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.
| ARCHER2 | |
|---|---|
| Name | ARCHER2 |
| Country | United Kingdom |
| Operator | UK National Supercomputing Service |
| Manufacturer | HPE Cray |
| Architecture | HPE Cray EX |
| Processors | AMD EPYC |
| Memory | 6 PB (aggregate) |
| Storage | Lustre-based high-performance storage |
| Peak | 28.0 PFLOPS (theoretical) |
| Site | Hartree Centre, Daresbury |
| Online | 2021 |
ARCHER2 ARCHER2 is a United Kingdom national high-performance computing system deployed to support scientific research across physics, climate science, chemistry, biology, and engineering. The system provides cluster-scale compute resources for academic consortia, national laboratories, and industrial partners, enabling simulation, data analysis, and modeling at scale. ARCHER2 is part of a lineage of UK supercomputing installations and integrates with regional research infrastructures and international collaborations.
ARCHER2 replaced an earlier national service and links to institutions such as University of Edinburgh, Science and Technology Facilities Council, Hartree Centre, EPSRC, UK Research and Innovation, and Jisc. The procurement and operation involved vendors and research bodies including HPE Cray, AMD, Dell Technologies, Intel (as competitor context), and service partners like Atos (as historical vendors). ARCHER2 supports research programs funded by European Research Council, Royal Society, Wellcome Trust, and national initiatives such as the National Innovation Centre and interacts with facilities like CERN, Met Office, NERC, and UKAEA.
The system is built on an HPE Cray EX architecture using AMD EPYC processors and a high-speed interconnect influenced by designs from Cray Inc. and modern interconnect vendors. Compute nodes are arranged in cabinets with dense CPU configurations, large aggregate memory, and a parallel file system derived from Lustre deployments common at facilities like Oak Ridge National Laboratory and Lawrence Livermore National Laboratory. Cooling and power provisioning reflect standards used at Daresbury Laboratory and comparable sites such as Jülich Research Centre and Barcelona Supercomputing Center. The system’s topology and network enable large-scale MPI workloads similar to deployments at NERSC and PRACE HPC centers.
ARCHER2’s theoretical peak performance approaches tens of petaflops, aligning it with contemporaneous systems at EPCC and European Tier-1 resources. Benchmark suites such as HPL, HPCG, and application-driven tests used by groups at Imperial College London, University of Cambridge, University of Oxford, and University College London gauge sustained performance for modeling codes from GROMACS, LAMMPS, Quantum ESPRESSO, and CP2K. Comparative studies reference results from Fugaku, Summit, Frontier, and Atlas to contextualize scaling, while profiling tools and performance teams collaborate with projects at STFC and Met Office to optimize throughput for climate, materials, and astrophysics workloads.
ARCHER2 provides a software stack including compilers, libraries, and resource managers consistent with ecosystems at PRACE and national computing centres: vendor compilers from AMD toolchains, open-source toolchains such as GCC, and ecosystem libraries like OpenMPI, MPI, OpenMP, CUDA (as GPU context in broader HPC), and OpenACC referenced in comparative environments. Batch scheduling and workflow tools mirror practices at Slurm Workload Manager deployments used by NERSC and CSC – IT Center for Science. Scientific applications supported include workflows from MOLPRO, ORCA, NAMD, VASP, and community packages utilized by teams at Max Planck Society and CNRS laboratories.
ARCHER2 serves a diverse user base across academia and industry, facilitating projects in climate modelling with partners at the Met Office and Hadley Centre, materials science with groups at Diamond Light Source and STFC, and life sciences collaborations supported by Wellcome Sanger Institute and European Bioinformatics Institute. Use cases span turbulence simulations used by Rolls-Royce and aerospace research at Airbus, computational chemistry for pharmaceutical partners including AstraZeneca, and fusion modeling with Culham Centre for Fusion Energy and UKAEA. The platform underpins national responses to crises, coordinating modeling efforts similar to collaborations among Public Health England and research consortia.
The procurement, funding, and governance for ARCHER2 involved bodies such as EPSRC, UKRI, STFC, Jisc, and the host institutions including University of Edinburgh and Sci-Tech Daresbury. Contracts and partnerships were established with HPE Cray and component suppliers like AMD and storage vendors similar to those supplying Cray installations. Governance frameworks drew on models from PRACE, EU Horizon 2020 consortia, and national research infrastructure programs administered by UKRI and associated councils. Advisory boards included representatives from universities, national labs, and industry stakeholders such as BP and Siemens for domain engagement.
ARCHER2 has accelerated scientific outputs at organizations like Imperial College London, University of Cambridge, University of Oxford, and labs across the UK, contributing to publications in venues such as Nature, Science, Physical Review Letters, and domain journals managed by societies like the Royal Society of Chemistry and Institute of Physics. Future directions involve integration with exascale initiatives, collaboration with EU and international efforts like PRACE and EuroHPC, and potential upgrades incorporating hardware from vendors including NVIDIA (GPU ecosystems), Intel (CPU and accelerator technologies), and next-generation interconnects demonstrated at centers such as TACC and Argonne National Laboratory. Ongoing training and community engagement continue through workshops with EPCC, UKRI training programs, and user support structures modeled on national supercomputing services.