Scalable Node Allocation for Improved Performance in Regular and Anisotropic 3D Torus Supercomputers
Scalable Node Allocation for Improved Performance in Regular and Anisotropic 3D Torus Supercomputers
复制标题
可扩展的节点分配可提高常规和各向异性 3D Torus 超级计算机的性能
DOI：
10.1007/978-3-642-24449-0_9
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
H. Mills
中科院分区：
文献类型：
--
作者：
Carl Albing;N. Troullier;Stephen Whalen;R. Olson;Joe Glenski;H. Pritchard;H. Mills
MPI 应用程序性能可能会根据调度程序的等级放置而有所不同，无论是在节点之间还是在同一多核芯片的内核上。默认情况下，MPI 应用程序受将节点分配给作业的应用程序放置软件决策的支配。我们在此描述了 3D 环面中分配节点排序的一般方法，以及它如何提高 MPI 应用程序性能，即使面对各向异性互连也是如此。我们定量地证明，基于拓扑的排序可以提高在 Top10 超级计算机上运行的多个 MPI 应用程序的性能。
MPI application performance can vary based on the scheduler’s placing of ranks, whether between nodes or on cores in the same multi-core chip. MPI applications, by default, are at the mercy of the application placement software decision that assigns nodes to a job. We describe herein the general approach of node ordering for allocation in a 3D torus, how it improved MPI application performance, even in the face of an anisotropic interconnect. We demonstrate, quantitatively, that our topologically-based ordering results in improved performance for several MPI applications running on a Top10 supercomputer.