CMSC 502
Parallel Programming

Fall, 2007.  Dr. Ju Wang

Class meetings: TR, 11:00AM-12:15PM, ENGRB 105.
Office Hours: TR 9:30 AM– 10:30 PM, ENGRB 406

TA: Jay N Vanzara, Room: B-5

Email: vanzarajn@vcu.edu

 

Office Hours: Wednesdays from 02.00pm to 04.00 pm.

§     Download: Lecture notes and Handout

o   Intro to Parallel Programming                                             

o   Matrix multiplications on Chapter 8, p345, p346, p347   

o   Linux/MPI howto                                                                 

o   MPI RUN HOWTO new version!!!        

o   Additional note1 for mapping                                               Oct 9

o   Additional note2 for mapping                                               Oct 9

o   An article about future parallel computing                          Oct 9

Final test


§     Project and assignment

Out                  Due                      Extra info

§       P1                   Sep 13                 Sep 25                 data generator

1024X1024 matrix:  A.dat, B.dat, C.dat

·                   P2               Sep 27                 Oct 9 Answer for P2

 

·                   P3               Oct 9                    Oct 30                 /home/project/p3

Extended to Nov, 6

     Submission guideline

Q&A about P3:

 

 when I do  mpic++ -g -o jpegtest jpegtest.cc for the test file, I get:

 

/usr/lib/gcc/i386-redhat-linux/4.0.0/../../../crt1.o(.text+0x18): In function `_start':

: undefined reference to `main'

/tmp/ccv7Uyxt.o(.text+0x2ba): In function `CJPEGTest::LoadJPEG(std::basic_st

 

You need to include a main.cc and the native jpeg library:

 

mpic++ -g -o jpegtest jpegtest.cc main.cc -ljpeg

 

here main.cc have the main() function, and the -ljpeg will link to a native jpeg library

which is from /usr/lib/libjpeg.a

--

 

 


·  Important class announcements!

·     Sep 4. Find you teammates and group information here.

·     Read: Section 2.1, 2.2, 2.4

·     Sep 6. You should be able to connect to the cluster by Friday,

·     Sep 7. Try to login in and walk through the self test steps in MPI RUN HOWTO new version!!!

o   Read: Chapter 6

·     Sep 27. Read 2.7: Process-Processor mapping.

o   Read page 36-39 about multistage networks (omega network).

·     Oct 17. No class on Thursday,

o   Midterm test will be Oct 25, next Thursday.

o   Check out the homework solution for HW2 above.

·     Oct 30, Read chap 4.

·     Nov 7, more about all-to-all personalized communication:

o   A better algorithm for ring (compared to figure 4.18, pg 172): two-stage forwarding

o   For D-hypercube: read section 4.5.3 (pg 175)

1.need D steps

2.Each node exchanges data with one node at a different dimension. (how many data is exchanged in one step?)

o   E-cube routing for all-to-all communication (figure 4.21, pg 178).

·     Distributed Dominant set Alg.

1.MPR

o   Notes on MPR

·     Topics after thanksgiving break:                  

·     Chapter 7

·     No 1 supercomputer whitepaper—blue gene and supercomputer slides                    

·     Sequence Alignment Alg                                                   

·     BLAST                                                    

·     Nov 26, I will be out of town this Tuesday. The class will be designated as an in-class discussion about supercomputer and BlueGene. Mr. Mbonisi will lead the discussion.

 

 


Course Description (syllabus)

(From VCU Bulletin) 3 credits
Software and hardware mechanisms for providing mutual exclusion in uniprocessor and multiprocessor environments. Architectures of multiprocessor systems and metrics for their evaluation. Design and use of parallel algorithms to solve concurrency problems in a distributed environment including message passing and remote procedure calls. Students will work in teams to design and implement parallel algorithms.

 

Distributed algorithms for networking problems. Distributed heuristic algorithms and their applications in wireless sensor networks.

    Prerequisites

  • CMSC312 (Operating Systems)
  • CMSC401 (Algorithm Analysis with Advanced Data Structures

   Textbook, Class Meetings, Method, Grading and Other Policies

  • The (required) textbook for this course is Parallel Computing by Grama, et al, Addision Wesley, 2nd edition, 2003. ISBN 0-201-64865-2

    Additional readings will provided by the instructor,.

 

Grade Determination Chart

Graded Component

% of
Final Grade

Comments

 

Test 1

20%

Tests are not comprehensive

 

Test 2

20%

 

 

 

 

 

 

 

 

 

Project 1

20%

All are group projects

 

Project 2

20%

 

 

Project 3

20%