Workshop on High Performance Computing and Applications

Tuesday, 19th of July 2016, European University Cyprus (link)

Christos Antonopoulos

  • webpage
  • University of Thesally, Greece
  • Disrupting the power/performance/quality tradeoff through approximate and error-tolerant computing
A major obstacle in the path towards exascale computing is the necessity to improve the energy efficiency of systems by two orders of magnitude. Embedded computing also faces similar challenges, in an era when traditional techniques, such as DVFS and Vdd scaling, yield very limited additional returns.
Heterogeneous platforms are popular due to their power efficiency. They usually consist of a host processor and a number of accelerators (typically GPUs). Alternatively, they may integrate multiple cores or processors with inherently different characteristics, or even just configured differently. Additional energy efficiency gains can be achieved for certain classes of applications by approximating computations, or in a more aggressive setting even tolerating errors. These opportunities, however, have to be exploited in a careful, educated manner, otherwise they may introduce significant development overhead and may also result to catastrophic failures or uncontrolled degradation of the quality of results. Introducing and tolerating approximations and errors in a disciplined and effective way requires rethinking, redesigning and re-engineering all layers of the system stack, from programming models down to hardware.
I will present the experiences of our group from this endeavor and discuss our perspective on the main obstacles preventing the wider adoption of approximate and error-aware computing and the necessary steps to be taken to that end.
Christos D. Antonopoulos, is Assistant Professor at the Department of Electrical and Computer Engineering of the University of Thessaly in Volos, Greece. He earned his PhD (2004), MSc (2001) and Diploma (1998) from the Department of Computer Engineering and Informatics of the University of Patras, Greece. His research interests span the areas of system and applications software for high performance computing, emphasizing on monitoring and adaptivity with performance and power/performance/quality criteria. He is the author of more than 50 refereed technical papers, and has been awarded two best-paper awards. He has been actively involved in several research projects both in the EU and in USA.

Pedro Moura Trancoso

  • webpage
  • University of Cyprus
  • Building Blocks for Petascale Computing: Scalable, Near-Memory, Approximate, and Reversible Computing
The increasing requirements for more complex and demanding applications spark the development of efficient and high performance systems. Therefore, the challenges are many. We are currently looking at different techniques, or building blocks, to achieve the Petascale goal.
Future systems will need to have both hardware and software that can scale effectively for a very large number of computation resources. Simpler scalable hardware needs to be combined with lightweight scalable runtime systems. We explore the use of a dataflow execution model to achieve the desired scalability.
The increasing scale of systems add more computation resources, which in turn put an increasing pressure in the large memory systems. In order to address this issue, we are looking at moving part of the computation to the memory system by exploiting the near-memory computation model. The feasibility of this approach is backed up by recent announcements of real products.
The increasing integration and reduced power operation levels will lead to more soft-errors in future systems. The traditional techniques to address these issues are costly in terms of hardware, latency, and power. Newer approaches include exposing these errors to the user-level, thus allowing the application to decide which errors need to be addressed and how. We are currently looking at these Approximate Computing techniques for applications that traditionally were error intolerant.
Finally, looking forward for new technologies, we are exploring the use of Reversible Computing, which is a paradigm that promises minimum power consumption and a greater flexibility in the execution of applications. This paradigm has the potential to improve not only the efficiency of the execution but also the productivity of the programmability of scalable systems.
In this presentation I will be addressing the above topics in the context of the research performed in our research group towards the goal of Extreme Scale or Petascale Computing.

Pedro Moura Trancoso is an Associate Professor at the Department of Computer Science at the University of Cyprus, which he joined in 2002. He has a PhD and MSc. in Computer Science from the University of Illinois at Urbana-Champaign, USA. His research interests are in the area of Computer Architecture and include Multicore Architectures, Memory Hierarchy, Parallel Processing and Programming Models, Database Workloads, High-Performance Computing, Approximate Computing, and Reversible Computing. Currently his research team, Computer Architecture, Systems and Performance Evaluation Research - CASPER ( is composed of 2 PhD students and 4 undergraduate students. The latest funding for his research include the participation in the UniServer EU H2020 project (3 years, starting 2/2016), Reversible Computation EU COST (4 years, started 4/2015), TERAFLUX EU FP7 IP project (4 years, completed in 2014) and the lending of a 48-core experimental processor, the Intel SCC, by the Intel Corporation. He is also a member of the IEEE, ACM, and the HiPEAC Network of Excellence.

