NDLTD Global ETD Search
Return to search

Communication/Computation Overlap in MPI

Description

This talk discusses optimized collective algorithms and the benefits of leveraging independent hardware entities in a pipelined manner. The resulting approach uses overlap of computation and communication to reach this task. Different examples are given.

Links & Downloads
  1. urn:nbn:de:swb:ch1-200600021
  2. https://monarch.qucosa.de/id/qucosa%3A18443
  3. https://monarch.qucosa.de/api/qucosa%3A18443/attachment/ATT-0/
  4. https://monarch.qucosa.de/api/qucosa%3A18443/attachment/ATT-1/
Tags
info:eu-repo/classification/ddc/004
ddc:004
MPI <Schnittstelle>
Parallelrechner
MPI_BARRIER
Non blocking
collective operations
kollektive Operationen
Additional Fields
Identiferoai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:18443
Date04 January 2006
CreatorsHoefler, Torsten
ContributorsRehm, Wolfgang
PublisherTechnische Universität Chemnitz
Source SetsHochschulschriftenserver (HSSS) der SLUB Dresden
LanguageEnglish
Detected LanguageEnglish
Typedoc-type:lecture, info:eu-repo/semantics/lecture, doc-type:Text
Rightsinfo:eu-repo/semantics/openAccess

Page generated in 0.0014 seconds