StarPU-MPI: Task Programming over Clusters of Machines Enhanced with Accelerators

Cédric Augonnet 1, 2 Olivier Aumage 1, 2 Nathalie Furmento 1, 2 Raymond Namyst 1, 2 Samuel Thibault 1, 2
2 RUNTIME - Efficient runtime systems for parallel architectures
Inria Bordeaux - Sud-Ouest, UB - Université de Bordeaux, CNRS - Centre National de la Recherche Scientifique : UMR5800
Abstract : GPUs clusters are becoming widespread HPC platforms. Ex- ploiting them is however challenging, as this requires two separate paradigms (MPI and CUDA or OpenCL) and careful load balancing due to node heterogeneity. Current paradigms usually either limit themselves to of- fload part of the computation and leave CPUs idle, or require static CPU/GPU work partitioning. We thus have previously proposed StarPU, a runtime system able to dynamically scheduling tasks within a single heterogeneous node. We show how we extended the task paradigm of StarPU with MPI to easily map the task graph on MPI clusters and automatically benefit from optimized execution.
Complete list of metadatas

Cited literature [2 references]  Display  Hide  Download

https://hal.inria.fr/hal-00725477
Contributor : Nathalie Furmento <>
Submitted on : Monday, August 27, 2012 - 11:07:15 AM
Last modification on : Tuesday, May 14, 2019 - 11:38:08 AM
Long-term archiving on : Friday, December 16, 2016 - 7:52:17 AM

File

submitted_poster.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00725477, version 1

Citation

Cédric Augonnet, Olivier Aumage, Nathalie Furmento, Raymond Namyst, Samuel Thibault. StarPU-MPI: Task Programming over Clusters of Machines Enhanced with Accelerators. The 19th European MPI Users' Group Meeting (EuroMPI 2012), Sep 2012, Vienna, Austria. ⟨hal-00725477⟩

Share

Metrics

Record views

813

Files downloads

757