Antonis Papadakis

  • webpage
  • Frederick University, Cyprus
  • High Performance Computing for modeling fluid-transport and electromagnetic phenomena
Many phenomena related to engineering and physics require the solution of partial differential equations for analyzing transport fluid phenomena such as conservation of mass, momentum and energy for multiple species coupled with Maxwell equations for modelling electromagnetic field propagation. In order to solve these multiple system of equations for various species, especially in 3D, it is necessary to revert to High Performance Computing (HPC). Due to the nature of these problems and the high RAM necessary, the only option is to revert to distributed computing. The solution of these partial differential equations always involves the breaking of the geometry into a computational mesh, which allows easy parallelization of such problems with good scalability by using mesh-partitioning algorithms. The partitioning of the mesh is performed according to the number of cores and the cores work almost independently, with very little inter-communication, only to combine the results and refresh during time-stepping operations. For the solution of these partial differential equations, the techniques developed are based on high-order, highly accurate finite element, finite volume and finite difference schemes in combination with adaptive mesh techniques in parallel algorithms to tackle these phenomena in possible applications of Physics and Engineering such as in Plasmas, Computational Fluid Dynamics, High Voltage Systems, Photovoltaics, Electromagnetics, Astrophysics and Space.
Antonis P. Papadakis was born in Limassol, Cyprus, on October 11, 1976. He graduated with a First Class Honors BEng degree in electrical engineering from the University of Warwick, UK in 1999, and received a PhD degree in electrical engineering, after a four year full time scholarship, from the University of Cambridge (Trinity College), UK, in 2004.
His employment experience includes a Postdoctoral Fellowship at the Department of Physics, University of Cyprus, High Energy Physics (HEP) group, from 2004-2007, for four years working on the Compact Muon Solenoid (CMS) experiment at the European Center for Particle Physics (CERN). Since October 2007, he is with the Department of Electrical Engineering and he is currently an Associate Professor. He is the coordinator of the MSc in Oil & Gas and Offshore Engineering programme and Vice-Chairman of the Department of Electrical Engineering.
His special fields of interest include Computational Fluid Dynamics, High Performance Computing, High voltage Systems, High Energy Physics, Plasma Physics, Renewable Energy Sources (Photovoltaics), Computational Modeling and Power System applications.
Prof. Papadakis is an author in 45 journal peer reviewed publications, 25 International Conference publications and 1 book chapter. He acted as Co-Chairman of the International Conference on Deregulated Electricity Market Issues in South‐Eastern Europe (DEMSEE 2014) held in Nicosia, Cyprus, Hilton Hotel on the 25 and 26th of September 2014.

Andreas Efstathiou

  • webpage
  • European University, Cyprus
  • Applications of High Performance Computing for modeling the panchromatic spectral energy distributions of a large sample of galaxies
It is now clear that in order to understand the numerous processes that control galaxy formation and evolution (e.g. quiescent star formation in disks, bursts of star formation in mergers, accretion of matter onto supermassive black holes and feedback associated with them) we need multi-wavelength observations of galaxies at all cosmic epochs. This has led to a series of surveys at all wavelengths from the X-rays to the radio of ever improving sensitivity, resolution and sky coverage. At the same time significant progress has been made in the development of models (stellar population synthesis models, radiative transfer models, N-body and hydrodynamic simulations) that aid the interpretation of these observations. The spectra of galaxies can be decomposed into a number of components. Models are available for all of these components but comparing them to the data is a very computing intensive operation as each model has at least 2 or 3 parameters. Professor Efstathiou has been involved for a number of years in the development of such models and their comparison with observations (e.g. Efstathiou & Rowan-Robinson 1995, Efstathiou et al. 2000, Efstathiou & Siebemorgen 2009, Efstathiou et al. 2013). In this talk he will review progress in this field and show how High Performance Computing can help us make further progress.
Prof. Andreas Efstathiou was educated at Queen Mary, University of London. He held appointments at Queen Mary & Westfield College (1989-1992), University of Hertfordshire (1992-1995) and Imperial College London (1995-2001) before joining Cyprus College which evolved into European University Cyprus (EUC).

He has published about 75 well-cited papers (h-index 40) in refereed journals and 30 papers in conference proceedings. His research interests lie in the area of infrared astronomy and cosmology, galaxy formation and evolution, star formation and active galactic nuclei. He has done pioneering work on radiative transfer modelling of the spectral energy distributions of galaxies and has developed models for the emission of almost every type of galaxy over their complete spectrum. His models have also been used for the interpretation of galaxies discovered in a number of extragalactic surveys (IRAS, ELAIS, FIRBACK, SWIRE, AKARI, SCUBA, SHADES, HerMES, Planck, WISE, ALMA).

He also participated in extragalactic surveys with the Infrared Space Observatory (ISO), the Spitzer Space Telescope and the submillimetre camera SCUBA. He is currently involved in surveys with the Herschel Space Observatory and a survey that searches for highly extinguished supernovae in luminous infrared galaxies using adaptive optics on 10-meter class telescopes. He also continues his work on radiative transfer modelling of galaxies. In the period 2005-2008 he was chairman of the Department of Computer Science and Engineering.

George Tsouloupas

  • webpage
  • The Cyprus Institute
  • Advanced Computing Infrastructures at the Cyprus Institute
The mission of the CyI High Performance Computing Facility (HPCF) is to provide ICT services and support the research activities of the Cyprus Institute and to provide compute and data resources (HPC, Cloud, Data Storage) to the research community of Cyprus and the Eastern Mediterranean region. It is also the mission of the HPCF to assist CyI in collaborating and providing ICT services to the national and regional/international academic research communities, and in certain cases to the public and private sectors. In this talk we will give an overview of the core infrastructures at the Facility as well as some of the operational aspects of HPC, Cloud and Storage Infrastructures.
Dr George Tsouloupas studied Computer Science and Mathematics at FDU and obtained his PhD from the University of Cyprus in the area of monitoring and performance evaluation in High Throughput and High Performance Computing. He as worked on several European infrastructure projects on Grid Computing and High Performance Computing. He has worked in industrial research and development on diverse subjects ranging from quality of drinking water to Electronic Health Records. George is the Technical Director of the High Performance Computing Facility at the Cyprus Institute.

About Us

The AHPC group was established in 2012 and carries out pioneering work in Astrophysics and Parallel and Distributed Computing.

The group has a prestigious network of international collaborators and is directed by Prof. Andreas Efstathiou, Vice Rector for Research and External Affairs.

Get In Touch

European University Cyprus

Department of Computer Science and Engineering

Address: 6, Diogenes, Engomi, 2404 Nicosia, Cyprus


euc logo new 2018