Avaliação de Desempenho de um Computador a Fluxo de Dados Massivamente Paralelo
Resumo
Este artigo apresenta a avaliação de desempenho de um Computador a Fluxo de Dados Massivamente Paralelo baseado em barramentos paralelos hierárquicos. Neste Computador, um programa a fluxo de dados é executado através da alocação das operações a fluxo de dados em vários elementos de processamento do computador. O computador é composto por um conjunto de barramentos paralelos dispostos de maneira hierárquica, que permite a execução de programas a fluxo de dados, é de fácil expansão e flexibilidade. O computador tem um mecanismo de tolerancia a falhas por software e hardware, tal que ele pode trabalhar como uma máquina confessável. Vários aspectos do computador tais como: arquitetura, estrutura do software, estratégias de tolerancia a falhas e avaliação da performance são discutidos neste artigo.
Palavras-chave:
Paralelismo, Arquitetura, Simulação
Referências
A. H. Veen, "Dataflow Machine ARchitecture, ACM Computing Surveys, Vol. 18, No. 4, Dec. 1986, pp. 365-396.
Arvind and R.S. Kikhil, "Executing a Program on the MIT Tagged-Token Dataflow Architecture", IEEE Transaction on Computer, Vol. 39, No. 3, 1990, pp. 300-318.
C. Kirner, "Design of a Recursively Structured Parallel Computer", Prod. of the 17th Annual Computer Science Conference. ACM, Louisville, USA, Feb. 1989, 1 pp.
C. Kirner. J.L. Silva, and S. Motoyama, "A Fault-Tolerant Massively Parallel Dataflow Computer", VI IASTED Jntemational Conference on Parallel and Distributed Computign Systems, Washington DC, USA, Oct. 3-5, 1994, pp.323-326
C. Kirner, E. Marques, "Design of Distributed System Suport Based in a Centralized Parallel Bus". Computer Architecture News. V. 14, No.4, Sep. 1986, pp. 1-26.
E. Marques and C. Kirner, "Design of the Matching Unit of a Massively Parallel Dataflow Computing System", Conference on Massively Parallel ompuling System, IEEE/EUROMICRO, lschia, ltaly, May 2-6, 1994, 19pp.
J.B. Dennis, "First Version of a Data Flow Produce Language". Lecture Notes in Computer Science. Vol. 19, 1974, pp. 362-372.
J.B. Dennis. & D.P. Misumas, " A Preliminary Architecture for a Basic Flow Processor". In: Annual Symp. Computer Architecture. 2. Proceedings, 915, pp, 126-132.
J.B. Dennis, "Data Flow Supercomputer". IEEE Compute Vol. 13, No. 11, Nov. 1980, pp. 48-56.
J. Gurd. & I. Watson, "Structuring software for Parallel Execution I. In: Data-Driven Systems for High Speed Parallel Computing". Computer Design, Vol. 19, No. 6, Jun. 1980, pp. 91-100.
J. Gurd, & I Watson, "Hardware Design. II. In: Data-Driven System for High Parallel Computing". Computer Design, Vol. 7, Jul. 1980, pp. 97-106.
M.G. Sami and N. Scarabotolo. "Fault Tolerance in Parallel Architecture", Lecture Notes in Computer Science. No. 272, Future Parallel Computer, pringer-Verlag, 1987, pp.82-152.
N. lto. et. al. "The Architecture and Preliminary Evaluation Results of the Experimental Parallel lnference Machine PIM-D". In Proc. of the 13th Anual International Symposium on Computer Architecture, Tokyo, Japan, Jul. 1986, pp. 149-156.
P.C. Trelevean, D.R. Brownbridge and R.P. Hopkins,"Data-Driven and Demand-Driven Computer Architecture", ACM Computing surveys, Vol. 14, No. 1 Marc. 1982, pp. 93-143.
S. Sakai, Y. Kodama, and Y. Yarnaguchi. "Prototype Implementation of a Highly Parallel Dataflow Machine EM-4". Proc. lnt. Parallel Processing Symposium. 1991.
H. Takagi, "Analysis of Polling Systems". London: MIT Press, 1985, 197p.
T. Shimada, K. Hiraki, K. Nishida and S. Sekigushi, Evaluation of a Prototype Dataflow Procesor of the Sigma-1 for Scientific Computations". Computer Architecture News, Vol. 15, No. 2, 1986, pp.226-234.
