Nanomagnet Logic (NML) - Field-Coupled Nanocomputing: Paradigms, Progress, and Perspectives - page 20

Nanomagnet Logic (NML)

Wolfgang Porod 1( & ) , Gary H. Bernstein 1 , György Csaba 1 ,
Sharon X. Hu 2 , Joseph Nahas 2 , Michael T. Niemier 2 ,
and Alexei Orlov 1

1

Department of Electrical Engineering, University of Notre Dame,
Notre Dame, IN, USA
[e-mail]

2

Department of Computer Science and Engineering,
University of Notre Dame, Notre Dame, IN, USA

Abstract. We describe the background and evolution of our work on magnetic
implementations of Quantum-Dot Cellular Automata (QCA), first called
Magnetic QCA (MQCA), and now known as Nanomagnet Logic (NML).

Keywords: Nanomagnet logic Quantum-Dot Cellular Automata Field-
coupled computing Cellular Automata

1 Introduction

We all are familiar with the fact that magnetic phenomena are widely used for data
storage, whereas electronic phenomena are used for information processing. This is
based on the fact that ferromagnetism is nonvolatile (i.e. magnetization state can be
preserved even without power), and that electrons and the flow of charge can
be effectively controlled to perform logic. However, charge-based logic devices are
volatile, which means that a power supply is needed to maintain the logic state and to
stop information (electrons) from leaking away.
Presently, there are multiple research efforts underway to harness magnetic phe-
nomena for logic in addition to storage. Motivation for this work is two-fold. First, the
amount of static power dissipation in CMOS chips now rivals the levels of dynamic
power dissipation. In other words, even if a chip does not perform any computation,
stand-by power (i.e., needed to maintain volatile logic state, etc.) is similar to the
power dissipated when performing useful work. Anecdotally, in 2004, Bernard
Myerson - then chief technologist for IBM's system and technology group - likened
the situation to ''a car with a 10-gallon gas tank losing 5 gallons while parked with its
motor turned off'' [ 2 ]. The nonvolatile nature of magnetic phenomena means that
there is no stand-by power dissipation. Second, the intrinsic switching energy of a
magnetic device can be orders of magnitude lower than a charge-based CMOS
transistor. While some drive circuitry overhead must be accounted for in magnetic
systems, this suggests that magnetic logic could help to minimize dynamic power
dissipation as well. Thus, while the multi-decade Moore's Law-based size scaling
trends may continue, associated performance scaling trends are threatened by energy-
related concerns that magnetic devices could help to alleviate.