CS4163: Parallel Algorithm Analysis & Design, Spring 2005

• Instructor: Ed Luke
• Office: Butler 330 (or Engineering Research Center 220)
• Office Hours: 2:00pm-4:00pm M W (Or by appointment)
• Meeting Place: Butler 104
• Meeting Time: 1pm-1:50pm M W F
• Web Site: http://www.cse.msstate.edu/~luke/Courses/sp04/CS4163
• Instructor E-Mail: luke@cse.msstate.edu
• Formal Content Description
• Course Topics:
• Parallel Programming and Performance Models
• Amdahl's Law
• PRAM (CREW and EREW)
• Network Models (Hypercube, Mesh, Fat Trees, Crossbar, Ring)
• Bulk Models (LogP and BSP)
• Parallel Programming Patterns
• Bag of Tasks
• Data Parallel
• Reduction (global and local)
• Pipelining
• Divide and Conquer
• Scalability Metrics
• Isoefficiency
• Cost Optimality
• Optimal Effectiveness
• Parallel Programming Algorithms
• Matrix Operations
• Sorting
• Sorting Networks
• Bubble Sort and variants
• Quicksort
• Bucket Sort
• Radix Sort
• Convex Hull
• Required Text: Introduction to Parallel Computing: Second Edition by Ananth Grama, Anshul Gupta, George Karypis, and Vipin Kumar
• Text and References
• Class Notes
• Grading:
• Theoretical Assignments: 10%
Each missed theoretical assignment will drop 1 point.
• Programming Projects : 30%
• First Project: Search for Intelligent Puzzles (SIP) [ Project1.tar] [ project1_extra_datafiles.tar.gz] Due: Wednesday, February 16th
• Second Project: OpenMP Parallelization of a Heat Equation Solver [ Project2.tar], Due Friday, March 28th
• Third Project: Parallel QuickSort [ project3.pdf], Due Friday, April 29th [ Dr. Carver's HPCS slides][ Defect_log.xls ]


• Project Report: This is a research project, not just a programming assignment. I expect to receive, in addition to any program source developed, a comprehensive report that documents:
• Background Material Researched in Support of the Project
• Techniques and Methods used in the Report
• Results presented in sufficient detail that the results are reproduce-able from the description
• Analysis: What are the lessons learned. What general principles can be observed from the results.
• ALL NON-ORIGINAL work must be cited!
• No mercy will be given to students that submit plagiarized material. Any student that submits the work of others represented as their own will face disciplinary sanctions!
• 3 Partial Exams: 40%
• First Partial Exam: Covering Parallel Architectures, Performance Models, Message Passing, Parallel Algorithm Design Principles, Friday, February 11th. Test #1 Outline
• Second Partial Exam: Covering Broadcast Algorithms and scalability analysis, Friday, March 11th, Test #2 Outline
• Third Partial Exam: Covering Parallel Sorting and Matrix Algorithms, Friday, April 22, Test #3 Outline
• Comprehensive Final Exam: 20%
• Final Examination on May 6th (Friday) 8:00am-11:00am Final Outline
• Programming Project Grading Policy:
• Difficulty: Undergraduates get a break ! Easier options/algorithms.
• Individual Efforts: Team participation is not encouraged, unless otherwise stated.
• Submission: E-mail attachments sent to instructor
• Submit all source and include files and makefiles
• Do not submit binaries PLEASE !!!. Repeat offenders will face penalties.
• Partial credit given to non-functional programs
• Academic Honesty:
• The Departmental policy is a must-reading for all students.
• All web material used should be acknowledged otherwise you are culpable
• Remember, if caught -- NO EXCEPTIONS, NO MERCY.