0   N/A  N/A         3133536      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          218MiB |
|    0   N/A  N/A         4005535      C   frigate.embeddings_manager             1024MiB |
|    0   N/A  N/A         4005996      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          115MiB |
+-----------------------------------------------------------------------------------------+

Relevant go2rtc log output (if applicable)

No response

Install method

Docker Compose

docker-compose file or Docker CLI command

services:
  frigate:
    container_name: frigate
    privileged: true 
    restart: always
    image: ghcr.io/blakeblackshear/frigate:0.17.0-rc2-tensorrt
    shm_size: 512mb
    networks:
      - caddynet
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities:
                - gpu
    devices:
      - /dev/bus/usb:/dev/bus/usb # Passes the USB Coral, needs to be modified for other versions
      - /dev/dri/renderD128:/dev/dri/renderD128 # For intel hwaccel, needs to be updated for your hardware
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /home/xxx/Frigate/config:/config
      - /home/xxx/Frigate/media:/media/frigate:rw
      - /home/rxxxobk/Frigate/models:/models
    expose:
      - 5000
      - 5001
      - 5002
      - 8971
      - 554
      - 8555
      - 8554
    ports:
      - 0.0.0.0:5000:5000
      - 8971:8971
      - 0.0.0.0:5001:5001
      - 0.0.0.0:554:554 # RTSP feeds
      - 0.0.0.0:8554:8554 # RTSP feeds
      - 0.0.0.0:8555:8555/tcp # WebRTC over tcp
      - 0.0.0.0:8555:8555/udp # WebRTC over udp
    environment:
      NVIDIA_VISIBLE_DEVICES: all
      NVIDIA_DRIVER_CAPABILITIES: compute,video,utility
      FRIGATE_RTSP_PASSWORD: xxx
      PLUS_API_KEY: xxx
networks:
  caddynet:
    external: true

Operating system

Ubuntu

CPU / GPU / Hardware

nvidia rtx a2000-12gb on Docker official images

Screenshots

see above

Steps to reproduce

reboot, wait a day

Any other information that may be helpful

nothing i can see that links with this easily. I've updated the nvidia drivers to the latest 590 ones but that doesn't make much of a difference (possibly in my head but the newer drivers make it happen but it takes 2 days instead of 1).

NickM-27 · 2026-02-22T21:31:48Z

NickM-27
Feb 22, 2026
Collaborator Sponsor

Which model type are you using? This is likely to be a driver issue or some other system problem, I have been running my 3070 with this for a long time without issues.

We would expect that this would include ONNX error logs if it was occurring. What does the nvidia-smi look like before and after this occurs?

0 replies

a10kiloham · 2026-02-22T22:07:40Z

a10kiloham
Feb 22, 2026
Author

normal nvidia-smi is below. i'll find the onnx logs - i'm using the largest one of the yolo models from Frigate Plus 640x480 rez. i need to figure out how to turn on onnx logging as it's not captured in my local mounts from docker. will do shortly thanks for the help.

+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 590.48.01              Driver Version: 590.48.01      CUDA Version: 13.1     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA RTX A2000 12GB          Off |   00000000:02:00.0 Off |                  Off |
| 51%   75C    P2             30W /   70W |    2716MiB /  12282MiB |      6%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A          776349      C   frigate.embeddings_manager             1060MiB |
|    0   N/A  N/A          776394      C   frigate.detector:onnx                   268MiB |
|    0   N/A  N/A          776757      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          115MiB |
|    0   N/A  N/A         3486471      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          125MiB |
|    0   N/A  N/A         3486478      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          138MiB |
|    0   N/A  N/A         3519579      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          328MiB |
|    0   N/A  N/A         3519584      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          424MiB |
|    0   N/A  N/A         3577052      C   /usr/lib/ffmpeg/7.0/bin/ffmpeg          226MiB |
+-----------------------------------------------------------------------------------------+

1 reply

NickM-27 Feb 22, 2026
Collaborator Sponsor

YOLOv9 or YOLO-NAS

a10kiloham · 2026-02-22T22:14:25Z

a10kiloham
Feb 22, 2026
Author

this one from Plus - yolov9s | 640x640

0 replies

a10kiloham · 2026-02-22T22:14:58Z

a10kiloham
Feb 22, 2026
Author

the card is a SFF A2000 but works pretty reliably in general on all tasks. Decently powered.

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[Beta Support]: random onnx decoder spikes on Nvidia GPU #22075

Uh oh!

{{title}}

Uh oh!

Replies: 4 comments 1 reply

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

Uh oh!

[Beta Support]: random onnx decoder spikes on Nvidia GPU #22075

Uh oh!

a10kiloham Feb 21, 2026

Describe the problem you are having

Beta Version

Issue Category

Frigate config file

Relevant Frigate log output

Relevant go2rtc log output (if applicable)

Install method

docker-compose file or Docker CLI command

Operating system

CPU / GPU / Hardware

Screenshots

Steps to reproduce

Any other information that may be helpful

Replies: 4 comments · 1 reply

Uh oh!

NickM-27 Feb 22, 2026 Collaborator Sponsor

Uh oh!

a10kiloham Feb 22, 2026 Author

Uh oh!

NickM-27 Feb 22, 2026 Collaborator Sponsor

Uh oh!

a10kiloham Feb 22, 2026 Author

Uh oh!

a10kiloham Feb 22, 2026 Author

a10kiloham
Feb 21, 2026

Replies: 4 comments 1 reply

NickM-27
Feb 22, 2026
Collaborator Sponsor

a10kiloham
Feb 22, 2026
Author

NickM-27 Feb 22, 2026
Collaborator Sponsor

a10kiloham
Feb 22, 2026
Author

a10kiloham
Feb 22, 2026
Author