In this bimonthly feature, HPCwire highlights newly published research in the high-performance computing community and related domains. From parallel programming to exascale to quantum computing, the details are here.
Particle physics experiments involving the Large Hadron Collider require a massive amount of computing power – a demand that is projected to increase with coming upgrades. In this paper, written by a team from the University of Freiburg, the authors discuss how the university has linked its NEMO HPC Cluster to the Worldwide LHC Computing Grid to augment the HPC resources available to the LHC.
Authors: Felix Bührer, Frank Fischer, Georg Fleig, Anton Gamel, Manuel Giffels, Thomas Hauth, Michael Janczyk, Konrad Meier, Günter Quast, Benoît Roland, Ulrike Schnoor, Markus Schumacher, Dirk von Suchodoletz and Bernd Wiebelt.
While operating, HPC systems generate large amounts of metadata. Existing systems do a decent job of managing some of the metadata, but the data known as “rich” metadata – which record running processes and jobs and the relationships between them – are mostly left unattended. In this study, written by a team from UNC Charlotte, Texas Tech University and Argonne National Laboratory, the authors propose a graph model for managing rich metadata. They evaluate the graph model on synthetic and real HPC workloads to demonstrate its advantages and scalability.
Authors: Dong Dai, Yong Chen, Philip Carns, John Jenkins, Wei Zhang and Rob Ross.
As modern technologies like the smart grid and renewable energy become more widely available, interconnected power systems are more complex and uncertain than ever. This paper, written by a team from China, presents a probabilistic study for such systems based on an HPC method supported by the Computational Shared Facility at the University of Manchester. Using this method, the researchers gained new insights into probabilistic studies on power systems and into running Monte Carlo simulations on HPC systems.
Authors: Pengyu Duan, Sheng Xu, Huijie Chen, Xiaotian Yang, Shuai Wang and Ende Hu.
Deep learning has dramatically advanced the capabilities of bioinformatics applications. In this study, conducted by a team from Oak Ridge National Laboratory, University of Memphis, the Louisiana State University Health Sciences Center and the National Cancer Institute, researchers trained a neural network to extract information from a massive dataset of cancer pathology reports. The researchers then evaluated its scalability and accuracy.
Authors: John X. Qiu, Hong-Jun Yoon, Kshtij Srivastava, Thomas P. Watson, J. Blair Christian, Arvind Ramanathan, Xiao C. Wu, Paul A. Fearn and Georgia D. Tourassi.
Incredibly intricate manufacturing and airframe assembly processes have the ability to generate enormous datasets as they go through various tests and simulations with many important parameters. In this paper, written by a team from Peter the Great St. Petersburg Polytechnic University discusses the use of HPC to accelerate this task, describing a specialized approach that combines variation simulation and HPC to optimize the process.
Authors: N. Zaitseva, S. Lupuleac, M. Petukhova, M. Churilova, T. Pogarskaia and M. Stefanova.
With open source software becoming crucial to the design and testing of quantum algorithms, this team of researchers from Toronto reviewed a wide range of open source software for quantum computing. They covered documentation, licenses, programming language, compliance, and project culture. The researchers find that despite the rich diversity of projects, many have shortcomings, including trouble attracting outside developers. The team then recommends best practices for improving the open source quantum computing software ecosystem.
Authors: Mark Fingerhuth, Tomáš Babej and Peter Wittek.
As supercomputers increase in size, abnormal events are more likely to occur. In this paper, a team of French researchers examined the problem of CPUs overheating in HPC systems, which can have a major impact on system efficiency. They analyze data collecting over a year on a Top500-ranked supercomputer demonstrating that overheating events are frequently associated with specific applications. They conclude by assessing the effect on system performance.
Authors: Marc Platini, Thomas Ropars, Benoit Pelletier and Noel De Palma.
Do you know about research that should be included in next month’s list? If so, send us an email at [email protected] We look forward to hearing from you.