DocumentCode :
914426
Title :
Latin squares for parallel array access
Author :
Kim, Kichul ; Prasanna, Viktor K.
Author_Institution :
Dept. of Comput. Res., ETRI, Daejon, South Korea
Volume :
4
Issue :
4
fYear :
1993
fDate :
4/1/1993 12:00:00 AM
Firstpage :
361
Lastpage :
370
Abstract :
A parallel memory system for efficient parallel array access using perfect latin squares as skewing functions is discussed. Simple construction methods for building perfect latin squares are presented. The resulting skewing scheme provides conflict free access to several important subsets of an array. The address generation can be performed in constant time with simple circuitry. The skewing scheme can provide constant time access to rows, columns, diagonals, and N1/2 ×N1/2 subarrays of an N× N array with maximum memory utilization. Self-routing Benes networks can be used to realize the permutations needed between the processing elements and the memory modules. Two skewing schemes that provide conflict free access to three-dimensional arrays are also discussed. Combined with self-routing Benes networks, these schemes provide efficient access to frequently used subsets of three-dimensional arrays
Keywords :
multiprocessor interconnection networks; shared memory systems; storage management; conflict free access; parallel array access; parallel memory system; perfect latin squares; self-routing Benes networks; skewing functions; skewing scheme; Bandwidth; Buildings; Cities and towns; Computer architecture; Concurrent computing; Helium; High performance computing; Multiprocessor interconnection networks; Routing;
fLanguage :
English
Journal_Title :
Parallel and Distributed Systems, IEEE Transactions on
Publisher :
ieee
ISSN :
1045-9219
Type :
jour
DOI :
10.1109/71.219753
Filename :
219753
Link To Document :
بازگشت