Advanced Parallel Programming with MPI（先进与MPI并行编程）.pptx

下载文档 降价啦

14
0
约6.85千字
约 262页
2017-07-29 发布于浙江
举报
版权申诉
保障服务

Advanced Parallel Programming with MPI（先进与MPI并行编程）.pptx

1、本文档共262页，可阅读全部内容。
2、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。

Advanced Parallel Programming with MPI（先进与MPI并行编程）

Advanced Parallel Programming with MPI;What this tutorial will cover;What is MPI?;Following MPI Standards;Important considerations while using MPI;Parallel Sort using MPI Send/Recv;;A Non-Blocking communication example;;MPI Collective Routines;MPI Built-in Collective Computation Operations;Introduction to Datatypes in MPI;Derived Datatype Example;Advanced Topics: Hybrid Programming with Threads and Shared Memory;MPI and Threads;Programming for Multicore;MPI’s Four Levels of Thread Safety;MPI+OpenMP;Specification of MPI_THREAD_MULTIPLE;Threads and MPI;An Incorrect Program;A Correct Example;The Current Situation;Performance with MPI_THREAD_MULTIPLE;Message Rate Results on BG/P ;Why is it hard to optimize MPI_THREAD_MULTIPLE;27;Ptolemy and Threads;An Example I encountered recently;2 Proceses, 2 Threads, Each Thread Executes this Code;What Happened;Hybrid Programming with Shared Memory;Advanced Topics: Nonblocking Collectives;Nonblocking Collective Communication;Nonblocking Communication;Nonblocking Communication;Collective Communication;Nonblocking Collective Communication;Nonblocking Collective Communication;Nonblocking Collectives Overlap; A Non-Blocking Barrier?;A Semantics Example: DSDE;Dynamic Sparse Data Exchange (DSDE);Using Alltoall (PEX) ;Reduce_scatter (PCX);MPI_Ibarrier (NBX);Parallel Breadth First Search;A Complex Example: FFT;FFT Software Pipelining;A Complex Example: FFT;Nonblocking And Collective Summary;Advanced Topics: One-sided Communication;One-sided Communication;Two-sided Communication Example;One-sided Communication Example;Comparing One-sided and Two-sided Programming;What we need to know in MPI RMA;Creating Public Memory;Window creation models;MPI_WIN_CREATE;Example with MPI_WIN_CREATE;MPI_WIN_ALLOCATE;Example with MPI_WIN_ALLOCATE;MPI_WIN_CREATE_DYNAMIC;Example with MPI_WIN_CREATE_DYNAMIC;Data movement;Data movement: Get;Data movement: Put;Data aggregation: Accumulate;Data aggregation: Get Accumulate;Ordering of Operations in MPI RMA;Additional