Title : Performance Comparison between GPU and CPU in CNN Learning Process
Published in : The 12th KIPS International Conference on Ubiquitous Information Technologies and Applications (CUTE 2017)
Author : EunKwang Jeon, JungYeon Seo, HwaMin Lee
Corresponding author : HwaMin Lee
Location : Providence University, Taichung City, Taiwan
The initial deep neural networks took a long time to learn and it was generally impractical to apply them to other areas. However, recent advances in computing performance and the ability to collect big data have re-emerged depth neural networks. GPU is used in the learning process of the GPU to reduce the learning time of the neural network. Using CUDA provided by NVIDIA, GPU enables quick learning. We used the GPU and CPU to perform the learning process in the CNN algorithm and confirmed the learning performance. As a result, the GPU used in the experiment was about 28 times faster than the CPU.
This research was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Science, ICT & Future Planning(NRF-2017R1A2B4010570) and the Ministry of Science and ICT, Korea, under the ITRC(Information Technology Research Center) support program (IITP-2017-2015-0-00403) supervised by the IITP(Institute for Information & communications Technology Promotion).