Everyone says you need a powerful GPU and tons of RAM to run local LLMs. I decided to challenge that idea.