Abstract
A pipelined multiprocessor interconnection method functionally equivalent to a full crossbar, but with a per processor cost proportional to the square of the log of the total number of processors, is presented.

This publication has 1 reference indexed in Scilit: