arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.12861cs.ARcs.ETcs.LG
收藏

Neural-PIM:高效的内存处理和外围设备的神经近似

Neural-PIM: Efficient Processing-In-Memory with Neural Approximation of Peripherals

Washington University in St. Louis(圣路易斯华盛顿大学);Shanghai Jiao Tong University(上海交通大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所)

Weidong Cao, Yilong Zhao, Adith Boloor, Yinhe Han, Xuan Zhang, Li Jiang

摘要:

内存处理(PIM)架构在加速众多深度学习任务方面表现出巨大的潜力。特别地,电阻式随机存取存储器(RRAM)设备提供了一种有前途的硬件基板,以构建PIM加速器,因为它们能够实现高效的原位矢量矩阵乘法(VMM)。然而,现有的PIM加速器遭受频繁和能量密集型模数(A/D)转换,严重限制了它们的性能。本文提出了一种新的PIM架构,通过使用模拟累积和神经近似外围电路最大限度地减少所需的A/D转换,有效地加速深度学习任务。我们首先描述了现有的PIM加速器所采用的不同的PLL,在此基础上,提出了一种新的PLL,通过在最终量化之前将移位和加法(S+A)操作扩展到模拟域来显著减少VMM所需的A/D转换。然后,我们利用神经近似方法设计模拟累加电路(S+A)和量化电路(ADC)与RRAM交叉阵列在一个高效的方式。最后,我们应用它们构建了一个基于RAM的PIM加速器(即,\textbf{Neural-PIM})对所提出的模拟电路进行仿真,并对其系统级性能进行评估。对不同基准的评估表明,与最先进的基于RRAM的PIM加速器相比,Neural-PIM可以将能效提高5.36倍(1.73倍),并将吞吐量提高3.43倍(1.59倍),而不会损失准确性,即,ISAAC(CASCADE).

英文摘要:

Processing-in-memory (PIM) architectures have demonstrated great potential in accelerating numerous deep learning tasks. Particularly, resistive random-access memory (RRAM) devices provide a promising hardware substrate to build PIM accelerators due to their abilities to realize efficient in-situ vector-matrix multiplications (VMMs). However, existing PIM accelerators suffer from frequent and energy-intensive analog-to-digital (A/D) conversions, severely limiting their performance. This paper presents a new PIM architecture to efficiently accelerate deep learning tasks by minimizing the required A/D conversions with analog accumulation and neural approximated peripheral circuits. We first characterize the different dataflows employed by existing PIM accelerators, based on which a new dataflow is proposed to remarkably reduce the required A/D conversions for VMMs by extending shift and add (S+A) operations into the analog domain before the final quantizations. We then leverage a neural approximation method to design both analog accumulation circuits (S+A) and quantization circuits (ADCs) with RRAM crossbar arrays in a highly-efficient manner. Finally, we apply them to build an RRAM-based PIM accelerator (i.e., \textbf{Neural-PIM}) upon the proposed analog dataflow and evaluate its system-level performance. Evaluations on different benchmarks demonstrate that Neural-PIM can improve energy efficiency by 5.36x (1.73x) and speed up throughput by 3.43x (1.59x) without losing accuracy, compared to the state-of-the-art RRAM-based PIM accelerators, i.e., ISAAC (CASCADE).

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。

↑