This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| TOP500 supercomputers | |
|---|---|
| Name | TOP500 supercomputers |
| Established | 1993 |
| Frequency | Semiannual |
| Domain | High-performance computing |
| Publisher | Gesellschaft für Wissenschaftliche Datenverarbeitung mbH Göttingen (GWDG), University of Mannheim |
TOP500 supercomputers
The TOP500 list is a semiannual compilation that ranks the world's most powerful supercomputers by performance on the LINPACK benchmark, used by institutions such as Lawrence Livermore National Laboratory, Oak Ridge National Laboratory, Los Alamos National Laboratory, Argonne National Laboratory, and National Energy Research Scientific Computing Center. The list is maintained by organizations with ties to University of Mannheim, Technische Universität München, Göttingen, and corporate partners including IBM, Cray (company), Hewlett Packard Enterprise, NVIDIA Corporation, and Intel Corporation. It is widely cited by agencies such as Department of Energy (United States), European Commission, Ministry of Science and Technology (China), and firms like Google LLC and Microsoft.
The TOP500 ranks installations at institutions such as Lawrence Berkeley National Laboratory, CERN, Riken, Fujitsu, and Tianhe centers by peak floating-point throughput measured in FLOPS using the LINPACK Benchmark developed at Dongarra's group, with results reported alongside system details from vendors including Dell Technologies, Lenovo Group, Fujitsu Limited, Huawei, and Sugon. Metrics reported often include sustained performance, theoretical peak, processor types such as Intel Xeon, AMD EPYC, ARM variants, and accelerators from NVIDIA Tesla, AMD Instinct, and custom designs like Fugaku's ARM A64FX. The list documents trends in deployments at research centers like Max Planck Society, Fraunhofer Society, Los Alamos National Laboratory, and national labs across United States, China, Japan, Germany, and France.
The project began in 1993 as a collaboration among experts at Lawrence Berkeley National Laboratory, University of Mannheim, and Göttingen to track supercomputing capability following milestones by systems such as Cray-1, Thinking Machines CM-5, and machines from IBM and Hitachi. Throughout the 1990s and 2000s the list reflected transitions from vector processors like NEC SX-series to massively parallel systems exemplified by K computer, Blue Gene, and clusters built with Intel Itanium and AMD Opteron processors. National programs by Japan Science and Technology Agency, National Natural Science Foundation of China, European High Performance Computing Joint Undertaking, and US National Nuclear Security Administration influenced procurement cycles and designs, with major shifts when accelerators from NVIDIA Corporation and custom chips from Fujitsu Limited appeared.
Rankings are determined by the LINPACK Benchmark performance submitted and validated by submitters, a methodology rooted in numerical linear algebra research led by Jack Dongarra and colleagues at University of Tennessee Knoxville and collaborators at University of Manchester. Submissions include system topology, interconnects such as InfiniBand, Omni-Path, Cray Aries, or proprietary networks like Slingshot (network), processor counts, memory, and storage details (e.g., Intel Optane, NVMe SSDs). The methodology has been augmented with alternative benchmarks and lists such as Green500 for energy efficiency and HPCG for sparse system performance promoted by organizations like Argonne National Laboratory and Lawrence Livermore National Laboratory.
Top-ranked systems illustrate architectures from vector processors (e.g., NEC Corporation designs) to massively parallel clusters using CPUs like Intel Xeon Phi or AMD EPYC and accelerators from NVIDIA Ampere and AMD Radeon Instinct. Interconnect technologies include InfiniBand, Intel Omni-Path, Cray Aries, and proprietary silicon interposers from vendors such as HPE Cray Systems. Storage and parallel file systems from IBM Spectrum Scale, Lustre (software), and Ceph support workflows for facilities like European Organization for Nuclear Research and Lawrence Livermore National Laboratory. Cooling innovations feature liquid cooling adopted by vendors including Fujitsu Limited, HPE, and specialized integrators like Asperitas.
Notable milestones include systems such as Fugaku (Fujitsu/RIKEN) achieving exascale-class performance on LINPACK, Summit (supercomputer) at Oak Ridge setting records with NVIDIA Tesla V100 GPUs, Sierra (supercomputer) at Lawrence Livermore used for simulations supporting National Nuclear Security Administration, and China's Sunway TaihuLight produced by National Research Center of Parallel Computer Engineering & Technology. Earlier record-holders include K computer by Riken, Blue Gene/Q installations at Lawrence Livermore National Laboratory, and IBM systems such as Sequoia (supercomputer). Entries often win awards at conferences like SC Conference and are profiled by publications such as Communications of the ACM and IEEE Spectrum.
High-ranking installations enable scientific programs at institutions like CERN, NASA, European Space Agency, National Oceanic and Atmospheric Administration, and universities such as Massachusetts Institute of Technology, Stanford University, and University of Cambridge. Applications include climate modeling for the Intergovernmental Panel on Climate Change, fusion research at ITER, drug discovery in collaboration with Pfizer, astrophysics simulations at Harvard–Smithsonian Center for Astrophysics, and machine learning workloads for companies such as DeepMind and OpenAI. National initiatives by agencies like DARPA and European Commission shape investments in exascale and post-exascale systems.
Critics from research groups at University of California, Berkeley, Princeton University, and think tanks such as Brookings Institution have argued that reliance on LINPACK favors architectures optimized for dense linear algebra and does not reflect real-world workloads common in industry players like Amazon Web Services and Google Cloud Platform. Debates involve procurement priorities influenced by vendors including IBM, HPE, NVIDIA Corporation, and national policy decisions by ministries such as Ministry of Industry and Information Technology (China) and United States Department of Energy. Issues around sustainability and energy consumption highlighted by Greenpeace and energy agencies have prompted attention to listings like Green500 and prompted research at institutions like Lawrence Berkeley National Laboratory into energy-efficient designs.
Category:Supercomputing