CS 3733 Operating Systems, Fall 1998 Assignment 4
Due November 19, 1998
This assignment is a continuation of Assignment 2. Use what you have learned
from that assignment. Although it is not due until November 19, you now know
all you need to in order to do this assignment. Assignment 5 will be given out
and will be due before this one. Plan your time accordingly.
We know that if all processes have the same CPU bursts then SJF and FCFS have
identical behavior. In this project you will design and carry out an
experiment which tests how the variability in the CPU burst distribution
can affect average waiting time for the two algorithms.
-
Design an experiment in which the load average is moderate and all CPU bursts
have the same value. Run this experiment with SJF and FCFS and confirm that they produce identical results.
-
Change only the distribution of the CPU bursts in such a way that you have
control over the variance and can keep the average the same. Do several
runs and compare the average waiting time for the different runs. Discuss
how your experiment supports or refutes the statement:
For FCFS, variability of the CPU burst time has little effect on the
average waiting time.
-
Use the same experiment as above, but with SJF. Discuss how your experiment
supports or refutes each of the two statements:
For SJF, the larger the variability in the CPU burst time, the greater
(smaller) the average waiting time.
Prepare a web page with a description of the problem,
a discussion of the experiments and why you chose the parameters as you did,
and a discussion of the results. Do not make this
web page accessible from outside your account at this time, but be
prepared to do so.
Extra Credit: Run your experiment with different seed values and without
a seed and discuss the results. For what range of parameters (duration and
number of processes) does it seem as if the seed value has little affect
on the results? Relate this the the validity of the results of the rest of
the assignment.
Late Breaking News ...
Late Breaking News ...
Late Breaking News ...
Late Breaking News ...
A new version of the simulator is now available.
You should use the new version for this assignment.
To access the new version, change the reference in your runit script
from runpsn2 to runpsn3.
There is now an easy way to install this version of the simulator at home
using a self-extracting zip file. You can find out how to do this
by clicking here.
Here are some of the differences besides the new look and feel:
- Some of the buttons have been moved.
- To change experiments, push the Change Experiment button.
The current experiment is displayed on the line above.
- You can now execute a single run from an experiment which contains
multiple runs.
To do this click on the Run All.
This will change it to Run Automatic and the run which will be
executed will be displayed a couple of lines above.
Keep clicking to get to different runs.
- There are now more graphing options.
The default is to graph the waiting time. Note that this is different
from before whan the waiting time ratio was graphed.
Click on the Graph: xxx button to scroll through the available
graphs. You can now graph turnaround time and CPU bursts.
- The Show Data button will pop up a window giving the table data
which appears in the log file. You can now view this info without having
to pull up a browser.
Let me know your comments on this new version.