Goals
CPS 210 is a multi-threaded course in
that it is meant to satisfy a number of different objectives:
- CPS 210 is a graduate level introduction
to the basic knowledge, skills, and research directions in the field of Operating
Systems. The overall course grade will reflect
mastery of this material.
- CPS 210 is a Quals Course.
This means that it can be used to meet one of the requirements for Ph.D. students.
To get quals credit, a student must pass the exams with a quals pass. A quals
pass is not necessarily directly related to the course grade.
- CPS 210 is a gateway course
for prospective systems students. This means it will present a view of the
research currently of interest to the systems faculty. Students who are considering
pursuing systems research (either Ph.D. or Masters level) within the systems
group are expected to demonstrate research potential by choosing to do a term
project (see discussion of E-track project below).
This year, the course content will focus
on a theme:
How would one design an energy-aware OS for mobile / embedded computing devices?
The traditional basics will be covered by considering how things change and what
things stay the same when one adopts different assumptions about the target architecture,
intended workload, and metrics of importance. This theme question is an open problem
and, hopefully, we will come up with some original contributions leading toward
a solution.
Resources
Continually under construction!
There is no textbook that fits the
theme of the course this term.
Any undergraduate textbook in Operating Systems will make an adequate reference
for the background in the traditional topics. At Duke, there are plenty of copies
of the following basic text around: Gary
Nutt, Operating Systems: A Modern Perspective.
Other basic texts include:
Silberschatz and Galvin, Operating System Concepts, Addison-Wesly.
A. Tanenbaum, Modern Operating Systems,Prentice-Hall.
Lecture notes will usually be made
available immediately before class and then revised (with content added in class
discussion) and reposted after class.
Workload
Base Workload. In addition
to the readings, there will be homework assignments involving some programming
(25%), a midterm exam (25%), and a written final exam (35%). Class participation
will be a significant portion (15%) of your grade. There is an underlying question
for most lectures: "How can this be used or how must a traditional approach be
changed for energy efficiency?" I am hoping for rich discussions in every class
period.
The E-track, Grading, and Quals.
This is a two-track class. Assigned work includes mandatory "G-track" components
(the base workload described above) and some optional "E-track" components.
The E-track work is necessary if you want to earn a grade of E or E+. The primary
component of the E-track work is the term project.
CPS 210 is a "quals course" for
CPS Ph.D. students. We assign quals passes based on demonstrated mastery of
the quals material, regardless of your grade. In particular, E-track work is
not necessary for a quals pass and will not improve your chances of earning
a quals pass. The E-track is intended for students with a serious interest in
systems or architecture. If you decide to take on E-track work, make sure you
also leave yourself enough time to learn the course material and earn your quals
pass. Students who passed the qualifying exam prior to taking the course are
waived from having to take the two exams.
Grades will be assigned by adding
your scores on the G-track work and exams, and assigning base grades curved
from G- to E-. Undergraduates are graded on a different scale since this course
is more advanced than typical undergraduate classes. Your scores for the E-track
work are then added and curved to improve your base grade from 0 to 3 notches
(e.g., 3 notches advances you from a base grade of G+ to a final grade of E+).
We give S grades only for barely satisfactory work, since graduate students
cannot receive credit for more than two S grades.
To summarize: if you do
an outstanding job on all the G-track work (which includes exams), you'll get
an E- and a quals pass; if you do a good job on the G-track work, you'll get
a G or a G+ and the quals pass will depend on your exam scores; if you additionally
do good work for the E-track, it is possible to get a grade higher than E-.
We reserve E+ grades for truly outstanding students.
Project.
The term project is the key component
of the E-track. This year, there are two possibilities:
Option 1: Making a contribution
to the Milly Watt group's SOSP (Symposium on Operating Systems Principles) submission.
The submission date is March 26th, so this means starting (and finishing) your
term project early in the term. The advantages are that you will be listed as
a co-author and may get a chance to attend the conference next fall, you will
be doing useful and valuable research with potential for expanding into second
year project topics (and beyond), and you will become a member (however temporary)
of our research group. Very soon, I will come out with a list of projects that
will add value to our SOSP paper.
Option 2: The usual plan
for term projects is to do projects of the students' own choosing. It is acceptable
to combine this project with project work from another course (e.g. networks)
as long as the topic of the project is appropriate (i.e., has something to do
with Operating Systems).
Both kinds of projects will be presented
in a mini-conference at the end of the reading period.
February 6: Submit a short
(maximum of 1 page) project proposal.
You are encouraged to form groups
of two or perhaps three students to collaborate on projects.
Projects and other E-track work are
held to a high standard; incomplete or incorrect work will receive little if
any extra credit.
Assignments
Problem sets and programming assignments
will be listed here.
January 18: Five presentations,
prepared by groups of 2 students each, on the following topics:
(1) Processors (StrongARM family, Transmeta Crusoe), (2) Memory and Secondary
Memory Devices (RAMBUS RDRAM, IBM microdisk, flash memory, MEMS disk alternatives),
(3) Input and Output (Display technology, data entry devices), (4) Power Sources
(batteries, Smart Batteries, alternatives), (5) Networking (Wireless LANs, Bluetooth,
cellular). Address characteristics of the technology, what to expect in the
near future, features to exploit for power/energy management, etc. Each talk
will last 10 to 15 minutes. Sign ups on the first day of class.
January 25: Six presentations,
prepared by groups of 2 students each, on the following topics:
(1) ACPI, (2) Symbian OS, (3) PalmOS, (4) embedded Java and Jini, (5) Microsoft
embedded solutions (winCE, PalmPC), (6) Embedded Linux. Address the programming
model supported, interesting features of the API, and especially the power management
features. Each talk will last 10 to 15 minutes. Sign ups on the first day of
class.
Policies
On collaboration. Collaboration
on problem sets and programming assignments is encouraged.
However, your final written work must be your own, and you may be called upon
to explain (alone) your choices and answers in more detail. How can you know when
you are crossing the line between leaning on your colleagues and learning with
your colleagues? Here's the Gilligan's Island Rule:
This rule says that you have truly
assimilated the knowledge gained by working with fellow students if you can
leave the session without a written record of the interactions (electronic or
otherwise), then engage in a half hour of mind-numbing activity (like watching
an episode of Gilligan's Island), before starting to work on your own.
This test assures that you are able to reconstruct what you learned from the
meeting, by yourself, using your own brain.
On giving credit where credit
is due. You must acknowledge the sources of your words and ideas, when
they are not your own, and you should list the names of anyone who helped you
with your work. Thus, I acknowledge that much of this web page content comes
from Professor Jeff Chase who developed the clever E-track system in previous
offerings of CPS 210. The Gilligan's Island Rule is an inspiration of Professor
Larry Ruzzo from University of Washington. Failing to give appropriate credit
is at best a violation of professional etiquette, and at worst it is plagiarism.
Plagiarism, like collaboration on exams, is cheating. Cheating is a very serious
offense and I do not expect it to occur in this class.
Previous CPS 210
Spring
2000
Spring
1999
Last Updated: 6-Jan-01