Abstract
The new generation of transputer components provides advanced communication functionality and performance supporting large systems of parallel processors and virtually full connectivity at high bandwidth and with short delays. We present the results of performance investigations carried out in the context of the ESPRIT PUMA project which predict and compare the communications performance of various network topologies, sizes and routing schemes for non-local communication patterns in order to assess the suitability of transputer networks for general-purpose computing and their scalability of performance with the network size. The results have been largely obtained through simulation, in combination with an extended analytical model for the assessment of very large networks. As a conclusion we found the Clos-type multistage networks together with an adaptive routing scheme to be the most universal solution for scalable high performance under non-local communication load.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
H. Hofestädt, A. Klein, E. Reyzl: Performance Benefits from Locally Adaptive Interval Routing in Dynamically Switched Interconnection Networks; Proc. 2nd Europ. Conf. on Distr. Memory Computing (EDMCC2), Munich, April 1991, pp. 193–202
H. Hofestädt: Preis-/Leistungsvergleich verschiedener skalierbarer Netzwerkklassen; presented at 2. PASA-Workshop “Parallele Systeme und Algorithmen”, Paderborn, Oktober 1991
INMOS Ltd.: The T9000 Transputer Products Overview Manual; Inmos Databook Series, 1991
A. Klein: Interconnection Networks for Universal Message-Passing Systems; Proc. ESPRIT Conference 1991, Brussels, November 1991
J. van Leeuwen, R.B. Tan: Interval Routing; The Computer Journal 30 (4) 1987, pp. 298–307
D. May, P. Thompson: Transputers and Routers: Components for Concurrent Machines; Proc. Japanese Occam User Group, 1990
E. Reyzl, H. Eckardt: Performance Evaluation for High-Performance Interconnection Networks; Proc. 11. ITG/GI-Fachtagung “Architektur und Betrieb von Rechensystemen”, Munich, March 1990, pp. 275–287
L.G. Valiant: A scheme for fast parallel communication; SIAM J. on Computing, 11, 1982, pp. 350–361
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1992 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Klein, A. (1992). Communication Performance of Advanced Transputer Networks for Universal Message-Passing. In: Jammel, A. (eds) Architektur von Rechensystemen. Informatik aktuell. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-77422-5_16
Download citation
DOI: https://doi.org/10.1007/978-3-642-77422-5_16
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-55340-3
Online ISBN: 978-3-642-77422-5
eBook Packages: Springer Book Archive