TY - GEN
T1 - UNO: Virtualizing and Unifying Nonlinear Operations for Emerging Neural Networks
AU - Wu, Di
AU - Li, Jingjie
AU - Behroozi, Setareh
AU - Kim, Younghyun
AU - Miguel, Joshua San
N1 - Funding Information:
ACKNOWLEDGMENTS This work is supported by the Wisconsin Alumni Research Foundation and NSF under award No. CNS-1845469.
Publisher Copyright:
© 2021 IEEE.
PY - 2021/7/26
Y1 - 2021/7/26
N2 - Linear multiply-accumulate (MAC) operations have been the main focus of prior efforts in improving the energy efficiency of neural network inference due to their dominant contribution to energy consumption in traditional models. On the other hand, nonlinear operations, such as division, exponentiation, and logarithm, that are becoming increasingly significant in emerging neural network models, have been largely underexplored. In this paper, we propose UNO, a low-area, low-energy processing element that virtualizes the Taylor approximation of nonlinear operations on top of off-the-shelf linear MAC units already present in inference hardware. Such virtualization approximates multiple nonlinear operations in a unified, MAC-compatible manner to achieve dynamic run-time accuracy-energy scaling. Compared to the baseline, our scheme reduces the energy consumption by up to 38.4% for individual operations and increases the energy efficiency by up to 274.5%
AB - Linear multiply-accumulate (MAC) operations have been the main focus of prior efforts in improving the energy efficiency of neural network inference due to their dominant contribution to energy consumption in traditional models. On the other hand, nonlinear operations, such as division, exponentiation, and logarithm, that are becoming increasingly significant in emerging neural network models, have been largely underexplored. In this paper, we propose UNO, a low-area, low-energy processing element that virtualizes the Taylor approximation of nonlinear operations on top of off-the-shelf linear MAC units already present in inference hardware. Such virtualization approximates multiple nonlinear operations in a unified, MAC-compatible manner to achieve dynamic run-time accuracy-energy scaling. Compared to the baseline, our scheme reduces the energy consumption by up to 38.4% for individual operations and increases the energy efficiency by up to 274.5%
U2 - 10.1109/ISLPED52811.2021.9502473
DO - 10.1109/ISLPED52811.2021.9502473
M3 - Conference contribution
AN - SCOPUS:85114294777
T3 - Proceedings of the International Symposium on Low Power Electronics and Design
SP - 1
EP - 6
BT - 2021 IEEE/ACM International Symposium on Low Power Electronics and Design, ISLPED 2021
PB - Institute of Electrical and Electronics Engineers
T2 - 2021 IEEE/ACM International Symposium on Low Power Electronics and Design, ISLPED 2021
Y2 - 26 July 2021 through 28 July 2021
ER -