The Journal of Korean Institute of Information Technology (한국정보기술학회논문지)

Korean Institute of Information Technology (한국정보기술학회)
권호 표지
DOI QR코드

DOI QR Code

CPU and GPU Performance Analysis for Convolution Neural Network

합성곱 신경망 사용을 위한 CPU와 GPU 성능 분석

  • 권대책 (경북대학교 대학원 기계공학과) ;
  • 강보영 (경북대학교 기계공학부)
  • Received : 2017.07.28
  • Accepted : 2017.08.13
  • Published : 2017.08.31
⟨ Previous Next ⟩

Abstract

In order to use deep learning method which uses vast amount of data specific to image data, a high-performance computer operation system is required hence a large operation system such as a supercomputer or mainframe was constructed and operated. However, recently, the development of a CPU and a GPU has led to the development of a parallel processing technique. In this paper we analyzed the performance of CPUs and GPUs, which are mainly used where high-resolution images of 8K and 4K are enlarged using convolutional neural network. Experimental results show that of various GPU and CPU configurations, TitanX (P) showed the best speed.

이미지 영상처리 기술은 해상도와 초당 프레임수가 높아짐에 따라 데이터의 양이 큰 폭으로 증가하게 되었으며, 이러한 고용량의 환경에서 딥러닝 학습법과 같이 방대한 데이터를 연산하기 위해서 고성능 컴퓨터 연산 시스템이 필요하게 되었다. 하지만, 이러한 이미지 처리를 위한 딥러닝 기술 적용의 발달에도 불구하고, 컴퓨터 하드웨어 구성 별 딥러닝 처리속도 차이에 대한 구체적인 성능결과 보고가 현재까지 미흡한 실정이다. 본 논문은 소규모 연구실에서 주로 사용하는 CPU와 GPU를 대표적인 사용처와 성능별로 구비하고, 합성곱 신경망에 기반하여 이미지 확대보간 결과를 연산장치 별로 보고함으로써, 소규모 연구실에서 합성곱 신경망 적용시 효율적인 연산장치 구성을 할 수 있도록 돕고자 한다. 실험결과 다양한 CPU, GPU 구성 중 TitanX(p)가 절대 성능면에서 우위를 차지하였다.

Keywords

Acknowledgement

Supported by : 경북대학교

References

  1. Yann LeCun, Yoshua Bengio, and Geoffrey Hinton, "Deep learning", Nature 521, pp. 436-444, May 2015. https://doi.org/10.1038/nature14539
  2. Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton, "Imagenet classification with deep convolutional neural networks", In: Advances in neural information processing systems, pp. 1097-1105, 2012.
  3. M. W. Gardnera and S. R. Dorlinga, "Artificial neural networks (the multilayer perceptron)-a review of applications in the atmospheric sciences" Atmospheric environment, Vol. 32, No. 14-15, pp. 2627-2636, Aug. 1998. https://doi.org/10.1016/S1352-2310(97)00447-0
  4. Kumar Chellapilla, Sidd Puri, and Patrice Simard, "High performance convolutional neural networks for document processing" In: Tenth International Workshop on Frontiers in Handwriting Recognition. Suvisoft, Oct. 2006.
  5. Xin Li and Michael T. Orchard, "New edgedirected interpolation", IEEE transactions on image processing, Vol. 10, No. 10, pp. 1521-1527, Oct. 2001. https://doi.org/10.1109/83.951537
  6. William T. Freeman, Egon C. Pasztor, and Owen T. Carmichael", International journal of computer vision, Vol. 40, No. 1, pp. 25-47, Oct. 2000. https://doi.org/10.1023/A:1026501619075
  7. Xianming Liu, Debin Zhao, Ruiqin Xiong, Siwei Ma, Wen Gao, and Huifang Sun, "Image interpolation via regularized local linear regression", IEEE Transactions on Image Processing, Vol. 20, No. 12, pp. 3455-3469, Dec. 2011. https://doi.org/10.1109/TIP.2011.2150234
  8. Ma Liyong, Shen Yi, and Ma Jiachen, "Local spatial properties based image interpolation scheme using SVMs", Journal of Systems Engineering and Electronics, Vol. 19, No. 3, pp. 618-623, Jun. 2008. https://doi.org/10.1016/S1004-4132(08)60129-2
  9. Guochun Shi, David Gohara, and John E. Stone, "OpenCL: A parallel programming standard for heterogeneous computing systems", Computing in science & engineering, Vol. 12, No. 3, pp. 66-73, Jun. 2010.
  10. Tianyi David Han and Tarek S. Abdelrahman, "hiCUDA: High-level GPGPU programming", IEEE Transactions on Parallel and Distributed Systems, Vol. 22, No. 1, pp. 78-90, Jan. 2011. https://doi.org/10.1109/TPDS.2010.62
  11. David Kirk NVIDIA and Santa Clara, CA, "NVIDIA CUDA software and GPU parallel computing architecture", In ISMM '07 Proceedings of the 6th international symposium on Memory management pp. 103-104, Oct. 2007.
  12. Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang, "Image super-resolution using deep convolutional networks", IEEE transactions on pattern analysis and machine intelligence, Vol. 38, No. 2, pp. 295-307, Feb. 2016. https://doi.org/10.1109/TPAMI.2015.2439281
  13. Hyeon-Joong Yoo, "Deep Convolution Neural Networks in Computer Vision", IEIE Transactions on Smart Processing & Computing, Vol. 4, No. 1, pp. 35-43, Feb. 2015. https://doi.org/10.5573/IEIESPC.2015.4.1.035

Cited by

  1. 효과적인 입력변수 패턴 학습을 위한 시계열 그래프 기반 합성곱 신경망 모형: 주식시장 예측에의 응용 vol.24, pp.1, 2018, https://doi.org/10.13088/jiis.2018.24.1.167
  2. Design of Convolutional Neural Network Structure for the Identification of Warhead and Debris in the Separation Phase vol.16, pp.6, 2018, https://doi.org/10.14801/jkiit.2018.16.6.81
  3. Classification of Warhead and Debris using CFAR and Convolutional Neural Networks vol.17, pp.6, 2017, https://doi.org/10.14801/jkiit.2019.17.6.85
  4. 그래픽 프로세서를 이용한 고차 유한 차분식 기반 수중채널모델 연구 vol.30, pp.1, 2017, https://doi.org/10.9709/jkss.2021.30.1.011