INTENDED AUDIENCE
This course is designed for CS majors with applied mathematics and applied science aptitude, and students from applications departments (engineering and the sciences) with computational aptitude and prior programming experience.
Parallel computing will be taught not in the abstract (though this is possible, and appropriate in some contexts), but in conjunction with scientific computing. The intractable problems of scientific computing are the prime motivating forces for the intense focus of manpower and money on the development of parallel computing in the 1990's under the federal High Performance Computing and Communication initiative. They are also principally responsible for the instructor's own interest in parallel computing.
LAB FACILITIES
Students will solve computational problems in parallel on distributed memory ensembles of SUNs and SGIs at Old Dominion University. External parallel processors will be utilized as available. The programming efforts will not all be "from scratch" efforts at coding. Where appropriate, debugged modules useful for various well-defined sub-tasks will be provided in order to free the student for creative and intellectually challenging exercises.
TEXTBOOK/READINGS
There is one required text: Scientific Computing: An Introduction with Parallel Processing by Gene Golub and James M. Ortega (Academic Press, 1993). Two somewhat more specialized monographs are helpful in extending this text: Solution of Partial Differential Equations on Vector and Parallel Computers by James M. Ortega and Robert G. Voigt (SIAM, 1985) and Solving Linear Systems on Vector and Shared Memory Computers by Jack J. Dongarra et al. (SIAM, 1991).
TOPICS
0. The "Why" of Parallel Computing 1. Numerical Algorithms 2. Elementary Parallel Computer Architecture 3. Algorithm-Architecture Interaction 4. Introduction to Parallel Performance Metrics and Benchmarks 5. Introduction to High-level and Portable Parallel Software Libraries 6. Some Scientific Computing Applications