Mapping schemes of the particle-in-cell method implemented on the PAX computer

Hoshino, Tsutomu; Hiromoto, Robert E.; Sekiguchi, Satoshi; Majima, Sumiko

doi:10.1016/0167-8191(88)90018-x

Cited by 15 publications

(9 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the remainder of this Section some approaches that have been taken for parallelizing PIC codes are reviewed. In discussing the decomposition of particles the nomenclature of Hoshino et al [31] will be adopted with a slight modification. A decomposition in which particles are assigned to processors on the basis of their position will be referred to as an Euferian decomposition.…”

Section: Basic Decomposition Strategiesmentioning

confidence: 99%

Characterizing the parallel performance of a large‐scale, particle‐in‐cell plasma simulation code

Walker

1990

Concurrency: Pract. Exper.

View full text Add to dashboard Cite

The efficient implementation of particle‐in‐cell (PIC) plasma simulation codes on distributed memory concurrent computers is made difficult by the conflicting aims of balancing the computational load and minimizing interprocessor communication. This paper surveys previous work on PIC plasma simulation codes on advanced architecture computers, identifies the main issues in parallelizing such codes, and discusses different decomposition schemes. For MIMD concurrent computers the adaptive Eulerian (AE) decomposition scheme is attractive because it seeks to maintain approximate load balance dynamically while avoiding excessive non‐local communication. The load balance and communication characteristics of a large‐scale AE/PIC code on the hypercube are investigated by simulating the behavior of the parallel code on a Cray‐2. The results show that for the three‐dimensional problems studied, in which the particle distribution is highly inhomogeneous, the communication of data between the particle and mesh decompositions dominates the communication overhead. Performing the load balancing may also make a significant contribution to the concurrent overhead if the grain size is sufficiently small. The advantages of the simulation approach in investigating the behavior of concurrent large‐scale applications are discussed, together with portability and software engineering issues.

show abstract

Section: Basic Decomposition Strategiesmentioning

confidence: 99%

Characterizing the parallel performance of a large‐scale, particle‐in‐cell plasma simulation code

Walker

1990

Concurrency: Pract. Exper.

View full text Add to dashboard Cite

show abstract

“…Vorher schon, 1956, hat sich Konrad Zuse bereits das Prinzip eines Feldrechners patentieren lassen [5]. Das herausragende erste wirklich reaüsierte Projekt (Anfang der 70iger Jahre) war der IL-LIA C-IV der University of Illinois mit 64 Prozessoren, die "im Gleichschritt" alle die gleiche Instruktion (aber mit verschiedenen Daten, z.B.…”

Section: Multiprozessoren Und Massivparallele Systemeunclassified

“…Since the first PAX Computer was constructed, the machines have been benchmarked for many typical scientific application models. 5. continuum models with regulär mesh, but with irregulär shapes of domain can be processed in high efficiencies, if they are mapped onto the PU array by the modular mapping.…”

Section: Application Summarymentioning

confidence: 99%

See 1 more Smart Citation

Vektor- und Parallel-Rechner: Vergangenheit, Gegenwart, Zukunft

Hertweck

1989

it - Information Technology

View full text Add to dashboard Cite

Die Fortschritte der Mikroelektronik haben zur Entwicklung einerseits von Supercomputern, andererseits von leistungsfähigen und preiswerten Mikroprozessoren geführt. Seit einigen Jahren versucht man, durch Zusammenschaltung von vielenMikroprozessoren supercomputer-ähnliche Verarbeitungsleistungen zu erzielen. Diese Entwicklung wird hier beschrieben, und es werden verschiedene neuere Rechnerarchitekturen vorgestellt. Besonders interessant sind die Parallel-Rechner mit "message-passing", da bei ihnen der Integrationsaufwand fär größere Systeme relativ klein ist. Rechner dieses Typs sind z.B. SUPRENUM und auf der Basis von Transputern entwickelte Systeme. Die Unterschiede und die Gemeinsamkeiten der Programmierung solcher Systeme werden an Hand der linearen Rekursion -ein Beispiel für einen nicht direkt vektorisierbaren oder parallelisierbaren Algorithmus -erläutert. Das Problem der Nutzung von massiv-parallelen Rechnern scheint vor allem eine Frage der Konstruktion geeigneter Algorithmen zu sein, erst in zweiter Linie eine Frage der Programmierung. Die erfolgreiche Verwendung von massiv-parallelen Rechnern ist eine große Herausforderung für Anwender, Mathematiker (insbesondere Numeriker), Informatiker und Computer-Architekten. Sie vrird alle diese Disziplinen mit Sicherheit im nächsten Jahrzehnt stark beeinflussen.The progress of microelectronics led to the development of both supercomputers and powerfui low cost microprocessors. For several years now the attempt has been made to achieve supercomputer-like Performance by connecting many microprocessors. This development is described and several novel Computer architectures are discussed. Of special interest are parallel Computers with message passing because they require only modest cost when integrating them into larger systems. Computers of this type are for instance SUPRENUM and systems developed on the basis of transputers. The differences and similarities of programming such systems are explained using the linear recursion -an example of a not directly vectorisable or parallelisable algorithm. The problem of utilising massivelyparallel Computers first of all seems to be a question of the design of suitable algorithms, and only then a question of programming. The sucsessful deployment of massively-parallel Computers is a great challenge for users from many areas of science, for mathematicians (especially those in the field of numerics), for Computer scientists, and for Computer architects. These developments will strongly influence these disciplines during the next decade. 70 75 80 85 90 95 Jahr der Erstinstallation Bild 1: Die Entwicklung der Supercomputer (verstanden als jeweils verfügbare leistungsfihigste Rechner). Die schwarzen Balken sind skalare (sequentielle) Rechner, die grauen Balken Vektorrechner. Bei Multiprozessoren ist die maximale Anzahl der Prozessoren angegeben. Die Zunahme der Leistung von Skalarrechnem wird immer langsamer, bei Vektorrechnem ist der Anstieg (noch) etwas steiler, größere Leistungssteigerungen lassen sich nur durch Parallelisie...

show abstract

“…Parallel PIC methods have been previously studied by a n umber of researchers on a wide variety of platforms using di erent parallel programming methodologies 11, 12,13,14,15,16].…”

Section: Introductionmentioning

confidence: 99%

Particle-in-cell simulation codes in High Performance Fortran

Akarsu

Dincer

Haupt

et al. 1996

Proceedings of the 1996 ACM/IEEE Conference on Supercomputing

View full text Add to dashboard Cite

Particle-in-Cell (PIC) plasma sim ulation codes model the interaction of charged particles with surrounding electrostatic and magnetic elds. PIC's computational requiremen ts are classi ed at as one of the grand-challenge problems facing the high-performance comm unity. In this paper we present the implemen tation of 1-D and 2-D electrostatic PIC codes in High Performance Fortran (HPF) on an IBM SP-2. W e used one of the most successful commercial HPF compilers currently available and augmented the compiler's missing HPF functions with extrinsic routines when necessary. W e obtained a near linear speed-up in execution time and a performance comparable to the nativ e message-passing implemen tations on the same platform.

show abstract

Mapping schemes of the particle-in-cell method implemented on the PAX computer

Cited by 15 publications

References 5 publications

Characterizing the parallel performance of a large‐scale, particle‐in‐cell plasma simulation code

Characterizing the parallel performance of a large‐scale, particle‐in‐cell plasma simulation code

Vektor- und Parallel-Rechner: Vergangenheit, Gegenwart, Zukunft

Particle-in-cell simulation codes in High Performance Fortran

Contact Info

Product

Resources

About