In recent years, traditional processors have not been able to directly translate chip fabrication technology advances intro performance gains. To keep satisfying the demand for computing power, there is a shift from homogeneous machines to heterogeneous architectures combining different kinds of processors (CPUs, GPUs, DSPs, FPGAs, and other accelerators). While this approach has allowed significant performance and energy efficiency benefits, heterogeneous systems are often highly difficult to program with existing tools. To reduce the cost of system development, reengineering techniques emerge as a solution which may help to balance ease-of-development with better performance, better reliability, and lower maintenance costs.

The RePara2016 workshop is organized in cooperation with the REPARA European Project and it aims to join experts from related disciplines to share recent advances in different areas contributing to better transformation of new and legacy applications to different programming models for diverse computing devices in the context of parallel heterogeneous architectures.