Multigrid Algorithms on the Hypercube Multiprocessor
IEEE Transactions on Computers
A parallel architecture comes of age at last
IEEE Spectrum
Hypernet: A communication-efficient architecture for constructing massively parallel computers
IEEE Transactions on Computers
IEEE Transactions on Computers
The connection machine
The cube-connected cycles: a versatile network for parallel computation
Communications of the ACM
Design, Analysis, and Simulation of I/O Architectures for Hypercube Multiprocessors
IEEE Transactions on Parallel and Distributed Systems
Extended Hypercube: A Hierarchical Interconnection Network of Hypercubes
IEEE Transactions on Parallel and Distributed Systems
Hi-index | 0.00 |
Hypercube has been the most popular topology for developing multiprocessor supercomputers because of its connectivity, regularity, symmetry and algorithmic mapping properties. However, if a hypercube needs to be expanded at some future time, both hardware configuration and communication software of each node has to be altered because its node degree is not constant. Hwang and Ghosh proposed a new interconnection network called hypernet which has a constant node degree and is easily expandable. However, its components count increases unsymmetrically with the increase in size. We propose a new topology called symmetric hypernet which grows symmetrically with the increase in dimension and hierachical level. The nodes in the symmetric net are functionally and physically separated as IO nodes and computation nodes to avoid traffic congestion and message delays. The IO nodes are placed uniformly among the processing nodes. The system can be dynamically partitioned into subsystems, each of which may be dedicated to serve a user's request(s). The system analysis can be done with the help of analytical models to serve a user's request(s). The system analysis can be done with the help of analytical models to determine how different structures would affect system performance.