To execute a parallel program on a multicomputer system, the tasks of the program have to be mapped to the particular processors of the parallel machine, The aim of the mapping is twofold: (i) to achieve a balanced lo...
详细信息
To execute a parallel program on a multicomputer system, the tasks of the program have to be mapped to the particular processors of the parallel machine, The aim of the mapping is twofold: (i) to achieve a balanced load on the processors (partitioning problem) and (ii) to keep communication delays low by placing communicating tasks closely together (mapping). Since both the communication structure of the program and the interconnection structure of the parallel machine can be represented as graphs, the mapping problem can be regarded as a graph embedding problem to minimize communication costs, As a new heuristic approach to this NP-hard problem we apply Kohonen's self-organizing maps to establish a topology-preserving embedding. Experimental results are presented and compared to other approaches to this problem, The most attractive feature of our new method is that it can be extremely well parallelized.
This work deals with the investigation of the performance of parallel iterative algorithms, used for the solution of linear systems obtained from the discretization of Elliptic PDEs using the Finite Element Collocatio...
详细信息
This work deals with the investigation of the performance of parallel iterative algorithms, used for the solution of linear systems obtained from the discretization of Elliptic PDEs using the Finite Element Collocation method. To increase parallelicity the initial matrix is reordered and then the algorithm is mapped on a distributedmemoryparallel computer. The case of the Star architecture with master-slave communication of the processors is studied. Earlier work of the authors is improved here and at the same time a new solution approach is developed so that the application of these iterative algorithms on more general problems is feasible. Speedup, processor utilization and efficiency measures are presented. The theoretical optimum is almost reached even though fixed number of processors was available.
To execute a parallel program on a multicomputer system, the tasks of the program have to be mapped to the particular processors of the parallel machine. The aim of the mapping is twofold: (i) to achieve a balanced lo...
详细信息
To execute a parallel program on a multicomputer system, the tasks of the program have to be mapped to the particular processors of the parallel machine. The aim of the mapping is twofold: (i) to achieve a balanced load on the processors (partitioning problem) and (ii) to keep communication delays low by placing communicating tasks closely together (mapping). Since both the communication structure of the program and the interconnection structure of the parallel machine can be represented as graphs, the mapping problem can be regarded as a graph embedding problem to minimize communication costs. As a new heuristic approach to this NP-hard problem we apply Kohonen's self-organizing maps to establish a topology-preserving embedding. Experimental results are presented and compared to other approaches to this problem. The most attractive feature of our new method is that it can be extremely well parallelized.
暂无评论