1) Effect on efficiency of decreasing the time quanta. Suppose that the overhead to task switch is 1 mS. Further suppose that there are three tasks, A, B, and C that require 30 mS each to complete and perform no IO or other waits during their execution. If the time quanta decreases from 10 mS to 5 mS, what is the impact on overhead (efficiency)?

a) There is no impact on overhead.

b) Overhead increases by 5 mS.

c) Overhead is reduced by 100%

d) Overhead increases by 100%

2) The XV6 scheduler scans the proc table (after first getting a lock) looking for a ready to run process. This is both slow and non scalable to thousands of processes. An alternative that is faster is to:

a) Maintain a stack of proc table pointers and pop process to run them and push them when they become ready to run.

b) Maintain a linked list of ready to run processes and always run the first process on the list, and place newly ready processes at the end.

c) Randomly select a task to run and run it.

d) Maintain a hash table or ready to run processes keyed on their priority.

3) CPU scheduling is the process of deciding:

a) When in the future to run a particular process.

b) What to do with a process when it completes IO.

c) Which process to run next from among the processes that are ready to run.

d) Estimating how long a CPU burst will be.

4) Shortest Job First scheduling in non-preemptive environments results in low average response times. However, it can result in:

a) Users changing their long running jobs into a sequence of short-running jobs to game the system.

b) Starvation of long running jobs.

c) Frequent context switches that drive overhead up and utilization down.