HPE leads industry with top-ranked supercomputers and AI servers

Hewlett Packard Enterprise (HPE) today announced its architecture continues to lead the bi-annual TOP500 and Green500 rankings of the world’s fastest and most power-efficient high performance computing (HPC) systems, while its artificial intelligence (AI) servers top industry benchmarks. These systems are designed with breakthrough innovation and differentiated services expertise to help organizations solve major challenges across a range of disciplines.  HPE has built its leadership through decades of expertise in the design, manufacturing, installation and management of large-scale infrastructure.

HPE built six of the 10 fastest supercomputers on the planet

The latest edition of the TOP500 list validated the world’s most powerful systems, including the top three – all built by HPE.

• No. 1 – El Capitan, built for Lawrence Livermore National Laboratory (LLNL), a U.S. Department of Energy national laboratory, and powered by AMD, maintained its position as the world’s fastest supercomputer and improved its performance by achieving 1.742 exaflops. El Capitan also debuted with a No. 1 ranking on both the High-Performance Conjugate Gradient (HPCG) and HPL-MxP benchmarks, which respectively measure the performance of scientific applications and the convergence of HPC and AI with mixed-precision computations. El Capitan supports U.S. national security efforts and ensures the safety, security and reliability of the nation’s nuclear stockpile. Advanced modeling and AI-driven research will also be dedicated to the study of climate science, materials discovery and fusion energy.
  

• No. 2 – Frontier, built for Oak Ridge National Laboratory and powered by AMD, was the very first exascale system to be verified. Frontier delivers 1.353 exaflops and continues to drive advancements in quantum physics, renewable energy and precision engineering. Most recently, researchers used Frontier to develop an eye-opening, more complete picture of nuclear properties at the atomic and subatomic levels. For the first time, this new model captures and combines vital details such as the nucleus shape, energy of rotation and the energy that holds the nucleus together. Researchers expect the resulting insights to advance the fields of quantum physics, energy production and national security.
  
• No. 3 – Aurora, built in collaboration with Intel for Argonne National Laboratory, reached 1.012 exaflops and is also the second-most powerful system on the HPL-MxP list at 11.6 exaflops for mixed-precision calculations. The system is available to researchers through an open science initiative and Aurora is being used to advance EvaCortex, an AI-driven project that aims to map the cerebral cortex. This work combines next-generation electron microscopes, deep learning models and accelerator-based computing to produce new insights into brain connectivity at a cellular level.

Rounding out the top 10 are three of the most powerful supercomputers in Europe, all of which contribute to diverse research initiatives, from climate modeling to energy transition engineering.

• No. 6 – HPC6, built for Eni, has been leveraged to optimize industrial plant operations, enhance the accuracy of geological and fluid dynamics studies for CO₂ storage, develop more efficient batteries and innovative materials, optimize biofuel supply chain and simulate plasma behavior in magnetic confinement fusion. Eni’s commitment to supercomputing is at the core of its decarbonization strategy, supporting the creation and growth of new businesses related to the energy transition.
  
• No. 8 – Alps, built for Swiss National Supercomputing Centre (CSCS), was developed to meet the extreme data and computing requirements of science and serves a broad community of researchers in Switzerland and around the world studying medicine, quantum chemistry, astrophysics and climatology as well as AI and machine learning.
  
• No. 9 – LUMI, built for EuroHPC JU and hosted by the LUMI consortium at CSC – IT Center for Science, is being used to train neural networks in Nordic languages.

The most energy efficient supercomputers in the world rely on liquid cooling

Not only are HPE-built supercomputers fast, they rank among some of the most energy efficient systems in the world. All of them are designed with HPE’s industry-first 100% fanless direct liquid cooling (DLC) system architecture that delivers eight elements of cooling throughout the cluster to offer unparalleled efficiency and performance. This approach is the reason why four HPE-built supercomputers are among the top 10 systems and 11 are among the top 20 on the Green500 list, which ranks systems based on performance per watt.

• No. 3 – Adastra 2, built for Grand Equipement National de Calcul Intensif - Centre Informatique National de l'Enseignement Suprieur (GENCI-CINES), advances France’s sovereign AI and HPC initiatives, supporting open science in the fields of climate research, new energies, material development, biology and personal medicine.
  
• No. 4 – Isambard-AI Phase 1, built for University of Bristol, supports AI research in robotics and climate science with a sustainable energy profile.
  
• No. 7 – SSC-24 Energy Module, built for Samsung Electronics, debuts on the list with a performance per watt of 67.25 petaflops. It is the most energy efficient supercomputer owned by an enterprise.
  
• No. 8 – Helios GPU, built for Academic Computer Centre Cyfronet AGH, is also Poland’s fastest supercomputer.
  
• No. 11 – Portage is a new benchmarking system built by HPE to test real-world HPC and AI workloads and workflows for the company its and customers. This system leverages next-generation HPE Cray technology, HPE Slingshot interconnect and benefits from HPE’s DLC leadership. Portage also ranked 65^th on the TOP500 list of the world’s fastest supercomputers.
  
• No. 12 – Hunter, built for the High-Performance Computing Center of the University of Stuttgart (HLRS), is part of the Gauss Centre for Supercomputing's national strategy to develop three research centers and Hunter is dedicated to a wide range of applications in engineering and applied sciences for industry and academia.
  
• No. 15 – rzAdams, built for LLNL, is a sibling system of the world’s fastest supercomputer, El Capitan, and is dedicated to classified research.
  
• No. 16 – Frontier TDS is a test and development system built for Oak Ridge National Laboratory.
  
• No. 17 –Tuolumne, built for LLNL, is a companion system to El Capitan that is dedicated to open science, providing computational resources for research projects in fields like earthquake modeling, computational drug discovery, advanced manufacturing and AI-assisted fusion.
• No. 18 – El Dorado, built for Sandia National Laboratories, is an application-readiness test system that supports the United States’ stockpile stewardship mission.
• No. 19 – Alps, built for CSCS, also appears on the top 10 of the TOP500 list.

HPE AI servers rank first in industry benchmarks

HPE set new AI inference records on the MLPerf Inference v5.0 benchmarks, an industry effort driven by MLCommons to help the industry standardize performance evaluation of AI technology. HPE ProLiant Compute DL380a Gen12 and DL384 Gen12 servers, along with the HPE Cray XD670, played pivotal roles in achieving top rankings in 50+ testing scenarios.

As AI continues to expand and accelerate classical computing workloads, these servers, equipped with the latest NVIDIA accelerated computing, achieved No. 1 rankings in tasks like image classification, object detection and 3D medical imaging. HPE also secured top industry positions in text generation and summarization benchmarks.

Enabling breakthroughs across disciplines

Supercomputers and industry-leading AI clusters are powerful systems designed to address some of the world’s most complex challenges and HPE is proud to play a critical role in helping these organizations make scientific breakthroughs, solve engineering problems and change the way people live. HPE works alongside research institutions, government agencies and private enterprises to deliver highly-performant and energy-efficient supercomputing solutions and AI systems that address some of society’s most pressing concerns.