The proposed visitors are leading researchers in the field of component based software engineering. The KeLP adn P3L work naturally complements the work at imperial on skeletons and run time optimisation. The proposed visits will help combine these activities adn take them forward to address new challenges relevant to the next generation of high performance computer systems. The visitors will bring with them key expertise in the KeLP P3L software.
We will build on KeLPs data structured irregular gridgs. The idea is to arrange for components to build a task graph data structure at run-time analogous to KeLPs motion plan. The task graph will represent a large-grain data-flow graph. We will use this to increase scope for adaptive run-time scheduling, and also investigate off-line optimisation. and automatic selection selection of placement for intermediate regions. We will develop imperials run-time optimisation work. By extending it from sequential to parallel composition we will integrate task adn data parallelism in a common optimisation framework which will naturally handle adaptive block-irregular applications. We will also extend the role of performance models in cross-component resource management, as developed in the P3L compiler at Pisa to deal not only with task graphs but also with heterogenous clusters fo processors and irregular data.