Subtractor-Based CNN Inference Accelerator

TL;DR

This paper introduces a subtractor-based CNN inference accelerator that replaces multiplications with subtractions, significantly reducing power and area consumption with minimal accuracy loss, by preprocessing weights through sorting, grouping, and rounding.

Contribution

It proposes a novel method to replace multiplications with subtractions in CNN inference, enabling power and area savings while controlling accuracy trade-offs.

Findings

32.03% power savings achieved

24.59% area reduction demonstrated

0.1% accuracy loss with proposed method

Abstract

This paper presents a novel method to boost the performance of CNN inference accelerators by utilizing subtractors. The proposed CNN preprocessing accelerator relies on sorting, grouping, and rounding the weights to create combinations that allow for the replacement of one multiplication operation and addition operation by a single subtraction operation when applying convolution during inference. Given the high cost of multiplication in terms of power and area, replacing it with subtraction allows for a performance boost by reducing power and area. The proposed method allows for controlling the trade-off between performance gains and accuracy loss through increasing or decreasing the usage of subtractors. With a rounding size of 0.05 and by utilizing LeNet-5 with the MNIST dataset, the proposed design can achieve 32.03% power savings and a 24.59% reduction in area at the cost of only 0.1% in terms of accuracy loss.