ST. LOUIS, Mo., Nov. 16, 2021 —HPCwire, the leading publication for news and information for the high performance computing industry, announced the winners of the 2021 HPCwire Readers’ and Editors’ Choice Awards at the Supercomputing Conference (SC21) taking place this week in St. Louis, MO. Tom Tabor, CEO of Tabor Communications Inc., unveiled the list of winners just before the opening gala reception.
“Every year we look forward to honoring the amazing accomplishments of the HPC community through our Readers’ and Editors’ Choice Awards,” said Tom Tabor, CEO of Tabor Communications, publisher of HPCwire. “These awards showcase the best and brightest research that can only be done with HPC-scale systems. To all of this year’s winners, I would like to extend a heartfelt congratulations and an appreciative thank you for the contributions you make to improving our quality of life on our planet.”
HPCwire has designated two categories of awards: (1) Readers’ Choice, where winners have been elected by HPCwire readers, and (2) Editors’ Choice, where winners have been selected by a panel of HPCwire editors and thought leaders in HPC. The process started with an open nomination process, with voting taking place throughout the month of September.
These awards, now in their 18th year, are widely recognized as being among the most prestigious recognition given by the HPC community to its own each year, and are the only awards that open voting to a worldwide audience of end users.
The 2021 HPCwire Readers’ and Editors’ Choice Award winners are:
Best Use of HPC in Life Sciences
Readers’ Choice: Working with the University of Birmingham School of Chemical Engineering, clinicians developed a physical model of an operating theatre with 3D measurements of equipment and an ultrasound anemometer to measure airflow. Using U-Birmingham’s BlueBEAR HPC (Lenovo, with Intel CPUs and IBM Spectrum Scale Storage, supplied by OCF Limited), within their model, they tracked air that emerges from the necks of surgical gowns that can contaminate surgical instruments, lead to infection, and increase patient suffering, which also places a greater financial burden on the United Kingdom’s National Health Service.
Editors’ Choice: The U.S. Department of Energy’s Joint Genome Institute (DOE/JGI) and the ExaBiome Project team at Lawrence Berkeley National Laboratory, supported by the DOE Exascale Computing Project, recently released the MetaHipMer (extension to JGI’s HipMer) end-to-end genome assembler that supports an unprecedented assembly of environmental microbiomes – evolutionary processes of microbial communities on decadal timescales (25 terabyte dataset). For seven years, JGI’s collaboration of computer scientists, biologists and bioinformaticians, with DOE HPC systems UPC++, Summit and Cori, has continuously improved ways to help researchers make meaningful discoveries from higher fidelity data.
Best Use of HPC in Physical Sciences
Readers’ Choice: An international collaboration of scientists from the Space Telescope Science Institute, Gemini Observatory, Caltech/IPAC, University of Chicago, Argonne National Laboratory, University of Illinois, Argelander-Institut für Astronomi, Zentrum für Astronomie der Universität Heidelberg, Australian National University, University of Alberta, and Max Planck Institut für Astronomie developed the first class of AI models that outperform humans in the classification of thousands of compact star clusters observed by the Hubble Space Telescope. Their work was enabled with Amazon cloud (AWS), Bridges-2 at the Pittsburgh Supercomputing Center, HAL at the National Center for Supercomputing Applications, Nvidia cuDNN-accelerated deep learning framework, TensorRT, and the Data and Learning Hub for Science.
Editors’ Choice: The Southern California Earthquake Center (SCEC) relies on the power of supercomputers to simulate why and how earthquakes occur, evaluate their effects, and help societies prepare for, survive and recover from quakes. SCEC used the Texas Advanced Computing Center’s Frontera supercomputer with Dell EMC PowerEdge servers, Intel Xeon Scalable processors and Intel Optane DC Persistent Memory, to produce dynamic models of earthquake processes, improving the predictability of earthquake system models, simulating earthquake events and promoting a safer society that’s more resilient to earthquakes.
Best Use of HPC in Response to Societal Plights (COVID-19 focused research/programs)
Readers’ Choice: The COVID-19 Genomics UK (COG-UK) Consortium expanded its CLIMB-COVID project, using hardware from DDN, Dell, Lenovo and others to sequence and aggregate over 675,000 coronavirus genomes and broadening the project to include samples from around the world.
