CSCI 8720 Advanced Computer Architecture and Parallel Processing
Instructor : Dr. Suchi Bhandarkar
Spring 2002

Text: High Performance Computer Architecture, by Harold S. Stone, Addison-Wesley, Third Edition.

Reference: Computer Architecture: A Quantitative Approach by J.L. Hennessy and D.A. Patterson, Second Edition, 1995, Morgan Kaufmann Publishers Inc.

Prerequisite: CSCI 6720 or a graduate-level course in Computer Architecture.

Course Description: The course will cover fundamental issues dealing with the analysis and design of high-performance and parallel/distributed computer architectures. Fundamental architecture design techniques such as hierarchical memory design, pipelining and multiprocessing will be covered in detail. Specific architectures will be studied and analyzed in terms of performance analysis parameters such as speedup, memory bandwidth, processing bandwidth, communication overhead, load balancing, fault tolerance and synchronization overhead. The specific list of topics covered in this course are as follows:

• Review of Conventional Architecture Design and Analysis

• Cache and Virtual Memory Design

• Pipelining

• Vector Processing

• Multiprocessing Techniques: SIMD and MIMD

• Interconnection Networks

In addition to these topics covered from the textbook, there will be papers assigned from the current literature. Students will be expected to read these papers and the material therein will be discussed in class and may also appear in the examinations and homework assignments.

Class Meeting Times: This course will be organized as a combination of formal classroom instruction and a student seminar. The class will meet on Tuesdays and Thursdays during the class period. Since this is a 4 credit hour course, the students are expected to devote the remaining class period and an additional 4--5 hours every week towards a self-study project of their choice. On March 5, Tuesday each student (or team) is expected to submit a formal project proposal. The students will be expected to present their work during the last week of the semester and also submit a technical report which will be due on the first day of the exam week.

Grading:

Test 1: 20%
Test 2: 20%
Homework Assignments: 20%
Project Report and Presentation: 40%

The test dates will be announced at least one week in advance. The tests will be held during the class period on the date scheduled. Students are expected to do their own work on the homework assignments and the tests. Students are expected to familiarize themselves and comply with the University of Georgia Policy on Academic Honesty detailed in http://www.uga.edu/ovpi/honesty/main.html.