ANCoEF: Asynchronous Neuromorphic Algorithm/Hardware Co-Exploration Framework with a Fully Asynchronous Simulator

TL;DR

ANCoEF introduces a co-exploration framework combining reinforcement learning-based hardware optimization and a fully asynchronous simulator, significantly improving neuromorphic hardware efficiency and accuracy for edge applications.

Contribution

It presents a novel asynchronous co-exploration framework with a faster simulator and RL-based optimization, advancing neuromorphic hardware design and performance.

Findings

RL-based optimization reduces hardware EDP by 1.81x on N-MNIST.

The framework improves SNN accuracy by 9.72% on DVS128Gesture.

The fully asynchronous simulator achieves over 2x speedup.

Abstract

Developing asynchronous neuromorphic hardware to meet the demands of diverse real-life edge scenarios remains significant challenges. These challenges include constraints on hardware resources and power budgets while satisfying the requirements for real-time responsiveness, reliable inference accuracy, and so on. Besides, the existing system-level simulators for asynchronous neuromorphic hardware suffer from runtime limitations. To address these challenges, we propose an Asynchronous Neuromorphic algorithm/hardware Co-Exploration Framework (ANCoEF) including multi-objective reinforcement learning (RL)-based hardware architecture optimization method, and a fully asynchronous simulator (TrueAsync) which achieves over 2 times runtime speedups than the state-of-the-art (SOTA) simulator. Our experimental results show that, the RL-based hardware architecture optimization approach of ANCoEF outperforms the SOTA method by reducing 1.81 times hardware energy-delay product (EDP) with 2.73 times less search time on N-MNIST dataset, and the co-exploration framework of ANCoEF improves SNN accuracy by 9.72% and reduces hardware EDP by 28.85 times compared to the SOTA work on DVS128Gesture dataset. Furthermore, ANCoEF framework is evaluated on external neuromorphic dataset CIFAR10-DVS, and static datasets including CIFAR10, CIFAR100, SVHN, and Tiny-ImageNet. For instance, after 26.23 ThreadHour of co-exploration process, the result on CIFAR10-DVS dataset achieves an SNN accuracy of 98.48% while consuming hardware EDP of 0.54 s nJ per sample.