CPT S 483 (section 02): Introduction to Parallel Computing

Fall 2015

(August 24 - December 18)
SCHOOL OF EECS
WASHINGTON STATE UNIVERSITY

TuTh 14:50-16:05
JKSN 55

Course Information        Policies      Lecture Notes        Course Schedule        Homeworks        Programming Projects        Survey Project    (old announcements archive)

CURRENT ANNOUNCEMENTS

• Welcome to the course website for Cpt S 483: Introduction to Parallel Computing, Fall 2015 edition. This website is under construction. So please look out for updates in the forthcoming weeks.  See you in class on the first day of class, Tuesday August 25.

COURSE INFORMATION

Instructor:

ANANTH KALYANARAMAN

EME 237, 335-6760

Weekly office hours:  Wednesdays 2-3pm

Instructor email contact: Through OSBLE (http://www.osble.org)

Syllabus:

    This course will serve as a basic introduction to the field of parallel computing. Emphasis will be on the fundamental principles for parallel algorithm design and analysis, and parallel programming for distributed and shared memory parallel machines. Topics to be covered (not necessarily in order) include (but not limited to):

1. Fundamentals in parallel algorithm design and analysis
2. Parallel computation models
3. Network intraconnects and embeddings
4. MPI programming
5. OpenMP shared memory multicore programming
6. Parallel reduction operations
7. Parallel prefix
8. Sorting
9. Matrix operations
10. Parallel scientific computing applications
11. MapReduce and cloud computing

Reference texts:

    There will *not* be any required text for this edition of the course. We will use the lecture notes available online on this website, along with lecture scribes, and a variety of online tutorials and reading materials identified by the instructor (also made available through the Lecture Notes link on this page). The following books will serve as references:

• "Introduction to Parallel Computing" by A. Grama, G. Karypis, V. Kumar, A. Gupta, Addison-Wesley, 2nd Edition
• "Introduction to Parallel Programming" by Peter S. Pacheco, Morgan Kaufmman
• "Parallel Programming with MPI" by Peter S. Pacheco, Morgan Kaufmman
• "Using OpenMP" by B. Chapman, G. Jost, A.R.van der Pas, The MIT Press
• "Introduction to Parallel Algorithms and Architectures" by F.T. Leighton, Morgan Kaufmann

Course Prerequisites:

• Familiarity with design and analysis of algorithms is required (CptS 223 or equivalent is required, Cpt S 350/algorithm design & analysis or equivalent is preferred although not required).
• Thorough programming experience in C or C/C++ is required.
• Basic knowledge and experience using Unix/Linux environments.

 Interested students who do not meet the above criteria are highly encouraged to contact the instructor to discuss the possibility of enrolling into the course.

Course Webpage & OSBLE:

    The course will use two web resources for two different purposes:

• The "course website" (i.e., this page you are reading now) is where lecture notes, all assignments will be posted, and the course schedule will be maintained.
• The OSBLE learning portal will be used for email exchanges, email announcements, and listing of useful web links/resources.
  Please see here for instructions on how to set up your OSBLE account and other personal settings. This is NOT an optional step. All students enrolled in this class should take care of this by the first week of the semester.

• Mail Alert:    Students are advised to check their OSBLE mailbox at least once a day.

COURSE POLICIES  

Grading policy:

• (15%) Homeworks
• (50%) Programming projects
• (15%) 1 midterm exam
• (15%) White paper 
• (5%) Classroom participation

Homework Policy:

By default, every problem in the homework should be done "individually". If "collaboration" is allowed for a subset of problems in a homework then it will be explicitly stated so in the homework. "Individual" assignments means that no collaborations are allowed. You are expected to work on the problems all by yourself. Consulting solutions online will also be considered "cheating" and is strictly prohibited. For "collaborative" assignments, strict adherence to the following guidelines is required:

• "Collaboration" is defined as a discussion with other students in the same class (no outsiders allowed) aimed at obtaining a better understanding of the problem question and/or exploring potential approaches at a very high level that can lead to a solution.
• All collaborative efforts should be explicitly acknowledged/cited in the answer sheet by all the participants using the cover sheet.
• Regardless of whether you collaborate or not, the final writing in the answer sheet should be *solely yours*.
• No points will be deducted for collaboration as defined above.

