Title: 	High Performance Computing

Units:	2

Qtr:	Winter 1993-94

Text:	High Performance Computing, Kevin Dowd, O'Reilly & Associates



Abstract:  This course covers effective techniques for mapping
numerically intensive algorithms to modern high performance
computer platforms such as the Cray C-90 or the IBM RS6000. 
This includes understanding hardware parallelism and software 
pipelining, compiler directives and limitations, source code
transformations to enhance register and memory usage, and when
and how to delve into assembly language for maximum
performance.  Hands-on exercises will develop skills in applying
general principles to a variety of architectures, including RISC,
vector, and VLIW machines.



Prerequisites:  Familiarity with Fortran and C are highly recommended. 
Prior assembly language experience helpful, but not required.



Maximum enrollment: 24 plus auditors



Labs:	Much of the course work will consist of computer labs requiring
implementation of carefully selected algorithms on various architectures.