I’ve got a PC the kids and I built years ago as a gaming machine that we don’t use anymore. Thinking of repurposing it for use with my home automation setup (replacing my current Raspberry Pi 4 with Home Assistant installed).
It’s certainly COMPLETELY overkill for that, so I’m curious what other ways I can make use of it as a home automation server of some sort. Or maybe there’s some reason it’s actually a bad idea to use it at all for this?
Specs…
- CPU: Intel i3-8100 3.6Ghz
- Memory: 8GB
- Storage: 2TB
- GPU: NVIDIA GeForce GTX 1050 Ti
- Motherboard: B360M Pro-VDH MS-7B24
Currently has Windows 11 installed on it.
If you want to do other things than home automation you could use Unraid or Open Media Vault and make a media server that has homeassistant running on a docker container. Plex can utilize the GPU for video transcoding. Just make sure your docker containers are configured to use the host network or you can run into issues where homeassistant can’t detect the other home automation devices on your network.
Put some tier 1 VM on it: ESXI, promox, etc. Now you can do a lot of stuff
I’d run proxmox on it and then from there you’ll have flexibility to spin up any number or vms or cts with any containers you want, loads of self hosted stuff and good isolation so if anything happens to any service you have full vm backups and can nuke them separately without compromising what’s stable, that’s my setup today have about 5 VMS running and some more that I plan to mess around
That’s only just slightly less powerful than the system I’m using for MSFS…
The 24/7/365 reliability, ie how long it goes without crashing or restarting from a power glitch, of a desktop PC would concern me. That and the electricity to power it.
I bet you could run a massive wyoming model on that, for super speedy voice control through HA
The GPU could be useful for object detection using an NVR like Frigate.
you can now use integrated intel GPU with Frigate, much more power efficient. I’d sell the GPU.
But can you, with reasonable latency, run speech to text or text to speech?
I’ve got a couple frigate cameras with object detection, STT and TTS running, and using like 2.8 gigs of VRAM. I might just bump up the quality on the STT and/or TTS actually…