On the effectiveness of combining in resolving ′Hot-Spot′ contention

Kyung Ho Lee, Clyde P. Kruskal, David J. Kuck

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Sharing the memory in a parallel computer suggests that there is a possibility of many processors requesting a shared variable at virtually the same time. This idea has been formalized in the “hot spot” traffic model, where a fixed fraction of memory requests is for a single shared variable. Under the model, it has been shown that such concurrent requests to shared variable can create contention serious enough to stall large machines. As an effective method of alleviating this type of contention, “combining,” in which several requests for the same variable can be combined into a single request, has been suggested. 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 of the processor-memory intereconnection. We study the effectiveness of combining. In particular, it turns out that pairwise combining is slightly too restrictive to handle hot spots if the machine size becomes very large. We suggest ways to overcome this weakness.

Original languageEnglish
Pages (from-to)136-144
Number of pages9
JournalJournal of Parallel and Distributed Computing
Volume20
Issue number2
DOIs
Publication statusPublished - 1994 Jan 1
Externally publishedYes

Fingerprint

Contention
Hot Spot
Data storage equipment
Pairwise
Switches
Traffic Model
Parallel Computers
Concurrent
Switch
Sharing

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Hardware and Architecture
  • Computer Networks and Communications
  • Artificial Intelligence

Cite this

On the effectiveness of combining in resolving ′Hot-Spot′ contention. / Lee, Kyung Ho; Kruskal, Clyde P.; Kuck, David J.

In: Journal of Parallel and Distributed Computing, Vol. 20, No. 2, 01.01.1994, p. 136-144.

Research output: Contribution to journalArticle

@article{15861f35bf5c4fcea55e8d53b4d1a742,
title = "On the effectiveness of combining in resolving ′Hot-Spot′ contention",
abstract = "Sharing the memory in a parallel computer suggests that there is a possibility of many processors requesting a shared variable at virtually the same time. This idea has been formalized in the “hot spot” traffic model, where a fixed fraction of memory requests is for a single shared variable. Under the model, it has been shown that such concurrent requests to shared variable can create contention serious enough to stall large machines. As an effective method of alleviating this type of contention, “combining,” in which several requests for the same variable can be combined into a single request, has been suggested. 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 of the processor-memory intereconnection. We study the effectiveness of combining. In particular, it turns out that pairwise combining is slightly too restrictive to handle hot spots if the machine size becomes very large. We suggest ways to overcome this weakness.",
author = "Lee, {Kyung Ho} and Kruskal, {Clyde P.} and Kuck, {David J.}",
year = "1994",
month = "1",
day = "1",
doi = "10.1006/jpdc.1994.1014",
language = "English",
volume = "20",
pages = "136--144",
journal = "Journal of Parallel and Distributed Computing",
issn = "0743-7315",
publisher = "Academic Press Inc.",
number = "2",

}

TY - JOUR

T1 - On the effectiveness of combining in resolving ′Hot-Spot′ contention

AU - Lee, Kyung Ho

AU - Kruskal, Clyde P.

AU - Kuck, David J.

PY - 1994/1/1

Y1 - 1994/1/1

N2 - Sharing the memory in a parallel computer suggests that there is a possibility of many processors requesting a shared variable at virtually the same time. This idea has been formalized in the “hot spot” traffic model, where a fixed fraction of memory requests is for a single shared variable. Under the model, it has been shown that such concurrent requests to shared variable can create contention serious enough to stall large machines. As an effective method of alleviating this type of contention, “combining,” in which several requests for the same variable can be combined into a single request, has been suggested. 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 of the processor-memory intereconnection. We study the effectiveness of combining. In particular, it turns out that pairwise combining is slightly too restrictive to handle hot spots if the machine size becomes very large. We suggest ways to overcome this weakness.

AB - Sharing the memory in a parallel computer suggests that there is a possibility of many processors requesting a shared variable at virtually the same time. This idea has been formalized in the “hot spot” traffic model, where a fixed fraction of memory requests is for a single shared variable. Under the model, it has been shown that such concurrent requests to shared variable can create contention serious enough to stall large machines. As an effective method of alleviating this type of contention, “combining,” in which several requests for the same variable can be combined into a single request, has been suggested. 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 of the processor-memory intereconnection. We study the effectiveness of combining. In particular, it turns out that pairwise combining is slightly too restrictive to handle hot spots if the machine size becomes very large. We suggest ways to overcome this weakness.

UR - http://www.scopus.com/inward/record.url?scp=0007740463&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0007740463&partnerID=8YFLogxK

U2 - 10.1006/jpdc.1994.1014

DO - 10.1006/jpdc.1994.1014

M3 - Article

AN - SCOPUS:0007740463

VL - 20

SP - 136

EP - 144

JO - Journal of Parallel and Distributed Computing

JF - Journal of Parallel and Distributed Computing

SN - 0743-7315

IS - 2

ER -