In this paper, we focus on the scheduling of Laplace task graph on a general platform where both communication links and processing units are heterogeneous. In this context, it is known that deriving optimal algorithm, in the sense of makespan …
This thesis deals with the high performance computation problems and more specifically with those of scientific parallel computation for irregular real-world applications. In the first part, we describe a method for overlapping communications on …
L'utilisation des machines parallèles ̀mémoire distribué apporte un gain important en performances et en taille m'óire mais am`è en contre partie un surcoût en communications. Afin d'obtenir des programmes performants et extensibles, il convient de …
Le cadre génŕal de ce travail concerne les machines parallèles M.I.M.D. ̀ m'óire distribu'e ́Dans ce contexte, il est admis que la bonne d'emćhe de gestion du parall'elié passe par l'utilisation syst'ematúe de bibliot`ùes performantes de …
In this paper, we present a method for overlapping communications on parallel computers for pipelined algorithms. We first introduce a general theoretical model which leads to a generic computation scheme for the optimal packet size. Then, we use the …