Are SNNs Truly Energy-efficient? $-$ A Hardware Perspective

TL;DR

This paper critically examines the actual energy efficiency of Spiking Neural Networks (SNNs) on hardware platforms, revealing significant discrepancies from estimated gains and identifying key hardware bottlenecks affecting deployment.

Contribution

It provides a hardware perspective analysis of SNN energy efficiency using SATA and SpikeSim platforms, highlighting practical bottlenecks and challenges.

Findings

Actual energy savings are lower than estimated due to hardware bottlenecks.

Repeated computations and data movements significantly impact energy efficiency.

Hardware non-idealities affect SNN performance and energy consumption.

Abstract

Spiking Neural Networks (SNNs) have gained attention for their energy-efficient machine learning capabilities, utilizing bio-inspired activation functions and sparse binary spike-data representations. While recent SNN algorithmic advances achieve high accuracy on large-scale computer vision tasks, their energy-efficiency claims rely on certain impractical estimation metrics. This work studies two hardware benchmarking platforms for large-scale SNN inference, namely SATA and SpikeSim. SATA is a sparsity-aware systolic-array accelerator, while SpikeSim evaluates SNNs implemented on In-Memory Computing (IMC) based analog crossbars. Using these tools, we find that the actual energy-efficiency improvements of recent SNN algorithmic works differ significantly from their estimated values due to various hardware bottlenecks. We identify and address key roadblocks to efficient SNN deployment on hardware, including repeated computations & data movements over timesteps, neuronal module overhead, and vulnerability of SNNs towards crossbar non-idealities.