Redução de Bits de Emparelhamento de Máquina de Fluxo de Dados de Manchester
Abstract
The dataflow model ia specially relevant to research in high-performance architectures. In this model, the execution control is done by taking into account only the data availability, thus allowing maximum exploitation of the parallelism implicit in programa. The present work is based on the Manchester dataflow machine, which, in order to handle the reentrant code, imposes the data token to have, in addition to the destination instruction field, a label. This additional information, which corresponds to 70% of the data token, compounds the machine implementation as it substantially bounds the execution speed and prevents the full model utilization. This work presents approaches for reducing the amount of information needed for proper machine operation in order to achieve simpler and more effective implementation.
References
Arvind & Gostelow,K.P. The U-interpreter IEEE Computer, 15(2): 42-49, fev. 1982.
Arvind & Nikhil,R.S. Executing a Program on the MIT Tagged-Token Dataflow Architecture IEEE Transactions on Computers, 39(3): 300-318, mar. 1990.
Arvind, K.P. & Thomas, R.E. I-Structures: An Efficient Data Structure for Funcional Languages Laboratory of Computer Science, MIT, out. 1981.
Barahona,P.M.C.C. & Gurd,J.R. Processor Allocation in a Multi-Ring Dataflow Machine Department of Computer Science, University of Manchester, 1987.
Gurd,J.R.; Kirkham,C.C. & Watson,I. The Manchester Prototype Dataflow Computer Communication of ACM, 28(1): 34-52, jan. 1985.
Gurd,J.R. & Watson,I. Data Driven System for High Speed Parallel Computing (Part 1) Computer Design, 19(6): 91-100,jun. 1980
Gurd,J.R. & Watson,I. Data Driven System for High Speed Parallel Computing (Part 2) Computer Design, 19(7): 97-106,jul. 1980
Kawakami,K. & Gurd,J.R. A Scalable Dataflow Structure Store Proceedings,13th Annual International Symposium on Computer Architecture,14(2):243-250, jun. 1986
McGraw,J. et. al. SISAL - Streams and Iteration in a Single Assigment Language Language Reference Manual, ver. 1.2, M-146, Lawrence Livermore National Laboratory, ago. 1984.
Padua,D.A. & Wolfe,M.J. Advanced Compiler Optimizations for Supercomputers Communications of ACM, 29(12): 1184-1201, dez. 1986.
Ruggiero,C.A. Throttle Mechanisms for Manchester Dataflow Machine Ph.D. Thesis, Department of Computer Science, University of Manchester, jul. 1987.
Sargeant,J. & Kirkham,C.C. Stored Data Structures on the Manchester Dataflow Machine Proceedings, 13th Annual International Symposium on Computer Architecture, 14(2): 235-242, jun. 1986.
Sargeant,J. Efficient Stored Data Structures for Dataflow Computing Ph.D. Thesis, Department of Computer Science, University of Manchester, ago. 1985.
Schönauer,W. Scientific Computing on Vector Computers Special Topics in Supercomputing, vol. 2, North-Holland, 1987.
Shimada,T.; Hiraki,K.; Nishida,K. & Sekigushi,S. Evaluation of a Prototype Dataflow Processor of the Sigma-1 for Scientific Computations Proceedings of the 13th International Symposium on Computer Architecture, 14(2): 226-234, jun. 1986.
da Silva,J.G.D. & Watson I. Pseudo-Associative Store with Hardware Hashing IEEE Proc., 130(1): 19-24, jan. 1983
Watson,I. & Gurd,J. A Practical Dataflow Computer IEEE Computer, 15(2): 51-57, fev. 1982.
