DocumentCode :
2790493
Title :
On the Power of the Multiple Associative Computing (MASC) Model Related to That of Reconfigurable Bus-Based Models
Author :
Jin, Mingxian ; Baker, Johnnie W.
Author_Institution :
Dept. of Math. & Comput. Sci., Fayetteville State Univ., NC
fYear :
2007
fDate :
26-30 March 2007
Firstpage :
1
Lastpage :
8
Abstract :
The MASC model is a multi-SIMD model that uses control parallelism to coordinate the interaction of data parallel threads. It supports a generalized associative style of parallel computation. The power of this model has been compared to that of priority CRCW PRAM and enhanced meshes. In this paper, represent the work on simulations between MASC and reconfigurable bus-based models, in particular, different versions of the reconfigurable multiple bus machine (RMBM). It is shown that MASC and the basic RMBM (B-RMBM) can simulate each other in constant time if the number of buses on the B-RMBM is Theta(j) where j is the number of MASC instruction streams. Thus, when these two models satisfy the preceding condition, they have the same power. Simulations of other stronger versions of RMBM using MASC are also considered. Since the RMBM model has been shown to be as powerful as a general reconfigurable mesh (RM), our simulations can be used to establish a relationship between MASC and RM. As RM has been widely accepted as an extremely powerful model, our work gives a better understanding of the MASC model and provides useful information concerning its power.
Keywords :
associative processing; multi-threading; basic reconfigurable multiple bus machine; multiSIMD model; multiple associative computing model; parallel computing; parallel threads; Broadcasting; Computational modeling; Computer aided instruction; Computer science; Concurrent computing; Mathematical model; Mathematics; Parallel processing; Phase change random access memory; Yarn; Multi-SIMD; associative computing; parallel computational model; reconfigurable buses; simulation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Parallel and Distributed Processing Symposium, 2007. IPDPS 2007. IEEE International
Conference_Location :
Long Beach, CA
Print_ISBN :
1-4244-0910-1
Electronic_ISBN :
1-4244-0910-1
Type :
conf
DOI :
10.1109/IPDPS.2007.370469
Filename :
4228197
Link To Document :
بازگشت