Architectures for nanoscale hybrid computing systems

Abstract

Spin torque oscillator (STO) nanodevices have been brought into focus of engineering hoping they could provide for a platform of computation beyond Moore's law. In this paper we propose hybrid-architectures (i.e. combining CMOS units and STO nanodevices) useful to realize Oscillatory Cellular Nonlinear Network (O-CNN) arrays that can be used for associative memory (AM) problem-solving. The fundamental components of the AM O-CNN are (1) a CMOS preprocessing unit generating input feature vectors from picture inputs, (2) an AM cluster generating signature outputs composed of spin torque oscillator (STO) cells and local spin-wave interactions, as an oscillatory CNN (O-CNN) array unit, applied several times arranged in space, and (3) a classification unit (CMOS). In this manuscript we focus on the AM cluster composed of several STO and we aim at showing how local spin-wave interactions lead to global indirect interactions. In addition, a mathematical methodology is proposed in order to design the fully-connected AM cluster of STO exploiting the local spin-wave interactions due to physical limits of the implementation.