15-740: Computer Architecture
Fall 2001
Syllabus

Course Details at a Glance

Lectures: Mondays, Wednesdays Fridays, 10:30-11:59 a.m., WeH 5409
Instructor: Todd C. Mowry, WeH 8123, 268-3725, tcm@cs.cmu.edu
Class Administrator: Maury Burgwin, WeH 8124, 268-4740, mburgwin@cs.cmu.edu
Web Page: www.cs.cmu.edu/afs/cs/academic/class/15740-f01/www/
Newsgroup: cmu.cs.class.cs740
Course Materials: /afs/cs.cmu.edu/academic/class/15740-f01/public

Textbooks

Main Text:
Hennessy, J. L, and Patterson, D. A., Computer Architecture: A Quantitative Approach, 2nd Edition. Morgan Kaufmann, 1996. (NOTE: don't try to get by with the first edition of this book; the second edition is much different and more aligned to the content of this course.)
Supplemental Text:
Culler, D., and Singh, J.P., with Gupta, A. Parallel Computer Architecture: A Hardware/Software Approach, Morgan Kaufmann, 1998. (Copies should be available in the ES Library by the time we need them.)

Course Overview and Objectives

This course attempts to provide a deep understanding of the issues and challenges involved in designing and implementing modern computer systems. Our primary goal is to help students become more skilled in their use of computer systems, including the development of applications and system software. Users can benefit greatly from understanding how computer systems work, including their strengths and weaknesses. This is particularly true in developing applications where performance is an issue.

The course material is divided evenly into two parts. The first half of the course covers systems based on a single processor, closely following the Hennessy and Patterson textbook. The second half of the course covers parallel systems containing multiple processors, with topics ranging from programming models to hardware realizations. The material for this latter half of the course can be found to some extent in the Hennessy and Patterson book, but is treated in much greater detail in the Culler, Singh and Gupta text.

Course Themes

An addition to our ``user-centric'' (vs. ``builder-centric'') approach, the course has several other themes. One theme is to emphasize the role of evolving technology in setting the directions for future computer systems. Computer systems, more than any other field of computer science, has had to cope with the challenges of exploiting the rapid advances in hardware technology. Hardware that is either technologically infeasible or prohibitively expensive in one decade, such as bitmapped full color displays or gigabyte disk drives, becomes consumer products in the next. Technology that seems to have a bright future, such as magnetic bubble memories, never becomes competitive. Others, such as CMOS, move from being a niche technology to becoming dominant. In addition, computer systems must evolve to support changes in software technology, including advances in languages and compilers, operating systems, as well as changing application requirements. Rather than teaching a set of facts about current (but soon obsolete) technology, we therefore stress general principles that can track evolving technology.

Another theme of the course is that ``hands-on'' exercises generally provide more insight regarding system behavior than paper-and-pencil exercises. Hence our assignments involve programming and using computer systems, although in a variety of different ways.

Finally, rather than stopping with state-of-the-art in computer architecture as of a decade ago, another theme of this course is looking at the state-of-the-art today as well as open research problems that are likely to shape systems in the future. Hence we will be discussing recent papers on architecture research in class, and students will perform a significant research project.

Prerequisites

This course is not intended to be your first course on computer architecture or organization; it is geared toward students who have already had such a course as undergraduates. For example, we expect that people are already at least somewhat familiar with assembly language programming, pipelining, and memory hierarchies. If you have not had such a course already, then it is still possible to take this course provided that you are willing to spend some additional time catching up on your own. If you feel uncertain about whether you have adequate preparation, please discuss this with the instructor.

In addition to an undergraduate computer organization course, here are some other topics which are helpful for this course (references are included for self study):

Non-CS Students

If you are not a graduate student in the Computer Science PhD program, you need permission to take this class. If you have not already done so, send a message to the instructor stating your status, why you want to take the class, and if you want to take the class for credit or as an auditor.

Placing Out

Students who have already taken graduate-level courses in computer architecture or parallel architecture may find that some of this course material is familiar. Although the course topics (especially in the first half of the course) may look familiar even to students who have taken an undergraduate computer architecture course, this course is designed to build on undergraduate material, and will cover this topics in much greater depth. It is likely that the focus and style of this course will be different from what you have experienced before, and that the pace will be fast enough that you will not be bored. However, if you feel strongly that you should be able to ``place out'' of all or part of this course, contact the instructor.

