- content:
The research programme Computer Structures and Systems is a combination of the advanced computing structures and efficient algorithms for complex-data processing that is the basis for reconfigurable computer systems. Reconfigurable systems are able to change their internal structure and function in response to external or internal stimuli. Increased performance and reduced power consumption are their widely recognized features, while their architectures provide the means for developing advanced computer systems that function autonomously most of the time without human intervention and have an ability to correct data, and to adapt and repair themselves.
Due to the explosion of data (e.g., complex massive-data) in real-life processing, re-configurable computer systems require new and innovative approaches to run, and manage processes. As a consequence, systems must become more versatile, resilient, dependable, energy-efficient, recoverable, customizable and adaptable to changing operational contexts, environments or system characteristics.
In high-performance/exascale computing the research involves the integration of reconfiguration and self-adaptivity with worst-case design principles in the development of the algorithms for distributed computing to be performed on clouds, clusters, multi-core CPUs, GPGPUs, FPGAs.
In algorithm engineering a multi-level approaches related to complex and massive-data are needed to allow the original complex problem to be decomposed into a set of low-order optimization sub-problems. Furthermore, many-objective approach should be introduced, since the state-of-the-art techniques often fail to find a solution of real-world problems with many objectives. When uncertainty affects the feasibility of a solution, resilience ensures a solution that will be feasible for any realization taken by the unknown conditions. To implement the solutions where the user does not need any specific knowledge about control parameter setting, the self-adaptive approaches for controlling the predictive and the optimisation algorithms are needed. The highest usability of the approach is seen in connection with coordination, where the collective functioning of the system is achieved by indirect interactions among the elements of the system.
The proposed research programme follows the main research objectives of the EU's Horizon 2020, within the mission of the Slovenian research agency. The results of the research will combine new and efficient computing structures, efficient systems for massive-data processing, and systems for effective human-computer interaction into a comprehensive unity to support the development of the most advanced computer systems. These systems will be used in production, transport, energy distribution, environmental sustainability, bioinformatics, health, and medicine, with the aim to ensure their social and technological progress.
|