Editors’ Choice: Using Pittsburgh Supercomputing Center’s Bridges and Bridges-2 systems (both built by HPE), supported by XSEDE, and working for the Louisiana Coastal Protection and Restoration Authority, a team of government, industry and academic scientists created simulations of storm effects on the state’s coast that will underlie Louisiana’s 2023 Coastal Master Plan.
Best Use of HPC in Energy
Readers’ Choice: A multinational collaboration of 50 researchers from 20 institutes, employing PRACE (Partnership for Advanced Computing in Europe) access to 35 million core hours on the Marconi HPC system at CINECA in Italy, and 32.6 million core hours on MareNostrum at the Barcelona Supercomputing Center in Spain – primarily leveraging Lenovo hardware – has successfully analyzed non-linear magnetohydrodynamic simulations to better understand the physics of fusion plasmas. Their discovery will lead to better control of plasma dynamics for fusion operation, and the production of abundant and safe nuclear energy.
Editors’ Choice: Vestas, a leading renewable energy company, used Deep Reinforcement Learning with Scalable Azure AI & HPC capabilities, including Nvidia GPUs, to develop an intelligent wind farm flow controller that dynamically yaws the turbines to mitigate the effect of wakes on downstream neighbors. Their innovation improves the energy yield of a wind farm by 2 percent over its lifetime (depending on site specific climate and park configuration), or around $1m per year (on a 100-megawatt wind farm at a capacity factor of 50 percent).
Best Use of HPC in Industry
(Automotive, Aerospace, Manufacturing, Chemical, etc.)
Readers’ Choice: Facing extreme regulations and HPC limits enforced by the Formula One racing organization, the Aston Martin Cognizant team was able to achieve 97 percent utilization of their HPC resources with Altair Grid Engine (within a limit of 30 teraflops – higher throughput, shorter wait-times, and reduced downtime), so they can edge out the competition even when HPC resources are finite. Modern Grand Prix racing has prioritized the construction of the most fuel- and energy-efficient cars, with a lower total cost, so running complex simulations using an optimized compute environment is a requirement long before the car hits the track.
Editors’ Choice: Aramco Americas collaborated with the U.S. Department of Energy’s Argonne National Laboratory (ANL) and Convergent Science to evaluate cold operations for propulsion systems. Model uncertainties at cold conditions were carefully characterized and resolved using the HPE Cray Theta supercomputer at ANL Leadership Computing Facility. Their research provides a better understanding of where pollutants originate, and will lead toward the development of zero climate impact propulsion systems.
Best Use of High Performance
Data Analytics & Artificial Intelligence
Readers’ Choice: Argonne National Laboratory, Parallel Works and Convergent Science developed a scalable, automated and adaptive machine learning genetic algorithm (ML-GA) workflow achieved with a novel design optimization “digital-twin” pipeline, which couples active ensemble ML-based surrogate models and GAs with computational fluid dynamics simulations, and accelerates the design optimization for virtual prototyping of an advanced heavy-duty engine by 10X over industry-standard optimization techniques – from a few months to a few days.
Editors’ Choice (TIE): The GatorTron research project at the University of Florida uses an academic hospital’s clinical notes to train a Natural Language Processing (NLP) model to recognize medical terminology. Using DDN’s A3I storage and Nvidia DGX SuperPOD technology, it’s the largest NLP trained model at this time.
A team led by Argonne National Laboratory, with scientists from the National Center for Supercomputing Applications and the University of Chicago, used the Summit supercomputer at Oak Ridge National Laboratory – with [email protected], funcX, Nvidia’s TensorRT and the Data and Learning Hub for Science – to create reproducible, physics-inspired AI models that search for and detect gravitational waves in one month’s worth of advanced LIGO data within seven minutes. It’s the first production-scale AI method that combines HPC and edge computing and enables orders of magnitude faster AI-driven inference of gravitational waves.
Best Use of HPC in Financial Services
Readers’ Choice: Research-based trading firms, like Jump Trading, use VAST Data’s Universal Storage, with Intel Optane SSDs and Intel QLC 3D NAND technology, for faster research pipelines and a competitive edge; superior to spinning media, this all-flash architecture provides faster access to algorithmic trading data for slight improvements in model accuracy that can mean the difference between windfall profits and catastrophic losses.
