CS665 --- Computer Architecture

Professor Alex Pothen 
Computer Science Department
Old Dominion University

ICASE
NASA Langley Research Center

Syllabus


CS 665 is a graduate level presentation of the fundamentals of the
architecture and design of digital computers.  
*Quantitative*  measures ofperformance and cost are emphasized, 
and design features and tradeoffs
are evaluated in relation to performance enhancement and cost reduction.
A regular pattern in the course is the identification of a performance
bottleneck and the resolution of the bottleneck.  What makes the subject
interesting is the conflict between individual design objectives.
CS 665 concentrates on the study and design of uniprocessor
machines.  

CS 635, Parallel Architecture, concentrates on additional
considerations that arise in the design and use of multiprocessor
computers. CS 635 will be taught by Professor David Keyes this 
semester as well.  

The intended audience for CS 665 is mainly CS and CE graduate students.
However, students (including those from applications departments)
lacking a formal CS or CE background, but possessing good computational
aptitude and prior programming experience, are welcome to enroll.

Topics to be covered include:  benchmarking (establishing useful metrics
for the measurement of performance), cost estimation, instruction sets
(both general principles and some specific examples), processor
implementation, pipelining, memory hierarchy, and an introduction to I/O. 
If time permits, we will briefly discuss  an
introduction to networks, and an introduction to parallelism.  
The focus in lecture and homework examples will be on RISC machines, because of
their prominence in today's markets, and their current favor as nodes of
multiprocessors.

The required textbook for CS 665 is the second edition of Computer
Architecture:  A Quantitative Approach by David Patterson and John
Hennessy, Morgan Kaufmann Publishers, 1996.  
Supplements from the World Wide Web will be incorporated as needed.

When:  Spring 1997
       Mondays,  7:10 -- 9:50 pm 
       first meeting: January 13

Where: Television Studio, Education Building, 2nd Floor
       (and three simulcast remote sites). 
       I will be teaching from the Peninsula Center. 

Credits: 3.0 hours


Contact: pothen@cs.odu.edu, 683-4414



Alex Pothen received his Ph.D. from Cornell 
in 1984. He taught at the Department of Computer Science 
at Penn State, and the University of Waterloo before coming to ODU in 1994. 
His research interests include parallel algorithms, 
scientific computing, and combinatorial algorithms.