EE 654  SIMD Parallel Processor Architectures and Applications

Fall  Term 2003w       last updated 2003-1-6
Calendar Description
General Course Info
Use the class newsgroup to ask questions, answer questions, or find project partners.
Use the email addresses below for questions of a confidential nature.
Professor Duncan Elliott 492-5357 ECERF w2-036


General Course Info

Lectures Mon-Wed-Thur 11:00 ECERF e1-018

Catalogue number 18342

SIMD parallel processors can offer a tremendous performance advantage for the right kind of applications.

Several examples of Single Instruction stream, Multiple Data stream (SIMD) parallel processor architectures with be described and analyzed in this course. Applications will be examined.

Background in computer architecture, and concurrent programming or operating systems would be an asset.
Concepts to be Learned
Basic elements of SIMD architecture processor element, ALU, registers, memory, communications network
Architectural case studies NASA MPP, CM-1, CM-2, MP-1, MP-2, GAPP, C-RAM, IMAP, EXECUBE
Programming languages languages with explicit parallel constructs vs. compiler-detected parallelism
Applications signal processing, scientific computation, graphics, image processing, database, CAD, optimization

Marking (tentative)

Lab Assignments 20 dates
Literature Review Lecture 15
Exam  15 last lecture, Apr 9
Project 50
Oral Presentations Mar 31, Apr 2, 4
Final Report Apr 11
All lab and project submissions are due in the first 5 minutes of the lecture on the due date.  Late submissions will be penalized 10% of the assigned mark per school day.  Submissions will not normally be re-graded more than two weeks after the first day these have been returned in class.

Grade Determination Method

See section 23.4 in the university calendar.  In this course, raw marks will be used up until after the final exam.  The resulting overall percentage mark will then be converted for each student to a grade on the nine point scale.  A standard expected distribution of grades, which is provided by the Faculty of Engineering, will be used as a rough guideline when mapping overall marks to grades.  Absolute merit of the work will also be taken into consideration.

Code of Student Behaviour

Refer to section 26 of the university calendar for a comprehensive discussion of what constitutes improper conduct for members of the University community and for a description of disciplinary procedures.  In particular, note the definitions of plagiarism and cheating in section 26.1.4 and the penalties for academic offenses specified in section 26.1.5.


Lab 1
Lab 2
Lab 3
Lab 4
Lab 5
Students are to work alone on assignments.  Reports should be brief but sufficiently clear and readable that you would be happy to submit them to a busy engineering manager.  If it only takes a sentence to describe something, then it should be described in a sentence.

Did reading this leave any questions unanswered?
Comments on the contents of these pages are welcome, - Duncan Elliott