Nonblocking conditions for multiring commutators and generalized hypercubes for arbitrary commutations. II. Generalized hypercubes. Intranode commutation

摘要

Necessary and sufficient conditions for implementing arbitrary commutations without blocking in generalized hypercubes are formulated. These conditions are shown to ensure fault tolerance to individual faults of nodes. Maximal speed and minimal complexity of generalized hypercubes are studied. Methods for resolving intranode conflicts and a functional scheme of nodes for a multiring commutator and a generalized hypercube are investigated.