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.
Expanding an HPC cluster to support the demands of digital pathology
A single clinical-grade image from a digital pathology scanner can range in size from hundreds of megabytes to five gigabytes. In this research, a team from Temple University set out to design a low-cost computing facility that could support the development of a repository for one million of these images. The researchers discuss the HPC cluster they expanded upon to accomplish this goal and evaluate its results.
Authors: C. Campbell, N. Mecca, T. Duong, I. Obeid and J. Picone.
Presenting a roadmap to exascale computing based on application performance goals
In this paper, a team of Swiss and British researchers step away from the hypothetical and plant an exascale roadmap firmly in the actual. Using two European numerical weather prediction models, the authors assessed their performance when run at high spatial resolution on existing supercomputers. Even using the state-of-the-art petascale Piz Daint supercomputer, the performance gap was 100-250x what it needs to be to meet real-world operational requirements. The authors present a path to improve the performance of hardware and software in order to meet the goal of 1 km-scale weather and climate simulation in the next decade.
Authors: Thomas C. Schulthess, Peter Bauer, Oliver Fuhrer, Torsten Hoefler, Christoph Schar and Nils Wedi.
Evaluating trends in training and education provided by the SciNet HPC Consortium
The SciNet HPC Consortium – a Canadian academic HPC center – has provided training and education in HPC and scientific computing over the last decade. In this paper, written by a team from the University of Toronto, the researchers evaluate how those efforts have changed over the last six years, evolving from basic, isolated training events into a broad range of workshops and courses that build to certificates. The researchers discuss overall trends and implications.
Authors: Ramses van Zon, Marcelo Ponce, Erik Spence and Daniel Gruner.
Paving a roadmap toward exascale computing for autonomous driving
With autonomous driving emerging as one of the most computationally demanding technologies of the era, this researcher outlines a high-level roadmap toward satisfying those computational demands at the exascale level. The author highlights limitations and opportunities for efficiency in the computational complexities of autonomous driving and recommends a path forward.
Author: Levent Gürel
Applying big data and HPC in drug discovery
The fields of genomics, biomolecular structure dynamics, and drug discovery can produce massive amounts of molecular data ranging into the petabytes, making analysis and visualization a significant challenge. In this paper, written by a team of researchers from Savitribai Phule Pune University, the authors discuss how advanced analytics platforms and HPC systems can help to accelerate the process of finding the appropriate molecular “poses” through intensive simulations.
Authors: Rajendra R. Joshi, Uddhavesh Sonavane, Vinod Jani, Amit Saxena, Shruti Koulgi, Mallikarjunachari Uppuladinne, Neeru Sharma, Sandeep Malviya, E. P. Ramakrishnan, Vivek Gavane, Avinash Bayaskar, Rashmi Mahajan and Sudhir Pandey.
Using HPC for energy system optimization models
Energy system optimization models help policymakers and planners understand how changes in the electric grid will affect the overall system. In this paper, written by a team from Ireland, Italy and Switzerland, the authors examine how these resource-intensive models – which can often take days to process a single inquiry – could be adapted to minimize solution time in an HPC environment. The authors discuss benefits and tradeoffs and outline a path forward.
Authors: Tarun Sharma, James Glynn, Evangelos Panos, Paul Deane, Maurizio Gargiulo, Fionn Rogan and Brian Brian Ó Gallachóir.
Applying an HPC framework for dynamic power grid security assessments
Another team of researchers also explored the use of HPC for electric grids. The team – a group from Pacific Northwest National Laboratory – examined dynamic security assessments, which help evaluate whether power grids can weather disturbances. Renewable energy and smart grid technologies have increased the uncertainty (and as a result, the computational burden) in these assessments. The authors discuss the advantages of applying a framework that links data from HPC to statistical analysis and visualization.
Authors: Yousu Chen, Bruce Palmer, Poorva Sharma, Yong Yuan, Bibi Mathew and Zhenyu Huang.
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.