Abstract
The mesh is one of the most popular interconnection networks for single-instruction multiple-data (SIMD) [1] array processors. For mesh-connected array processors, the communication latency is proportional to the diameter (about two times the linear size) of the array, which is the maximum number of nodes a message has to travel to reach its final destination. To decrease the diameter of the mesh, we have developed an optoelectronic reduced cellular hypercube (RCH) interconnection [2] which is combined with the mesh to form a combined network called M-RCH. In this paper, we present a time multiplexing scheme for the optoelectronic RCH and discuss the communication speedup of the M-RCH over the mesh for some common operations and applications.
© 1995 Optical Society of America
PDF ArticleMore Like This
Bogdan Hoanca and A. A. Sawchuk
OWC.4 Optics in Computing (IP) 1997
C. B. Kuznia and A. A. Sawchuk
MC2 Optical Computing (IP) 1991
Kuang-Yu J. Li and B. Keith Jenkins
OThA3 Optical Computing (IP) 1995