# What should I do with 32 GPUs collecting dust? What would you do? — retained record

> Retained reddit record from reddit:homelab.

Canonical: https://highsignal.app/data/records/acd2a5cfc25cb9e2

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:homelab)  published 2026-11-13 14:14 ingested 2026-11-14 02:40

[original source ↗](https://www.reddit.com/r/homelab/comments/1xv8yx6/what_should_i_do_with_32_gpus_collecting_dust/)

## retained text



```
Hey everyone!  
  
Back in 2020, I got caught up in the Ethereum mining hype, and stayed in it until Ethereum switched to Proof of Stake (PoS) and GPU mining was no longer an option.  
  
From January 2020 to September 2022, I had a blast diving into hardware, crypto, and blockchain. The whole learning process was incredibly fun, and I really enjoyed building mining rigs on all kinds of budgets for friends and people I knew. (Even tried putting a mining rig inside a portable grow tent for plants... it turned out like absolute shit...)  
  
In October 2022, I shut down all my rigs, checked and serviced every GPU, PSU, and other component, labeled everything, packed it all into boxes, and \*\*NEVER touched any of it again.\*\* Some pretty rough stuff happened in my personal life not long after that, and well... it's only now that I'm finally looking into what to do with all this hardware.  
  
Lately, I've been having fun experimenting with local AI models using an RTX 3070 Ti 8GB and an RTX 3060 Ti 8GB (in separate setups).  
  
I'm probably going a little overboard with this list, but I figured I'd give you guys an idea of the kind of \*\*firepower\*\* I've got sitting around collecting dust. xD  
\---  
\- \*\*2x\*\* X99 + Xeon E5-2680 v4 + 16GB DDR4 + PSUs  
\- \*\*3x\*\* Chinese B250 motherboards (12x PCIe 1.0 slots) + Intel G3900 + 8GB DDR4 + PSUs  
\- \*\*8x\*\* RTX 3070 Ti 8GB — Colorful iGame Ultra  
\- \*\*12x\*\* RX 580 8GB — XFX GTS  
\- \*\*12x\*\* RX 580 8GB — Sapphire  
\---  
  
I'm not entirely sure whether there's an efficient way to put all these GPUs to good use for running local AI models.  
  
From what I've been reading, there are ways to split workloads across two or more GPUs, but I'm still pretty unsure about how efficient or practical that would actually be, especially with this mix of older hardware.  
  
So, what would you guys do if you were in my shoes?  
  
I'm genuinely curious to hear your thoughts, especially from anyone who's experimented with multi-GPU setups for running AI locally! If splitting the workload across multiple GPUs actually works well, I'll definitely run some tests just out of curiosity...
```



document fetched 2026-11-14 00:39 · canonical https://reddit.com/r/homelab/comments/1xv8yx6/what_should_i_do_with_32_gpus_collecting_dust
