What would you do with this old PC if you had €0 to invest in it?
I have an older desktop sitting around with:
- AMD FX-6100, 6 cores at 3.3 GHz
- NVIDIA GeForce GT 730
- 24 GB RAM
- ASRock 970 Extreme R2.0 motherboard
- One SSD and one HDD
It still works, but I currently have no money to upgrade it or turn it into an ambitious homelab project.
I am trying to decide what would be the most useful and least wasteful thing to do with it.
A few possibilities I have considered:
- Install a lightweight Linux distribution and use it for experiments
- Turn it into a local server, NAS, self-hosting box, or automation playground
- Use it to learn more about local AI, although I assume the GT 730 is far too limited for most modern models
- Keep it until I have money for a few upgrades
- Sell it cheaply and put the money towards a newer, more energy-efficient machine
My main concern with using it as a server is electricity consumption. Keeping an older FX-based desktop running all day might cost more, both financially and environmentally, than the practical value it provides.
What would you do in my situation?
Are there any genuinely useful €0 experiments this hardware is still suitable for? Is it worth keeping, or would selling it or reusing the parts elsewhere be the more sensible choice?
I don’t think that GPU is for modern PC gaming as one of my machines with close specs struggles with DOTA 2, but older games and emulation is going to be okay.
I’d look into BIOS in that thing and if you can reliably downvolt, dumb it down, so it doesn’t have a lot of overhead over the tasks you give it.
I doubt you can use it for local LLMs, but I do think if you’d get a grip of server/vm tooling, it’d be a huge benefit outweighting it’s energy consumption, that’s you investing into your experience and knowledge. It’s golden. Even if tasks you ran on it are not efficient or needed, you’d get an understanding of how to do a lot of things that can come up handy.
Your point about treating the electricity use as an investment in learning was probably the most useful perspective for me.
I am interested in automation, local AI and self-hosting, so even if this is not an efficient permanent server, it could still be a very useful experimental machine. I think I will keep it off most of the time and switch it on specifically for learning, backups, AI experiments or older games.
I will also look into undervolting once I have checked the exact motherboard and measured how stable the machine is. Thanks for pushing me to think about the value of the skills rather than only the efficiency of the hardware.
A stretch goal might be 24/7 running a pi zero or openwrt router next to it so you can turn the machine on remotely.
The benefit of doing your work or learning on that machine, even if your laptop is faster, is that you can offload heavy or long running things and save your laptop battery when mobile. Want to compile something? Run some local llm job? Build a new docker image? Start it, close your laptop and go about your day. Open it later to see the thing is done and your laptop still has 99% battery. (This will require, at minimum, learning about SSH, tmux, networking, wake on lan, and probably VPNs like wireguard or tailscale, or maybe dynamic DNS - lots of fun).
You do need something running all the time to do remote WOL because those packets don’t work over the Internet, but there are lots of Linux computers that use less electricity than your typical bedside alarm clock. Even an old phone might work.