Search

Parallel programming has proved to be an effective technique to increase the performance of computationally intensive applications. However, writing parallel programs is not easy, and activities such as debugging are usually hard and time consuming. To cope with these difficulties, skeletal parallel programming has been widely explored in recent years with very promising results. However, prototypal skeletal systems developed so far tend to be rather inflexible and difficult to adapt to many practical parallelization scenarios. For instance, many systems restrict all the substructures of an application being encapsulated together in term of possibly nested skeletons, which may be cumbersome when parallelizing some large and complex applications. Moreover, it is usually difficult to share resources among different skeleton instances and to reuse the same instance of a skeleton in different parts of the code. This paper reports on the current status of the OcamIP3I (2.0) system, which sensibly changes the skeletal model of the previous versions, making it more usable and flexible. In particular, we describe the new skeletons, the new skeletal execution model as well as related issues on design and implementation. [ABSTRACT FROM AUTHOR]

We introduce a very simple, yet powerful, distribution calculus aimed at describing different strategies that can be used to distribute (ie, partition and replicate) multidimensional dense arrays over a set of processors, like what is done in the implementation of the P3L Map skeleton. We give a formal semantics, that allows to prove equations between distributions, and we show how to associate a cost to such distributions, allowing to choose between semantically equivalent distribution strategies. [ABSTRACT FROM AUTHOR]