High Performance Computation Based on Semantic P2P Network

High performance computations are widely used in scientific research and industries. As supercomputers are expensive, how to use cheap computers such as clusters and personal computers is important issue. In this paper we describe a novel distributed computation based on semantic P2P network, in which the peers can be grouped virtually into hierarchical classified domains and the problems are partitioned into subproblems and scheduled to those peers. This strategy is scalable to millions of computers effectively in theory. We have implemented α version distributed knapsack problem solution in our semantic P2P network platform.