Reconsidering the energy efficiency of spiking neural networks
This work addresses the problem of misleading energy efficiency claims in neuromorphic computing for researchers and engineers, providing a more rigorous framework for evaluation, though it is incremental in refining existing methodologies.
The paper re-evaluates the energy efficiency of Spiking Neural Networks (SNNs) compared to Quantized Artificial Neural Networks (QNNs) by developing a detailed analytical model that includes data movement and memory access, finding that SNNs only outperform QNNs under specific conditions, such as moderate time windows and low spike rates below 6.4%.
Spiking Neural Networks (SNNs) promise higher energy efficiency over conventional Quantized Artificial Neural Networks (QNNs) due to their event-driven, spike-based computation. However, prevailing energy evaluations often oversimplify, focusing on computational aspects while neglecting critical overheads like comprehensive data movement and memory access. Such simplifications can lead to misleading conclusions regarding the true energy benefits of SNNs. This paper presents a rigorous re-evaluation. We establish a fair baseline by mapping rate-encoded SNNs with $T$ timesteps to functionally equivalent QNNs with $\lceil \log_2(T+1) \rceil$ bits. This ensures both models have comparable representational capacities, as well has similar hardware requirement, enabling meaningful energy comparisons. We introduce a detailed analytical energy model encompassing core computation and data movement (sparse and dense activations, weights). Using this model, we systematically explore a wide parameter space, including intrinsic network characteristics ($T$, spike rate $s_r$, QNN sparsity $γ$, model size $N$, weight bit-level) and hardware characteristics (memory system and network-on-chip). Our analysis identifies specific operational regimes where SNNs genuinely offer superior energy efficiency. For example, under typical neuromorphic hardware conditions, SNNs with moderate time windows ($T \in [5,10]$) require an average spike rate ($s_r$) below 6.4% to outperform equivalent QNNs. These insights guide the design of genuinely energy-efficient neural network solutions.