Design and analysis of scheduling strategies for multi-CPU and multi-GPU architectures

Abstract : In this paper, we present a comparison of scheduling strategies for heterogeneous multi-CPU and multi-GPU architectures. We designed and evaluated four scheduling strategies on top of XKaapi runtime: work stealing, data-aware work stealing, locality-aware work stealing, and Heterogeneous Earliest-Finish-Time (HEFT). On a heterogeneous architecture with 12 CPUs and 8 GPUs, we analysed our scheduling strategies with four benchmarks: a BLAS-1 AXPY vector operation, a Jacobi 2D iterative computation, and two linear algebra algorithms Cholesky and LU. We conclude that the use of work stealing may be efficient if task annotations are given along with a data locality strategy. Furthermore, our experimental results suggests that HEFT scheduling performs better on applications with very regular computations and low data locality.
Complete list of metadatas

Cited literature [29 references]  Display  Hide  Download

https://hal.inria.fr/hal-01132037
Contributor : Joao Vicente Ferreira Lima <>
Submitted on : Monday, March 23, 2015 - 1:11:21 AM
Last modification on : Tuesday, July 9, 2019 - 1:26:59 AM

File

parco2014.pdf
Files produced by the author(s)

Identifiers

Collections

Citation

Joao Vicente Ferreira Lima, Thierry Gautier, Vincent Danjean, Bruno Raffin, Nicolas Maillard. Design and analysis of scheduling strategies for multi-CPU and multi-GPU architectures. Parallel Computing, Elsevier, 2015, 44, pp.37-52. ⟨10.1016/j.parco.2015.03.001⟩. ⟨hal-01132037⟩

Share

Metrics

Record views

519

Files downloads

501