Computer Science ›› 2015, Vol. 42 ›› Issue (11): 101-103.doi: 10.11896/j.issn.1002-137X.2015.11.021

Previous Articles     Next Articles

Towards Unified Description for Reduction Algorithms

XIONG Yu-qing   

  • Online:2018-11-14 Published:2018-11-14

Abstract: Reduction algorithms are used widely in parallel computing.There are many reduction algorithms applied to different situations.These reduction algorithms are different from each other.Logical topology is the key cause to make the differences.In order to unify the description of reduction algorithms,and uncover their common feature,a definition of logical topology and its properties were presented in this paper.Based on it,a unified description for reduction algorithms was proposed.The description contributes to understand reduction algorithms and thus to design efficient reduction algorithms for different computational applications and environments.It can also be regarded as a reduction algorithmic framework for integrating different semantics and thus maybe guide to design new reduction algorithms with the semantics.Essentially,the unified description is a definition of reduction algorithms formally,and so it helps to verify correctness of reduction algorithms.

Key words: Reduction algorithm,Logic topology,Algorithm description,Parallel computing

[1] Balaji P,Kimpe D.On the Reproducibility of MPI ReductionOperations:ANL/MCS-P4093-0713[R].Argonne,IL,USA:Argonne National Laboratory,2013
[2] Thakur R,Rabenseifner R,Gropp W.Optimization of Collective Communication Operations in MPICH: ANL/MCS-P1140-0304[R].Argonne,IL,USA:Argonne National Laboratory,2004
[3] Rabenseifner R.New optimized MPI reduce algorithm.http://www.hlrs.de/organization/par/services/models/mpi/myreduce.html
[4] Dongarra J J,Whaley R C.A User’s Guide to the BLACS v1.0.1:UT CS-95-281[R]LAPACK Working Note #94,University of Tennessee,1995

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!