CS 3733 Operating Systems, Spring 2000 Assignment 2 Comments
This assignment was graded on a basis of 30 points with 10 point for the
first two parts (Part 0 and Part 1) and 10 points for each of parts 2 and 3.
Comments on Part 2:
- The load average is the ratio of the total waiting time and the total
running time.
- The total waiting time is gotten by multiplying the average waiting time
by the number of processes.
- The total running time is given in the Time column of the table.
- In the first run, the load average is 11.5. This means that on the
average there are about 11 processes in the ready queue. This is about
half of the processes. The CPU
utilization is 1.0, meaning the CPU is always busy. This is because the
I/O bursts are small, and a process returns to the ready queue before
too many other processes have executed.
- In the second run, the load average is about .12, indicating that most of
the time the ready queue is empty. The CPU utilization is low, about 20%.
By the time a process finishes its I/O burst, all of the other processes
that were in the ready queue have used the CPU and started doing I/O.
- This shows that the I/O time is important when designing an experiment.
If there is never more than one process in the ready queue, then all
non-preemptive algorithms behave the same.
Comments on Part 3:
- Most people were able to design an experiment in which the CPU utilization
was high and the number of proceses varied.
- The variable y is proportional to x if there is some constant, k, such that y = kx.
- This is not the same as saying that an increase in x causes an increase
in y.
- One way to tell if two things are proportional, it to draw a graph and see
if the points fall on a straight line through the origin.
- Most people who did this did not check to see if the line went through
the origin.
- Probably the simplest way was to calculate the ratio of the average
waiting time to the number of processes, and check to see if it is
constant.
- Many people said that it was obvious that they were proportional by
looking at the number in the table. This is not an analysis.
You must show what you did.
- You cannot prove a statement like the one given is true by doing an
experiment. You can show that it might be true, or you can show that
it is false.