Neural network architectures for systolic arrays

Summary form only given, as follows. The authors propose the use of an artificial neural network model to simulate the functions and operations of a systolic array. A systolic array and a neural network are both easy to implement and easy to configure. Since a neural network can be implemented on a programmable VLSI chip, it is very fast, easy to reconfigure, and cost-effective. It is concluded that a wide variety of designs of systolic arrays can easily be simulated on neural networks.<
>