In: Janicki, R.; et al.: Computing and Information. Proceedings of the International Conference ICCI'89, 1989, Toronto, Ont., Canada, pages 169-176. Amsterdam, The Netherlands: North-Holland, 1989.
Abstract: The problem of modeling and evaluation of performance and reliability of a hypercube architecture is addressed. A detected and isolated hardware failure causing a decrease in the number of processors and/or the number of communication paths available in the hypercube may lead to an increase in the computation time component and/or the communication time component of the parallel tasks in an application program. A generalized stochastic Petri Net is used to model the component failure and repair events in the hypercube. This model generalizes a previously published Markov chain model for the reliability of the hypercube.
Keywords: performability (of the) hypercube; generalized stochastic net; repair event; reliability.