Topic 3: Scheduling and Load Balancing

More than ever, parallelism is available today at every level of computing systems, including dedicated embedded systems, basic instructions and registers, hardware accelerators, multi-core platforms, computational grids, etc. Despite of lot of efforts and nice positive results obtained during the past years, such systems are still not fully exploited. Scheduling represents the use or optimization of resources allocation in parallel and distributed systems. There are many issues to study for a better share of the load, a better reliability, a better adaptivity under computing, bandwidth or memory constraints. They are all crucial for obtaining a better use of parallel and distributed systems. It is a big challenge to study related techniques provided at both application and system levels. At the application level, the choice of the adequate computational model, the design of dynamic algorithms that are able to adapt to the particular characteristics, the mapping of applications onto the underlying computing platforms and the actual utilization of the systems are particularly relevant.