| Instructor: | Richard Hughey |
| Office: | 315-A Applied Sciences |
| Phone: | 459-2939 |
| e-mail: | rph@ce.ucsc.edu |
Tuesdays and Thursdays, 2:00-3:45, Stevenson 221.
May 26 is an exchange day -- no class.
This course will focus on parallel algorithms. We will be using the department's AlphaServers (and other machines) with the Message Passing Interface (MPI), a standard course-grain parallel programming library, as well as the Kestrel hardware (hopefully) and simulator. The differences parallelization and performance between fine-grain and course-grain parallel processors will form part of our focus. We may write a few small programs for the department's aging MasPar as well.
The 290M course alternates with the 220 course (Parallel Processing). The focus of both courses changes from year to year, but 290M is generally a software-oriented course, while 220 is a hardware-oriented course. The previous two 290M courses have considered parallel programming languages and numerical algorithms.
Just last week, I asked the bookstore to order copies of Introduction to Parallel Computing but Kumar, Grama, Gupta, and Karypis. A copy will be available to read in 316 until it comes in.
I have put many of my books on the bookshelf in 316 (including copies of the course text). These books are not under any circumstances to leave 316 or to be left anywhere in 316 apart from the bookcase. 316 is a combination Kestrel lab and teaching and research workstation lab. CE124 will be using the primary users of right half of the room and have priority for the workstations there. You should feel free to read the books in the Kestrel. If we get a working Kestrel system together by the end of the quarter, there will be located in PCs on the left half of the room.
Journals of interest include the Journal of Parallel and Distributed Processing, IEEE Transactions on Parallel and Distributed Systems, IEEE Parallel and Distributed Technology magazine, and Computer. Conferences include ICPP (Int. Conf. Parallel Processing), IPPS (Int. Parallel Processing Symposium), and SPDP (Symposium on Parallel and Distributed Processing), among others.
Subscription to comp.parallel and comp.super is encouraged. You
should also look around David Bader's parallel site:
http://www.umiacs.umd.edu/~dbader/sites.html.
I've started up a class WWW page that you can access from page. I'll frequently update it.
Evaluations will be based on attendance and participation, homework assignments, a midterm, and a class project and presentation. Satisfactory performance in each categories will be needed to pass the course. As this course works well in seminar format, you will be called on to present problem solutions from time to time.
For the project, a brief (paragraph) project proposal will at the end of the fifth week (April 30), and a short 3-5 page discussion of the problem (suitable for the introductory section of your project report) in the 7th week of class (May 14). The presentations during the last two weeks should be detailed, and you should provide reading material during the period before your talk. The final report will be due on Wednesday, June 12, at 5pm.
Programming and written homework assignments are to be your own work and handed in on time. As this is a seminar, it is certainly appropriate to explore solutions to problems with your classmates, but the final programming or writeup must be solely your own work. All collaborations must be acknowledged. Academic dishonesty will result in failure in the course.
| Week | Topics | Reading |
| 4/7? | Intro, Theory, Networks | 1 |
| 4/14 | Communication, MPI | 2, handouts |
| 4/21 | Systolic | 12 |
| 4/28 | Kestrel | handouts |
| 5/5? | Dynamic Programming | 9, handouts |
| 5/12 | Graph Algorithms | 7 |
| 5/19 | Neural Networks | handouts |
| 5/28 | FFT (short week) | 10 |
| 6/2? | Project Talks | |
| 6/9? | Project Talks |