In this paper, we obtain the lower bounds on total energy dissipation of deep submicron (DSM) VLSI circuits via an informationtheoretic framework. This framework enables the derivation of lower bounds under the constraint of reliable operation in presence of DSM noise. We employ noise-tolerance via coding to approach the lower bounds on energy dissipation. It is shown that the lower bounds on energy for an off-chip I/O signaling example are a factor of 24X below present day systems. It is also shown that reduction in energy consumption by a factor of 4X can be obtained by employing Reed-Muller codes to achieve the desired bit-error-rate (BER) in presence of DSM noise.
Rajamohana Hegde, Naresh R. Shanbhag