Before LLMs, the use of parallelism on your typical laptop was limited to application level parallelism, e.g. one thread for Outlook and one for each tab in Chrome.