- In the realm of sequential computing the random access machine model
has played a successful role in bridging the gaps among algorithm
designers, computer architects and language experts. While the drive
for performance in parallel computing strongly motivates the
development of an analogous abstract machine model, in this field
there has been no corresponding success. Indeed, the modeling of
parallel computing is mired in controversy and chaos.
In this paper we selectively examine models of parallel computation which have been used in various disciplines of computer science to better understand what performance characteristics are crucial in each of these disciplines. Our objective is to foster the development of an appropriate computational model by providing an overview and synthesis of the characteristics of proposed models within a simple logical framework. As an impetus for discussion and development we conclude by suggesting a parallel computational model which is consistent with a model design philosophy which balances simplicity and descriptivity with prescriptivity.