Abstract
Quality of applications that includes robustness, real time response, and accurate performance has become a vital property for current applications. Among other existing solutions for such problems, parallel computing is a trending solution. Sorting is one of the main parts in almost every algorithm and various parallel sorting techniques have employed parallel architectures to do a qualified sorting. Gaining the best based on various different factors where speedup is the premier, is a topic of discussion. In this paper, we have issued the generalization of SOCD sort on the novel Centralized Diamond architecture which benefits from Single Instruction Multiple Data (SIMD) architecture with a time complexity of O (logn) on PRAM EREW(Parallel Random Access Machine Exclusive Read Exclusive Write). The results of conducted simulations of the algorithm prove the results of theoretical analysis of the algorithm. The findings of this research can be exploited in developing faster embedded systems. Using an appropriate interconnection network for achieving reasonable speedup in the execution of applications is important especially in embedded systems. Keywords: Parallel sorting, Diamond architecture, SIMD, Generalized SOCD
Highlights
One of the most exciting research areas in computer science is parallel processing that has gained a lot of interest in the past decade
Jana proposed Multi-Mesh of Trees (MMT) architecture, which is a combination of multi mesh and mesh of trees and uses n4 processors
Hayashi et al, (1998) provided an EREWPRAM and both the CREW-PRAM and the CRCW merge sorts are issued with a total cost of O(log n) and O(log log n +log k) respectively
Summary
One of the most exciting research areas in computer science is parallel processing that has gained a lot of interest in the past decade. We have generalized the new born SOCD sorting algorithm on Centralized Diamond architecture. This specific design leads to a tradeoff between the number of each type of processor element and the related complexity and cost This architecture has been employed by NOD and ENOD to sort input data elements (Damrudi et al, 2009; Damrudi et al, 2010; Jadidy Aval et al, 2010). Generalized SOCD sort on centralized diamond architecture is presented. There is a direct connection between every two neighboring PEs in this level that is formalized as following: As it is explained by Damrudi et al, (2011), the relation between data elements and PEs of this architecture is N 7 4 n 1.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
