Message-passing concurrent computers such as the Caltech Cosmic Cube(1) and its commercial descendents consist of many computing nodes that interact with each other by sending and receiving messages over communication channels between the nodes. For finer grain message-passing concurrent machines, such as the Caltech Mosaic(2), it becomes progressively more difficult to achieve the performance required of the communication network. The bisection capacity of physically realizable communication networks grow more slowly than the expected message traffic, which is the least linear in the number of nodes.