Putra Rachmad Vidya Wicaksana, Hanif Muhammad Abdullah, Shafique Muhammad
Embedded Computing Systems, Institute of Computer Engineering, Technische Universität Wien, Vienna, Austria.
eBrain Lab, Division of Engineering, New York University Abu Dhabi (NYUAD), Abu Dhabi, United Arab Emirates.
Front Neurosci. 2022 Aug 10;16:937782. doi: 10.3389/fnins.2022.937782. eCollection 2022.
Spiking Neural Networks (SNNs) have shown capabilities of achieving high accuracy under unsupervised settings and low operational power/energy due to their bio-plausible computations. Previous studies identified that DRAM-based off-chip memory accesses dominate the energy consumption of SNN processing. However, state-of-the-art works do not optimize the DRAM energy-per-access, thereby hindering the SNN-based systems from achieving further energy efficiency gains. To substantially reduce the DRAM energy-per-access, an effective solution is to decrease the DRAM supply voltage, but it may lead to errors in DRAM cells (i.e., so-called ). Toward this, we propose , a novel design framework that provides a solution for resilient and energy-efficient SNN inference using reduced-voltage DRAM for embedded systems. The key mechanisms of our EnforceSNN are: (1) employing quantized weights to reduce the DRAM access energy; (2) devising an efficient DRAM mapping policy to minimize the DRAM energy-per-access; (3) analyzing the SNN error tolerance to understand its accuracy profile considering different bit error rate (BER) values; (4) leveraging the information for developing an efficient fault-aware training (FAT) that considers different BER values and bit error locations in DRAM to improve the SNN error tolerance; and (5) developing an algorithm to select the SNN model that offers good trade-offs among accuracy, memory, and energy consumption. The experimental results show that our EnforceSNN maintains the accuracy (i.e., no accuracy loss for ≤ 10) as compared to the baseline SNN with accurate DRAM while achieving up to 84.9% of DRAM energy saving and up to 4.1x speed-up of DRAM data throughput across different network sizes.
脉冲神经网络（SNN）由于其具有生物合理性的计算方式，已展现出在无监督设置下实现高精度以及低运算功耗/能量的能力。先前的研究表明，基于动态随机存取存储器（DRAM）的片外内存访问在SNN处理的能量消耗中占主导地位。然而，当前的前沿研究并未对DRAM每次访问的能量进行优化，从而阻碍了基于SNN的系统在能源效率上取得进一步提升。为了大幅降低DRAM每次访问的能量，一个有效的解决方案是降低DRAM的电源电压，但这可能会导致DRAM单元出现错误（即所谓的 ）。针对这一问题，我们提出了EnforceSNN，这是一种新颖的设计框架，它为使用低电压DRAM的嵌入式系统提供了一种实现弹性且节能的SNN推理的解决方案。我们的EnforceSNN的关键机制包括：（1）采用量化权重以降低DRAM访问能量；（2）设计一种高效的DRAM映射策略，以最小化DRAM每次访问的能量；（3）分析SNN的容错能力，以了解在考虑不同误码率（BER）值时其精度概况；（4）利用这些信息来开发一种高效的故障感知训练（FAT），该训练考虑DRAM中不同的BER值和误码位置，以提高SNN的容错能力；（5）开发一种算法，以选择在精度、内存和能耗之间提供良好权衡的SNN模型。实验结果表明，与具有精确DRAM的基线SNN相比，我们的EnforceSNN保持了精度（即对于≤10没有精度损失），同时在不同网络规模下实现了高达84.9%的DRAM节能以及高达4.1倍的DRAM数据吞吐量加速。
