CSE4163: Parallel Algorithm Analysis & Design, Spring 2007
- 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: 1:00pm-1:50pm M W F
- Web Site: http://www.cse.msstate.edu/~luke/Courses/sp07/CSE4163
- 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: Collective Communication Operations project1.tar Due Monday, February 26th Updated Job Scripts: job_scripts.tar
- Second Project: Parallel Nearest Neighbors project2.1.tar Due Thursday May 3rd (final exam day) Note: Updated project2 tar file includes fixed verification routine! Please download the 2.1 verion of the file.
Updated jobscripts for project 2: jobscripts_p2.tar
- 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: Architecture, MPI and Communications Algorithms, outline , Friday, Febuary 16th
- Second Partial Exam: Parallel Algorithm Design Principles, Analysis of Parallel Algorithms, outline , Friday, March 30th
- Third Partial Exam: Sorting and Matrix Algorithms, outline , Monday, April 23rd
- Comprehensive Final Exam: 20%
-
Final Examination on May 3rd (Thursday) 8:00am-11:00am
-
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.
luke@cse.msstate.edu
Last modified: Thu Apr 19 14:49:56 CDT 2007