ACM Transactions on Programming Languages and Systems (TOPLAS)
Universal schemes for parallel communication
STOC '81 Proceedings of the thirteenth annual ACM symposium on Theory of computing
A logarithmic time sort for linear size networks
Journal of the ACM (JACM)
Randomized parallel communications on an extension of the omega network
Journal of the ACM (JACM)
Fast algorithms for bit-serial routing on a hypercube
SPAA '90 Proceedings of the second annual ACM symposium on Parallel algorithms and architectures
Fast-fault-tolerant parallel communication and on-line maintenance using information dispersal
SPAA '90 Proceedings of the second annual ACM symposium on Parallel algorithms and architectures
Methods for message routing in parallel machines
STOC '92 Proceedings of the twenty-fourth annual ACM symposium on Theory of computing
PODC '92 Proceedings of the eleventh annual ACM symposium on Principles of distributed computing
An Efficient General-Purpose Parallel Computer
Journal of the ACM (JACM)
N-processors graphs distributively achieve perfect matchings in O(log2N) beats
PODC '82 Proceedings of the first ACM SIGACT-SIGOPS symposium on Principles of distributed computing
A logarithmic time sort for linear size networks
STOC '83 Proceedings of the fifteenth annual ACM symposium on Theory of computing
Optimality of a Two-Phase Strategy for Routing in Interconnection Networks
IEEE Transactions on Computers
SFCS '87 Proceedings of the 28th Annual Symposium on Foundations of Computer Science
Hi-index | 0.00 |
A fundamental problem in the theory of parallel computation is to find an efficient interconnection pattern between N processors that minimizes the number of lines entering or leaving each processor while enabling fast communication between the processors. A family of Balanced communication schemes for connecting N processors with only a constant number of lines entering or leaving each processor is defined. It is proved that this network topology enables a fully distributed algorithm to route in parallel N packets each located in distinct processors to their distinct destinations in O(log2N) steps. Thus we give an optimal solution to the above problem.