The IST research project ADVANCE (Asynchronous and Dynamic Virtualisation through performance ANalysis to support Concurrency Engineering) addresses the dynamic adaptation of concurrent applications based on statistical performance feedback.
Multi-core/many-core systems offer potential, both for cheap, scalable high-performance computing and also for significant reductions in power consumption compared with conventional processor technologies. At the same time, cores are becoming increasingly complex and heterogeneous. While current programming technologies are (just) coping with the small-scale homogeneous dual-core and quad-core processors available today, new programming paradigms are needed to deal with the massive numbers of heterogeneous cores that will become available in future. The Advance project will tackle this important problem by developing a new and advanced cost-directed hardware virtualisation technology to map programs onto emerging hardware architectures in a way that is both flexible and transparent to the programmer, but which, nevertheless, respects the programmers’ expectations and requirements on extra-functional properties, such as resource utilisation or power consumption. This interdisciplinary project will exploit leading work on computer architecture, probabilistic resource usage analysis, heuristic placement and mapping, programming languages and compilation methods to develop new cost-directed stream-processing models for parallel execution, and to apply these to commercial problems taken from a range of sectors: enterprise business software, image/voice/video processing, computational healthcare, etc.