Efficiency Issues in Parallel Coarsening Schemes

Gallivan, K A ; Yang, U M

Lawrence Livermore National Laboratory

关键词: Storage;    Implementation;    42;    Efficiency;    Algorithms;   

DOI  :  10.2172/15004536
RP-ID  :  UCRL-ID-153078
RP-ID  :  W-7405-ENG-48
RP-ID  :  15004536

美国|英语

来源: UNT Digital Library

【 摘 要 】

Various options for sequential, shared memory and distributed memory implementations for the CLJP algorithm, a parallel coarsening scheme within algebraic multigrid, are discussed. The use of different data structures as well as different approaches of implementing the actual algorithm are investigated, and experimental results illustrating the results are presented.