paper-with-me

홈 › Papers

MATIC: Learning Around Errors for Efficient Low-Voltage Neural Network Accelerators

2017-06-14 · Sung Kim, Patrick Howe, Thierry Moreau, Armin Alaghi, Luis Ceze, Visvesh Sathe

As a result of the increasing demand for deep neural network (DNN)-based services, efforts to develop dedicated hardware accelerators for DNNs are growing rapidly. However,while accelerators with high performance and efficiency on convolutional deep neural networks (Conv-DNNs) have been developed, less progress has been made with regards to fully-connected DNNs (FC-DNNs). In this paper, we propose MATIC (Memory Adaptive Training with In-situ Canaries), a methodology that enables aggressive voltage scaling of accelerator weight memories to improve the energy-efficiency of DNN accelerators. To enable accurate operation with voltage overscaling, MATIC combines the characteristics of destructive SRAM reads with the error resilience of neural networks in a memory-adaptive training process. Furthermore, PVT-related voltage margins are eliminated using bit-cells from synaptic weights as in-situ canaries to track runtime environmental variation. Demonstrated on a low-power DNN accelerator that we fabricate in 65 nm CMOS, MATIC enables up to 60-80 mV of voltage overscaling (3.3x total energy reduction versus the nominal voltage), or 18.6x application error reduction.

📄 PDF Abstract BibTeX arXiv:1706.04332

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Random and Adversarial Bit Error Robustness: Energy-Efficient and Secure DNN Accelerators

2021-04-16 · David Stutz, Nandhini Chandramoorthy, Matthias Hein, Bernt Schiele

Deep neural network (DNN) accelerators received considerable attention in recent years due to the potential to save energy compared to mainstream hardware. Low-voltage operation of DNN accelerators allows to further redu…

Quantization

Bit Error Robustness for Energy-Efficient DNN Accelerators

2020-06-24 · David Stutz, Nandhini Chandramoorthy, Matthias Hein, Bernt Schiele

Deep neural network (DNN) accelerators received considerable attention in past years due to saved energy compared to mainstream hardware. Low-voltage operation of DNN accelerators allows to further reduce energy consumpt…

Quantization

An Experimental Study of Reduced-Voltage Operation in Modern FPGAs for Neural Network Acceleration

2020-05-04 · Behzad Salami, Erhan Baturay Onural, Ismail Emir Yuksel, Fahrettin Koc 외

We empirically evaluate an undervolting technique, i.e., underscaling the circuit supply voltage below the nominal level, to improve the power-efficiency of Convolutional Neural Network (CNN) accelerators mapped to Field…

image-classificationImage ClassificationQuantization

Bias-Compensated State of Charge and State of Health Joint Estimation for Lithium Iron Phosphate Batteries

2024-01-16 · Baozhao Yi, Xinhao Du, Jiawei Zhang, Xiaogang Wu 외

Accurate estimation of the state of charge (SOC) and state of health (SOH) is crucial for the safe and reliable operation of batteries. Voltage measurement bias highly affects state estimation accuracy, especially in Lit…

State Estimation

ThUnderVolt: Enabling Aggressive Voltage Underscaling and Timing Error Resilience for Energy Efficient Deep Neural Network Accelerators

2018-02-11 · Jeff Zhang, Kartheek Rangineni, Zahra Ghodsi, Siddharth Garg

Hardware accelerators are being increasingly deployed to boost the performance and energy efficiency of deep neural network (DNN) inference. In this paper we propose Thundervolt, a new framework that enables aggressive v…

General Classification