CSE721: Introduction to Parallel Computing (Wi 2006)

Instructor

Grader

Course Description

Course Web site

Meeting time and place

Level and Credits

• UG 4

Prerequisites

• CIS 621.

Objectives

• Mastery of basic concepts of parallel computing;
• Mastery of parallel algorithm design and analysis;
• Familiarity of performance evaluation for parallel programs;
• Familiarity of problem solving on parallel computers;
• Exposure to several advanced parallel computers, such as, Cray T94, SGI Origin 2000, and cluster of PCs.

Texts

• [GGKK03] A. Grama, A. Gupta, G. Karypis, V. Kumar, Introduction to Parallel Computing, 2nd edition, Addison-Wesley, 2003.
  

Topics (tentative)

Grading Plan

NOTE: there will be no final exam for this class, and the final grading will be based only on the project, assignement and midterm grades according to the above weights.
 

Exam Schedule

Project Schedule

Handouts

• Myricom, Inc. Guide to Myrinet switches - info on Clos networks.
• Fabrizio Petrini and Marco Vanneschi. k-ary n-trees: High Performance Networks for Massively Parallel Architectures. In Proceedings of the 11th International Parallel Processing Symposium, IPPS'97, pages 87-93, Geneva, Switzerland, April 1997.
• Ed Huai-hsin Chi, Elizabeth Shoop, John Carlis, Ernest Retzel, and John Riedl. Efficiency of Shared-Memory Multiprocessors for a Genetic Sequence Similarity Search Algorithm. Number TR97-05. University of Minnesota, Computer Science Department, Minneapolis, MN, Janurary, 1997.
• E. Smirni and D.A. Reed, Lessons from Characterizing the Input/Output Behavior of Parallel Scientific Applications
  Performance Evaluation , 33, 1998, pp. 27-44.

Useful Links

Misc

PROJECT.  For your term project, you can choose any topic pertaining to the subject matter of this course, i.e. algorithmic or performance issues in parallel computing. Here are some possibilities:

• Study of how to augment an existing MPI application so to use MPI-IO for I/O operations and report the effect on the performance. Here is a list of  MPI codes that can have a significant volume of I/O and therefore are appropriate for this type of project. For these project a group of no more than two students would be acceptable, with the understanding that both students get the same grade for the project.
  
• Implementation of a distributed application using the BOINC software platform using a suitable application (e.g. NCBI's BLAST).
  
• If you are not able to perform one of the projects above (please send me a brief explaination), here are some more suggestions.
  
• Implementation, performance evaluation, and analysis of experimental performance measurement for a comparative study of algorithms for some computation/application. At least two OSC systems should be used, and preferably at least two alternative implementations should be compared.
• Detailed study on the cost of one or more collective communication algorithms: this can either be analytical (for some context not covered in the text, e.g. remove single-port assumption) or experimental (e.g. evaluation of different alternatives for a system such as the SGI Origin2000). 
• Survey of available public domain parallel software for some chosen computation/application, and experimental comparison on one or more OSC systems.
• Here is a list of projects done by CIS721 students in the past.

READING ASSIGNMENTS. Some light readings will be assigned from the text book for self reading.

HANDOUTS. Right now I don't anticipate any handouts in addition to those listed above; however, if I come across something interesting and relevant to the course, its copies will be distributed in the class.

MAKE-UP EXAMS. Make-up examination will only be given under circumstances where a student is prevented from taking the regular examination due to unforeseen circumstances (e.g., sickness, flat tire), and a formal proof will be required to prove the prevailing circumstance.

STUDENTS WITH DISABILITY. Any student who feels he or she may need an accomodation based on the impact of a disability should contact me privately to discuss his or her specific needs. Please contact the Office for Disability Services at (614) 292-3307, or visit 150 Pomerene Hall, to coordinate reasonable accommodations available for students with documented disabilities.

ACADEMIC HONESTY. Students are expected to know and abide by OSU's policy on academic integrity. Any work you submit, whether in a homework set or on an exam, must be your own creation, unless exclusions are explicitly described. Collaboration is allowed for working out general principles and establishing your knowledge of the material presented in the course, but you should not show your work to another student, receive a copy of work done by another student, or copy any work created by another student - whether a paper or electronic copy. All instances of suspected academic misconduct will be reported to the department chairperson and the Committee on Academic Misconduct. Violations of the student code of conduct often result in the student receiving a failing grade in the course. The best way to avoid the temptation to cheat is to start on your assignments in time to ask for help from the instructor or grader. It is in your best interest to learn the material included on the homeworks in order to perform well on the exams. Don't jeopardize your GPA by flouting the standards of academic integrity expected of OSU students.

Homeworks

• Homework #1 - (assigned Jan 12 - due Jan 19)
• Homework #2 - (assigned Jan 30 - due Feb 7)
  
• Homework #3 - (assigned Feb 16 - due Feb 28)
  
• Homework #4 - (assigned Feb 28 - due Mar 7) - CANCELLED
  

Lecture slides

• Mar 7:

• K.A.
• N.A.
• E.B.
• J.B.
• R.C.
• Y.D.
• M.I.

• Mar 9:
• B.L.
• G.L., Y.Y.
• W.M., Y.T.
• C.O.
• N.S.
• S.T., G.Z.
• J.X.

•   Final grades

Last updated: Mar 15