Dynamic Load Balancing with Pair Potentials

被引:0
作者
Papin, Jean-Charles [1 ,2 ]
Denoual, Christophe [2 ]
Colombet, Laurent [2 ]
Namyst, Raymond [3 ]
机构
[1] ENS Cachan, CMLA, F-94235 Cachan, France
[2] CEA, DAM, DIF, F-91297 Arpajon, France
[3] Univ Bordeaux, F-33400 Talence, France
来源
EURO-PAR 2014: PARALLEL PROCESSING WORKSHOPS, PT II | 2014年 / 8806卷
关键词
Simulation; dynamic load-balancing; tasks; many-core; pair potential;
D O I
暂无
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
We present a new load balancing algorithm inspired by Molecular Dynamics Simulations. Our main motivation is to anticipate the rising costs of tasks-scheduling caused by the growth of the number of available cores on chips. This algorithm is based on a virtual decomposition of workload in Vorono cells centered around computing units. The method used in this paper allows cores to virtually move in order to change their computing load. Cores displacements are result of forces computation (with pair potential): attractive or repulsive forces between cores are balanced by the cores computing load (total cost of Vorono cell). Over-charged cores are more attractive than under-charged cores (which are then more repulsive). In this paper, we demonstrate the relevance of our approach by experimenting our algorithm with a high number of automatically-generated test cases, ranging from almost stable to quickly-evolving scenarii. In all cases, our algorithm is able to quickly converge to a distribution which maintains good locality properties.
引用
收藏
页码:462 / 473
页数:12
相关论文
共 13 条
  • [1] [Anonymous], 2001, CILK 5 4 6 REF MAN
  • [2] Augonnet C., 2010, 16 INT C PAR DISTR S
  • [3] StarPU: a unified platform for task scheduling on heterogeneous multicore architectures
    Augonnet, Cedric
    Thibault, Samuel
    Namyst, Raymond
    Wacrenier, Pierre-Andre
    [J]. CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE, 2011, 23 (02) : 187 - 198
  • [4] Aurenhammer F, 2000, HANDBOOK OF COMPUTATIONAL GEOMETRY, P201, DOI 10.1016/B978-044482537-7/50006-1
  • [5] A portable programming interface for performance evaluation on modern processors
    Browne, S
    Dongarra, J
    Garner, N
    Ho, G
    Mucci, P
    [J]. INTERNATIONAL JOURNAL OF HIGH PERFORMANCE COMPUTING APPLICATIONS, 2000, 14 (03) : 189 - 204
  • [6] PT-SCOTCH: A tool for efficient parallel graph ordering
    Chevalier, C.
    Pellegrini, F.
    [J]. PARALLEL COMPUTING, 2008, 34 (6-8) : 318 - 331
  • [7] Green HPC From Nice to Necessity Introduction
    Hemmert, Scott
    [J]. COMPUTING IN SCIENCE & ENGINEERING, 2010, 12 (06) : 8 - 10
  • [8] NOVEL COMPUTING ARCHITECTURES
    Kindratenko, Volodymyr V.
    [J]. COMPUTING IN SCIENCE & ENGINEERING, 2009, 11 (03) : 54 - 57
  • [9] Kukanov A., 2007, INTEL TECHNOLOGY J, V11
  • [10] OpenMP-Committee, TECHNICAL REPORT