News @ Tabor Communications

The 2024 HPCwire Readers’ and Editors’ Choice Award winners are:

Best Use of HPC in Life Sciences

Readers’ Choice:

The University of Birmingham’s BlueBEAR cluster (Lenovo, with Intel® Xeon® processors and IBM Storage Scale software-defined file and object storage) used Nanopore sequencing on plasma samples from pediatric cancer patients to accurately determine cancer type and recurrence without invasive biopsies. High-performance computing resources enabled efficient processing of multi-petabyte datasets, potentially revolutionizing cancer detection and monitoring in both children and adults.

Editors’ Choice:

Researchers at KAUST developed HPC-GVCW, an open-source parallel implementation for processing 20,000 rice genomes on the Shaheen-III HPE Cray EX supercomputer. This breakthrough accelerates the discovery of genetic diversity in Asian rice, supporting global food sustainability efforts and paving the way for creating the world’s first “digital gene bank” for a major food crop. 

Best Use of HPC in Physical Sciences

Readers’ Choice:

The Destination Earth initiative uses EuroHPC supercomputers, including LUMI, an HPE Cray supercomputer at CSC, to create digital twins of the Earth system. DestinE provides a novel capability to perform bespoke high-resolution simulations to investigate possible evolutions of the climate and its extreme weather events, supporting decision-makers to better respond and adapt to these challenges.

Editors’ Choice:

(TIE): Scientists at Argonne National Laboratory, University of Chicago, University of Illinois at Urbana-Champaign, NCSA, and the University of Minnesota used the Globus platform for data management in developing a physics-informed transformer model to predict gravitational wave evolution for spinning binary black hole mergers, including higher-order modes. This AI approach dramatically reduces simulation time from days to seconds, handling terabyte-scale datasets with high accuracy.

(TIE): Cerebras and researchers at Sandia National Laboratory, Lawrence Livermore National Laboratory, Los Alamos National Laboratory and the National Nuclear Security Administration used the WSE-2 chip to perform molecular dynamics simulations 457x faster than the Frontier supercomputer. By mapping each atom to individual WSE-2 cores in a 2D grid, they simulated 800,000 atoms for milliseconds, a major advance over previous microsecond-scale simulations. This breakthrough enables new materials science research.

Best HPC Response to Societal Plight

Readers’ Choice:

Nvidia Earth-2 is an open, full-stack platform that accelerates climate and weather predictions using AI-augmented, high-resolution simulations. It combines physical simulations, machine learning models, and visualization tools to provide unprecedented speed and scale in global atmospheric modeling. This platform is accessible to users, enabling faster and more detailed climate and weather forecasting.

Editors’ Choice:

The Open Science Platform OSPREY aims to enhance pandemic response by enabling health officials to utilize HPC resources and data-driven decision-making. With support from Argonne National Lab and the University of Chicago, the researchers used Globus, Parsl, and EMEWS (Extreme-scale Model Exploration with Swift) to integrate automated workflows, data curation, and model management to facilitate rapid collaboration and development during health crises.

Best Use of HPC in Energy

Readers’ Choice:

Carnegie Clean Energy uses supercomputing and AI to optimize ocean wave energy capture and conversion efficiency. With successful deployments in Europe and Australia, their technology offers a viable clean energy source, contributing to the diversification of renewable energy and combating climate change.

Editors’ Choice:

Nvidia collaborated with Shell Oil to develop Fourier Neural Operator-based surrogate models using Nvidia Modulus for real-time simulation of CO2 plume migration in carbon capture and storage. This work scales scientific machine learning to realistic 3D subsurface systems, advancing digital twin technology for CCS applications.

Best Use of HPC in Industry (Automotive, Aerospace, Manufacturing, Chemical, etc.)

Readers’ Choice:

Argonne National Laboratory and RTX Technology Research Center used HPC and CFD modeling to simulate gas turbine film cooling with surface roughness defects. They employed Argonne’s GPU-accelerated NekRS solver on DOE supercomputers, providing high-fidelity data for developing surrogate models to optimize thermal management in next-gen aircraft engines.

Editors’ Choice:

Carnegie Mellon researchers used Pittsburgh Supercomputer Center Bridges-2 system to create a virtual water treatment plant, enabling AI to learn from veteran engineers’ responses to simulated breakdowns. This human-AI knowledge sharing aims to train new engineers and address potential staff shortages in water treatment facilities.

Best Use of HPC in Financial Services

Readers’ Choice:

Citadel Securities partnered with Google Cloud to enhance its quantitative research platform, leveraging high-performance computing for deeper data insights. This collaboration enabled massive scalability, tailored hardware solutions, and significantly reduced costs-per-research-hour through improved resource utilization and flexible consumption models.

Editors’ Choice:

DBS Bank transformed its quant pricing and risk engines using in-house capabilities and HPC+AI technologies from Amazon Web Services and Intel. The cloud-native solution efficiently handles Monte Carlo simulations for derivative trading, integrating advanced models and APIs. Using AWS services powered by Intel® Xeon® processors, built-in accelerators and software, the bank processes millions of risk pricings daily, improving customer response, risk management, and product delivery.

