I placed a low bid on an auction for 25 Elitedesk 800 G1s on a government auction and unexpectedly won (ultimately paying less than $20 per computer)
In the long run I plan on selling 15 or so of them to friends and family for cheap, and I’ll probably have 4 with Proxmox, 3 for a lab cluster and 1 for the always-on home server and keep a few for spares and random desktops around the house where I could use one.
But while I have all 25 of them what crazy clustering software/configurations should I run? Any fun benchmarks I should know about that I could run for the lolz?
Edit to add:
Specs based on the auction listing and looking computer models:
- 4th gen i5s (probably i5-4560s or similar)
- 8GB of DDR3 RAM
- 256GB SSDs
- Windows 10 Pro (no mention of licenses, so that remains to be seen)
- Looks like 3 PCIe Slots (2 1x and 2 16x physically, presumably half-height)
Possible projects I plan on doing:
- Proxmox cluster
- Baremetal Kubernetes cluster
- Harvester HCI cluster (which has the benefit of also being a Rancher cluster)
- Automated Windows Image creation, deployment and testing
- Pentesting lab
- Multi-site enterprise network setup and maintenance
- Linpack benchmark then compare to previous TOP500 lists
Run 70b llama3 on one and have a 100% local, gpt4 level home assistant . Hook it up with coqui.Ai xttsv2 for mind baffling natural language speech (100% local too ) that can imitate anyone’s voice. Now, you got yourself Jarvis from Ironman.
Edit : thought they were some kind of beast machines with 192gb ram and stuff. They’re just regular middle-low tier pcs.
These are 10 year old mid range machines. Llama 7b won’t even run well
The key is quantized models. A full model wouldn’t fit but a 4bit 8b llama3 would fit.
It would fit but it would be very slow
No. Quantization make it go faster. Not blazing fast, but decent.
I tried doing that on my home server, but running it on the CPU is super slow, and the model won’t fit on the GPU. Not sure what I’m doing wrong
Sadly, can’t really help you much. I have a potato pc and the biggest model I ran on it was Microsoft phi-2 using the candle framework. I used to tinker with Llama.cpp on colab, but it seems they don’t handle llama3 yet. ollama says it does , but I’ve never tried it before. For the speed, It’s kinda expected for a 70b model to be really slow on the CPU. How much slow is too slow ? I don’t really know…
You can always try the 8b model. People says it’s really great and even replaced the 70b models they’ve been using.
Show as in I waited a few minutes and finally killed it when it didn’t seem like it was going anywhere. And this was with the 7b model…
It shouldn’t happen for a 8b model. Even on CPU, it’s supposed to be decently fast. There’s definitely something wrong here.
Hm… Alright, I’ll have to take another look at it. I kinda gave up, figuring my old server just didn’t have the specs for it
Specs? Try mistral with llama.ccp.
It has a Intel Xeon E3-1225 V2, 20gb of ram, and a Strix GTX 970 with 4gb of VRAM. I’ve actually tried Mistral 7b and Decapoda Llama 7b, running them in Python with Huggingface’s Transformers library (from local models)
Yeah, it’s not a potato but not that powerful eaither. Nonetheless, it should run a 7b/8b/9b and maybe 13b models easily.
That’s your problem right here. Python is great for making llms but is horrible at running them. With a computer as weak as yours, every bit of performance counts.
Just try ollama or llama.ccp . Their github is also a goldmine for other projects you could try.
Llama.ccp can partially run the model on the gpu for way faster inference.
Piper is a pretty decent very lightweight tts engine that can be directly run on your cpu if you want to add tts capabilities to your setup.
Good luck and happy tinkering!