Chyong K. Topologies of wide-scale fault-tolerant parallel computing systems

This thesis is dedicated to a problems of research the ways for increase of efficiency and fault tolerance of scalable computing systems oriented on mass multi sequencing. The new method for synthesizing of topologies is proposed and based on the mirror-image presentations and operations of inverting of code notation. Two classes of topologies are designed. The capability of the synthesised topology, such like degree, diameters etc, can be simple changed by variation on the system radix of base clusters. The proposed topologles is compared to the well-know topologies and pretend to be very good solutions for wide-scale fault-tolerant parallel computing systems with thousands and more processor elements.