Modular Dissipation Analysis for QCA - Field-Coupled Nanocomputing: Paradigms, Progress, and Perspectives - page 354

from irreversible information loss may be substantial in FCN circuits with densi-
ties and computational throughputs far exceeding “end-of-the-roadmap” CMOS
(e.g. [ 3 , 4 ]). This motivates investigation of fundamental lower bounds on dis-
sipation in FCN circuits, and of ultimate performance limits that follow from
these bounds under various assumptions regarding heat removal capability.
In previous work [ 3 , 5 ], we introduced a very general methodology that enables
determination of fundamental heat dissipation bounds for concrete and non-
trivial nanocomputing scenarios. We showed how irreversible information loss
can be isolated in any clocked quantum-cellular automata (QCA) circuit, and
how lower bounds on the resulting energy dissipation can be obtained. The
bounds resulting from this approach are truly fundamental, in that they depend
only on the circuit structure, clocking scheme, and temperature of the circuit's
environment; implementation-specific quantities such as the kink energy do not
appear in these energy bounds. Combining results from such analyses with
assumptions on circuit scale and clock rate for specified QCA circuits, lower
bounds on areal heat dissipation can be obtained at any desired computational
throughput for arbitrary QCA circuits.
In this work, we “modularize” this methodology to facilitate - and possibly
automate - the determination of fundamental dissipation bounds for large, com-
plex QCA circuits designed according to specified rules. We begin in Sect. 2 with
a review of our general methodology, which enables fundamental lower bounds
on heat dissipation to be determined for nanoelectronic circuits designed within
various paradigms through a manual analysis (hereafter the “general approach”).
We then show how the general approach, applied to QCA, can be “modularized”
by decomposing a circuit into smaller zones - such that dissipative contributions
can be evaluated separately for each zone and summed - and how this can sim-
plify dissipation analysis of QCA circuits (hereafter the “modular approach”).
We emphasize the enabling feature of this decomposition; that it preserves the
effects of field interactions across tile boundaries that influence the reversibility
of information loss while avoiding spurious contributions that can be introduced
if boundaries are improperly placed. Comparative application of the general and
modular analyses to a QCA adder circuit is presented in Sect. 3 . In Sect. 4 ,we
briefly discuss prospect for automation of QCA dissipation analysis using the
modular approach. Automated analysis could, for example, enable evaluation of
dissipation bounds for full QCA-based processor architectures [ 6 ] via the app-
roach of [ 7 ] by simplifying circuit-level dissipation analyses of the constituent
circuit blocks. We conclude this work in Sect. 5 .
2 Dissipation Analysis
2.1 General Approach
We begin by sketching our general approach for determination of fundamen-
tal lower bounds on the dissipative costs of digital computation. The approach
allows fundamental bounds to be obtained for specified circuits realized in a