CSC369H5 Chapter Notes - Chapter 3: Performance Metric, Context Switch, Cpu Cache
Document Summary
Need basic framework for implementing scheduling policies. Each job runs for same amount of time. Turnaround time : t_turnaround = t_completion - t_arrival. If jobs arrive at same time, t_arrival = 0, t_turnaround = t_completion. Balancing performance/fairness is key to scheduling properly. A, b ,c jobs arrive at same time (t_arrival = 0) Fifo makes a = 1st, b = 2nd, c = 3rd. T_completion = 10 seconds for a, 20 seconds for b, 30 seconds for c. Relax assumption 1, each job has variable runtime. A, b, c at same time , as before. A runs for 100 seconds, b = 10 secs, c = 10 secs. A runs for 100 seconds before b, c. Queueing small processes behind large runtime ones is bad practice, leads to large. Avg t_turnaround = (100 + 110 + 120)/3 = 110 turnaround time. Idea is simple, run jobs in increasing order of runtime schedulers are now.
