DocumentCode :
416876
Title :
Performance study of load balancing strategies for simulators of nanoscale measurements of material properties
Author :
Song, Yingwen ; Watanabe, Satoshi
Author_Institution :
CREST, Tokyo Univ., Japan
fYear :
2004
fDate :
20-22 July 2004
Firstpage :
203
Lastpage :
209
Abstract :
We have compared various load balancing strategies for the program which was developed by our group to simulate nanoscale measurements of material properties. Among the strategies we have examined, static block distributions, static cyclic distributions, self-adapting block distributions, dynamic load balancing, and hybrid load balancing, the hybrid approach has been proven to give the best performance for the program on Hitachi SR8000.
Keywords :
computational complexity; digital simulation; measurement systems; parallel algorithms; resource allocation; Hitachi SR8000; dynamic load balancing; hybrid load balancing; material properties; nanoscale measurements simulation; self-adapting block distributions; static block distributions; static cyclic distributions; Computational modeling; Electrodes; Electrons; Equations; Grid computing; High performance computing; Load management; Material properties; Nanostructured materials; Scattering;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
High Performance Computing and Grid in Asia Pacific Region, 2004. Proceedings. Seventh International Conference on
Print_ISBN :
0-7695-2138-X
Type :
conf
DOI :
10.1109/HPCASIA.2004.1324036
Filename :
1324036
Link To Document :
بازگشت