Parallelisation d'applications pour des reseaux de processeurs homogenes ou heterogenes

Abstract : The aim of this thesis is to study and develop efficient methods for parallelization of scientific applications on parallel computers with distributed memory. In the first part we present two libraries of PVM((\it Parallel Virtual Machine)) and MPI ((\it Message Passing Interface)) communication tools. They allow implementation of programs on most parallel machines, but also on heterogeneous computer networks. This chapter illustrates the problems faced when trying to evaluate performances of networks with heterogeneous processors. To evaluate such performances we modified and adapted the concepts of speed-up and efficiency to account for heterogeneity. The second part deals with a study of parallel application libraries such as ScaLAPACK and with the development of communication masking techniques. The general concept is based on communication anticipation, in particular by pipelining message sending operations. Experimental results on Cray T3D and IBM SP1 machines validates the theoretical studies performed on basic algorithms of the libraries discussed above.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00005084
Contributor : Thèses Imag <>
Submitted on : Wednesday, February 25, 2004 - 11:32:02 AM
Last modification on : Thursday, January 11, 2018 - 6:14:33 AM
Long-term archiving on : Friday, September 14, 2012 - 11:40:50 AM

Identifiers

  • HAL Id : tel-00005084, version 1

Collections

IMAG | UGA

Citation

Laurent Colombet. Parallelisation d'applications pour des reseaux de processeurs homogenes ou heterogenes. Réseaux et télécommunications [cs.NI]. Institut National Polytechnique de Grenoble - INPG, 1994. Français. ⟨tel-00005084⟩

Share

Metrics

Record views

300

Files downloads

272