Fast deterministic simulation of computations on faulty parallel machines

Bogdan S. Chlebus, Leszek Gąsieniec, Andrzej Pelc

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

Abstract

A method of deterministic simulation of fully operational parallel machines on the analogous machines prone to errors is developed. The simulation is presented for the exclusive-read exclusive-write (EREW) PRAM and the Optical Communication Parallel Computer (OCPC), but it applies to a large class of parallel computers. It is shown that simulations of operational multiprocessor machines on faulty ones can be performed with logarithmic slowdown in the worst case. More precisely, we prove that both a PRAM with a bounded fraction of faulty processors and memory cells and an OCPC with a bounded fraction of faulty processors can simulate deterministically their fault-free counterparts with O(logn) slowdown and preprocessing done in time O(log2 n). The fault model is as follows. The faults are deterministic (worst-case distribution) and static (do not change in the course of a computation). If a processor attempts to communicate with some other processor (in the case of an OCPC) or read a memory word (in the case of a PRAM) then it is immediately notified whether the operation was successful (fault-free addressee) or failed (faulty addressee). This is for the first time that a general fast deterministic simulation technique is designed for the EREW PRAM with the worst-case fault distribution. The simulation is designed in such a way that it relies only on a fraction of all the operational processors. During preprocessing, the active processors retrieve the original input provided to all processors before the simulation started. This is accomplished by adapting the information-dispersal method.

Original languageEnglish (US)
Title of host publicationAlgorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings
EditorsPaul Spirakis
PublisherSpringer Verlag
Pages89-101
Number of pages13
ISBN (Print)3540603131, 9783540603139
StatePublished - Jan 1 1995
Externally publishedYes
Event3rd Annual European Symposium on Algorithms, ESA 1995 - Corfu, Greece
Duration: Sep 25 1995Sep 27 1995

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume979
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference3rd Annual European Symposium on Algorithms, ESA 1995
CountryGreece
CityCorfu
Period9/25/959/27/95

Fingerprint

Parallel Machines
Optical communication
Parallel Computers
Fault
Optical Communication
Simulation
Data storage equipment
Preprocessing
Multiprocessor
Immediately
Logarithmic
Cell

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Chlebus, B. S., Gąsieniec, L., & Pelc, A. (1995). Fast deterministic simulation of computations on faulty parallel machines. In P. Spirakis (Ed.), Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings (pp. 89-101). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 979). Springer Verlag.

Fast deterministic simulation of computations on faulty parallel machines. / Chlebus, Bogdan S.; Gąsieniec, Leszek; Pelc, Andrzej.

Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings. ed. / Paul Spirakis. Springer Verlag, 1995. p. 89-101 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 979).

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

Chlebus, BS, Gąsieniec, L & Pelc, A 1995, Fast deterministic simulation of computations on faulty parallel machines. in P Spirakis (ed.), Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 979, Springer Verlag, pp. 89-101, 3rd Annual European Symposium on Algorithms, ESA 1995, Corfu, Greece, 9/25/95.
Chlebus BS, Gąsieniec L, Pelc A. Fast deterministic simulation of computations on faulty parallel machines. In Spirakis P, editor, Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings. Springer Verlag. 1995. p. 89-101. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
Chlebus, Bogdan S. ; Gąsieniec, Leszek ; Pelc, Andrzej. / Fast deterministic simulation of computations on faulty parallel machines. Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings. editor / Paul Spirakis. Springer Verlag, 1995. pp. 89-101 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{7c3bb165afbf4ebfaebbd18ada6e4dfd,
title = "Fast deterministic simulation of computations on faulty parallel machines",
abstract = "A method of deterministic simulation of fully operational parallel machines on the analogous machines prone to errors is developed. The simulation is presented for the exclusive-read exclusive-write (EREW) PRAM and the Optical Communication Parallel Computer (OCPC), but it applies to a large class of parallel computers. It is shown that simulations of operational multiprocessor machines on faulty ones can be performed with logarithmic slowdown in the worst case. More precisely, we prove that both a PRAM with a bounded fraction of faulty processors and memory cells and an OCPC with a bounded fraction of faulty processors can simulate deterministically their fault-free counterparts with O(logn) slowdown and preprocessing done in time O(log2 n). The fault model is as follows. The faults are deterministic (worst-case distribution) and static (do not change in the course of a computation). If a processor attempts to communicate with some other processor (in the case of an OCPC) or read a memory word (in the case of a PRAM) then it is immediately notified whether the operation was successful (fault-free addressee) or failed (faulty addressee). This is for the first time that a general fast deterministic simulation technique is designed for the EREW PRAM with the worst-case fault distribution. The simulation is designed in such a way that it relies only on a fraction of all the operational processors. During preprocessing, the active processors retrieve the original input provided to all processors before the simulation started. This is accomplished by adapting the information-dispersal method.",
author = "Chlebus, {Bogdan S.} and Leszek Gąsieniec and Andrzej Pelc",
year = "1995",
month = "1",
day = "1",
language = "English (US)",
isbn = "3540603131",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "89--101",
editor = "Paul Spirakis",
booktitle = "Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings",

