Supercomputing and Massive Parallelism

1989 ◽  
pp. 503-511
Author(s):  
Bill Buzbee
Keyword(s):  
1996 ◽  
Vol 5 (2) ◽  
pp. 121-135
Author(s):  
Jack B. Dennis

Complex signal-processing problems are naturally described by compositions of program modules that process streams of data. In this article we discuss how such compositions may be analyzed and mapped onto multiprocessor computers to effectively exploit the massive parallelism of these applications. The methods are illustrated with an example of signal processing for an optical surveillance problem. Program transformation and analysis are used to construct a program description tree that represents the given computation as an acyclic interconnection of stream-processing modules. Each module may be mapped to a set of threads run on a group of processing elements of a target multiprocessor. Performance is considered for two forms of multiprocessor architecture, one based on conventional DSP technology and the other on a multithreaded-processing element design.


1996 ◽  
Vol 5 (3) ◽  
pp. 203-217
Author(s):  
John Gustafson ◽  
Srinivas Aluru

A number of "tricks" are known that trade multiplications for additions. The term "tricks" reflects the way these methods seem not to proceed from any general theory, but instead jump into existence as recipes that work. The Strassen method for 2 × 2 matrix product with seven multiplications is a well-known example, as is the method for finding a complex number product in three multiplications. We have created a practical computer program for finding such tricks automatically, where massive parallelism makes the combinatorially explosive search tolerable for small problems. One result of this program is a method for cross products of three-vectors that requires only five multiplications.


Sign in / Sign up

Export Citation Format

Share Document