I added a second GPU just for local AI workloads, and it cost less than upgrading my main one
…You might think that running an LLM locally requires significant compute, but even an old GPU can handle several smaller models. I can't expect Claude-like intelligence from my local AI…