Course Work

Grades will be based on homeworks, a research project, two exams, and class participation.

Homeworks:
There will be roughly two homework assignments. Each assignment involves a non-trivial amount of programming. You will work in groups of two or three (preferably three) people on the assignments. (Turn in a single writeup per group.)

Project:
A major focus of this course is the project. We prefer that you work in groups of two on the project, although groups of up to three may be permitted depending on the scale of project (ask the instructor for permission before forming a group of three). The project is intended to be a scaled-down version of a real research project. The project must involve an experimental component---i.e. it is not simply a paper and pencil exercise. We encourage you to come up with your own topic for your project, although we will be posting suggested projects to the class web page at a future date. You will have six weeks to work on the project. You will present your findings in a written report (the collected reports may be published as a technical report at the end of the semester), and also during a poster session during the last day of class. Start thinking about potential project ideas soon!

Exams:
There will be two exams, each covering its respective half of the course material. Note that the second exam is not cumulative, and is weighted equally with the first exam. Both exams will be closed book, closed notes.

Class Participation:
In general, we would like everyone to do their part to make this an enjoyable interactive experience (one-way communication is no fun). Three classes are set aside entirely for student-led in-class discussions on active areas of research and innovation in computer architecture. All students are expected to lead one of these discussions.

Grading Policy

To pass this course, you are expected to demonstrate competence in the major topics covered in the course. Your overall grade is determined as follows:

Exams: 40% (20% each)
Homework: 15%
Project: 35%
Class Participation: 10%

Late assignments will not be accepted without prior arrangement.

Schedule

Table 1 shows the tentative schedule. There might be some variations.


Table 1: CS 740, Fall 2001, Tentative Schedule.
Class Date Day Topic Reading Homework
1 9/12 Wed Performance Technology HP Ch. 1 #1 Out
2 9/14 Fri Alpha Programming HP Ch. 2  
3 9/17 Mon Instruction Set Comparison HP App. C D  
4 9/19 Wed Basic Pipelining HP Ch. 3  
5 9/21 Fri Advanced Pipelining HP Ch. 4 #1 Due, #2 Out
6 9/24 Mon Superscalar Processing Concepts HP Ch. 4  
7 9/26 Wed Superscalar Processing Practice HP Ch. 4  
8 9/28 Fri The Memory Hierarchy HP Ch. 5  
9 10/1 Mon Programming for Performance    
10 10/3 Wed Virtual Memory HP Ch. 5  
11 10/5 Fri Recent Research on Uniprocessors I Handouts #2 Due
12 10/8 Mon Recent Research on Uniprocessors II Handouts  
  10/10 Wed Exam I
13 10/12 Fri Intro to Parallel Architecture HP Ch. 8,CSG Ch. 1  
14 10/15 Mon Parallel Programming I CSG Ch. 2  
15 10/17 Wed Parallel Programming II CSG Ch. 3 4  
  10/19 Fri Mid-Semester Break (No Class)   Project Proposal
  10/22 Mon Mid-Semester Break (No Class)    
16 10/24 Wed Cache Coherence I CSG Ch. 5  
17 10/26 Fri Cache Coherence II CSG Ch. 5  
18 10/29 Mon Memory Consistency CSG Ch. 6  
19 10/31 Wed Synchronization CSG Ch. 5  
20 11/2 Fri Interconnection Networks CSG Ch. 10  
21 11/5 Mon Recent Research on Multiprocessors Handouts  
  11/19 Mon     Project Milestone
  11/28 Wed Exam II
  12/3 Mon     Project Due
  12/10 Mon Project Poster Session

Bibliography

Aho+86
Aho, A. V. and Sethi, R. and Ullman J. D., Compilers. Addison-Wesley, 1986. Background.

CulSin98
Culler, D., and Singh, J. P., with Gupta, A., Parallel Computer Architecture: A Hardware/Software Approach. Morgan Kaufman, 1998.

HenPat96
Hennessy, J. L, and Patterson, D. A., Computer Architecture A Quantitative Approach, 2nd Edition. Morgan Kaufman, 1996. Main textbook.

Kernighan+88
Kernighan, B. W., and Ritchie, D. M., The C Programming Language, 2nd edition, Prentice Hall, 1988. Background.

Silberschatz+91
Silberschatz, A., Peterson, J., and Galvin, T., Operating System Concepts, 3rd Edition, Addison-Wesley, 1991. Background.