Midway evaluation in Computational Engineering - Joseph Arnold Xavier

Event on Zoom: https://eu01web.zoom.us/j/61227812032

Title of thesis: Optimization of parallel algorithms for heterogeneous high performance computing systems

Student: Joseph Arnold Xavier

Doctoral committee:
Dr. Morris Riedel, Professor, Faculty of Industrial Engineering, Mechanical Engineering and Computer Science, University of Iceland
Dr. Gabriele Cavallaro, Associate Professor, Faculty of Electrical and Computer Engineering, University of Iceland
Dr. Hemanadhan Myneni, Research Specialist at the Faculty of Industrial Engineering, Mechanical Engineering and Computer Science, University of Iceland
Dr. Rocco Sedona, Postdoctoral Researcher, Forschungszentrum Jülich
Dr. Markus Goetz, Group Leader, Karlsruhe Institute of Technology (KIT)

Abstract

High Performance Computing (HPC) is essential for the detailed analysis of large and complex Earth Observation (EO) datasets. However, the adoption of HPC within the EO community faces challenges due to the increasing heterogeneity of HPC systems and a lack of specialized expertise in the optimization and the development of scalable parallel software that can fully harness the capabilities of emerging computing architectures and trends in parallel computing. Hence, the primary goal of the PhD research is to develop efficient algorithms that efficiently process large-scale remote sensing satellite data, utilizing parallel machine learning and deep learning algorithms scalable to modular supercomputing architectures. Besides optimization, the benchmarking of the applications finetuned for earth observation datasets for performance and energy efficiency will an area of focus in the research. The optimization of the applications will also be tested on the pilot system of ‘European Pilot for Exascale Computing’ (EUPEX).