I'm a fairly new convert to running TrueNAS at home, after nearly a decade using Synology boxes. I'd outgrown the locked-in ecosystem, and recent business decisions were less than agreeable, so I now have a powerful home server with TrueNAS virtualized on top of a Proxmox host. This gives me the best of both worlds, while enabling backup practices that keep the VM and the drive's data safe from all but the worst failures.

I also have a large assortment of PC hardware that has been collected over the years, some of it working and some in various states of disrepair. That includes a stack of GPUs, and one of them no longer outputs a signal over the display outputs. It still powers on, though, and I was trying to think of what to do with it until I realized it could be a transcoding beast for my NAS.

My display outputs stopped working one day

Thankfully, running a headless server doesn't need any

The GPU in question is an Nvidia GeForce RTX 5060 Ti, so it's recent enough that it shouldn't have had any issues, except it did. I was going to use it to build an SFF system to put next to the TV for gaming, but that dream is gone. Having no signal from the display outputs is not something I'd experienced before, but maybe it got damaged in transit to me, as it wasn't in the original packaging.

But by putting it into my server and using PCIe pass-through to pipe the GPU into my TrueNAS VM, I can still leverage the CUDA cores for TV use. They can sit there idling until I watch something that's either too high resolution or the wrong codec for my client app to stream across my network, and then it'll jump into action. That takes away some of the strain from the CPU for when the file also has 7.1 audio tracks that need downmixing, making for an overall smoother stream.

I could use it for other home server needs, like upscaling images, categorizing and tagging my disorganized photo stream, and running Moonlight and Sunshine for remote game streaming. The point is that just because the GPU doesn't have display output from its ports, it doesn't mean it can't be used for other tasks.

Using the GPU for transcoding in Jellyfin

Or anything else that needs an accelerated workflow

Using a GPU for transcoding in Jellyfin (or any other self-hosted media platform that can use hardware acceleration) makes sense no matter what generation it's from. Still, you want to use a relatively low-powered one because transcoding doesn't need a flagship, and you'll just waste electricity costs.

And for that use case, this 50-series Nvidia card is perfect, as it supports all the recent video codecs, including HEVC and AV1. Not having display outputs doesn't matter in this case because it's going in a headless server anyway, and the video files are being sent over the network to the Jellyfin client on an Apple TV 4K for display.

The 180W TGP of the RTX 5060 Ti is a little high for transcoding only, but it'll be idling at single-digits most of the time, so I'm not too worried. If I'm not using Jellyfin at the time, I can use it for CCTV processing through Frigate, or re-encoding video files for mobile playback, or a whole bunch of other home server tasks that the hardware acceleration will chew through in no time at all.

Or I could have used it for AI

The RTX 5060 Ti might not be the most astonishing GPU for performance, but this model has 16GB of VRAM, which is perfect for many of the optimized LLM models. The faster GDDR7 will also bring some performance gains over older GPUs, and that's before connecting some MCP servers for even better results.

Adding Qwen3 to SearXNG, for example, gives my local AI model access to the most important of things — the Internet. That extends its usefulness past the "knowledge cut-off," aka the last data used to train the model. Larger, online models like GPT-5 or Gemini 2.5 Pro might have significantly newer training data, and significantly more data to begin with. SearXNG reduces the gap, making my local, private, (mostly) secure LLM more useful when I ask it questions or to .

My GPU is still useful, even without working ports

It's all too easy to think that because part of a piece of PC hardware has stopped working, that it's no longer useful. I certainly thought it was over for this RTX 5060 Ti when the display outputs stopped working, because it could no longer be used to play games or output graphics, and getting it repaired would probably cost more than the card did new. But my NAS doesn't need a display output as it's done over the web GUI, and the encoding power of the Nvidia card means it can live on, transcoding my media streams and running LLM queries.