Programming Project Policy:

The statement of the programming project will clearly mention if it is intended to be an "individual" or "team" project. Each team can contain no more than 3 students, with the average team size being 2. Teams need to be identified before starting work on a project, and team projects will be evaluated no differently from individual projects - i.e., the grade given to a team will apply to all individual members of that team.
All team members of a team will receive the same grade on each programming project. Therefore, when working in teams, it is expected that all of the team members contribute equally to the project. If there is a disparity in the levels of contribution, the student members are expected to bring that to the attention of the instructor.

Also, all the source code pertaining to each programming project will have to be solely written from scratch by the project team. Reuse of any code that is supplied by the instructor along with the programming assignment is allowed.

However, use of any code snippets from the internet or from other project teams or elsewhere is strictly prohibited, and if evidence is found it will be considered "cheating".

Academic Integrity Policy:
*Please read carefully*

• What constitutes cheating? Any deviation from the above homework and programming project policy will be considered "cheating" and will be subject to academic dishonesty code. Note that this includes sharing or even showing your solutions with others in the class. Also, looking up solutions on the web should be strictly avoided and if evidence is found, it will be considered cheating.
• Consequence of cheating: A student who has been conclusively caught cheating in the class will be given an F grade and will also be reported to the WSU academic integrity/student conduct office.  In other words, there will be Zero Tolerance toward cheating.
  

Academic integrity is the cornerstone of the university. You assume full responsibility for the content and integrity of the academic work you submit. You may or may not collaborate with classmates on assignments as stated in each assignment explicitly. For collaborative assignments, however, the guiding principle of academic integrity shall be that your submitted work, examinations, reports, and projects must be your own work. Any student who violates the University's standard of conduct relating to academic integrity will be referred to the Office of Student Conduct and may fail the assignment or the course. You can learn more about Academic Integrity on your campus at  http://conduct.wsu.edu. Please also read this link carefully: EECS Academic Integrity Policy   Please use these resources to ensure that you don’t inadvertently violate WSU's standard of conduct.  
 

Graduate students: please also refer to http://gradschool.wsu.edu/FacultyStaff/Committee/

Late submission policy:

• No late submissions will be allowed on homeworks unless the instructor grants an extension to the entire class.
• For programming projects, a 24-hour grace period will be provided with 10% penalty in points. Projects submitted after the 24-hour grace period will not be graded.
• Exceptions to the above rules may be made under extraordinary circumstances upon obtaining prior permission from the instructor, at least 1 week prior to the submission date. Last minute permissions will not be entertained.

Exam policy: Closed-book, closed-notes, comprehensive

PROGRAMMING ASSIGNMENTS  - click on this link

HOMEWORKS

• All homework problems will be posted/updated in this PDF throughout the course. Look for the problems in the "Active problems" section for problems that are currently due. The "Retired problems" is a running archive of problems that have been completed.

Click here to download the cover sheet for use in all the homeworks and programming projects 

LEARNING OUTCOMES

• Obtain a conceptual understanding of the fundamental principles behind parallel computing, parallel programming models, and parallel algorithm design and analysis.

• Obtain proficiency in parallel programming using MPI and OpenMP (for distributed memory and shared memory multicore machines, respectively), while also being able to identify other emerging parallel computing models and environments.

• Obtain proficiency in parallel algorithm design by demonstrating the ability to design parallel algorithms , analyze their complexities, estimate/predict performance, implement and explain empirical observations.

• Demonstrate the ability to differentiate among parallel architectures and network intraconnect models.

• Obtain proficiency in reporting of empirical tests and programming exercises in a manner that is consistent with scientific literature.

SAFETY ON CAMPUS

Washington State University is committed to enhancing the safety of the students, faculty, staff, and visitors. It is highly recommended that you review the Campus Safety Plan (http://safetyplan.wsu.edu/) and visit the Office of Emergency Management web site (http://oem.wsu.edu/) for a comprehensive listing of university policies, procedures, statistics, and information related to campus safety, emergency management, and the health and welfare of the campus community

STUDENTS WITH DISABILITIES

Reasonable accommodations are available for students with a documented disability. If you have a disability and need accommodations to fully participate in this class, please either visit or call the Access Center (Washington Building 217; 509-335-3417) to schedule an appointment with an Access Advisor. All accommodations MUST be approved through the Access Center.

OLD ANNOUNCEMENTS ARCHIVE

COURSE ARCHIVES