D-NAT: Data-Driven Non-Ideality Aware Training Framework for Fabricated Computing-In-Memory Macros
Journal
IEEE Journal on Emerging and Selected Topics in Circuits and Systems
Journal Volume
12
Journal Issue
2
Pages
381-392
Date Issued
2022
Author(s)
Lin M.-G
Huang C.-T
Chuang Y.-C
Chen Y.-T
Hsu Y.-T
Chen Y.-K
Chou J.-J
Abstract
To enable energy-efficient computation for deep neural networks (DNNs) at edge, computing-in-memory (CIM) is proposed to reduce the energy costs during intense off-chip memory access. However, CIM is prone to multiply-accumulate (MAC) errors due to non-idealities of memory crossbars and peripheral circuits, which severely degrade the accuracy of DNNs. In this work, we propose a Data-Driven Non-ideality Aware Training (D-NAT) framework to compensate for the accuracy degradation. The proposed D-NAT framework has the following contributions: 1) We measured a fabricated SRAM-based CIM macro to obtain a data-driven MAC error model (D-MAC-EM). Based on the derived D-MAC-EM, we analyze the impact of the non-idealities on DNN's accuracy. 2) To make DNNs robust to the non-idealities of CIM macros, we incorporate the measured D-MAC-EM into DNN's training procedure. Moreover, we propose a statistical training mechanism to better estimate the gradients of the discrete D-MAC-EM. 3) We investigate trade-offs between quantization range and quantization errors. To mitigate the quantization errors in activations, we introduce extended PACT (E-PACT) that adaptively learns the upper and lower bounds of input activations for each layer. Simulation results show that our proposed D-NAT improves the accuracy of ResNet20, VGG8, ResNet34, and VGG16 by 78.98%, 71.8%, 72.04%, and 57.85%, respectively, which reaches the ideal upper bound of the quantized model. Lastly, the D-NAT framework is validated on an FPGA platform with the fabricated SRAM-based CIM macro chip. Based on the measurement results, D-NAT successfully recovers the accuracy under non-idealities of a real SRAM-based CIM macro. © 2011 IEEE.
Subjects
Computing-in-memory (CIM); deep neural network (DNN); non-ideality aware training
SDGs
Other Subjects
Chemical activation; Deep neural networks; Economic and social effects; Energy efficiency; Errors; Static random access storage; Timing circuits; Circuits and systems; Common information model (computing); Computational modelling; Computing-in-memory; Deep neural network; Hardware; Information Modeling; Non-ideality aware training; Nonideality; Quantization (signal); Semiconductor device measurements; Quantization (signal)
Type
journal article