Parallel and distributed architectures are emerging as commonplace platforms for high performance computing. The existence of standard parallel programming libraries facilitates the use of such platforms for running challenging applications in science or engineering. However, as compared to the homogeneous case, a number of additional factors should be taken into account in order to ensure an efficient execution on heterogeneous computing environments. In such systems one must consider not only the application's dynamic behavior but also the environment's dynamics. We dealt with the dynamic load balancing for parallel adaptive simulations in heterogeneous computing environments. Such applications are usually characterized by the fact that the workloads assigned to processors may change significantly and unpredictably. Our work followed two main directions. In a first step we developed a theoretical heterogeneous computational model and we extended and analyzed the well-known diffusion algorithms with respect to such a model. Afterwards, our variant of generalized diffusion was further compared with other approaches and was shown to be faster than another popular technique, the hydrodynamic algorithm. The possibility of using polynomial acceleration schemes in a dynamic context was further investigated. Other faster dynamic load balancing schemes were proposed and investigated. An incremental algorithm that is capable of taking advantage of an already computed fair distribution for computing a new one was described. In a second step, we designed and implemented a testbed for adaptive numerical simulations in heterogeneous computing environments. A tool, called HeRMeS, which allows to perform such simulations and that incorporates the methods developed at the first step, was designed and implemented. Experiments with the discussed methods and with HeRMeS were performed.

This book constitutes the thoroughly refereed proceedings of the 18th International Conference, Euro-Par 2012, held in Rhodes Islands, Greece, in August 2012. The 75 revised full papers presented were carefully reviewed and selected from 228 submissions. The papers are organized in topical sections on support tools and environments; performance prediction and evaluation; scheduling and load balancing; high-performance architectures and compilers; parallel and distributed data management; grid, cluster and cloud computing; peer to peer computing; distributed systems and algorithms; parallel and distributed programming; parallel numerical algorithms; multicore and manycore programming; theory and algorithms for parallel computation; high performance network and communication; mobile and ubiquitous computing; high performance and scientific applications; GPU and accelerators computing.

This book constitutes the refereed proceedings of the 19th International Conference on Parallel and Distributed Computing, Euro-Par 2013, held in Aachen, Germany, in August 2013. The 70 revised full papers presented were carefully reviewed and selected from 261 submissions. The papers are organized in 16 topical sections: support tools and environments; performance prediction and evaluation; scheduling and load balancing; high-performance architectures and compilers; parallel and distributed data management; grid, cluster and cloud computing; peer-to-peer computing; distributed systems and algorithms; parallel and distributed programming; parallel numerical algorithms; multicore and manycore programming; theory and algorithms for parallel computation; high performance networks and communication; high performance and scientific applications; GPU and accelerator computing; and extreme-scale computing.

Branch and Bound (B&B) algorithms are exact methods used to solve combinatorial optimization problems (COPs). The computation process of B&B is extremely time-intensive when solving large problem instances since the algorithm must explore a very large space which can be viewed as a highly irregular tree. Consequently, B&B algorithms are usually parallelized on large scale distributed computing environments in order to speedup their execution time. Large scale distributed computing environments, such as Grids and Clouds, can provide a huge amount of computing resources so that very large B&B instances can be tackled. However achieving high performance is very challenging mainly because of (i) the irregular characteristics of B&B workload and (ii) the heterogeneity exposed by large scale computing environments. This thesis addresses and deals with the above issues in order to design high performance parallel B&B on large scale heterogeneous computing environments. We focus on dynamic load balancing techniques which are to guarantee that no computing resources are underloaded or overloaded during execution time. We also show how to tackle the irregularity of B&B while running on different computing environments, and consider to compare our proposed solutions with the state-of-the-art algorithms. In particular, we propose several dynamic load balancing algorithms for homogeneous, node-heterogeneous and link-heterogeneous computing platforms. In each context, our approach is shown to perform much better than the state-of-the-art approaches.

Euro-ParConferenceSeries The European Conference on Parallel Computing (Euro-Par) is an international conference series dedicated to the promotion and advancement of all aspects of parallel and distributed computing. The major themes fall into the categories of hardware, software, algorithms, and applications. This year, new and interesting topicswereintroduced,likePeer-to-PeerComputing,DistributedMultimedia- stems, and Mobile and Ubiquitous Computing. For the ?rst time, we organized a Demo Session showing many challenging applications. The general objective of Euro-Par is to provide a forum promoting the de- lopment of parallel and distributed computing both as an industrial technique and an academic discipline, extending the frontiers of both the state of the art and the state of the practice. The industrial importance of parallel and dist- buted computing is supported this year by a special Industrial Session as well as a vendors’ exhibition. This is particularly important as currently parallel and distributed computing is evolving into a globally important technology; the b- zword Grid Computing clearly expresses this move. In addition, the trend to a - bile world is clearly visible in this year’s Euro-Par. ThemainaudienceforandparticipantsatEuro-Parareresearchersinaca- mic departments, industrial organizations, and government laboratories. Euro- Par aims to become the primary choice of such professionals for the presentation of new results in their speci?c areas. Euro-Par has its own Internet domain with a permanent Web site where the history of the conference series is described: http://www.euro-par.org. The Euro-Par conference series is sponsored by the Association for Computer Machinery (ACM) and the International Federation for Information Processing (IFIP).