Optimizing Local LLMs for Low-End Hardware: 8GB GPU Guide