In general, on a multi$-$core machine, a process will tend to run on that core most of the time. Yet it could in theory be allowed to run on any core. Given that, what makes the OS prefer to keep processes running on the core they ran on last?