The continuous development of hardware makes it possible to significantly increase the software complexity, reaching a high performance scale. In addition to development of a standard computer platform (CPU, memory), different hardware technologies dedicated to a particular type of computation with a dramatic increase of performance capacity and decreased production costs have been developed. These dedicated computational resources are starting to be used in general-purpose applications. The next generation of embedded systems will be built on heterogeneous platforms, i.e. platforms consisting of different types of computational units: Field-Programmable Gate Arrays (FPGAs), Graphics Processing Units (GPUs) and MultiCore Central Processing Units (CPUs). The problem that appears in using these computational units today is an inefficient and ineffective development of such systems. A question that appears is: which system configuration is the best in respect to the performance, energy consumption, or similar requirements. HELPING project will focus on building formalisms that enable specifications of such systems in an early development phase and on reasoning about their optimal deployment. In addition, the project will develop connectors that will be automatically built in code and will enable a standardized communication between software components allocated on different computational units. HELPING project will focus on embedded systems with their specific requirements.
Professor vid Chalmers, Data- och informationsteknik, Software Engineering
Finansierar Chalmers deltagande under 2014–2016