On the performance of parallel matrix factorisation on the hypermesh

A. Al-Ayyoub, M. Ould-Khaoua, K. Day

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

Most common multicomputer networks, e.g. d-ary h-cubes, are graph topologies where an edge (channel) interconnects exactly two vertices (nodes). Hypergraphs are a generalisation of the graph model, where a channel interconnects an arbitrary number of nodes. Previous studies have used synthetic workloads (e.g. statistical distributions) to stress the superior performance characteristics of regular multi-dimensional hypergraphs, also known as hypermeshes, over d-ary h-cubes. There has been, however, hardly any study that has considered real-world parallel applications. This paper contributes towards filling this gap by providing a comparative study of the performance of one of the most common numerical problems, namely matrix factorisation, on the hypermesh, hypercube, and d-ary h-cube, To this end, the paper first introduces orthogonal networks as a unified model for describing both the graph and hypergraph topologies. It then develops a generalised parallel algorithm for matrix factorisation and evaluates its performance on the hypermesh, hypercube and d-ary h-cube. The results reveal that the hypermesh supports matrix computation more efficiently, and therefore provides more evidence of the hypermesh as a viable network for future large-scale multicomputers.

Original languageEnglish
Pages (from-to)37-53
Number of pages17
JournalJournal of Supercomputing
Volume20
Issue number1
DOIs
Publication statusPublished - Aug 2001

Fingerprint

Matrix Factorization
Factorization
Hypergraph
Regular hexahedron
Multicomputers
Interconnect
Hypercube
Topology
Cube
Matrix Computation
Statistical Distribution
Parallel Applications
Graph Model
Graph in graph theory
Vertex of a graph
Real-world Applications
Parallel algorithms
Parallel Algorithms
Comparative Study
Workload

Keywords

  • Gaussian elimination
  • Hypermeshes
  • Interconnection networks
  • Matrix factorization
  • Multicomputers
  • Orthogonal networks
  • Performance analysis

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Hardware and Architecture
  • Theoretical Computer Science
  • Computational Theory and Mathematics

Cite this

On the performance of parallel matrix factorisation on the hypermesh. / Al-Ayyoub, A.; Ould-Khaoua, M.; Day, K.

In: Journal of Supercomputing, Vol. 20, No. 1, 08.2001, p. 37-53.

Research output: Contribution to journalArticle

@article{19ea79b4f6d545ada639122e972f5f70,
title = "On the performance of parallel matrix factorisation on the hypermesh",
abstract = "Most common multicomputer networks, e.g. d-ary h-cubes, are graph topologies where an edge (channel) interconnects exactly two vertices (nodes). Hypergraphs are a generalisation of the graph model, where a channel interconnects an arbitrary number of nodes. Previous studies have used synthetic workloads (e.g. statistical distributions) to stress the superior performance characteristics of regular multi-dimensional hypergraphs, also known as hypermeshes, over d-ary h-cubes. There has been, however, hardly any study that has considered real-world parallel applications. This paper contributes towards filling this gap by providing a comparative study of the performance of one of the most common numerical problems, namely matrix factorisation, on the hypermesh, hypercube, and d-ary h-cube, To this end, the paper first introduces orthogonal networks as a unified model for describing both the graph and hypergraph topologies. It then develops a generalised parallel algorithm for matrix factorisation and evaluates its performance on the hypermesh, hypercube and d-ary h-cube. The results reveal that the hypermesh supports matrix computation more efficiently, and therefore provides more evidence of the hypermesh as a viable network for future large-scale multicomputers.",
keywords = "Gaussian elimination, Hypermeshes, Interconnection networks, Matrix factorization, Multicomputers, Orthogonal networks, Performance analysis",
author = "A. Al-Ayyoub and M. Ould-Khaoua and K. Day",
year = "2001",
month = "8",
doi = "10.1023/A:1011140203528",
language = "English",
volume = "20",
pages = "37--53",
journal = "Journal of Supercomputing",
issn = "0920-8542",
publisher = "Springer Netherlands",
number = "1",

}

TY - JOUR

T1 - On the performance of parallel matrix factorisation on the hypermesh

AU - Al-Ayyoub, A.

AU - Ould-Khaoua, M.

AU - Day, K.

PY - 2001/8

Y1 - 2001/8

N2 - Most common multicomputer networks, e.g. d-ary h-cubes, are graph topologies where an edge (channel) interconnects exactly two vertices (nodes). Hypergraphs are a generalisation of the graph model, where a channel interconnects an arbitrary number of nodes. Previous studies have used synthetic workloads (e.g. statistical distributions) to stress the superior performance characteristics of regular multi-dimensional hypergraphs, also known as hypermeshes, over d-ary h-cubes. There has been, however, hardly any study that has considered real-world parallel applications. This paper contributes towards filling this gap by providing a comparative study of the performance of one of the most common numerical problems, namely matrix factorisation, on the hypermesh, hypercube, and d-ary h-cube, To this end, the paper first introduces orthogonal networks as a unified model for describing both the graph and hypergraph topologies. It then develops a generalised parallel algorithm for matrix factorisation and evaluates its performance on the hypermesh, hypercube and d-ary h-cube. The results reveal that the hypermesh supports matrix computation more efficiently, and therefore provides more evidence of the hypermesh as a viable network for future large-scale multicomputers.

AB - Most common multicomputer networks, e.g. d-ary h-cubes, are graph topologies where an edge (channel) interconnects exactly two vertices (nodes). Hypergraphs are a generalisation of the graph model, where a channel interconnects an arbitrary number of nodes. Previous studies have used synthetic workloads (e.g. statistical distributions) to stress the superior performance characteristics of regular multi-dimensional hypergraphs, also known as hypermeshes, over d-ary h-cubes. There has been, however, hardly any study that has considered real-world parallel applications. This paper contributes towards filling this gap by providing a comparative study of the performance of one of the most common numerical problems, namely matrix factorisation, on the hypermesh, hypercube, and d-ary h-cube, To this end, the paper first introduces orthogonal networks as a unified model for describing both the graph and hypergraph topologies. It then develops a generalised parallel algorithm for matrix factorisation and evaluates its performance on the hypermesh, hypercube and d-ary h-cube. The results reveal that the hypermesh supports matrix computation more efficiently, and therefore provides more evidence of the hypermesh as a viable network for future large-scale multicomputers.

KW - Gaussian elimination

KW - Hypermeshes

KW - Interconnection networks

KW - Matrix factorization

KW - Multicomputers

KW - Orthogonal networks

KW - Performance analysis

UR - http://www.scopus.com/inward/record.url?scp=0035425980&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035425980&partnerID=8YFLogxK

U2 - 10.1023/A:1011140203528

DO - 10.1023/A:1011140203528

M3 - Article

VL - 20

SP - 37

EP - 53

JO - Journal of Supercomputing

JF - Journal of Supercomputing

SN - 0920-8542

IS - 1

ER -