[adossi]

The catch is you need to be safe:

- Where is the heat going? Are you exhausting it out the window? Are you going to burn your house down? - How are you distributing power? Are you using a PDU and a 240V 30A breaker or are you maxing out several 15A 120V lines? - Is the humidity dropping rapidly to dangerous levels, risking shock? Are you going to fry that brand new $10,000 ASIC miner because you dried out the air too much? - Did you short out the GPU pins removing a GPU from the socket? Did you just cost yourself thousands of dollars?

Safety is important not just for yourself and family but you need the hardware to actually survive. If you have a bunch of 3090s that just finally reached ROI 9 or 10 months in but the cards themselves just died, you’ve effectively broken even after all that effort.

Surprisingly GPUs are quite resilient so long as you’re not massively overheating them, and you open them up every now and then to replace the thermal paste and perhaps thermal pads. I have a monitor visible at all times that displays all the temperatures, and I have one eye on a humidistat making sure I don’t suddenly drop below 30% relative humidity on an unseasonably dry day.

There are lots of resources available. If you’ve ever built your own PC you already have like 80% of the knowledge you need.

[quickthrower2]

Thanks. This is actually what has put me off doing it. I don’t want an unsafe thing in my house! Also its the type of engineering problem i hate most - one where you meed to monitor 24/7 and all sorts of things can give you down time, but you cant really make it perfect because that would be too expensive.

[raegis]

Do you shut it down when on vacation? Or hire someone to babysit your machines when you're away?

[adossi]

I left it running while on vacation for a week recently, I was incredibly paranoid the entire time and I probably wouldn’t do that again.