DocumentCode :
2866537
Title :
Grid Based Analysis Toolkit for Partial Wave Analysis
Author :
Yang, Zhi ; Jones, Richard T. ; Yin, Changqin
Author_Institution :
Coll. of Software Eng., Tongji Univ., Shanghai, China
fYear :
2011
fDate :
20-22 Oct. 2011
Firstpage :
161
Lastpage :
166
Abstract :
To get the physics out of the data, GlueX relies entirely on an amplitude-based analysis-PWA (Partial Wave Analysis). We build a grid test platform to verify how computational and data grid can be used to process large scale dataset in PWA, and make PWA toolkit to be grid-enable. The work we did has demonstrated that grid is a promising computing architecture for PWA in GlueX project. We setup grid computational platform, which consists of three clusters. We also setup data grid platform using dCache and evaluate the file access performance of dCache. A new Condor wrapper for Open MPI is presented in this grid platform. Through revising ruby-PWA package, we parallel ruby-PWA in grid environment. We run some scaling test benchmarks, and the results show grid-enabled ruby-PWA gets high performance in multi-node amplitude analysis.
Keywords :
application program interfaces; benchmark testing; grid computing; information retrieval; message passing; open systems; quantum computing; Condor wrapper; GlueX project; PWA toolkit; amplitude-based analysis; computational platform; dCache; data grid platform; file access performance; grid based analysis toolkit; grid test platform; large scale dataset; multinode amplitude analysis; parallel ruby-PWA package; partial wave analysis; scaling test benchmark; Collaboration; Computer architecture; Educational institutions; Memory; Mesons; Physics; Security; Grid; MPI; Partial Wave Analysis; framework;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Parallel and Distributed Computing, Applications and Technologies (PDCAT), 2011 12th International Conference on
Conference_Location :
Gwangju
Print_ISBN :
978-1-4577-1807-6
Type :
conf
DOI :
10.1109/PDCAT.2011.43
Filename :
6118945
Link To Document :
بازگشت