Testing turing-unlock on my triple 2080ti LLM server. Will report back. EDIT: I've never seen power max like this. Still testing. I need to repaste one. EDIT 2: Prompt ingestion is much faster. Regular inference is the same. So prompt to answering is now faster, but it draws some power! EDIT 3: Holy moly. Digital Spaceport's bench script result now uploaded. These results are 🤯 #llm #homelab #localai