Projects
Development of geotechnical discrete element calculations on GPU Ghent University
Developments in the graphical cards for computers (GPU) have led to a calculation power in these cards that significantly exceeds the calculation power of the central processing unit (CPU). This project aims to utilize this power to perform geotechnical calculations using discrete elements by adapting existing software originally developed for the gaming industry.
Magnetism in nano-pore graphene University of Antwerp
Image reconstruction for in situ Computed Tomography University of Antwerp
Parallel algorithms for cutting and packing problems KU Leuven
System-on-Chip architectures for view interpolation in computational multi-camera systems. Hasselt University
Fault-tolerant network-on-chip design Ghent University
Network-on-chip (NoCs) possess high flexibility and scalability and thus represent an efficient approach for solving the communication problem in multi-core chips. However, it brings about reliability challenges. We focus on providing efficient strategies for mitigating both transient and hard errors in the NoC, for which we will develop fault-tolerant cache coherence protocols.
Photonic Network-on-Wafer for Multi-Tile GPUs: From Architecture to Hardware Implementation Ghent University
A new compute paradigm is proposed, based on large arrays of graphics processing units (GPUs) that are interconnected via a wafer-scale photonic network. This results in unprecedented performance at low power. A holistic study of this new paradigm will be carried out, ranging from the design of a suitable computer and network architecture to the implementation of key hardware components.