Editors’ Choice: Scotiabank used Microsoft Azure HPC & AI, with Nvidia GPUs and Intel CPUs, to develop a new risk engine that runs nearly a billion risk calculations per second, with dynamic computational scaling allowing traders and risk managers to better understand exposure across asset classes, and improve bank response to market gyrations during volatile periods (for example, the height of the pandemic); they reported a 33 percent year-on-year rise in revenues for its wholesale banking business to end October 2020, including a 38 percent increase in capital markets revenues and an 87 percent increase in rates and credit trading revenues.
Best AI Product or Technology
Readers’ Choice: Lenovo ThinkSystem SR670 V2
Editors’ Choice: Nvidia DGX A100
Best Use of HPC in the Cloud (Use Case)
Readers’ Choice: A team at the National Center for Supercomputing Applications (NCSA) used Amazon Web Services, PostgreSQL and open-source tools to develop a custom, HIPAA-compliant, Laboratory Information Management System (LIMS) for the University of Illinois COVID SHIELD saliva test. The NCSA COVID LIMS system integrates with multiple healthcare systems to accept test orders and report results, and is now used by diagnostic laboratories across the country to test K-20 students so they can safely return to in-person learning.
Editors’ Choice: In their quest to find the best position for their artificial heart valve used by surgeons to repair cardiac valve leakage, 3DT Holdings (Drugs and Devices for Diagnostics and Therapeutics) performed more than 3,000 heart simulations using Dassault Systemes’ living heart geometry with Abaqus to train multi-stage machine learning with XGBoost, Cubist, and Feed-forward DL. With UberCloud’s multi-cloud platform on Google Cloud Platform HPC spot instances, and Kubernetes Management by SUSE Rancher on Azure, 3DT Holdings was able to reduce the 6–10-hour simulation times to just 2 seconds real-time prediction with more than 95 percent accuracy, for under $20k.
Best HPC Cloud Platform
Readers’ Choice: Amazon Web Services
Editors’ Choice: Microsoft Azure Cloud
Best HPC Server Product or Technology
Readers’ Choice: NVIDIA A100 GPU
Editors’ Choice: AMD 3rd generation 7003 Epyc “Milan” processors
Best HPC Storage Product or Technology
Readers’ Choice: BeeGFS
Editors’ Choice (TIE): HPE Cray ClusterStor E1000, DDN EXAScaler
Best HPC Programming Tool or Technology
Readers’ Choice: oneAPI
Editors’ Choice: Python
Best HPC Interconnect Product or Technology
Readers’ and Editors’ Choice: Nvidia Mellanox HDR/NDR
Best HPC Collaboration (Academia/Government/Industry)
Readers’ Choice: Within weeks of the Arecibo telescope’s collapse, a broad coalition of collaborators from the Arecibo Observatory, the Texas Advanced Computing Center, University of Central Florida, the Engagement and Performance Operations Center, the NSF Cyberinfrastructure Center of Excellence Pilot (CICoE), and Globus at the University of Chicago convened to transfer petabytes of irreplaceable observation data to a safe place in proximity to capability-class computing to foster analysis.
Editors’ Choice: After years of planning and investment, the EU’s EuroHPC Joint Undertaking successfully launched four of its first eight planned supercomputers, with more on the way in the coming months.
Best Sustainability Innovation in HPC
Readers’ Choice: University of Florida’s HiPerGator AI supercomputer placed second on the June 2021 Green500 list with 29.52 gigaflops-per-watt power efficiency. With 140 Nvidia DGX A100 nodes, this SuperPod utilizes the AMD Epyc 7742 “Rome” processor, and ranked 22 on the corresponding Top500 list.
Editors’ Choice: The LUMI datacenter in Kajaani, Finland, is among the world’s most eco-efficient datacenters. By utilizing 100 percent carbon-free energy and rechanneling waste heat, it can operate a 10-megawatt supercomputer with -13,500 tons of CO2 emissions per year (corresponds to removal of 4,000 cars from traffic).
