Log In Sign Up

Bit Error Robustness for Energy-Efficient DNN Accelerators

by   David Stutz, et al.

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 consumption significantly, however, causes bit-level failures in the memory storing the quantized DNN weights. In this paper, we show that a combination of robust fixed-point quantization, weight clipping, and random bit error training (RandBET) improves robustness against random bit errors in (quantized) DNN weights significantly. This leads to high energy savings from both low-voltage operation as well as low-precision quantization. Our approach generalizes across operating voltages and accelerators, as demonstrated on bit errors from profiled SRAM arrays. We also discuss why weight clipping alone is already a quite effective way to achieve robustness against bit errors. Moreover, we specifically discuss the involved trade-offs regarding accuracy, robustness and precision: Without losing more than 1 trained 8-bit DNN, we can reduce energy consumption on CIFAR-10 by 20 energy savings of, e.g., 30 for 4-bit DNNs.


page 1

page 2

page 3

page 4


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

Deep neural network (DNN) accelerators received considerable attention i...

DNN Quantization with Attention

Low-bit quantization of network weights and activations can drastically ...

Sharpness-Aware Training for Accurate Inference on Noisy DNN Accelerators

Energy-efficient deep neural network (DNN) accelerators are prone to non...

MATIC: Adaptation and In-situ Canaries for Energy-Efficient Neural Network Acceleration

- The primary author has withdrawn this paper due to conflict of interes...

MIMHD: Accurate and Efficient Hyperdimensional Inference Using Multi-Bit In-Memory Computing

Hyperdimensional Computing (HDC) is an emerging computational framework ...