Locality-Aware Work Stealing on Multi-CPU and Multi-GPU Architectures

Abstract : Most recent HPC platforms have heterogeneous nodes com- posed of a combination of multi-core CPUs and accelerators, like GPUs. Scheduling on such architectures relies on a static partitioning and cost model. In this paper, we present a locality-aware work stealing scheduler for multi-CPU and multi-GPU architectures, which relies on the XKaapi runtime system. We show performance results on two dense linear algebra kernels, Cholesky (POTRF) and LU (GETRF) factorization, to evaluate our scheduler on a heterogeneous architecture composed of two hexa-core CPUs and eight NVIDIA Fermi GPUs. Our experiments show that an online locality-aware scheduling achieve performance results as good as static strategies, and in most cases outperform them.
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.inria.fr/hal-00780890
Contributor : Joao Vicente Ferreira Lima <>
Submitted on : Thursday, January 24, 2013 - 11:34:10 PM
Last modification on : Thursday, October 11, 2018 - 8:48:03 AM
Long-term archiving on : Thursday, April 25, 2013 - 3:57:10 AM

File

joao-lima-multiprog2013.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00780890, version 1

Collections

Citation

Thierry Gautier, Joao Vicente Ferreira Lima, Nicolas Maillard, Bruno Raffin. Locality-Aware Work Stealing on Multi-CPU and Multi-GPU Architectures. 6th Workshop on Programmability Issues for Heterogeneous Multicores (MULTIPROG), Jan 2013, Berlin, Germany. ⟨hal-00780890⟩

Share

Metrics

Record views

1291

Files downloads

1377