Multi-Mode Inference Engine for Convolutional Neural Networks

TL;DR

This paper introduces a multi-mode inference engine for CNNs that significantly improves hardware efficiency, reducing latency and energy consumption while maintaining high performance across popular CNN architectures.

Contribution

The paper proposes a novel dataflow and a multi-mode inference engine that efficiently performs both convolutional and fully-connected operations using the same processing elements.

Findings

Achieves over 84% performance efficiency on AlexNet, VGGNet, and ResNet.

Reduces energy consumption, latency, and silicon area compared to state-of-the-art architectures.

Supports flexible CNN layer computations with a unified hardware approach.

Abstract

During the past few years, interest in convolutional neural networks (CNNs) has risen constantly, thanks to their excellent performance on a wide range of recognition and classification tasks. However, they suffer from the high level of complexity imposed by the high-dimensional convolutions in convolutional layers. Within scenarios with limited hardware resources and tight power and latency constraints, the high computational complexity of CNNs makes them difficult to be exploited. Hardware solutions have striven to reduce the power consumption using low-power techniques, and to limit the processing time by increasing the number of processing elements (PEs). While most of ASIC designs claim a peak performance of a few hundred giga operations per seconds, their average performance is substantially lower when applied to state-of-the-art CNNs such as AlexNet, VGGNet and ResNet, leading to low resource utilization. Their performance efficiency is limited to less than 55% on average, which leads to unnecessarily high processing latency and silicon area. In this paper, we propose a dataflow which enables to perform both the fully-connected and convolutional computations for any filter/layer size using the same PEs. We then introduce a multi-mode inference engine (MMIE) based on the proposed dataflow. Finally, we show that the proposed MMIE achieves a performance efficiency of more than 84% when performing the computations of the three renown CNNs (i.e., AlexNet, VGGNet and ResNet), outperforming the best architecture in the state-of-the-art in terms of energy consumption, processing latency and silicon area.