Development of efficient numerical programs for large distributed parallel computers is a challenging problem. Many programming languages, systems and libraries exist and evolve to help with it, yet the problem is far...
详细信息
Development of efficient numerical programs for large distributed parallel computers is a challenging problem. Many programming languages, systems and libraries exist and evolve to help with it, yet the problem is far from being solved. Of interest are particular application implementations' studies, which reveal actual capabilities of a system in the real computation. In this paper, the implementation of an indicative 3D model heat equation parallel solver using fragmentedprogramming technology and lunasystem is investigated. A comparative testingwith conventional MPI implementation is presented. The pros and cons of the approach are analyzed for corresponding applications class.
Dataflow-based systems of parallel programming, such as luna fragmented programming system, often lack efficiency in high performance computations due to a high degree of non-determinism of a parallel program executio...
详细信息
ISBN:
(纸本)9783319219097;9783319219080
Dataflow-based systems of parallel programming, such as luna fragmented programming system, often lack efficiency in high performance computations due to a high degree of non-determinism of a parallel program execution and execution overhead it causes. The authors concern defining control flow in luna programs in order to optimize their execution performance. The basic idea is to aggregate several fragments of a program and to execute them under control flow, thus reducing both surplus parallelism and system overhead. Tests presented show effectiveness of the proposed approach.
暂无评论