Best Use of High Performance Data Analytics & Artificial Intelligence

Readers’ Choice:

Argonne National Laboratory and Dow Inc. developed a framework combining computational fluid dynamics with an active machine learning optimizer (ActivO) for efficient turbulent jet mixer design. This novel approach optimized jet-mixing technology, potentially reducing reliability issues and costs associated with traditional agitators, with estimated savings of up to $6.1 million per year per plant.

Editors’ Choice:

The San Diego Supercomputer Center and Scientific Computing and Imaging Institute at the University of Utah launched the $6 million NSF-funded National Data Platform to create an equitable data ecosystem. This initiative aims to enhance access to open data, fostering innovation and collaboration while addressing global challenges like climate change through AI-integrated solutions.

Best HPC Storage Product or Technology

Readers’ Choice:

BeeGFS

Editors’ Choice:

DDN Exascaler

Best AI Product or Technology

Readers’ Choice:

Nvidia GH200

Editors’ Choice:

Cerebras Systems CS-3 AI System

Best Use of AI Methods for Augmenting HPC Applications

Readers’ Choice:

A multi-institutional team from the Carnegie Mellon University, UC Berkeley, Pittsburgh Supercomputer Center, and ACCESS used machine learning on PSC’s Bridges-2 system, allocated via U.S. NSF’s ACCESS program, to identify 50 gene regulatory elements in the brains of humans, bats, whales, and seals associated with vocal learning across mammals. This research revealed parallel genetic evolutionary paths for vocalization and a potential link to autism spectrum disorder in humans.

Editors’ Choice:

(TIE): Researchers from Carnegie Mellon and the University of British Colombia used AI and HPC to identify potential inhibitors for LRRK2, the primary genetic cause of familial Parkinson’s disease. Using computing resources at the Pittsburgh Supercomputing Center and the San Diego Supercomputer Center, their method narrowed 4.5 billion molecules to 14 promising candidates, potentially accelerating drug discovery for the disease affecting over 500,000 Americans.

(TIE): Google DeepMind’s GNoME, a deep learning model using graph neural networks, has accelerated materials discovery by predicting stability of new compounds. It identified 380,000 stable crystals with potential applications in green technologies, from improved batteries to efficient superconductors.

Best Use of HPC in the Cloud (Use Case)

Readers’ Choice:

Globus enabled near real-time data analysis at Argonne National Laboratory by connecting APS instruments with ALCF supercomputers. This automated pipeline allows scientists to adjust experiments on the fly, potentially accelerating scientific breakthroughs by delivering rapid results while researchers still have facility access.

Editors’ Choice:

The Institut Pasteur’s “IndexThePlanet” project aims to analyze and map the DNA of all living organisms to identify potential pandemic threats. Partnering with Amazon Web Services, they processed 20 petabytes of data using a cluster of 2.18 million vCPUs, expanding on their previous Serratus project to catalog viruses and prepare for future pandemics.

Best HPC Cloud Platform

Readers’ Choice:

Amazon Web Services

Editors’ Choice:

Parallel Works ACTIVATE

Best HPC Server Product or Technology

Readers’ Choice:

Nvidia Grace CPU Superchip

Editors’ Choice:

Intel® Xeon® 6 processors

Best HPC Programming Tool or Technology

Readers’ Choice:

European Environment for Scientific Software Installations (EESSI)

Editors’ Choice:

Spack package management tool

Best HPC Interconnect Product or Technology

Readers’ Choice:

Nvidia Quantum-X800

Editors’ Choice:

HPE Slingshot

Best HPC Collaboration (Academia/Government/Industry)

Readers’ Choice:

As one of the nation’s many contributors to the National Artificial Intelligence Research Resource (NAIRR) Pilot, including academic institutions such as Indiana University, NCSA, PSC, Purdue, TACC, and more, the San Diego Supercomputer Center uniquely makes available to the national science community AI research resources from commercial providers (AWS, AZURE, Google and Nvidia), as well as on-premise resources including NAIRR Classroom allocations via the National Research Platform, Expanse, and Voyager.

Editors’ Choice:

DigiFarm uses AI and supercomputing to revolutionize agriculture globally. Utilizing LUMI, an HPE Cray EX 4000 supercomputer and ClusterStor E1000 at CSC, they process high-resolution satellite data to accurately detect field boundaries and seeded areas. This technology helps farmers and agribusinesses optimize resource allocation, potentially reducing costs by 10% while increasing crop yields and efficiency.

Top Energy-Efficient HPC Achievement

Readers’ Choice:

The Massachusetts Green High Performance Computing Center (MGHPCC), which achieved LEED Platinum Certification for sustainability, houses multiple TOP500 systems for leading research institutions. The facility uses predominantly carbon-free energy sources, including local hydroelectric and solar power, and employs liquid cooling to further reduce its environmental impact for both its existing HPC installations and the future quantum computing complex that was recently announced.

