Title :
Parallel Text Matching Using GPGPU
Author :
Takahashi, Ryosuke ; Inoue, Ushio
Author_Institution :
Grad. Sch. of Eng., Tokyo Denki Univ., Tokyo, Japan
Abstract :
This paper studies implementation methods of parallel text matching using General Purpose computing on Graphics Processing Unit (GPGPU). It is necessary to accelerate text matching in applications of real-time processing, such as anomaly-detection and decision-making. GPGPU is a technology that can be used to accelerate a variety of applications with highly parallel processing elements in GPUs. Recently, the Parallel-Failure-less Aho-Corasick (PFAC) algorithm has been developed, and an open-source PFAC library is currently available. However, there are several different implementation methods in the host-side, and choosing a good combination of these methods is important to improve the performance. We implemented a prototype system, and evaluated the performance and power consumption varying the implementation methods and input data. The evaluation results show that the performance of the system using GPGPU is better than a system using 4-core CPU with smaller power consumption.
Keywords :
decision making; general purpose computers; graphics processing units; multiprocessing systems; parallel processing; pattern matching; public domain software; security of data; software libraries; text analysis; 4-core CPU; GPGPU; PFAC algorithm; anomaly-detection; decision-making; general purpose computing on graphics processing unit; open-source PFAC library; parallel processing element; parallel text matching; parallel-failure-less Aho-Corasick algorithm; power consumption; real-time processing; Arrays; Graphics processing unit; Instruction sets; Libraries; Power demand; Prototypes; Throughput; GPGPU; parallel processing; text matching;
Conference_Titel :
Software Engineering, Artificial Intelligence, Networking and Parallel & Distributed Computing (SNPD), 2012 13th ACIS International Conference on
Conference_Location :
Kyoto
Print_ISBN :
978-1-4673-2120-4
DOI :
10.1109/SNPD.2012.28