Shenjing: A low power reconfigurable neuromorphic accelerator with partial-sum and spike networks-on-chip

TL;DR

Shenjing is a configurable, low-power neuromorphic accelerator that efficiently maps various neural network models onto hardware without retraining, significantly reducing energy consumption for on-device AI applications.

Contribution

It introduces Shenjing, a reconfigurable SNN architecture with fully software-exposed communication, enabling direct mapping of diverse neural networks without model modification or retraining.

Findings

Achieved 96% accuracy on MNIST with 1.26mW power consumption.

Successfully mapped ANNs like CNNs and residual networks onto Shenjing.

Demonstrated energy-efficient inference suitable for on-device AI.

Abstract

The next wave of on-device AI will likely require energy-efficient deep neural networks. Brain-inspired spiking neural networks (SNN) has been identified to be a promising candidate. Doing away with the need for multipliers significantly reduces energy. For on-device applications, besides computation, communication also incurs a significant amount of energy and time. In this paper, we propose Shenjing, a configurable SNN architecture which fully exposes all on-chip communications to software, enabling software mapping of SNN models with high accuracy at low power. Unlike prior SNN architectures like TrueNorth, Shenjing does not require any model modification and retraining for the mapping. We show that conventional artificial neural networks (ANN) such as multilayer perceptron, convolutional neural networks, as well as the latest residual neural networks can be mapped successfully onto Shenjing, realizing ANNs with SNN's energy efficiency. For the MNIST inference problem using a multilayer perceptron, we were able to achieve an accuracy of 96% while consuming just 1.26mW using 10 Shenjing cores.