Novel architecture of multiplier for Accuracy-scalable approximate computing

Abstract

Many error-tolerant applications rely on multiplication as a fundamental function.
Approximate multiplication is regarded as an efficient method for balancing performance and accuracy
against energy consumption. As a result, this research presents a low-power and compact multiplier for
accurate, scalable approximate computation. A carry-maskable adder and an approximate tree
compressor (ATC) make up the proposed approximate multiplier (CMA). By adopting a simple circuit
topology and a simple carry-masking technique, ATC compresses partial products while the CMA
provides precision scaling. This approximate multiplier reduces both power consumption and circuit
area when compared to the Wallace tree multiplier, according to the experimental designs.