Top Supercomputing Achievement
Readers’ and Editors’ Choice: The Perlmutter supercomputer at the National Energy Research Scientific Computing Center (NERSC), Lawrence Berkeley National Laboratory, built by HPE using AMD Epyc 7763 “Milan” processors and Nvidia A100 GPUs, is capable of 64 petaflops of ‘standard’ Linpack performance, and nearly four exaflops of AI performance, making it the world’s fifth-fastest supercomputer on the June 2021 Top500 list. It also ranked third on HPCG, fourth on HPL-AI, and sixth on the Green500 lists.
Top 5 New Products or Technologies to Watch
AMD 3rd Generation Epyc 7003 series “Milan” processors
AMD MI200 Instinct GPUs
Intel 3rd Generation Xeon Scalable “Ice Lake” processor
Intel Xe-HPC Ponte Vecchio GPU
Nvidia BlueField 2/3 DPU
AMD Instinct MI200 GPUs
Cornelis Networks Omni-Path
DAOS with Intel Optane Persistent Memory
Intel Xe-HPC Ponte Vecchio GPU
Nvidia BlueField 2/3 DPU
Top 5 Vendors to Watch
Amazon Web Services
Workforce Diversity & Inclusion Leadership Award
Readers’ Choice: The Sustainable Research Pathways (SRP) program, managed jointly by Sustainable Horizons Institute and the U.S. Department of Energy’s Lawrence Berkeley National Laboratory, engages a diverse community of faculty and students from backgrounds that are traditionally underrepresented in HPC academics and careers. Now in its seventh year, SRP has changed the lives of hundreds who have participated in workshops and supported computationally-intensive research led by Berkeley Lab’s world-class scientists.
Editors’ Choice: NSF XSEDE EMPOWER (U.S. National Science Foundation, Extreme Science and Engineering Discovery Environment, Expert Mentoring Producing Opportunities for Work, Education and Research), led by Shodor Education, recruited and enabled a diverse group of ~200 student apprentices and interns who have contributed to XSEDE computationally-enabled research at U.S. universities and national laboratories.
Outstanding Leadership in HPC
Readers’ Choice: Lisa Su, AMD Chief Executive Officer, masterminded the comeback of AMD as a leading supplier of HPC processors, bringing to market AMD EPYC Rome and Milan CPUs that deliver industry leading price/performance across a wide range of workloads, while executing strongly on roadmap schedules and performance goals. Su is serving as a member of the President’s Council of Advisors on Science and Technology (PCAST) and in 2021 became the first woman to receive the Robert N. Noyce Medal from the Institute of Electrical & Electronics Engineers (IEEE).
Editors’ Choice (TIE):
Buddy Bland retired in February 2021 after 40 years of faithful service to the U.S. Department of Energy and the nation. As Project Director of the Oak Ridge Leadership Computing Facility, Bland delivered and deployed the Jaguar, Titan and Summit supercomputers that each ranked number one, at different times, on the Top500 list of fastest supercomputers in the world. In 2020, Bland oversaw the expansion of Summit to support COVID-19 research.
Valerie Taylor, Division Director/Argonne Distinguished Fellow at Argonne National Lab, is an award-winning computer scientist known for distinguished research, leadership and efforts to increase diversity and participation in computing. Taylor has authored or co-authored more than 100 papers in the area of high-performance computing. Taylor is an IEEE Fellow, ACM Fellow, and serves as the Executive Director of the Center for Minorities and People with Disabilities in IT (CMD-IT).
More information on these awards can be found at the HPCwire website at https://www.hpcwire.com/2021-hpcwire-awards-readers-editors-choice/ and on Twitter through the hashtag: #HPCwireAwards.
HPCwire is the #1 news and information resource covering the fastest computers in the world and the people who run them. With a legacy of world-class editorial and journalism dating back to 1987, HPCwire is the news source of choice for science, technology and business professionals interested in high performance and data-intensive computing. Visit HPCwire at www.hpcwire.com.
About Tabor Communications Inc.
Tabor Communications Inc. (TCI) is a media and services company dedicated to high-end, performance computing. As publisher of a complete advanced scale computing portfolio that includes HPCwire, Datanami, EnterpriseAI, and HPCwire Japan, TCI is the market-leader in online journalism covering emerging technologies within the high-tech industry, and a services company providing events, audience insights, and other services for companies engaged in performance computing in enterprise, government, and research. More information can be found at www.taborcommunications.com.