HIGHLY PARALLEL ARCHITECTURE BASED ON A DISTRIBUTED SHARED MEMORY.

A. Bode, A. Fritsch, G. Haendler, W. Henning, J. Volkert

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

For a large number of scientific computations (e. g. multigrid algorithms) an enormous amount of data has to be handled and processed. Therefore in the context of a parallel computing system the organization and access to primary memory and the close coupling of secondary memories is crucial to the overall system performance. For applications with local character we propose the concept of a distributed shared memory as the basis for a hierarchical NN (nearest neighbour) multiprocessor system. The systems architecture, applications and first results on the EGPA and DIRMU multiprocessors are presented.

Original languageEnglish
Title of host publicationUnknown Host Publication Title
EditorsGerard L. Reijns, Michael H. Barton
PublisherNorth-Holland
Pages19-28
Number of pages10
ISBN (Print)0444701311
StatePublished - 1987
Externally publishedYes

Fingerprint

Dive into the research topics of 'HIGHLY PARALLEL ARCHITECTURE BASED ON A DISTRIBUTED SHARED MEMORY.'. Together they form a unique fingerprint.

Cite this