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.
| HPE Cray EX | |
|---|---|
| Name | HPE Cray EX |
| Manufacturer | Hewlett Packard Enterprise |
| Family | Cray |
| Type | Supercomputer |
| Introduced | 2020s |
| Os | Compute Linux |
| Cpu | AMD EPYC, Intel Xeon (config dependent) |
| Accelerator | NVIDIA GPU, AMD Instinct |
| Interconnect | Slingshot |
| Memory | High-bandwidth memory options |
| Storage | Lustre, ClusterStor |
| Purpose | High-performance computing, AI, simulation |
HPE Cray EX
The HPE Cray EX is a family of high-performance computing systems designed for large-scale scientific simulation, artificial intelligence, and data analytics. It integrates technologies from Hewlett Packard Enterprise, Cray Inc., AMD, NVIDIA, and other suppliers to target national laboratory, university, and enterprise deployments. The design emphasizes scalability, energy efficiency, and low-latency networking for exascale-class workloads.
The platform represents a collaboration between Hewlett Packard Enterprise and the legacy of Cray Inc., aligning with procurement programs at Oak Ridge National Laboratory, Argonne National Laboratory, Lawrence Livermore National Laboratory, and international centers such as Barcelona Supercomputing Center and Forschungszentrum Jülich. HPE marketed the line to participants in initiatives like the Exascale Computing Project and procurement efforts led by United States Department of Energy and other national agencies. The systems aim to serve projects in climate modeling associated with European Centre for Medium-Range Weather Forecasts, astrophysics linked to Max Planck Society, and genomics efforts parallel to Broad Institute initiatives.
The EX line employs modular cabinets combining compute blades, accelerators, memory, and storage controllers. Compute nodes have been offered with processors from AMD EPYC series and options for Intel Xeon families, paired with accelerators such as NVIDIA A100 GPUs and AMD Instinct accelerators. The interconnect fabric is based on the Cray Slingshot architecture developed post-acquisition, designed to reduce tail latency for MPI workloads and integrate congestion control mechanisms used in large deployments at National Energy Research Scientific Computing Center and Deployment Programmes within European High Performance Computing Joint Undertaking. Storage subsystems include scalable parallel filesystems like Lustre and HPE's ClusterStor designs, and node-local persistent memory options compatible with technologies pioneered by Intel Optane initiatives. Cooling strategies reference liquid-cooling heritage from Cray-1 engineering and modern solutions used at Argonne Leadership Computing Facility.
Software stacks for EX systems include vendor-supported distributions such as SUSE Linux Enterprise Server or Red Hat Enterprise Linux derivatives, HPC middleware like OpenMPI, MPICH, and programming models including CUDA, HIP, OpenACC, and standards-driven approaches like OpenMP and MPI. Development toolchains integrate compilers from GNU Compiler Collection, LLVM Project/Clang (compiler), and vendor compilers from AMD and Intel Corporation. Performance analysis and tuning tools referenced include TAU Performance System, HPCToolkit, and vendor suites used at Los Alamos National Laboratory. Resource managers and schedulers common in EX deployments encompass SLURM Workload Manager and integrations with science gateways such as those linked to XSEDE infrastructures.
EX-class systems target scalability for benchmarks like High Performance LINPACK (HPL) and High Performance Conjugate Gradients (HPCG), as well as application benchmarks used in Top500 and Green500 lists. Vendors highlight performance for deep learning workloads through metrics derived from MLPerf and domain-specific kernels used in computational fluid dynamics related to ANSYS workflows. Published performance claims often reference comparisons to predecessors used at Oak Ridge National Laboratory and Lawrence Berkeley National Laboratory, and to architectures featured in procurements by European Centre for Medium-Range Weather Forecasts.
Configurations range from mid-sized clusters suitable for university centers like University of Cambridge or University of Oxford HPC groups to large-scale systems deployed at national labs such as Oak Ridge National Laboratory and Argonne National Laboratory. Variants differ in CPU choice (AMD EPYC vs Intel Xeon), accelerator presence (NVIDIA A100 vs AMD Instinct), and storage scale (Lustre vs ClusterStor). Rack-level options include air-cooled and liquid-cooled enclosures similar to those adopted by Jülich Supercomputing Centre and cabinet designs comparable to installations at National Center for Supercomputing Applications.
EX deployments support use cases across computational science domains: climate and weather modeling used by European Centre for Medium-Range Weather Forecasts and Met Office; fusion simulations linked to ITER research; materials science pursued by groups at Lawrence Berkeley National Laboratory and Max Planck Society; genomics workloads undertaken by consortia like Broad Institute; and AI workloads similar to initiatives at DeepMind and OpenAI research collaborations. Installations have been configured for national research infrastructure programs such as PRACE and regional supercomputing centers like National Computational Infrastructure in Australia. Partnerships for applications and optimization include collaborations with Ansys, Siemens, and national laboratories such as Los Alamos National Laboratory.
The EX family builds on the acquisition of Cray Inc. by Hewlett Packard Enterprise and subsequent integration of design expertise from legacy systems like the Cray XC and networking advances following projects funded by the United States Department of Energy. Development timelines intersect with exascale initiatives such as the Exascale Computing Project and collaborations with national facilities including Oak Ridge National Laboratory and Argonne National Laboratory. Roadmaps for EX systems reflect broader industry shifts toward heterogeneous computing exemplified by partnerships with NVIDIA, AMD, and ecosystem contributors including Intel Corporation and software communities like the OpenACC and MPI standards groups.