HUN-REN Expands Its Cloud Infrastructure with Next-Generation GPU Servers

The upgraded infrastructure delivers more efficient resource utilisation, supports computationally intensive modelling and data analytics, and facilitates the development and deployment of advanced artificial intelligence applications. Its primary purpose is to support AI-based and other compute-intensive research while significantly expanding the computational resources available to HUN-REN researchers.

The investment will also strengthen the ability of HUN-REN researchers to participate in international scientific collaborations by providing access to cutting-edge computing infrastructure. This will enable them to contribute more effectively to both national and international research projects requiring high-performance computing (HPC) resources.

The HUN-REN cloud service is a federated research cloud infrastructure jointly operated by HUN-REN SZTAKI and the HUN-REN Wigner Research Centre for Physics, meeting European standards. The service is currently used by researchers from 31 research institutes, higher education institutions and national laboratories. Since its launch in 2016, it has supported more than 400 research projects and contributed to the publication of over 400 scientific papers, including more than 100 articles published in Q1-ranked journals.

The two newly commissioned systems—an HPE Cray XD670 ROT H200 CTO and a Lenovo ThinkSystem SR680a V3—represent the latest generation of AI/HPC servers. Each is equipped with two 5th Generation Intel Xeon Scalable processors and supports up to eight NVIDIA H200 SXM5 GPUs, providing more than 1 TB of total HBM3e memory. The systems also feature 400 Gb/s InfiniBand or Ethernet networking, high-density NVMe storage, and architectures optimised for AI, generative AI, large language model (LLM), and high-performance computing (HPC) workloads. In addition, they provide enterprise-grade redundancy and management capabilities while delivering outstanding computational performance.

The deployment of these new GPU servers further expands HUN-REN's computational research capacity and contributes to strengthening the competitiveness of Hungary's research infrastructure. The investment also establishes a solid foundation for future compute-intensive research and ensures that the HUN-REN cloud platform continues to provide a modern, reliable infrastructure for artificial intelligence, data-intensive research, and high-performance computing workloads.

The GPU servers were procured by the HUN-REN Hungarian Research Network as part of the project “Service Development of the HUN-REN Hungarian Research Network” (project ID: 2025-1.3.2-HUN-REN-2025-00001), supported by the National Research, Development and Innovation Office.