DocumentCode :
3149083
Title :
Parallel research on matrix chain multiplication problem based on multi-core
Author :
Li, Hongyi ; Wu, Suping
Author_Institution :
Sch. of Math. & Comput. Sci., Ningxia Univ., Yinchuan, China
fYear :
2011
fDate :
16-18 April 2011
Firstpage :
3324
Lastpage :
3327
Abstract :
Now the computer is developing towards the multi-core and multi-CPU, but some quintessence serial algorithm can not be well transplanted to multi-core computer, so it´s necessary to make parallel analysis and improvement on the original serial algorithm. This paper analyzes matrix chain multiplication problem, and gives three algorithms based on the multi-core and multi-CPU - parting matrix chain, balancing load among nucleuses, optimal algorithm at the initial stage, theoretically analyzing and implementing each one and getting good speedup.
Keywords :
matrix multiplication; multiprocessing systems; parallel processing; resource allocation; load balancing; matrix chain multiplication problem; multi-CPU computer; multi-core computer; optimal algorithm; parallel analysis; parting matrix chain; serial algorithm; Algorithm design and analysis; Computer science; Computers; Dynamic programming; Educational institutions; Parallel algorithms; Presses; load balancing; matrix chain multiplication; multi-core; parallel;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Consumer Electronics, Communications and Networks (CECNet), 2011 International Conference on
Conference_Location :
XianNing
Print_ISBN :
978-1-61284-458-9
Type :
conf
DOI :
10.1109/CECNET.2011.5768270
Filename :
5768270
Link To Document :
بازگشت