I believe the unblocking order in the clarifications is intended for a single time step, meaning unblocking is ordered first by time the task is completed, and secondly by the type of system call it was (and thirdly by the time the system call was called). Chris' sample seems to follow the pattern of doing work and then incrementing time by one in a loop until there are no processes alive. So the sleeping and then the waiting process would both become unblocked during one time step, and not until another time step in the next 10usecs would the second sleeper become unblocked. Hope it isn't too cheeky to imagine that Chris might say: a sleeping process cannot 'advance the time'. :P Even though your implementation may work like that.