ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Measurement of Operator Utilization in DNN HW Accelerator based on SoC Virtual Platform
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
Sungjae Yoon, Hanjin Cho, Wonjong Kim
Issue Date
2019-06
Citation
International Conference on Consumer Electronics (ICCE) 2019 : Asia, pp.1-3
Publisher
IEEE
Language
English
Type
Conference Paper
Abstract
Deep neural network (DNN) HW accelerator is being studied to accelerate the process of DNN. Zero-skipping is one of the methods to optimize the DNN HW accelerator. To optimize DNN HW accelerators, a simulation environment is required to monitor the unnecessary calculations. In this paper, we propose the method to measure unnecessary computation of DNN HW accelerator using SoC virtual platform. Experimental results show that the proposed method can be used during DNN HW accelerator optimization.
KSP Keywords
Deep neural network(DNN), HW Accelerators, Simulation Environment, Virtual platform, operator utilization