High Performance Computing
High-Performance Computing (HPC) refers to the practice of aggregating computing power in a way that delivers much higher performance than one could get out of a typical desktop computer. HPC is used to solve complex or large problems in science, engineering, business or the industry.
Indeed, the intensive growth of processing power, data storage and transmission capabilities has revolutionized many aspects of science and these resources are essential to achieve high-quality results in many application areas.
- On the one hand, HPC provides a key-tool for the development of high-level and excellent research in different scientific fields.
- On the other hand, Luxembourg’s data-driven innovation strategy, conducted by the government, calls for more HPC and HPC-related experts in Data Analytics and Artificial Intelligence (AI).
In line with its research and teaching missions, the University of Luxembourg has made HPC one of its top priorities.
The University of Luxembourg can draw on:
- a multidisciplinary team dedicated to HPC;
- more than 650 data scientists, engineers, designers, physicists, material scientists, biologists, bioinformatics, historians and economists using HPC and continuously gaining expertise and knowledge on the way to run “compute- and storage-intensive computations” sustaining research excellence while achieving faster time-to-solution;
- an extended HPC, High-Performance Data Analytics (HPDA) and AI training offer addressing Bachelor’s, Master’s or doctoral students, as well as University researchers and professionals of Luxembourg and the Greater Region;
- on-site supercomputers hosted in the premises of the University’s Centre de Calcul (CDC), which implements cutting-edge approaches for data centre management, as well as the capacity to host state-of-the-art liquid-cooled HPC systems. As of 2021, the University HPC facility features two supercomputers (Iris and Aion) totalling a computing capacity of 2,76 PetaFlops and a shared storage capacity of 10,68 PetaBytes.
- The large installed storage capacity of the University HPC facility is moreover crucial for research in polymer physics or material science, to perform simulations in economic research, as well as biomedical research on neurodegenerative diseases like Parkinson’s or Alzheimer’s or to fight against COVID-19.
For more details, see the last ULHPC Annual Report (2020)
A platform for research: what University researchers can do with supercomputers?
Modern research requires again significant computing power and generates large amounts of data. While High-Performance Computers – also called supercomputers – will enable to run fast complex and large-scale calculations to make simulations or models of physical systems. HPC-related technologies offer additional opportunities that may be combined. HPDA, which refers to the analysis in real time of huge amounts of data, is crucial nowadays in the era of Big Data. AI, which has the ability to identify patterns, learn and make decisions in different problems, has already drawn the attention of scientists from a broad range of disciplines. HPC provides the necessary tool for the development of training and testing of the most advanced large-scale AI applications requiring to perform trillions of calculations per second for processing huge amounts of data (words, images, videos …). The High-Performance Computing team of the University of Luxembourg (UL HPC) has built an online repository to host a number of demonstrators highlighting the capabilities and capacities of HPC in different scientific projects.
Computing and storage capacity outlook
The HPC facility is one of the digital platforms of the University. It is managed by an expert team under the responsibility of Dr. Sebastien Varrette.
Composed by two clusters totalling hundreds of compute nodes, the UL HPC supercomputing platform has kept growing over time thanks to the continuous efforts of the Research Computing and HPC operations team (Dr. S. Varrette, H. Cartiaux, T. Valette, A. Olloh and S. Peter) completed by the core contributions from several HPC researchers and experts (in particular Dr. F. Pinel, Dr. E. Kieffer, Dr. E. Krishnasamy, Dr. L. Koutsantonis). It provides in 2021 a total computing capacity of 2,76 PetaFlops (1 PetaFlops = 1015
floating point operations per second) and around 10 PetaByte of shared data storage. This places the HPC facility of the University of Luxembourg as one of the major actors in HPC and Big Data for the Greater Region Saar-Lor-Lux.
In 2020, 667 520 jobs were scheduled by 650 distinct researchers on the Uni.lu HPC facility, totalling 4319 CPU Years (37 835 309 CPU hours) of computation.
For more information, see hpc.uni.lu and hpc-docs.uni.lu.
Fully involved in the development of HPC across the country and the European Union
From its reputation and national expertise in the HPC and Big Data domains, the University of Luxembourg has been involved since 2008 to support the government in the HPC developments of the country. In 2017, it was selected by the ministry to represent the country within PRACE (Partnership for Advanced Computing in Europe). Later, the University was involved to support the EuroHPC development in the country, initially with key contributions to the design and architecture of the Luxembourg MeluXina supercomputer (in particular from V. Plugaru and Dr. S. Varrette) or in the establishment of LuxProvide, the company piloting MeluXina and the National HPC supercomputing Center (with Prof. P. Bouvry acting as half-time co-CEO of the company or V. Plugaru who joined as CTO).
Nowadays, the University pursue these efforts to further develop the HPC ecosystem in the country through strategic European projects. Outside PRACE (Partnership for Advanced Computing in Europe) previously mentioned or ETP4HPC (European Technology Platform for High Performance Computing), a strategic initiative was held in 2020 where Luxinnovation, LuxProvide and the University of Luxembourg have composed a consortium which is actively participating in the EuroCC project. It is aiming to set up the Luxembourg National Competence Centre (NCC) in HPC and contribute -with the support of the CASTIEL initiative- to a network of 33 European NCCs. As a result, the Luxembourg National Competence Centre in HPC encourages the use of High Performance Computing by academics, the industry as well as public administrations, and offer a portfolio of services in HPC, HPDA and AI meeting the local needs. This project is coordinated for the University by A. Moinier-Vandeventer with the support of Dr. L Koutsantonis, Dr. A. Roy. and Dr. E. Krishnasamy.
Other ongoing initiatives aim at consolidating the HPC and Data Science environment provided within the University. This includes the Luxembourg NVidia AI Technology Center (NVAITC) coordinated since 2019 by Dr. F. Pinel, Slices-DS, an experimental research infrastructure for Computing/Communication Experimental Studies managed for Luxembourg by Dr. E. Kieffer and completing the existing Grid5000 Luxembourg site deployed in 2008 by Dr. Varrette and maintained since 2011 by H. Cartiaux.
Other actions and initiatives led by Prof. Bouvry have been developed since 2020, typically part of the Digital Strategy program to offer HPC research support (Dr. T. Carneiro Pessoa, Dr. O.-C. Marcu, Dr. J. Sirajzade) or Partnerships developments (Dr. J. Pecero).
More recently (in Oct 2021), the University of Luxembourg was selected by EuroHPC to pilot the EuroHPC Master program and will be led by Prof. Bouvry.
HPC: a booming topic at the University of Luxembourg since the beginning
Since its creation in 2003, the University of Luxembourg has been continuously expanding its pool of HPC-related researchers. Impulsed by Prof. Pascal Bouvry and further continuously supported at the rectorate level thanks to the seminal efforts of Prof. Franck Leprevost, the University of Luxembourg has invested tens of millions of euro into its own HPC facilities to responds to the growing number of domain-specific researchers in need for increased high-performance computing and storage.
Driven by Dr Sebastien Varrette, the development of this platform kept on growing over time to offers a cutting-edge research infrastructure to Luxembourg public research.
In parallel, the training offer in HPC and HPDA was consolidated outside the Parallel Computing lectures, especially with the online ULHPC tutorials and the bi-annual ULHPC School launched in 2014, which offer instructions, hands-on and guided sessions on a variety of topics representative of research activities and domains present at the university.
The recently awarded EuroHPC Master program will expand this offer starting 2022.
Stay connected!
#HPC #HPDA #AI