Editors’ Choice:

Shearwater Geoservices migrated their UK HPC infrastructure to atNorth’s ICE02 data center in Iceland. This move resulted in a 92% reduction in CO2 emissions and 85% cost savings. Despite initial concerns about latency, comprehensive testing and collaboration ensured a successful transition. The migration was driven by rising energy costs in the UK and atNorth’s environmentally responsible approach, with ICE02’s PUE of 1.2 compared to 1.8 at the previous UK provider.

Top HPC-Enabled Scientific Achievement

Readers’ Choice:

The Destination Earth Climate Change Adaptation Digital Twin operationalizes global multi-decadal climate projections, and in 2024 it produced its first projections at fine spatial (5km) resolution. Such simulations enable evaluating impacts of climate change from global to local scales, supporting decision-making for climate change adaptation.

Editors’ Choice:

Microsoft and Pacific Northwest National Laboratory used AI and cloud computing to rapidly identify a potential new battery material. The Azure Quantum Elements platform screened 32 million candidates in just 80 hours, putting researchers on the path to synthesize and characterize a novel solid-state electrolyte. This AI-driven approach demonstrates potential to accelerate materials discovery for safer, more sustainable batteries. Researchers believe this method could compress centuries of chemistry and materials research into decades, addressing urgent energy and environmental challenges.

Top Supercomputing Achievement

Readers’ Choice:

The Aurora supercomputer at Argonne National Laboratory showcases AI’s growing impact in supercomputing. With its massive GPU cluster and advanced interconnect, Aurora enables AI-driven research across fields like neuroscience, particle physics, and drug discovery. This system demonstrates how AI can be effectively implemented for high-performance computing, accelerating scientific breakthroughs.

Editors’ Choice:

ORBIT, the Oak Ridge Base AI Foundation Model for Earth System Predictability, is a 113-billion-parameter AI spatiotemporal transformer model that leverages the Frontier supercomputer’s 49,152 AMD GPUs to achieve a sustained computing performance of 1.6 exaFLOPs. Using novel parallelism techniques, it sets new benchmarks in AI-driven climate modeling, enhancing both computing efficiency and prediction accuracy to support more informed decision-making for climate change challenges.

Top 5 New Products or Technologies to Watch

Readers’ Choice:

AMD Instinct MI300 Series Accelerators

Nvidia Blackwell GPU

Nvidia Quantum-X800

Nvidia Spectrum-X

Open OnDemand

Editors’ Choice:

AMD Instinct MI300 Series Accelerators

AWS Parallel Computing Service

Micron Multiplexed Rank Dual Inline Memory Module (MR DIMM)

CUDA-Q

Nvidia Blackwell GPU

Top 5 Vendors to Watch

Readers’ Choice:

Adaptive Computing

AMD

Ansys

Amazon Web Services

Nvidia

Editors’ Choice:

AMD

Amazon Web Services

Hammerspace

Intel Corporation

Penguin Solutions

Workforce Diversity & Inclusion Leadership Award

Readers’ Choice:

STEM-Trek is a 501.c.3 public charity that supports STEM workforce development through scholarly travel, mentoring, and training. It encourages beneficiaries to pay it forward and bridges gaps between high- and low-tech job seekers. STEM-Trek traditionally hosts pre-conference workshops ahead of SC, with ART@SC24 supporting 40 participants from 10 countries and 13 U.S. states.

Editors’ Choice:

The Technology Training for Non-Traditionals (TNTs) program, funded by the National Science Foundation and hosted by Texas Advanced Computing Center (TACC), is a year-long initiative to broaden participation in the cyberinfrastructure workforce. The NSF-supported program offers mentoring, hands-on experience, and networking opportunities, including participation in the Supercomputing Conference’s SCinet.

Outstanding Leadership in HPC

Readers’ Choice:

Bryan Johnston’s leadership of HPC community development efforts at the Centre for High-Performance Computing has been instrumental in advancing HPC across Africa. He has developed crucial infrastructure, fostered pan-African collaboration, implemented training programs, and promoted HPC awareness. His efforts have helped to position Africa as a significant player in the global HPC community.

Editors’ Choice:

Since 2022, David Keyes has been a finalist for the ACM Gordon Bell Prize utilizing leadership-scale supercomputers such as Fugaku, Frontier, and Shaheen, and partnering with industry leaders like Cerebras, Nvidia, and HPE. These achievements, assisted by Hatem Ltaief, were demonstrated across various real scientific applications, including geostatistics, seismic analysis, climate modeling, and genomics.

More information on these awards can be found at the HPCwire website at https://www.hpcwire.com/2024-hpcwire-awards-readers-editors-choice/ and on Twitter through the hashtag: #HPCwireAwards.

About HPCwire

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, BigDATAwire, AIwire, 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.