Parallel execution of an irregular algorithm depth first search (DFS) on heterogeneous clusters of workstation

This paper presents a status report on our efforts and experiences with heterogeneous clusters of workstation based on dynamic load balancing for parallel tree computation depth-first-search (DFS) project at the High Performance Computing laboratory in Shanghai Jiaotong University. We describe an implementation of one parallel search algorithm DFS, running under the MPI message passing interface and Solaris operating system on heterogeneous workstation clusters. The main goal of this paper is to demonstrate the speed gained by the heterogeneous workstation clusters platform computing to solve large search problem, distribute the tree search space among the processors and show through a parallel simulation application, the maturity of the more recent technologies. We have run the parallelism and distribution of the search space of the DFS algorithm among the processors successfully found that the critical issue in parallel DFS algorithm is the distribution of the search space among the processors. First experimental results of parallel DFS are given for tests that will serve as a starting point for further development of our project. We are presenting our preliminary progress here, and we expect in the near future to demonstrate a real dynamic load balancing for DFS algorithm running on heterogeneous clusters of workstation computing platform resulting in a good load balance among all the processors.