Abstract
The author advocates a concentration on two main classes of highly parallel architecture rather than a proliferation, which confuses the direction of software engineering developments. He discusses the two classes. DAP and supernode transputer arrays, and compares their performance. He concludes that the idea of welding together the SIMD and MIMD architectures to provide a very general capability machine should be discouraged. Instead, their individual versatility suggests that either could be used in many roles, and the complications of data interfacing and dual standards of programming data interfacing are unnecessary.

This publication has 9 references indexed in Scilit: