[HW Accel Support]: ARC A310 Support? Working Setup Questions.

[moogoos]

Describe the problem you are having

I dont have a problem, necessarily. Curious about my upgraded setup.

I upgraded from an Intell 14500 Q670 platform, using an M.2 Coral card which worked pretty well and iGPU. I've had mostly success with A+E and M.2 Coral cards in a couple builds. I recently decided to make some changes, and consolidate to the newer z890 Intel Ultra Core platform and see if it would work. Overall the consolidation went well. The only issue I had was with the M.2 device. It has some problem potentially with power management. I keep getting a device inaccessible and some other errors so a driver or BIOS problem perhaps. So I became stuck. I could have reverted to the Tensor setup I had a config for and hardware for but want to keep the power down as much as possible, which is why I prefer Coral with iGPU.

My setup is Proxmox >> VM >> Docker with PCIe passthrough.

I grabbed a sparkle A310 card to play around with for transcoding and decided to just move my containers to the VM that has this card passed to it as passthrough. It shows up as /dev/dri is passed to the container from the VM. The VM has the IOMMU group passed to it from the host, so the A310 card is passed to the VM. That is what makes /dev/dri and card0 card1 and renderD128 available, I think. The iGPU is NOT passed to the VM.

I then switch to use CPU type detection instead of coral in my frigate config.

detectors: cpu1: type: cpu num_threads: 3
docker-compose device pass through

`

• /dev/dri/renderD128 # for intel hwaccel, needs to be updated for your hardware
  `

So if im understanding this correctly, this should then be using the A310 for detect.

Either way the setup is working extremely well. It's just as fluid and responsive as it was using the Coral card, I was going to maybe get the USB Coral but I'm not a fan of USB and at about $80 bucks I really didn't want another card even though it probably would have worked. What im trying to understand is how can I confirm that the GPU is working detect? I'm running 8 detect sub streams. I just dont see a reason to use a coral card now at all?:

width: 1280 height: 720 fps: 5

detected inference is about 24ms

`
Vainfo Output
Return Code: 0

Process Output:

vainfo: VA-API version: 1.22 (libva 2.10.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 24.3.3 ()
vainfo: Supported profile and entrypoints
VAProfileNone : VAEntrypointVideoProc
VAProfileNone : VAEntrypointStats
VAProfileMPEG2Simple : VAEntrypointVLD
VAProfileMPEG2Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointEncSliceLP
VAProfileH264High : VAEntrypointVLD
VAProfileH264High : VAEntrypointEncSliceLP
VAProfileJPEGBaseline : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointEncPicture
VAProfileH264ConstrainedBaseline: VAEntrypointVLD
VAProfileH264ConstrainedBaseline: VAEntrypointEncSliceLP
VAProfileHEVCMain : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointEncSliceLP
VAProfileHEVCMain10 : VAEntrypointVLD
VAProfileHEVCMain10 : VAEntrypointEncSliceLP
VAProfileVP9Profile0 : VAEntrypointVLD
VAProfileVP9Profile0 : VAEntrypointEncSliceLP
VAProfileVP9Profile1 : VAEntrypointVLD
VAProfileVP9Profile1 : VAEntrypointEncSliceLP
VAProfileVP9Profile2 : VAEntrypointVLD
VAProfileVP9Profile2 : VAEntrypointEncSliceLP
VAProfileVP9Profile3 : VAEntrypointVLD
VAProfileVP9Profile3 : VAEntrypointEncSliceLP
VAProfileHEVCMain12 : VAEntrypointVLD
VAProfileHEVCMain422_10 : VAEntrypointVLD
VAProfileHEVCMain422_10 : VAEntrypointEncSliceLP
VAProfileHEVCMain422_12 : VAEntrypointVLD
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
VAProfileAV1Profile0 : VAEntrypointEncSliceLP
VAProfileHEVCSccMain444_10 : VAEntrypointVLD
VAProfileHEVCSccMain444_10 : VAEntrypointEncSliceLP
`

Version

15-beta

Frigate config file

NA

docker-compose file or Docker CLI command

NA

Relevant Frigate log output

NA

Relevant go2rtc log output

NA

FFprobe output from your camera

NA

Install method

Docker Compose

Object Detector

Other

Network connection

Wired

Camera make and model

NA

Screenshots of the Frigate UI's System metrics pages

NA

Any other information that may be helpful

No response

[NickM-27]

using the cpu detector does not use the intel GPU, you need openvino for that

[moogoos]

What is the significance of width and height here?

model:
width: 300
height: 300

for openvino config?

[NickM-27]

The size of the model

[moogoos]

ill ask a seperate question on that width height settings.

[moogoos]

openvino looks MUCH better and the inference time is extremely reasonable.