DocumentCode
1914021
Title
Applying Multi-cell MD Method to Shared-Memory Multi-core Systems
Author
Kim, Myongchan ; Yao, Shucai ; Go, Gwangsu
Author_Institution
Dept. of Comput. Sci. & Technol., Univ. Sci. & Technol. Beijing, Beijing, China
fYear
2012
fDate
20-23 Sept. 2012
Firstpage
164
Lastpage
169
Abstract
For decades, the main parallelization systems have been message passing ones, in which there are many parallel methods whose efficiencies are very high. The multi-cell MD (Molecular Dynamics) method is also a parallel one that is widely used for molecular dynamics simulation on the message passing systems. Cluster is still the main environment for parallel computing and every node in the cluster is generally the shared-memory multi-core computer. In this paper, we propose a data structure and a method so that the multi-cell MD method can be also used on the shared-memory multi-core systems. The proposed data structure and method are designed on the basis of the Torus network structure, the characteristic of the multi-cell MD method and a work sharing structure of OpenMP. Experimental results show that our method allows the multi-cell MD method to be used on the shared-memory multi-core systems and the parallel efficiency per a time step of that is increased about 10% than the HS (Half-Shell) method.
Keywords
application program interfaces; biology computing; message passing; molecular biophysics; parallel processing; pattern clustering; shared memory systems; HS method; OpenMP work sharing structure; Torus network structure; cluster; data structure; half-shell method; message passing systems; molecular dynamics simulation method; multicell MD method; parallel computing; parallel methods; parallelization systems; shared-memory multicore systems; Computational modeling; Computers; Data structures; Force; Message passing; Optimization; Parallel processing; Data Structure; Interaction Path; Multi-Cell MD Method; Shared-Memory Multi-Core System; Torus Network;
fLanguage
English
Publisher
ieee
Conference_Titel
ChinaGrid Annual Conference (ChinaGrid), 2012 Seventh
Conference_Location
Beijing
Print_ISBN
978-1-4673-2623-0
Electronic_ISBN
978-0-7695-4816-6
Type
conf
DOI
10.1109/ChinaGrid.2012.34
Filename
6337292
Link To Document