Limits on high-performance applications

Abstract

There is evidence that some high performance applications such as Air Traffic Control and other real time dynamic IT and database tasks are NP-hard. This paper proposes that it is the nature of the basic computing model, not the tasks themselves that results in intractable situations, and that it is the communication component of many high performance IT applications that makes them intractable. Consequently, bigger, faster, higher performance hardware using this model cannot solve these tasks. Networks and clusters of computers simply move the communication bottleneck, ultimately aggravating rather than helping the situation. New models of computing are needed. 'It is no easy matter to root out old prejudices, or to overturn opinions which have acquired an establishment by time, custom and great authorities.' It is proposed that the associative model can significantly reduce if not eliminate this bottleneck for many commercial high-performance applications. Moreover, the associative model is well suited for the coming generation of high performance optical, biological and polymer computers.