My laptops screen has been on the fritz lately so I figured with prime day/ summer laptop deals coming I’d convert this laptop (asus rog with nvidia 3060) into my HA rig. I’ve copied over the backup and all seems to be running well.
However it seems there’s no local/on machine support for gpus? So I’ve got this decent gpu I can’t really use now? Is there any hope that’ll change with how powerful gpu are now and how HA wants to be local and private?
Thanks for the input!
I don’t think HA is capable of making use of the GPU. However, if it’s Nvidia you can do some interesting things with voice (providing you don’t mind flashing the laptop with Ubuntu):
Not being facetious, what do you expect the HA SERVER (it’s a server and intended to run headless in a corner.) to need the GPU for.
I do admit the GPU is helpful for LLM and tensor type loads but in most cases those workloads should be on another box. (or another container) which itself does have GPU/NPU access.
If you want it on the same box then use something like Proxmox to run multiple vms or containers and punch the GPU through to the VM and then load your drivers.
Basically they don’t because they don’t need to.
I just repurposed my old laptop with a 3060 as my HA OS machine so wanted to take advantage of it. I have a 3070 with 8gb in my Unraid server and use that for some ollama stuff too. I just find asking my s3 box to turn things on/off to be inconsistent. I pay for nabu casa so I’ll just use that for now.