Skip to Main content Skip to Navigation
Theses

Approches algorithmiques pour l'ordonnancement d'applications parallèles avec communications

Abstract : This thesis focus on the problem of scheduling the tasks of a parallel application taking into account the impact of communications. On distributed memory computers such as PC clusters communication time are important. The objectives of this thesis is to study models allowing to take into account efficiently those communications and to study the scheduling problems under those models. We study the large communication delay model based on an explicit modeling of communications and the malleable tasks model where task may be executed on a variable number of processors. oure contributions goes to the three following directions. For scheduling malleable tasks with precedence constraints we proposed approximation algorithm (polynomial algorithm offering guarantees relatively to an optimal solution) for the case of trees and for the case of an arbitrary precedence task graph. An original approach is proposed for the clustering problem (scheduling on an unbounded nimber of processors). This approach is based on a recursive decomposition of
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00010476
Contributor : Grégory Mounié <>
Submitted on : Friday, October 7, 2005 - 3:25:50 PM
Last modification on : Friday, November 6, 2020 - 4:38:59 AM
Long-term archiving on: : Tuesday, September 7, 2010 - 5:29:27 PM

Identifiers

  • HAL Id : tel-00010476, version 1

Collections

INRIA | CNRS | IMAG | UGA

Citation

Renaud Lepère. Approches algorithmiques pour l'ordonnancement d'applications parallèles avec communications. Réseaux et télécommunications [cs.NI]. Institut National Polytechnique de Grenoble - INPG, 2001. Français. ⟨tel-00010476⟩

Share