TVHeadend with NVIDIA support on Docker Desktop

any chance to build this image with NVIDIA support on Docker Desktop? current version is not able to pickup GPU, no problem on other images that have support for NVIDIA included. anyone with experience or suggestion?

i am running docker-desktop on WIN11, wsl2, ubuntu 22.04 LTS.
my docker-compose:
version: “2.1”
#privileged: true
container_name: TVHjan-server
#- PUID=0
#- PGID=0
- TZ=Europe/Prague
- RUN_OPTS=–satip_xml
deploy: # <------------- Add this section
- driver: nvidia
#device_ids: [‘0’] # this is only needed when using multiple GPUs
capabilities: [gpu]
- /home/jan-server-ubuntu/tvheadend/config:/config
- /mnt/d/Recordings:/recordings
- 9981:9981
- 9982:9982
- /dev/dri:/dev/dri #optional
#- /dev/dvb:/dev/dvb #optional
restart: unless-stopped

We have no intention of doing anything specifically to support docker desktop (and obviously not windows) or any other ui (barring our already automated unraid templates). We suggest using docker run or docker compose, that is what we test with and what we support.

that all said, i am just glancing at tvheadend’s repo, but im not seeing anywhere that they indicate they support GPUs? is there a link?

i would expect you could do this with ffmpeg flags, but im not seeing anything from the devs beyond that capability

1 Like

thx for quick feed, i understand your motivation for as much automation as possible. pity for me…
tvh does support hardware acceleration for sure, as long as HW allows it and it is compiled with that feature.
i was able to find two images with NVIDIA support built in (probably based on LSIO), but there are some other problem in those and i could not get hold of the authors. Docker

I have just been told the following by @j0nnymoe one of my teammates:
tvheadend wouldn’t get Nvidia support until it’s swapped to Ubuntu base

i do not have an ETA or any planning info for this.

(for more info, this is due to nvidia requiring glibc and alpine being musl based.)

1 Like