| Hi-index | 0.00 |
Parallel programming models based on a mixture of task and data parallelismhave shown to be successful in addressing the increasing communicationoverhead of distributed memory platforms with a large number of processors.In these models, an application is decomposed into a set of parallel tasksthat can run on an arbitrary number of processors. The communicationbetween different tasks is allowed only at the start and the end of atask, thus limiting the possible communication patterns and the potentialgranularity of the tasks.In this paper, we consider an extended parallel programming model thatadditionally supports communication between running parallel tasks.We describe a specification language for applications in the newprogramming model and propose a transformation framework for a step-wisederivation of an executable message passing program from the specificationlanguage. The advantages of the approach are demonstrated forsolution methods for ordinary differential equations.