Abstract
This paper presents the network interface for the Data Driven Network Of Workstations (D2NOW), a multithreaded architecture that uses the decoupled data driven model of execution. D2NOW is built using commodity workstations. The support for the data driven synchronization of threads, is provided by the Thread Synchronization Unit (TSU). The TSU is attached in the COAST (Cache On A STick) L2 Cache slot of Pentium workstations and thus it has an implicit interface, using snooping, to the Pentium microprocessor. Workstations are connected via a Telegraphos interconnection network. The D2NOW supports fine grain threads and thus is coupled to a fine grain network, the Telegraphos, which is a high throughput ATM-like network. Telegraphos uses short packets and guarantees no packet-drop, which is a must for fine grain data-driven computation. The Network Interface Unit (NIU) is incorporated on the TSU board. The NIU is implemented using two FPGAs and it has very low hardware overhead. Remote writes are carried out in less than a microsecond.
Preview
Unable to display preview. Download preview PDF.
References
Robert A. Iannucci et al. Multithreaded Computer Architecture a Summary of the State of the Art. Kluwer Academic Publishers, 1994.
Panel Discussion. Architectural implementation issues for multithreadeding. In R. Iannucci et al., editor, Mutlithreaded Computer Architecture a Summary of the State of the Art. Kluwer Academic Publishers, 1994.
Jack Dennis and Guang Gao. Multithreaded Architectures: Principles, Projects and Issues. In R. Iannucci et al., editor, Mutlithreaded Computer Architecture a Summary of the State of the Art. Kluwer Academic Publishers, 1994.
P. Evripidou. Thread Synchronization Unit (TSU): A building block for High Performance Computers. In Proceedings of the International Symposium on High Performance Computing, Fukuoka, Japan, Nov. 1997.
E. Markatos and M. Katevenis. Telegraphos: High-performance networking for parallel processing on workstation clusters. In Symposium on High Performance Computer Architectures, 1996.
J. Heinlein, K. Gharachorloo, S. Dresser, and A. Gupta. Integration of message passing and shared memory in the stanford flash multiprocessor. In 6th International Conference on Architectural Support for Programming Languages, and Operating Systems, 1994.
J. Kuskin et. al. The stanford flash multiprocessor. In 21st International Symposium on Computer Architecture, 1995.
Anant Agarwal et. al. The mit alewife machine: Architecture and performance. In 21st International Symposium on Computer Architecture, 1995.
David Kranz et. al. Integrating message passing and shared memory. In Fourth Symposium on Principles and Practices of Parallel Programming, May 1993.
T. Anderson, D. Culler, and D. Patterson. A case for now (networks of workstations). IEEE Micro, 1995.
Arvind and K.P. Gostelow. The U-Interpreter. IEEE Computer, pages 42–49, February 1982.
Intel Architecture Developer's Manual. Intel, 1997.
P. Evripidou and J-L. Gaudiot. A Decoupled Graph/Computation Data-Driven Architecture with Variable Resolution Actors. In Proceedings of the 1990 International Conference on Parallel Processing, August 1990.
M. Katevenis, P. Votsalaki, A. Efthymiou, and M. Stratakis. Vc-level flow control and shared buffering in the telegraphos switch. In IEEE Hot Interconnects III Symposium, 1995.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Kyriacou, C., Evripidou, P. (1999). Network interface for a data driven network of workstations (D2NOW). In: Polychronopoulos, C., Fukuda, K.J.A., Tomita, S. (eds) High Performance Computing. ISHPC 1999. Lecture Notes in Computer Science, vol 1615. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0094927
Download citation
DOI: https://doi.org/10.1007/BFb0094927
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-65969-3
Online ISBN: 978-3-540-48821-7
eBook Packages: Springer Book Archive