Task migration of DSP application specified with a DFG and implemented with the BSP computing model on a CPU-GPU cluster

Farouk Mansouri 1 Sylvain Huet 1 Vincent Fristot 1 Dominique Houzet 1
1 GIPSA-AGPIG - AGPIG
GIPSA-DIS - Département Images et Signal
Abstract : Nowadays computer applications are becoming heavier and require, at the same time, real-time results. The Heterogeneous clusters with their computing power represent a good solution to this request. However, it is possible that during the execution, a computing element of the cluster becomes defaulting, needs maintenance, or that the load needs to be re-balanced. . . In this paper, we propose a migration strategy for relocating the execution of a task to another computing element. In particular, we are interested in remap nodes of Data Flow Graph (DFG), representing Digital Signal Processing (DSP) application, onto heterogeneous (CPU-GPU) clusters while keeping up the flow of data and minimizing the temporal perturbation. For our approach, we give a lower bound for the flow of data after the migration and, validate it by the real-time construction of visual saliency map from video input.
Complete list of metadatas

Cited literature [17 references]  Display  Hide  Download

http://hal.univ-grenoble-alpes.fr/hal-00937725
Contributor : Farouk Mansouri <>
Submitted on : Tuesday, January 28, 2014 - 5:15:55 PM
Last modification on : Thursday, October 25, 2018 - 9:34:01 AM
Long-term archiving on : Sunday, April 9, 2017 - 1:54:47 AM

File

mansouri.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00937725, version 1

Citation

Farouk Mansouri, Sylvain Huet, Vincent Fristot, Dominique Houzet. Task migration of DSP application specified with a DFG and implemented with the BSP computing model on a CPU-GPU cluster. 2013 Conference on Design and Architectures for Signal and Image Processing (DASIP 2013), Oct 2013, Cagliari, Italy. pp.326-333. ⟨hal-00937725⟩

Share

Metrics

Record views

742

Files downloads

220