Yes, I use this extensively in a web crawler -- network I/O is done in a main thread that's cpu bound doing networking stuff, and cpu-burning tasks like html pa

Yes, I use this extensively in a web crawler -- network I/O is done in a main thread that's cpu bound doing networking stuff, and cpu-burning tasks like html parsing are done in child threads done by run_in_executor. Without trying very hard I can get to a load of 3-4 during ordinary web crawling.