"Tools to schedule and simulate parallel computations expressed as directed graphs"
Arturo González Escribano, Arjan J. C. van Gemund, Valentín Cardeñoso Payo
Informe técnico: IT-DI-2007-0001
Documento completo: 
Pdf (109 Kb).
Resumen: Some parallel computations, i.e. sparse-matrix factorizations, may be expressed as directed graphs. Once the graphs are obtained, they may be scheduled or mapped to a parallel machine in different ways. This document describes tools developed to experiment with these kind of computations, creating trivial schedulings, and generating MPI programs which simulate their behavior when executed on real machines. The tools have been used to experiment with some sparse-matrix factorization graphs on a Beowulf cluster.
|
Referencia bibliográfica (formato BibTeX) @TECHREPORT { IT-DI-2007-0001, author = { Arturo González Escribano and Arjan J. C. van Gemund and Valentín Cardeñoso Payo }, title = { Tools to schedule and simulate parallel computations expressed as directed graphs }, number = { IT-DI-2007-0001 }, year = { 2007 }, institution = { Dep. Informatica, Universidad de Valladolid } } |