2015 IEEE International Conference on Smart City/SocialCom/SustainCom (SmartCity) (2015)
Dec. 19, 2015 to Dec. 21, 2015
In this paper, we present cuLib, a R package that provides an easy-to-access interface for utilizing the computing power of NVIDIA GPU. The cuLib package aims to make GPU-based parallel programming easier, flexible and high-performance. It allows the use of GPU computing in R without further knowledge because the syntax for definition and manipulation of GPU data is similar to formal R language. cuLib is compatible with device and operation system. The data interface is very flexible, enabling the users manipulate data freely. The cuLib package provides an R wrapper for libraries of NVIDIA's CUDA toolkit and numerous operations. More importantly, it is not only a mathematical tool but also practicable for the algorithm in dealing with data intensive computation.
Graphics processing units, Sparse matrices, Parallel processing, Acceleration, Libraries, Computer architecture, Central Processing Unit
J. Huang, B. Zhuang and F. Su, "A General Accelerated R Package Using GPU," 2015 IEEE International Conference on Smart City/SocialCom/SustainCom (SmartCity)(SMARTCITY), Chengdu, China, 2015, pp. 605-608.