OPENCL and VAAPI setup

[ceewanna]

In my setup in debian linux, I have tried the followings.

1. Intel CPU integrated graphics: it can detect opencl. (device /dev/dri) but I don't see any vaapi used.
2. AMD CPU with Nvidia graphic cards: it can detect CUDA. (NVIDIA Docker is deprecated so NVIDIA Container Toolkit was used.)
3. AMDCPU integrated graphics: nothing available eventhough I had opencl and vaapi installed and linux command can detect its presence. Any extra settings need to be done?

Note: In viseron document, it says.
VAAPI hardware acceleration support is built into every amd64 container.
To utilize it you need to add --device /dev/dri to your docker command.
I have added above line but so far I have never seen VAAPI used in the log.

[bsyomov]

In a CUDA container, to use VAAPI, you must use raw_command or hwaccel_args.
Example from my config(CUDA container, Intel iGPU via VAAPI for transcode):

      recorder:
        segments_folder: /tmp/cam1
        retain: 30
        idle_timeout: 30
        lookback: 15
        codec: hevc_vaapi
        hwaccel_args:
          - -hwaccel
          - vaapi
          - -hwaccel_device
          - /dev/dri/renderD128
          - -hwaccel_output_format
          - vaapi

[ceewanna]

@bsyomov Thanks for your input.
I had a scan on the log output after adding your recommended settings. This is what I got from my intel i5 13500 integrated graphics with intel-media-va-driver-non-free installed.

`
/mnt/d11/viseron$ ./detect.sh VA-API
VA-API cannot be used

/mnt/d11/viseron$ ./detect.sh hwacc
[2024-09-02 16:59:09] [DEBUG ] [viseron.components.ffmpeg.recorder.cam2] - Concatenation command: ffmpeg -hide_banner -loglevel error -y -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format vaapi -protocol_whitelist file,pipe -f concat -safe 0 -i - -c:v copy -c:a copy -movflags +faststart /recordings/cam2/2024-09-02/165845.mp4

[2024-09-02 17:00:38] [DEBUG ] [viseron.components.ffmpeg.recorder.cam1] - Concatenation command: ffmpeg -hide_banner -loglevel error -y -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format vaapi -protocol_whitelist file,pipe -f concat -safe 0 -i - -c:v copy -c:a copy -movflags +faststart /recordings/cam1/2024-09-02/165842.mp4
`

`
vainfo
libva info: VA-API version 1.17.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_17
libva info: va_openDriver() returns 0
vainfo: VA-API version: 1.17 (libva 2.12.0)
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

`

[roflcoopter]

Which version of Viseron are you running? The v3 beta has upgraded drivers which might resolve this if you are on v2.x

[ceewanna]

I am still on good old v2.

So far I have seen these.

1. OpenCL working under intel integrated GPU (i5 13500),
2. OpenCL and CUDA working under AMD CPU/Integrated GPU (Ryzen 5 4600G) + NVIDIA 1650.

I have 12 cameras under 2), each running at 3fps (a few 4fps) motion and darknet object scan. CPU load on htop is normally around 2-5 (out of 12 threads max) and NVIDIA GPU takes only 1 core while the other core seems to be idle most of the time (the screen is hooked up with integrated GPU).

For 1), it only runs 3 cameras (each with motion and darknet object) and I hope to explore how far it could go. From htop, it takes only 1-2 out of 19 threads.

[roflcoopter]

Ahh okay that is probably why then. v2.x is based on Ubuntu 20.04 which has old drivers.
V3 is based on Ubuntu 22.04 where the drivers are much more recent