arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.03491cs.LGcs.AR
收藏

当微小变化成为重大失败:计算-内存神经加速器的可靠性挑战

When Small Variations Become Big Failures: Reliability Challenges in Compute-in-Memory Neural Accelerators

Yifan Qin, Jiahao Zheng, Zheyu Yan, Wujie Wen, Xiaobo Sharon Hu, Yiyu Shi

AI总结:

本文研究了计算-内存神经加速器的可靠性挑战,提出SWIM机制和学习导向方案,通过跨层协同设计提升系统可靠性与效率。

摘要:

计算-内存(CiM)架构通过缓解冯·诺依曼瓶颈,为深度神经网络加速带来了显著的能效和吞吐量提升。然而,其对新兴非易失性存储器设备的依赖引入了设备层面的非理想特性,如写入变化性、电导漂移和随机噪声,这些从根本上挑战了可靠性、可预测性和安全性,尤其是在安全关键应用中。本文探讨了基于CiM的神经加速器的可靠性极限,并提出了一系列技术,将设备物理、架构和学习算法结合起来以解决这些挑战。我们首先证明,即使设备变化很小,也会导致在安全关键的推断工作负载中出现不成比例的精度下降和灾难性故障,揭示了平均情况评估与最坏情况行为之间的关键差距。基于这一见解,我们引入了SWIM,一种选择性写入-验证机制,仅在最需要的地方应用验证,显著提高了可靠性,同时保持了CiM的效率优势。最后,我们探索了一种以学习为中心的解决方案,通过训练神经网络时引入右截断高斯噪声,使训练假设与硬件引起的变异性对齐,从而在不增加过多硬件开销的情况下实现稳健部署。这些工作突显了对CiM加速器进行跨层协同设计的必要性,并为在新兴存储技术上实现可靠、高效的神经推断提供了系统性的路径,为它们在安全性和可靠性关键系统中的应用铺平了道路。

英文摘要:

Compute-in-memory (CiM) architectures promise significant improvements in energy efficiency and throughput for deep neural network acceleration by alleviating the von Neumann bottleneck. However, their reliance on emerging non-volatile memory devices introduces device-level non-idealities-such as write variability, conductance drift, and stochastic noise-that fundamentally challenge reliability, predictability, and safety, especially in safety-critical applications. This talk examines the reliability limits of CiM-based neural accelerators and presents a series of techniques that bridge device physics, architecture, and learning algorithms to address these challenges. We first demonstrate that even small device variations can lead to disproportionately large accuracy degradation and catastrophic failures in safety-critical inference workloads, revealing a critical gap between average-case evaluations and worst-case behavior. Building on this insight, we introduce SWIM, a selective write-verify mechanism that strategically applies verification only where it is most impactful, significantly improving reliability while maintaining CiM's efficiency advantages. Finally, we explore a learning-centric solution that improves realistic worst-case performance by training neural networks with right-censored Gaussian noise, aligning training assumptions with hardware-induced variability and enabling robust deployment without excessive hardware overhead. Together, these works highlight the necessity of cross-layer co-design for CiM accelerators and provide a principled path toward dependable, efficient neural inference on emerging memory technologies-paving the way for their adoption in safety- and reliability-critical systems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