Home > Archive > 2012 > Volume 2 Number 2 (Apr. 2012) >
IJIET 2012 Vol.2(2): 185-187 ISSN: 2010-3689
DOI: 10.7763/IJIET.2012.V2.106

Comparison and Analysis of GPGPU and Parallel Computing on Multi-Core CPU

Hong Zhang, Da-Fang Zhang, and Xia-An Bi

Abstract—There are two ways to improve the performance of the algorithm computing, which are general purpose of computation and parallel computation of multi-core CPU. By comparison and analysis, contrast the main difference between them, we reach a conclusion that GPU is suitable for processing large-scale data-parallel load of high-density computing but relatively simple branching logic , however, the CPU is more suitable for processing complex logic computation. Now, the appearance of the CUDA makes GPU architecture more suitable for general purpose of computation. Cryptographic algorithm is typical compute-intensive algorithm, this paper take the modular exponentiation of RSA algorithms for example, through the comparison and analysis of GPU implementation and CPU implementation, the experiment results show: that the GPU implementation can achieve more than 45 times speedup in comparison with multi-core CPU implementation of RSA.

Index Terms—GPU, GPU multi-core, CUDA, RSA, cryptographic algorithm.

The authors are with School of information Science and Engineering Hunan University, China (e-mail: zhanghonghaiqin@163.com).

[PDF]

Cite: Hong Zhang, Da-fang Zhang, and Xia-an Bi, "Comparison and Analysis of GPGPU and Parallel Computing on Multi-Core CPU," International Journal of Information and Education Technology vol. 2, no. 2, pp. 185-187, 2012.

General Information

  • ISSN: 2010-3689 (Online)
  • Abbreviated Title: Int. J. Inf. Educ. Technol.
  • Frequency: Monthly
  • DOI: 10.18178/IJIET
  • Editor-in-Chief: Prof. Jon-Chao Hong
  • Managing Editor: Ms. Nancy Y. Liu
  • E-mail: editor@ijiet.org
  • Abstracting/ Indexing: Scopus (CiteScore 2023: 2.8), INSPEC (IET), UGC-CARE List (India), CNKI, EBSCO, Google Scholar
  • Article Processing Charge: 800 USD

 

Article Metrics in Dimensions