[HW Accel Support]: UnRaid 7.1.4 - frigate:stable-tensorrt does not detect NVIDIA GeForce GTX 1650 #20052
Replies: 3 comments 1 reply
-
|
Based on the error messages showing "Cannot load libcuda.so.1" and "Could not dynamically load CUDA", your Docker container cannot access the NVIDIA GPU properly. Here are the configuration steps needed: Docker Configuration for NVIDIA GPUsFor NVIDIA GPUs in Docker, you need to use the NVIDIA Container Toolkit and specify the GPU deployment configuration (1). Your current Docker command is missing the proper GPU configuration. Correct Docker Compose Configurationservices:
frigate:
...
image: ghcr.io/blakeblackshear/frigate:stable-tensorrt
deploy: # <------------- Add this section
resources:
reservations:
devices:
- driver: nvidia
device_ids: ['0'] # this is only needed when using multiple GPUs
count: 1 # number of GPUs
capabilities: [gpu]Alternative Docker Run CLI ConfigurationHardware Acceleration ConfigurationOnce Docker can access the GPU, configure hardware acceleration using the preset: ffmpeg:
hwaccel_args: preset-nvidiaVerificationAfter proper configuration, you should see TensorRT Detector ConfigurationSince you're using the detectors:
tensorrt:
type: tensorrt
device: 0 # This is the default, select the first GPU
model:
path: /config/model_cache/tensorrt/yolov7-320.trt
labelmap_path: /labelmap/coco-80.txt
input_tensor: nchw
input_pixel_format: rgb
width: 320 # MUST match the chosen model
height: 320 # MUST match the chosen modelThe key issue is that your current Docker configuration doesn't properly expose the NVIDIA GPU to the container. Remove the 📚 Sources: Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
@inkeep I am only able to Add another Path, Port, Variable, Label or Device |
Beta Was this translation helpful? Give feedback.
-
|
For Unraid, you need to install the NVIDIA Container Toolkit first, as the Docker Compose Based on the discussion in the sources, Docker device passthrough in Windows requires WSL2, and the documentation shows that you need For Unraid specifically, you should:
The key issue is that Unraid's Docker interface doesn't support the Docker Compose If you continue having issues after installing the NVIDIA Container Toolkit and adding 📚 Sources: Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
We've seen some reports of issues with the latest drivers and using multiple Nvidia GPUs |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Describe the problem you are having
After searching and seeing others with similar issue I list the following
Using "hwaccel_args: preset-nvidia" as Global or Per-Camera causes many errors and cameras will not display video.
Tried swapping two different GPU cards in the Unraid server.
NVIDIA GeForce GTX 1650 and NVIDIA MSI GTX 1650
NVIDIA driver versions I have tried are
575.57.08 - CUDA 12.9 and 580.82.09 - CUDA 13.0
NOTE: The system is great and will run fine using CPU only for acceleration / detection but with higher CPU usage
As per Additional requirements on UnRaid App I have also tried adding "--runtime=nvidia" as extra parameter under advanced view but the Frigate docker will not start with this added.
I have included both good & Error logs below.
(I have had both GPU cards working with Tdarr in the same server, and used 'NVIDIA_VISIBLE_DEVICES' without an issue when this docker was installed)
Version
0.16.1-e664cb2
Frigate config file
docker-compose file or Docker CLI command
Command /usr/local/emhttp/plugins/dynamix.docker.manager/scripts/docker create --name='frigate' --net='bridge' --pids-limit 2048 -e TZ="Pacific/Sydney" -e HOST_OS="Unraid" -e HOST_HOSTNAME="Teho" -e HOST_CONTAINERNAME="frigate" -e 'FRIGATE_RTSP_PASSWORD'='enterpassword' -e 'PLUS_API_KEY'='' -e 'LIBVA_DRIVER_NAME'='iHD' -e 'NVIDIA_VISIBLE_DEVICES'='GPU-ece94afd-afae-ed12-5537-051e8583ea2f' -e 'NVIDIA_DRIVER_CAPABILITIES'='compute,utility,video' -e 'YOLO_MODELS'='yolov4-416,yolov4-tiny-416' -e 'USE_FP16'='false' -e 'TRT_MODEL_PREP_DEVICE'='0' -l net.unraid.docker.managed=dockerman -l net.unraid.docker.webui='http://[IP]:[PORT:8971]' -l net.unraid.docker.icon='https://raw.githubusercontent.com/yayitazale/unraid-templates/main/frigate.png' -p '8971:8971/tcp' -p '8554:8554/tcp' -p '5000:5000/tcp' -p '8555:8555/tcp' -p '8555:8555/udp' -v '/mnt/user/appdata/frigate':'/config':'rw' -v '/mnt/disks/security/cctv/':'/media/frigate':'rw' -v '/etc/localtime':'/etc/localtime':'rw' --device='/dev/bus/usb' --device='/dev/dri/renderD128' --shm-size=256m --mount type=tmpfs,target=/tmp/cache,tmpfs-size=1000000000 --restart unless-stopped 'ghcr.io/blakeblackshear/frigate:stable-tensorrt'Relevant Frigate log output
Relevant go2rtc log output
FFprobe output from your camera
Install method
Docker Compose
Object Detector
TensorRT
Network connection
Wired
Camera make and model
Dahua - HDW3649H
Screenshots of the Frigate UI's System metrics pages
Any other information that may be helpful
Beta Was this translation helpful? Give feedback.
All reactions