Implementation of parallel control algorithms on a hypercube
Eric Rogers · 1990
Summary form only given. The author focuses on one particular aspect of parallel processing applied to control problems in the form of hypercube multiprocessors. The so-called distributed-memory hypercubes are exclusively considered in which the processors execute programs independently and communicate only by passing messages. The material presented begins with a review of the architecture and programming modes. Following this, algorithms and codes for solving a representative control systems problem, a Riccati equation are presented. A comparison of the performance of these against solutions run on other machines is given in the form of timing results. Finally, some on going work and future research areas are highlighted.