A. De Luca¹ & A. Jiménez^2,3

¹ CINVESTAV-IPN. Electric Engineering. Computation Section. e-mail: dlap@delta.cs.cinvestav.mx

² Universidad Autónoma Metropolitana. Electronic Department. e-mail: ajf@correo.azc.uam.mx

³ ESIME-IPN. Electric Engineering Department.

Received: January 13^h, 2003. ]]> Accepted: April 2^th, 2003.

Abstract

The present article contains a high performance buffer design, useful for message flow control in parallel systems. The appropriate handling of the buffers is an important activity for the flow control function. The proposed buffer, which we have denominated as Self-Compacting Buffer (SCB), reduces the communication latency through a highly efficient management of their space, and a hardware implementation. The SCB was designed with a parallel-distributed control unit, using a control cell for each storage locality. Their capacity is expandable: it can grow width and length, preserving the complexity of its control cells. The SCB allows simultaneous writing and reading operations in a single clock cycle.

Keywords: Routers, Flow Control, Parallel systems, Self-compacting buffer, VHDL.

Resumen

El presente artículo contiene el diseño de un buffer de alto desempeño, útil para el control de flujo de mensajes en sistemas paralelos. El manejo adecuado de buffers es una actividad importante de la función de control de flujo. El buffer propuesto, el cual hemos denominado como SCB (Buffer AutoCompactante), permite reducir la latencia de comunicación mediante una administración altamente eficiente de su espacio y una implantación de hardware. El SCB fue diseñado con una unidad de control distribuida paralela, utilizando una celda de control para cada localidad de almacenamiento. Su capacidad es expandible: puede crecer en ancho y largo, manteniendo constante la complejidad de sus celdas de control. El SCB permite la realización simultánea de operaciones de escritura y lectura en un solo ciclo de reloj.

DESCARGAR ARTÍCULO EN FORMATO PDF

References

[1] Hwang K. Advanced Computer Architecture: Parallelism, Scalability, Programmability. McGraw-Hill, 1993.         [ Links ]

[2] Thin-fries J. G., Vassiliadis S., Pechanek G. G., Johnson H. D. & Green D. M. To Processing Unit Flexible for Multiprocessor Machine Organizations. Instrumentation and Development, vol. 3, not. 4, pages 20-32, 1994.         [ Links ]

[3] Jiménez-Flores A. & De Luca P. A., Design of a Flexible and Adaptive Router for Parallel Systems. Computation International Congress CIC'99. IPN, México.         [ Links ]

[4] Delgado-Frias J., Vassiliadis S, Johnson H. D., Summerville D. & De Luca A. A Processing Unit for Multiprocesor Organizations. Department of Electrical Engineering, State University of New York at Binghamton. 1996.         [ Links ]

[5] Duato, J. & Sudhakar Y. Interconnection Networks: an engineering approach. IEEE Computer Society Press, 1997.         [ Links ]

[6] McHugh J. A. Algorithmic Graph Theory. Prentice-Hall, 1990.         [ Links ]

[7] Park J., Vassiliadis S. & Cold Thin J. G. Router Architecture for Oblivious Routing Algorithms. Parallel Computing Technologies (PaCT-93), Obninsk, Russia, 1993.         [ Links ]

[8] Culler D. E., Singh J. P., and Gupta A. Parallel Computer Architecture: A hardware/Software Approach. Morgan Kaufmann Publishers Inc., 1999, pp. 1025.         [ Links ]

[9] Tanenbaum A. S., Structured Computer Organization. Prentice Hall Inc., 1999.         [ Links ]

[10] Jay C. VHDL and Synthesis Tools Provide to Generic Design Entry Platform Into FPGAs, PLDs and ASICs. Microprocessors and Microsystems, Volume 17, Issue 7, September 1993, 391-398.         [ Links ]

[11] Dally W. J., Chao L., Chien A. A., Hassoun S., Horwat W., Kaplan J., Song P., Scott B. T. Architecture of to Message-Driven Processor. 25 Years ISCA: Retrospectives and Reprints, pp 337-344, 1998.         [ Links ]

[12] Tamir Y. and Frazier G.L. Dynamically Allocated Multi-queue Buffers for VLSI Communication Switches. IEEE Trans. On Computers, volume 41, Not. 6 pp. 725-737, June, 1992.         [ Links ]

[13] VHDL Reference Guide. Xilinx Inc., 1999.         [ Links ]

[14] Pellerin D., Taylor D. VHDL. Made Easy. Prentice Hall PTR., 1997.         [ Links ]