I have ZimaOS 1.6.1 and an Nvidia GTX 1050 graphics card. The card appears on the dashboard and in nvidia-smi as well. However, when I try transcoding to Jellyfin with Nvidia NVENC, I always get an error. Can you help me, please?
If that is not already the case, you most probably need to enable GPU access to your container… It’s explained here : https://jellyfin.org/docs/general/post-install/transcoding/hardware-acceleration/nvidia.
Cheers,
J!
If the GTX 1050 appears in the ZimaOS dashboard and nvidia-smi, that is a good sign, but it does not yet confirm that the Jellyfin container can access the GPU correctly.
The first thing I would check is whether the NVIDIA devices are visible inside the Jellyfin container and what error Jellyfin/FFmpeg is producing during transcoding.
Can you please run these and paste the output?
docker ps --format "table {{.Names}}\t{{.Image}}\t{{.Status}}\t{{.Ports}}"
Then replace jellyfin with the actual container name if it is different:
docker exec -it jellyfin nvidia-smi
Also check whether the NVIDIA device files exist inside the container:
docker exec -it jellyfin ls -l /dev/nvidia*
And please also paste the Jellyfin transcode error log from Jellyfin Dashboard → Logs, especially the FFmpeg error section.
The GTX 1050 should support NVENC, but on ZimaOS the important part is confirming the container has the correct NVIDIA runtime/device access, not just that the host can see the card.
``
Thanks for the information.
I successfully installed Jellyfin-nvidia via the App Store. I just needed to uncheck the codecs that aren’t supported by the Nvidia GTX 1050 (Google is your friend). I launched a video on my Jellyfin client and the transcoding worked perfectly. Also, on nvidia-smi, I noticed that ffmpeg is indeed active. However, just one last question: Why does the GPU widget on the ZimaOS dashboard always display 0W when the graphics card is transcoding, Logically, it should be consuming power, right?
Great, that confirms the important part: Jellyfin is now using the GPU properly.
If nvidia-smi shows ffmpeg active while transcoding, then NVENC is working.
About the ZimaOS dashboard showing 0W: that does not necessarily mean the GPU is not consuming power. It usually means the dashboard is not receiving a valid power draw value from the NVIDIA driver for that card.
Some older NVIDIA cards, especially consumer GTX cards like the GTX 1050, may not expose full power telemetry in the same way newer cards do. ZimaOS can show the GPU and usage, but the power widget may still stay at 0W if the driver/card does not report power.draw correctly.
You can check directly with:
nvidia-smi --query-gpu=name,utilization.gpu,power.draw,power.limit,temperature.gpu --format=csv
If power.draw shows N/A, 0W, or does not change during transcoding, then it is only a reporting limitation, not a Jellyfin problem.
In your case, because ffmpeg appears in nvidia-smi and transcoding works, I would treat this as a dashboard/telemetry limitation rather than a GPU transcoding issue.