}

TY - GEN

T1 - Fast deterministic simulation of computations on faulty parallel machines

AU - Chlebus, Bogdan S.

AU - Gąsieniec, Leszek

AU - Pelc, Andrzej

PY - 1995/1/1

Y1 - 1995/1/1

N2 - A method of deterministic simulation of fully operational parallel machines on the analogous machines prone to errors is developed. The simulation is presented for the exclusive-read exclusive-write (EREW) PRAM and the Optical Communication Parallel Computer (OCPC), but it applies to a large class of parallel computers. It is shown that simulations of operational multiprocessor machines on faulty ones can be performed with logarithmic slowdown in the worst case. More precisely, we prove that both a PRAM with a bounded fraction of faulty processors and memory cells and an OCPC with a bounded fraction of faulty processors can simulate deterministically their fault-free counterparts with O(logn) slowdown and preprocessing done in time O(log2 n). The fault model is as follows. The faults are deterministic (worst-case distribution) and static (do not change in the course of a computation). If a processor attempts to communicate with some other processor (in the case of an OCPC) or read a memory word (in the case of a PRAM) then it is immediately notified whether the operation was successful (fault-free addressee) or failed (faulty addressee). This is for the first time that a general fast deterministic simulation technique is designed for the EREW PRAM with the worst-case fault distribution. The simulation is designed in such a way that it relies only on a fraction of all the operational processors. During preprocessing, the active processors retrieve the original input provided to all processors before the simulation started. This is accomplished by adapting the information-dispersal method.

AB - A method of deterministic simulation of fully operational parallel machines on the analogous machines prone to errors is developed. The simulation is presented for the exclusive-read exclusive-write (EREW) PRAM and the Optical Communication Parallel Computer (OCPC), but it applies to a large class of parallel computers. It is shown that simulations of operational multiprocessor machines on faulty ones can be performed with logarithmic slowdown in the worst case. More precisely, we prove that both a PRAM with a bounded fraction of faulty processors and memory cells and an OCPC with a bounded fraction of faulty processors can simulate deterministically their fault-free counterparts with O(logn) slowdown and preprocessing done in time O(log2 n). The fault model is as follows. The faults are deterministic (worst-case distribution) and static (do not change in the course of a computation). If a processor attempts to communicate with some other processor (in the case of an OCPC) or read a memory word (in the case of a PRAM) then it is immediately notified whether the operation was successful (fault-free addressee) or failed (faulty addressee). This is for the first time that a general fast deterministic simulation technique is designed for the EREW PRAM with the worst-case fault distribution. The simulation is designed in such a way that it relies only on a fraction of all the operational processors. During preprocessing, the active processors retrieve the original input provided to all processors before the simulation started. This is accomplished by adapting the information-dispersal method.

UR - http://www.scopus.com/inward/record.url?scp=84947710857&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84947710857&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84947710857

SN - 3540603131

SN - 9783540603139

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 89

EP - 101

BT - Algorithms - ESA 1995 - 3rd Annual European Symposium, Proceedings

A2 - Spirakis, Paul

PB - Springer Verlag

ER -