DocumentCode :
2556597
Title :
Molecular dynamics simulation: Implementation and optimization based on Hadoop
Author :
Jiao, Shanfei ; He, Chen ; Dou, Yusheng ; Tang, Hong
Author_Institution :
Coll. of Comput. Sci. & Technol., Chongqing Univ. of Posts & Telecommun., Chongqing, China
fYear :
2012
fDate :
29-31 May 2012
Firstpage :
1203
Lastpage :
1207
Abstract :
At first the paper introduced the basic concept and mechanism of Hadoop architecture and the current parallel research situation of molecular dynamics simulation, then three solutions were proposed for implementing the parallel algorithm using atom-decomposition method on Hadoop to overcome shortcomings of fast iteration and communication among sub-tasks. After that tested and analyzed program´s speed-up ratio, expansibility and cost-time of various parts, then did performance tuning in many ways, finally obtained good results as high as 28× speedup in large-scale system simulations. The test showing molecular dynamics simulation based on Hadoop is more economical and practical.
Keywords :
iterative methods; optimisation; parallel algorithms; parallel architectures; public domain software; Hadoop architecture; atom decomposition method; iteration algorithm; molecular dynamics simulation; optimization; parallel algorithm; performance tuning; Cloud computing; Computational modeling; Educational institutions; Load modeling; Scientific computing; Synchronization; Hadoop; MapReduce; cloud computing; molecular dynamics; optimization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Natural Computation (ICNC), 2012 Eighth International Conference on
Conference_Location :
Chongqing
ISSN :
2157-9555
Print_ISBN :
978-1-4577-2130-4
Type :
conf
DOI :
10.1109/ICNC.2012.6234529
Filename :
6234529
Link To Document :
بازگشت