Hybrid MPI-thread parallelization of adaptive mesh operations

作     者：Ibanez, Dan Dunn, Ian Shephard, Mark S. 

作者机构：Rensselaer Polytech Inst Sci Computat Res Ctr 110 Eighth St Troy NY 12180 USA 

出 版 物：《PARALLEL COMPUTING》 (并行计算)

年 卷 期：2016年第52卷

页      面：133-143页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Office of Advanced Scientific Computing Research, Office of Science, U.S. Department of Energy [DE-SC00066117] National Science Foundation (FASTMath SciDAC Institute) [OCI-1126125, IIP-1237555] 

主　　题：Hybrid system Shared memory Parallel programming models MPI Termination detection Non-blocking 

摘      要：Many of the world s leading supercomputer architectures are a hybrid of shared memory and network-distributed memory. Such an architecture lends itself to a hybrid MPI-thread programming model. We first present an implementation of inter-thread message passing based on the MPI and pthread libraries. In addition, we present an efficient implementation of termination detection for communication rounds. We use the term phased message passing to denote the communication interface based on this termination detection. This interface is then used to implement parallel operations for adaptive unstructured meshes, and the performance of resulting applications is compared to pure MPI operation. We also present new workflows enabled by the ability to vary the number of threads during run-time. (C) 2016 Elsevier B.V. All rights reserved.