High Level Support for Distributed High Performance Computing

Recent trends in hardware, in particular in interconnection technologies, have paved the way to the exploitation of heterogeneous, distributed computing platforms for advanced scientific applications. This infrastructure enables the building of meta-applications that are composed of several modules which may be implemented in different languages, exploit heterogeneous platforms, and employ several forms of parallelism. However, software development has not kept pace with the advances in hardware. Despite the existence of high level parallel languages and meta-computing frameworks there is, for the time being, a lack of a unified framework that efficiently deals with heterogeneity, interoperability, and hybrid parallelism at a high level