An Accuracy Reconfigurable Vector Accelerator based on Approximate Logarithmic Multipliers for Energy-Efficient Computing

The approximate logarithmic multiplier proposed by Mitchell provides an efficient alternative for processing dense multiplication or multiply-accumulate operations in applications such as image and real-time robotics. It offers the advantages of small area, high energy efficiency is suitable that do not necessarily achieve accuracy. However, its maximum error 11.1% makes it challenging to deploy requiring relatively This paper proposes a novel operand decomposition method (OD) decomposes one into sum multiple multiplications widely reduce errors while taking full advantage area savings. Based on OD method, this also accuracy reconfigurable (MAC) unit accuracies with parallelism. Compared MAC consisting accurate multipliers, significantly reduced less than half, improving hardware parallelism satisfying required various scenarios. experimental results show excellent applicability our smoothing robot localization mapping application. We have designed prototype processor integrates minimum functionality vector accelerator implemented software-level reconfiguration form instruction set extension. experimentally confirmed correct operation accelerator, which different degrees at software level.