Optimal parallel merging algorithms on BSR

Author

Xiang, Limin ; Ushijima, Kazuo

Author_Institution

Dept. of Comput. Sci. & Commun. Eng., Kyushu Univ., Fukuoka, Japan

fYear

2000

fDate

2000

Firstpage

12

Lastpage

17

Abstract

Merging is one of the most fundamental problems in computer science. It is well known that Ω(N/p+loglogN) time is required to merge two sorted sequences each of length N on CRCW PRAM with p processors, where p⩽N log^αN for any constant α. We describe two optimal O(1) time solutions to the problem for p=N on BSR (Broadcasting with Selective Reduction). They are the first constant time solutions to the problem on any model of computation. We also give an optimal solution to the problem for p<N on BSR with O(N/p) time, which is the first improvement with non-constant time but still better than the lower bound for PRAM

Keywords

computational complexity; concurrency theory; merging; parallel algorithms; BSR; Broadcasting with Selective Reduction; PRAM; computer science; constant time solutions; fundamental problems; non-constant time; optimal parallel merging algorithms; optimal solution; optimal time solutions; sorted sequences; Bismuth; Broadcasting; Chromium; Computational modeling; Computer science; Erbium; Merging; Phase change random access memory; Read-write memory;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel Architectures, Algorithms and Networks, 2000. I-SPAN 2000. Proceedings. International Symposium on

Conference_Location

Dallas, TX

ISSN

1087-4089

Print_ISBN

0-7695-0936-3

Type

conf

DOI

10.1109/ISPAN.2000.900253

Filename

900253