[HW Accel Support]: Unable to get GPU status under sriov

[xemxx]

Describe the problem you are having

I started frigate with a simple configuration with only one camera providing h265 streams and configured hardware acceleration.

But I found some error logs about getting the GPU state

I observe that my GPU is actually working, via the command

intel_gpu_top -d sriov

My OS environment is lxc deployed based on pve and running docker, configured sriov support on pve and passthrough raw gpu to lxc, then running frigate on docker.

I'll make sure no other programs are using the gpu on my machine.

I think this should be fixable.

Your help would be appreciated, thanks!

Version

0.14.1-f4f3cfa

Frigate config file

mqtt:
  enabled: false

record:
  enabled: true
  retain:
    days: 5
    mode: all

ffmpeg:
  hwaccel_args: preset-vaapi
  output_args:
    record: preset-record-generic-audio-aac

snapshots:
  enabled: false
  retain:
    default: 30

detect:
  enabled: false

detectors:
  ov:
    type: openvino
    device: GPU

model:
  width: 300
  height: 300
  input_tensor: nhwc
  input_pixel_format: bgr
  path: /openvino-model/ssdlite_mobilenet_v2.xml
  labelmap_path: /openvino-model/coco_91cl_bkgr.txt
cameras:
  tp-link:
    enabled: true
    ffmpeg:
      inputs:
        - path: hidden
          roles:
            - record
            - detect

docker-compose file or Docker CLI command

services:
  frigate:
    container_name: frigate
    privileged: true # this may not be necessary for all setups
    restart: unless-stopped
    image: ghcr.io/blakeblackshear/frigate:stable
    shm_size: "128mb" # update for your cameras based on   calculation above
    devices:
      - /dev/dri/renderD128:/dev/dri/renderD128 # For intel hwaccel, needs to be updated for your hardware
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - ./config:/config
      - ./storage:/media/frigate
      - type: tmpfs # Optional: 1GB of memory, reduces SSD/SD Card wear
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    ports:
      - "8971:8971"

Relevant Frigate log output

frigate  | 2024-10-15 17:02:18.434368645  [2024-10-15 17:02:18] frigate.util.services          ERROR   : Unable to poll intel GPU stats: Failed to detect engines! (No such file or directory)
frigate  | 2024-10-15 17:02:18.434374816  (Kernel 4.16 or newer is required for i915 PMU support.)
frigate  | 2024-10-15 17:02:18.434377845  timeout: the monitored command dumped core

Relevant go2rtc log output

nothing

FFprobe output from your camera

ok

Install method

Proxmox via Docker

Object Detector

OpenVino

Network connection

Wired

Camera make and model

TP-link TL-IPC44AW-COLOR 6.0

Screenshots of the Frigate UI's System metrics pages

Any other information that may be helpful

My PVE

My docker lxc

[NickM-27]

what does the hardware info button show in the system UI?

[xemxx]

Sorry for taking so long to answer as I found out that the GPU is under actual load so I haven't paid any more attention to this, but of course it would be best if you could fix it.

Here's my output, and I've noticed that sometimes this button disappears and you have to refresh the page to get it

Return Code: 0

Process Output:

vainfo: VA-API version: 1.17 (libva 2.10.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 23.1.1 ()
vainfo: Supported profile and entrypoints
VAProfileNone : VAEntrypointVideoProc
VAProfileNone : VAEntrypointStats
VAProfileMPEG2Simple : VAEntrypointVLD
VAProfileMPEG2Simple : VAEntrypointEncSlice
VAProfileMPEG2Main : VAEntrypointVLD
VAProfileMPEG2Main : VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointFEI
VAProfileH264Main : VAEntrypointEncSliceLP
VAProfileH264High : VAEntrypointVLD
VAProfileH264High : VAEntrypointEncSlice
VAProfileH264High : VAEntrypointFEI
VAProfileH264High : VAEntrypointEncSliceLP
VAProfileVC1Simple : VAEntrypointVLD
VAProfileVC1Main : VAEntrypointVLD
VAProfileVC1Advanced : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointEncPicture
VAProfileH264ConstrainedBaseline: VAEntrypointVLD
VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
VAProfileH264ConstrainedBaseline: VAEntrypointFEI
VAProfileH264ConstrainedBaseline: VAEntrypointEncSliceLP
VAProfileHEVCMain : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointEncSlice
VAProfileHEVCMain : VAEntrypointFEI
VAProfileHEVCMain : VAEntrypointEncSliceLP
VAProfileHEVCMain10 : VAEntrypointVLD
VAProfileHEVCMain10 : VAEntrypointEncSlice
VAProfileHEVCMain10 : VAEntrypointEncSliceLP
VAProfileVP9Profile0 : VAEntrypointVLD
VAProfileVP9Profile0 : VAEntrypointEncSliceLP
VAProfileVP9Profile1 : VAEntrypointVLD
VAProfileVP9Profile1 : VAEntrypointEncSliceLP
VAProfileVP9Profile2 : VAEntrypointVLD
VAProfileVP9Profile2 : VAEntrypointEncSliceLP
VAProfileVP9Profile3 : VAEntrypointVLD
VAProfileVP9Profile3 : VAEntrypointEncSliceLP
VAProfileHEVCMain12 : VAEntrypointVLD
VAProfileHEVCMain12 : VAEntrypointEncSlice
VAProfileHEVCMain422_10 : VAEntrypointVLD
VAProfileHEVCMain422_10 : VAEntrypointEncSlice
VAProfileHEVCMain422_12 : VAEntrypointVLD
VAProfileHEVCMain422_12 : VAEntrypointEncSlice
VAProfileHEVCMain444 : VAEntrypointVLD
VAProfileHEVCMain444 : VAEntrypointEncSliceLP
VAProfileHEVCMain444_10 : VAEntrypointVLD
VAProfileHEVCMain444_10 : VAEntrypointEncSliceLP
VAProfileHEVCMain444_12 : VAEntrypointVLD
VAProfileHEVCSccMain : VAEntrypointVLD
VAProfileHEVCSccMain : VAEntrypointEncSliceLP
VAProfileHEVCSccMain10 : VAEntrypointVLD
VAProfileHEVCSccMain10 : VAEntrypointEncSliceLP
VAProfileHEVCSccMain444 : VAEntrypointVLD
VAProfileHEVCSccMain444 : VAEntrypointEncSliceLP
VAProfileAV1Profile0 : VAEntrypointVLD
VAProfileHEVCSccMain444_10 : VAEntrypointVLD
VAProfileHEVCSccMain444_10 : VAEntrypointEncSliceLP

[fay000fay]

I have the same problems with u. the same configurations, and environments. waiting for the answers...

[mm2293]

Same here

[letrain02]

this was added. but i still can't get it to work. even though I added it to config.
telemetry:
stats:
sriov: true
intel_gpu_stats: true

#15796

maybe @blakeblackshear can tell me I did it wrong.

[milch]

I can get the GPU stats working if I run a command like this in the container:

intel_gpu_top -d drm:/dev/dri/card0

But I'm getting the same error as everyone else no matter if I set sriov to true or false.

[Calimerorulez]

Same here...

[hawkeye217]

We (myself and the others on the dev team) don't have the ability to test SRIOV stats support, so we consider it a community-supported feature. There was a PR that a user submitted here that could be problematic:

#18236

If someone could test and/or suggest a better solution, we'll gladly accept a PR for it.

[milch]

Would you accept a PR that either makes the sriov accept a string with the device path, or a separate device path parameter? It seems different setups need different values there, and a "one size fits all" Boolean is not going to work

[hawkeye217]

Feel free to submit a PR and we can review it and comment.