Which part of asyncio manages CPU-bound tasks?

In asyncio, to manage CPU-bound tasks (i.e., performing heavy computations), a mechanism is used to run such tasks in a separate thread or process, since asyncio itself is based on an event loop and is not suitable for CPU-bound tasks.

For this, the following are used:

• run_in_executor() — allows running blocking or CPU-intensive functions in a thread or process pool without blocking the main event loop.

• ThreadPoolExecutor and ProcessPoolExecutor from the concurrent.futures module — used as executors to run tasks in separate threads or processes.

Example of usage:

import asyncio
from concurrent.futures import ProcessPoolExecutor

def cpu_bound_task(x):
    # Some heavy computational task
    return sum(i * i for i in range(x))

async def main():
    loop = asyncio.get_running_loop()
    with ProcessPoolExecutor() as pool:
        result = await loop.run_in_executor(pool, cpu_bound_task, 10**7)
        print(result)

asyncio.run(main())

Thus, asyncio itself does not directly manage CPU-bound tasks but provides an interface to run them in separate threads or processes via executors.