- 1、本文档共119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)。
查看更多
Matrix-vector Multiplication参考
Parallel Programmingin C with MPI and OpenMP Michael J. Quinn Chapter 8 Matrix-vector Multiplication Chapter Objectives Review matrix-vector multiplicaiton Propose replication of vectors Develop three parallel programs, each based on a different data decomposition Outline Sequential algorithm and its complexity Design, analysis, and implementation of three parallel programs Rowwise block striped Columnwise block striped Checkerboard block Sequential Algorithm Storing Vectors Divide vector elements among processes Replicate vector elements Vector replication acceptable because vectors have only n elements, versus n2 elements in matrices Rowwise Block Striped Matrix Partitioning through domain decomposition Primitive task associated with Row of matrix Entire vector Phases of Parallel Algorithm Agglomeration and Mapping Static number of tasks Regular communication pattern (all-gather) Computation time per task is constant Strategy: Agglomerate groups of rows Create one task per MPI process Complexity Analysis Sequential algorithm complexity: ?(n2) Parallel algorithm computational complexity: ?(n2/p) Communication complexity of all-gather: ?(log p + n) Overall complexity: ?(n2/p + log p) Isoefficiency Analysis Sequential time complexity: ?(n2) Only parallel overhead is all-gather When n is large, message transmission time dominates message latency Parallel communication time: ?(n) n2 ? Cpn ? n ? Cp and M(n) = n2 System is not highly scalable Block-to-replicated Transformation MPI_Allgatherv MPI_Allgatherv MPI_Allgatherv in Action Function create_mixed_xfer_arrays First array How many elements contributed by each process Uses utility macro BLOCK_SIZE Second array Starting position of each process’ block Assume blocks in process rank order Function replicate_block_vector Create space for entire vector Create “mixed transfer” arrays Call MPI_Allgatherv Function read_replicated_vector Process p-1 Opens file Reads vector length Broadcast vector length (root process = p-
您可能关注的文档
- Libnids在商用多核系统上的并行化详细设计说明书参考.doc
- linux下文件压缩与解压的分析研究参考.doc
- Introduction to CloudStack 3.0 Technical Presentation参考.pptx
- Loan Amortization;Mortgage loan amortization schedule参考.ppt
- Linux进程间通信的研究与实现参考.doc
- loan balances and details of individual payments参考.ppt
- LPG储罐码头环境风险评价参考.doc
- Long-Term Debt参考.ppt
- LED灯具生产基地建设项目可行性研究报告参考.doc
- Long-Term Debt and Lease Financing参考.ppt
- GB/T 40096.6-2024就地化继电保护装置技术规范 第6部分:母线保护.pdf
- 《GB/T 40096.6-2024就地化继电保护装置技术规范 第6部分:母线保护》.pdf
- GB/T 43980-2024口译服务 医疗口译要求.pdf
- 中国国家标准 GB/T 43980-2024口译服务 医疗口译要求.pdf
- 《GB/T 43980-2024口译服务 医疗口译要求》.pdf
- GB/T 17215.301-2024电测量设备(交流) 特殊要求 第1部分:多功能电能表.pdf
- 《GB/T 17215.301-2024电测量设备(交流) 特殊要求 第1部分:多功能电能表》.pdf
- 中国国家标准 GB/T 17215.301-2024电测量设备(交流) 特殊要求 第1部分:多功能电能表.pdf
- 中国国家标准 GB/Z 43973-2024非介入式负荷监测(NILM)系统用感知装置.pdf
- GB/Z 43973-2024非介入式负荷监测(NILM)系统用感知装置.pdf
文档评论(0)