What is the way that parallel programming models use to know that they receive/finish sending the message if it is asynchronous? Is it implemented using for example interrupts or by repeatedly polling?