DreamCloud: Dynamic Resource Allocation in Embedded and High-Performance Computing
The main objective addressed by DreamCloud is to enable dynamic resource allocation in many- core embedded and high performance systems while providing appropriate guarantees on performance and energy efficiency.
DreamCloud will address techniques to allocate computation and communication workloads onto computing platforms during run-time whilst respecting performance and energy budget requirements. To achieve this, DreamCloud brings together teams from embedded computing and high- performance computing (HPC) to address these common ambitious yet achievable goals:
- Provide complex embedded systems with cloud-like capabilities such as those available in today’s high-performance computing, allowing them to dynamically tune resource usage but without sacrificing critical application-specific constraints in performance and energy.
- Enable HPC and cloud computing systems to balance workload and manage resources so that they can offer more meaningful guarantees in terms of performance and energy, focussing not only on improving average behaviour but also on reducing variability and upper bounds of timing and energy metrics.
DreamCloud brings together industrial partners from deeply embedded systems (e.g. automotive), consumer embedded systems (e.g. household media), and high performance computing (e.g. HPC platforms); academic partners from embedded systems, real-time systems and HPC. This will enable DreamCloud to develop better approaches by cross-fertilising expertise and experience from multiple industrial domains and academic communities.
|UoY Lead||Leandro Soares Indrusiak|
|UoY People on Project||Neil Audsley, Amit Singh, Piotr Dziurzanski, Andrew Burkimsher|
|Partners||Universitaet Stuttgart, AICAS GMBH, Robert Bosch GMBH, Centre National De La Recherche Scientifique, University of York, Rheon Media Limited, Universite Montpellier 2 Sciences et Techniques|