Beyond-CMOS Device Benchmarking for Boolean and Non-Boolean Logic Applications

TL;DR

This paper benchmarks various beyond-CMOS charge- and spin-based devices, extending methodologies to interconnect and non-Boolean logic, revealing potential advantages of spintronic devices in non-Boolean applications.

Contribution

It introduces new benchmarking methods for beyond-CMOS devices, including non-Boolean logic, and compares device performance with a focus on interconnect and neural network applications.

Findings

Spintronic devices can outperform CMOS in non-Boolean logic.

Deep pipelining boosts throughput of low-power devices.

Extended benchmarking includes interconnect-centric and neural network applications.

Abstract

The latest results of benchmarking research are presented for a variety of beyond-CMOS charge- and spin-based devices. In addition to improving the device-level models, several new device proposals and a few majorly modified devices are investigated. Deep pipelining circuits are employed to boost the throughput of low-power devices. Furthermore, the benchmarking methodology is extended to interconnect-centric analyses and non-Boolean logic applications. In contrast to Boolean circuits, non-Boolean circuits based on the cellular neural network demonstrate that spintronic devices can potentially outperform conventional CMOS devices.