KirJul 31, 2023

Proxmox HDMI output for one VM-container

https://feddit.it/post/913318

Proxmox HDMI output for one VM-container - Feddit.it

Hi everyone, I’m planning to experiment with my first home server (jellyfin,nextcloud,bitwarden etc) running on a “old” Thinkcentre i5-7th. I’m thinking using proxmox in order to experiment with different configuration and I was wondering if it’s possible to have a single container/vr (with Libreelec) output HDMI to my TV, and keep other VM’s headless and controlled from another PC when needed. Are there some particular setting or nwebie suggestion that could help me achieve that? Also, do you think proxmox is a good choice, or it’s better stick to a single debian/ubuntu server OS?

Show threadsnekerpimpJul 31, 2023

What you are looking for is gpu pass through. You can do it with integrated graphics as well as dedicated. There is a lot of good documentation out there for proxmox to do this.

I would totally start with a hypervisor. The ability to spin up a new VM for an experiment or project far outweighs any downside. I use proxmox, but that’s mainly because it was the thing to do when I was starting out. There are a few options out there, including installing Debian and running kvm/qemu, which is basically what proxmox is. They all have the same functionality, just different UI/features. Lxc containers in proxmox help a lot as well, when you have a small service you want to run but don’t want to take up a VMs worth or resources. These also helped a lot.

Good luck

Proxmox VE Helper-Scripts

Scripts for Streamlining Your Homelab with Proxmox VE

Show threaddragnet
Just gonna add a bit here. I haven't used proxmox, but I do have a GPU I assign to QEMU/KVM virtual machines. The arch wiki has an article that helps a lot, for anyone who wishes to try this. I have also found that the virgl drivers allow for sufficient graphical performance for video decoding at high resolution, so that is another potential option to explore here that doesn't involve allocating the GPU to a VM.
Jul 31, 2023 at 1:25pmWeb
Show threadSafetyGogglesJul 31, 2023

The main reason why OP need GPU passthrough is so that Jellyfin can do hardware-accelerated transcoding, I think.

I presume virgl doesn’t provide hardware-acceleration for encoding (and probable decoding too)?