Information Technology Reference
In-Depth Information
4 Prospects for Automation
The modular dissipation analysis presented here is much better suited to automa-
tion than is the general approach. For easy comparison, the flow of the general
and modular dissipation analysis procedures is shown schematically in Fig.
9
.
An algorithm could certainly be devised that would enable automation of the
general approach, but it would be complex and dicult to formulate and imple-
ment. The general approach requires that data zones and subzones be identified
at each step, which requires that the flow of input information be tracked in space
and time throughout the computational cycle. Irreversible information loss and
associated energy dissipation depend upon changes in the amount of correlation
between the states of these zones during each step.
It would be comparatively straightforward to formulate an algorithm for
modular dissipation analysis of QCA circuits designed according to the design
rules of this work and driven by a random input string with specified input pmf.
The first step is simply to identify the device cells in the circuit, which could
easily be performed by searching a simple matrix representation of the circuit
layout. The second step is to determine the required joint pmfs and marginal
(input and output) pmfs for the gates corresponding to the dissipation zones
associated with each device cell. This could be achieved, perhaps in a simulator
embedding QCADesigner simulation of the circuit, by building appropriately
weighted input-output histograms for all gates in simulations that step through
all adder inputs. The third step is to evaluate the conditional entropies, and
thus the corresponding lower bounds on the dissipative contributions, for each
dissipation zone, which is easily done once the joint input-output pmfs have
Fig. 9.
Schematic representation of the general (top) and modular (bottom) dissipation
analysis procedures discussed in this work.
Search WWH ::
Custom Search