Binary Adder Circuits of Asymptotically Minimum Depth, Linear Size, and Fan-Out Two

TL;DR

This paper introduces a new family of binary adder circuits that achieve near-optimal depth of log_2 n with linear size and fan-out two, improving upon longstanding theoretical bounds.

Contribution

It presents the first adder circuits that combine minimal depth, linear size, and fan-out two, surpassing previous constructions in theoretical efficiency.

Findings

Achieve asymptotically optimal depth of log_2 n + o(log_2 n)

Maintain linear size O(n)

Ensure fan-out is bounded by two

Abstract

We consider the problem of constructing fast and small binary adder circuits. Among widely-used adders, the Kogge-Stone adder is often considered the fastest, because it computes the carry bits for two $n$-bit numbers (where $n$ is a power of two) with a depth of $2\log_2 n$ logic gates, size $4 n\log_2 n$, and all fan-outs bounded by two. Fan-outs of more than two are avoided, because they lead to the insertion of repeaters for repowering the signal and additional depth in the physical implementation. However, the depth bound of the Kogge-Stone adder is off by a factor of two from the lower bound of $\log_2 n$. This bound is achieved asymptotically in two separate constructions by Brent and Krapchenko. Brent's construction gives neither a bound on the fan-out nor the size, while Krapchenko's adder has linear size, but can have up to linear fan-out. With a fan-out bound of two, neither construction achieves a depth of less than $2 \log_2 n$. In a further approach, Brent and Kung proposed an adder with linear size and fan-out two, but twice the depth of the Kogge-Stone adder. These results are 33-43 years old and no substantial theoretical improvement for has been made since then. In this paper we integrate the individual advantages of all previous adder circuits into a new family of full adders, the first to improve on the depth bound of $2\log_2 n$ while maintaining a fan-out bound of two. Our adders achieve an asymptotically optimum logic gate depth of $\log_2 n + o(\log_2 n)$ and linear size $\mathcal {O}(n)$.