1dOptimze It for My GPU20rncodeCan LLMs write better GPU kernals if you keep asking them to “Optimze it for my GPU”?