GPU-based heterogeneous architectures have been given much attention recently. How to get optimal performance out of those architectures with affordable programming effort remains a complex challenge. The PEPPHER framework is one possible solution. Within the PEPPHER framework, the StarPU run-time system is used to decrease such programming efforts, and at the same time to ensure near optimal performance by efficient scheduling over such architectures. However, adapting a normal C/C++ application to the StarPU runtime system requires additional programming effort. This thesis implements and tests a composition tool for automatic adaptation of normal C/C++ applications withPEPPHER components to StarPU. This composition tool requires XML annotation for applications and several trivial changes to applications, which take limited efforts. Our results obtained by three test cases (vector scale, sorting, andmatrix multiplication) show that automatic adaptation works well on different platforms that StarPU supports. It is also shown that besides StarPU’s dynamic composition, this tool facilitates static composition to improve performance portability of normal C/C++ applications.

Software development processes incorporate a variety of different artifacts (e.g., source code, models, and documentation). For multiple reasons the data that is contained in these artifacts does expose some degree of redundancy. Ensuring global consistency across artifacts during all stages in the development of software systems is required, because inconsistent artifacts can yield to failures. Ensuring consistency can be either achieved by reducing the amount of redundancy or by synchronizing the information that is shared across multiple artifacts. The discipline of software engineering that addresses these problems is called Round-Trip Engineering (RTE). In this thesis we present a conceptual framework for the design RTE systems. This framework delivers precise definitions for essential terms in the context of RTE and a process that can be used to address new RTE applications. The main idea of the framework is to partition models into parts that require synchronization - skeletons - and parts that do not - clothings. Once such a partitioning is obtained, the relations between the elements of the skeletons determine whether a deterministic RTE system can be built. If not, manual decisions may be required by developers. Based on this conceptual framework, two concrete approaches to RTE are presented. The first one - Backpropagation-based RTE - employs change translation, traceability and synchronization fitness functions to allow for synchronization of artifacts that are connected by non-injective transformations. The second approach - Role-based Tool Integration - provides means to avoid redundancy. To do so, a novel tool design method that relies on role modeling is presented. Tool integration is then performed by the creation of role bindings between role models. In addition to the two concrete approaches to RTE, which form the main contributions of the thesis, we investigate the creation of bridges between technical spaces. We consider these bridges as an essential prerequisite for performing logical synchronization between artifacts. Also, the feasibility of semantic web technologies is a subject of the thesis, because the specification of synchronization rules was identified as a blocking factor during our problem analysis. The thesis is complemented by an evaluation of all presented RTE approaches in different scenarios. Based on this evaluation, the strengths and weaknesses of the approaches are identified. Also, the practical feasibility of our approaches is confirmed w.r.t. the presented RTE applications.