V.P. Srini. "A Fault-Tolerant Dataflow System". IEEE Computer, Vol. 18, No. 11, Mar. 1985, pp.54-68.
V.P. Srini, Än Architectureal Comparation Of Dataflow Systems". Computer, Vol. 19, No. 3, 1986, pp. 68-88.
Arvind and R.S. Kikhil, "Executing a Program on the MIT Tagged-Token Dataflow Architecture", IEEE Transaction on Computer, Vol. 39, No. 3, 1990, pp. 300-318.
C. Kirner, "Design of a Recursively Structured Parallel Computer", Prod. of the 17th Annual Computer Science Conference. ACM, Louisville, USA, Feb. 1989, 1 pp.
C. Kirner. J.L. Silva, and S. Motoyama, "A Fault-Tolerant Massively Parallel Dataflow Computer", VI IASTED Jntemational Conference on Parallel and Distributed Computign Systems, Washington DC, USA, Oct. 3-5, 1994, pp.323-326
C. Kirner, E. Marques, "Design of Distributed System Suport Based in a Centralized Parallel Bus". Computer Architecture News. V. 14, No.4, Sep. 1986, pp. 1-26.
E. Marques and C. Kirner, "Design of the Matching Unit of a Massively Parallel Dataflow Computing System", Conference on Massively Parallel ompuling System, IEEE/EUROMICRO, lschia, ltaly, May 2-6, 1994, 19pp.
J.B. Dennis, "First Version of a Data Flow Produce Language". Lecture Notes in Computer Science. Vol. 19, 1974, pp. 362-372.
J.B. Dennis. & D.P. Misumas, " A Preliminary Architecture for a Basic Flow Processor". In: Annual Symp. Computer Architecture. 2. Proceedings, 915, pp, 126-132.
J.B. Dennis, "Data Flow Supercomputer". IEEE Compute Vol. 13, No. 11, Nov. 1980, pp. 48-56.
J. Gurd. & I. Watson, "Structuring software for Parallel Execution I. In: Data-Driven Systems for High Speed Parallel Computing". Computer Design, Vol. 19, No. 6, Jun. 1980, pp. 91-100.
J. Gurd, & I Watson, "Hardware Design. II. In: Data-Driven System for High Parallel Computing". Computer Design, Vol. 7, Jul. 1980, pp. 97-106.
M.G. Sami and N. Scarabotolo. "Fault Tolerance in Parallel Architecture", Lecture Notes in Computer Science. No. 272, Future Parallel Computer, pringer-Verlag, 1987, pp.82-152.
N. lto. et. al. "The Architecture and Preliminary Evaluation Results of the Experimental Parallel lnference Machine PIM-D". In Proc. of the 13th Anual International Symposium on Computer Architecture, Tokyo, Japan, Jul. 1986, pp. 149-156.
P.C. Trelevean, D.R. Brownbridge and R.P. Hopkins,"Data-Driven and Demand-Driven Computer Architecture", ACM Computing surveys, Vol. 14, No. 1 Marc. 1982, pp. 93-143.
S. Sakai, Y. Kodama, and Y. Yarnaguchi. "Prototype Implementation of a Highly Parallel Dataflow Machine EM-4". Proc. lnt. Parallel Processing Symposium. 1991.
H. Takagi, "Analysis of Polling Systems". London: MIT Press, 1985, 197p.
T. Shimada, K. Hiraki, K. Nishida and S. Sekigushi, Evaluation of a Prototype Dataflow Procesor of the Sigma-1 for Scientific Computations". Computer Architecture News, Vol. 15, No. 2, 1986, pp.226-234.
V.P. Srini. "A Fault-Tolerant Dataflow System". IEEE Computer, Vol. 18, No. 11, Mar. 1985, pp.54-68.
V.P. Srini, Än Architectureal Comparation Of Dataflow Systems". Computer, Vol. 19, No. 3, 1986, pp. 68-88.
Publicado
29/07/1995
Como Citar
LUIZ E SILVA, Jorge; MOTOYAMA, Shusaburo; KIRNER, Cláudio.
Avaliação de Desempenho de um Computador a Fluxo de Dados Massivamente Paralelo. In: INTERNATIONAL SYMPOSIUM ON COMPUTER ARCHITECTURE AND HIGH PERFORMANCE COMPUTING (SBAC-PAD), 7. , 1995, Canela.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
1995
.
p. 439-448.
DOI: https://doi.org/10.5753/sbac-pad.1995.19879.
