Toward Functional Safety of Systolic Array-Based Deep Learning Hardware Accelerators
Toward Functional Safety of Systolic Array-Based Deep Learning Hardware Accelerators
复制标题
实现基于脉动阵列的深度学习硬件加速器的功能安全
DOI：
10.1109/tvlsi.2020.3048829
复制
发表时间：
2021
影响因子：
2.8
通讯作者：
K. Basu
中科院分区：
文献类型：
--
作者：
Shamik Kundu;Suvadeep Banerjee;Arnab Raha;S. Natarajan;K. Basu
高精度和不断增长的计算能力使深度神经网络（DNN）成为计算领域各种机器学习、计算机视觉和图像处理应用的首选算法。为此，Google开发了张量处理单元（TPU），以加速DNN在其脉动阵列架构上的计算密集型矩阵乘法运算。由于潜在的制造缺陷或单事件效应而在此类脉动阵列的数据路径中出现的故障可能会导致功能安全（FuSa）违规。虽然DNN具有固有的容错特性，可以抵抗微小的扰动，但我们发现，该模型的分类准确度从97.4%骤降到7.75%，加速器中的最小故障率为0.0003%，这意味着在关键任务系统中部署时会出现灾难性的情况。因此，为了确保这种加速器的FuSa，本文提供了一个广泛的FuSa评估的加速器暴露在数据路径中的故障，通过不同的网络参数，位置和特性的诱导错误跨多个详尽的数据集。此外，我们提出了两种新的策略，以获得一个小型的功能测试模式，以检测FuSa违反DNN加速器。我们的实验结果表明，所获得的测试集可以达到平均92.63%（在某些情况下，高达100%）的故障覆盖率与基数低至0.1%的整个测试数据集。
High accuracy and ever-increasing computing power have made deep neural networks (DNNs) the algorithm of choice for various machine learning, computer vision, and image processing applications across the computing spectrum. To this end, Google developed the tensor processing unit (TPU) to accelerate the computationally intensive matrix multiplication operation of a DNN on its systolic array architecture. Faults manifested in the datapath of such a systolic array due to latent manufacturing defects or single-event effects may lead to functional safety (FuSa) violation. Although DNNs are known to resist minor perturbations with their inherent fault-tolerant characteristics, we show that the classification accuracy of the model plummets from 97.4% to 7.75% with a minimal fault rate of 0.0003% in the accelerator, implying catastrophic circumstances when deployed across mission-critical systems. Hence, to ensure FuSa of such accelerators, this article provides an extensive FuSa assessment of the accelerator exposed to faults in the datapath, by varying the network parameters, position, and characteristics of the induced error across multiple exhaustive data sets. Furthermore, we propose two novel strategies to obtain a diminutive set of functional test patterns to detect FuSa violation in a DNN accelerator. Our experimental results demonstrate that the obtained test sets can achieve an average of 92.63% (in some cases, up to 100%) fault coverage with cardinality as low as 0.1% of the entire test data set.