Simulating CPU scheduling

In order to simulate the scheduling of processes among CPUs, one would have to model things like which CPUs are currently in use and how many processes are waiting for a CPU. In other words, one would create variables to store this information. Things that have no effect on CPU scheduling, such as whether a particular process was generated by user a148pita or user a108zeme, need not be modelled.

The output from such a simulation would include statistics such as how long, on average, processes waited for a CPU. Input to the simulation would allow one to compare the system's performance in different situations. For instance, we might run the simulation with different numbers of CPUs so that we could determine how many CPUs are required in order for the average wait time to be within acceptable limits.

Attached to this handout are excerpts showing what a single run of such a simulation might look like.

Part A: Program

• the items in the queue have no priorities - they are all served on a purely first-come first-served basis. In our simulation, the priority of a process determines whether it gets a CPU before or after other processes. (If two processes have the same priority, they are served on a first-come first-served basis.)
• the simulation is event-driven, meaning that the clock moves ahead in leaps according to the time of the next event that will happen. In our simulation, the clock moves ahead in equal, fixed-length ticks. This is called a time-driven simulation.