Skip to content
Log in
Open access

A Hardware-Aware Analysis of Energy Efficiency and Quantization Trade-Offs in Spiking Neural Networks

A. Castagnetti Alain Pegatoquet Benoît Miramond
2026 · IEEE Journal on Selected Areas in Information Theory · Vol 7, pp. 411-430 · 0 citations · 63 references

TL;DR

An in depth analysis of the different trade-offs between quantization, generalization performance, and energy efficiency between binary SNNs, multi-level SNNs and ANNs for two different applications scenarios: image classification and image denoising and results show that multi-level spiking neurons provide better information compression, allowing therefore a reduction in latency without performance loss for classification tasks.

Abstract

Spiking Neural Networks (SNNs) are one of the most promising bio-inspired neural networks models and have drawn increasing attention in recent years. The event-driven communication mechanism of SNNs allows for sparse and theoretically low-power operations on dedicated neuromorphic hardware. However, the binary nature of instantaneous spikes also leads to considerable information loss in SNNs, resulting in performance degradation. To address this issue, multi-level spiking neuron models have been proposed to provide both low-quantization error and minimal inference latency while approaching the performance of full precision Artificial Neural Networks (ANNs). In this paper, we provide an in depth analysis of the different trade-offs between quantization, generalization performance, and energy efficiency between binary SNNs, multi-level SNNs and ANNs for two different applications scenarios: image classification and image denoising. Our goal is to determine analytical thresholds on SNNs activity to understand when SNNs can be more energy efficient than ANNs but also to explain why SNNs cannot provide energy gains in other scenarios. Our analysis is based on an hardware execution model, which allows us to take into account the energy costs related to synaptic operations as well as memory accesses. Our experimental results show that multi-level spiking neurons provide better information compression, allowing therefore a reduction in latency without performance loss for classification tasks. When compared to binary SNNs on image classification scenarios, multi-level SNNs indeed allow reducing by 2 to 3 times the energy consumption depending on the number of quantization intervals. On neuromorphic data, multi-level SNNs allows us to drastically reduce the inference latency to 1 timestep, which corresponds to a compression factor of 10 compared to previously published results on binary SNNs. However, for the image denoising task we observe that neither the binary nor the multi-level SNNs can match the ANNs energy efficiency and performance. Based on a hardware-aware execution model of a neuromorphic SNN accelerator, we then provide analytical upper-bounds on SNNs activity and energy-efficiency. These theoretical bounds are then compared against the experimental results and allows us to understand, from a system level point of view, the relation between energy efficiency, quantization schemes and finally the performance of the SNNs.

DOI Semantic Scholar
Read PDF

Similar papers

Preprint Jul 2026

A Hardware-Aware Open-Source Framework for Design Space Exploration of Mixed-Signal Spiking Neural Networks

Energy-efficient neuromorphic computing at the edge requires simulation tools that can capture the non-ideal behavior of mixed-signal spiking neural network (SNN) hardware while supporting system-level design exploration. This work presents an open-source hardware-aware simulation framework for mixed-signal SNNs that enables comparative analysis across neuron, synapse and architecture choices. The framework supports multiple neuron models, including Leaky Integrate-and-Fire (LIF), Hodgkin-Huxley (HH) and Axon-Hillock (AH), together with non-volatile analog synapses based on floating-gate transistors and ReRAM devices. By incorporating device-level nonlinearities directly into PyTorch-based training and inference, the tool enables optimization of physical synaptic parameters rather than idealized abstract weights. The framework is evaluated on standard neuromorphic benchmarks, including N-MNIST, DVS Gesture and Spiking Heidelberg Digits (SHD). For each model dataset configuration, it reports classification accuracy together with hardware-oriented metrics such as silicon area, power consumption and quantization sensitivity. These capabilities enable cross-layer design space exploration and help identify neuron-synapse configurations that best satisfy application-specific constraints on accuracy, energy efficiency, area and hardware fidelity.

Sayma Nowshin Chowdhury, V. Nair, Taseen Forhad et al. · 0 citations
Preprint Aug 2026

BASC : Behavior-Aligned Quantization and Pruning for Low-Bit Spiking Neural Networks

Extensive experiments on static and neuromorphic benchmarks show that lower-bit BASC models match or outperform higher-bit baselines and retain this accuracy advantage after structured pruning, while further reducing model storage and synaptic operations.

Linliang Chen, Yan Zhong, Xin Liu et al. · 0 citations
Review Jul 2026

Spiking Neural Networks: A Computational Paradigm for Neuromorphic Computing

It is aimed at proving that SNNs have potential in such areas as computer vision, robotics, and speech recognition, and their role in overcoming the barrier between artificial and biological neural systems is proved.

Mesala Sravani, K. Kumari, S. M. Reddy · 0 citations
Open access Jul 2026

Spiking Neural Network Based on Hierarchical Residual Quantization and Temporal Error Compensation for Remote Sensing Object Detection

A spiking neural network based on hierarchical residual quantization and temporal error compensation (HRQ-TEC-SNN) and used for remote sensing object detection tasks is proposed and the model’s detection accuracy has been further improved while reducing the number of model parameters and computational costs.

Yaming Cao, Yukai Xing, Liqun Kuang et al. · 0 citations
Jul 2026

A Fair Energy Comparison of Spiking and Artificial Neural Networks on a General-Purpose Microcontroller.

Energy efficiency is a key constraint for deploying machine learning models on resource-limited internet of things (IoT) and edge devices. Spiking Neural Networks (SNNs) are often proposed as an energy-efficient alternative to conventional Artificial Neural Networks (ANNs) due to their event-driven computation paradigm. However, direct and fair energy comparisons between SNNs and ANNs on inexpensive general-purpose microcontrollers remain limited. In this work, we present a controlled experimental comparison of inference-time energy consumption between an SNN and an ANN with identical multilayer perceptron architectures deployed on an ESP32-C6 microcontroller. Both networks are implemented in C/C++ without relying on optimized inference libraries, use int8 quantization, and are evaluated on the MNIST dataset. Energy consumption is measured directly during inference under identical implementation and measurement conditions to isolate the impact of the computational paradigm. Quantitatively, the ANN requires 11.01 mJ per inference, whereas the SNN consumes 2.31 mJ, 2.34 mJ, and 3.88 mJ for 1, 3, and 10 simulation timesteps, respectively. This corresponds to an approximate reduction in energy consumption of about [Formula: see text] for a single timestep and still about [Formula: see text] even at ten timesteps, demonstrating that the SNN maintains a substantial efficiency advantage under the evaluated conditions. These findings indicate that spiking neural networks can provide significant inference-time energy savings on widely available microcontroller units even without specialized hardware acceleration, while emphasizing the importance of controlled and reproducible comparison conditions.

Jonas Hansert, Franck Gechter, Peter Offermann · 0 citations