Systematic Reduction of Data Movement in Algebraic Multigrid Solvers
Systematic Reduction of Data Movement in Algebraic Multigrid Solvers
复制标题
系统地减少代数多重网格求解器中的数据移动
DOI:
10.1109/ipdpsw.2013.164
复制
发表时间:
2013
期刊:
影响因子:
--
通讯作者:
U. Yang
中科院分区:
文献类型:
--
作者:
Hormozd Gahvari;W. Gropp;K. E. Jordan;M. Schulz;U. Yang
Algebraic Multigrid (AMG) solvers find wide use in scientific simulation codes. Their ideal computational complexity makes them especially attractive for solving large problems on parallel machines. However, they also involve a substantial amount of data movement, posing challenges to performance and scalability. In this paper, we present an algorithm that provides a systematic means of reducing data movement in AMG. The algorithm operates by gathering and redistributing the problem data to reduce the need to move it on the communication-intensive coarse grid portion of AMG. The data is gathered in a way that ensures data locality by keeping data movement confined to specific regions of the machine. Any decision to gather data is made systematically through the means of a performance model. This approach results in substantial speedups on a multicore cluster when using AMG to solve a variety of test problems.
DOI:
--
发表时间:
2012
期刊:
影响因子:
--
作者:
Nakajima;K
通讯作者:
K
DOI:
10.1137/1.9780898718003
发表时间:
2003-05
期刊:
--
影响因子:
--
作者:
Y. Saad
通讯作者:
Y. Saad