EFFECTIVENESS OF COMBINING IN SHARED MEMORY PARALLEL COMPUTERS IN THE PRESENCE OF 'HOT SPOTS'.

Gyungho Lee, Clyde P. Kruskal, David J. Kuck

Research output: Chapter in Book/Report/Conference proceedingConference contribution

34 Citations (Scopus)

Abstract

Concurrent requests to a shared variable by many processors on a shared-memory machine can create contention serious enough to stall large machines. This idea has been formalized in the hot spot traffic model in which a fixed fraction of memory requests is for a single shared variable. Combining, in which several requests for the same variable can be combined into a single request, has been suggested as an effective method of alleviating this contention. The NYU Ultracomputer and the IBM RP3 machine use pairwise combining, in which only two requests for the same variable can be combined at a switch. A study is made of the effectiveness of combining. It turns out that pairwise combining cannot handle hot spots if the machine size is large enough. Suggestion are put forth to overcome this weakness.

Original languageEnglish
Title of host publicationProceedings of the International Conference on Parallel Processing
EditorsKai Hwang, Steven M. Jacobs, Earl E. Swartzlander
PublisherIEEE
Pages35-41
Number of pages7
ISBN (Print)0818607246
Publication statusPublished - 1986 Dec 1

Publication series

NameProceedings of the International Conference on Parallel Processing
ISSN (Print)0190-3918

ASJC Scopus subject areas

  • Hardware and Architecture

Fingerprint Dive into the research topics of 'EFFECTIVENESS OF COMBINING IN SHARED MEMORY PARALLEL COMPUTERS IN THE PRESENCE OF 'HOT SPOTS'.'. Together they form a unique fingerprint.

  • Cite this

    Lee, G., Kruskal, C. P., & Kuck, D. J. (1986). EFFECTIVENESS OF COMBINING IN SHARED MEMORY PARALLEL COMPUTERS IN THE PRESENCE OF 'HOT SPOTS'. In K. Hwang, S. M. Jacobs, & E. E. Swartzlander (Eds.), Proceedings of the International Conference on Parallel Processing (pp. 35-41). (Proceedings of the International Conference on Parallel Processing). IEEE.