AMD and HPE Forge Ahead in AI Infrastructure with Helios Rack-Scale Platform and Next-Gen Supercomputer Herder
the landscape of Artificial Intelligence (AI) is rapidly evolving, demanding increasingly powerful and efficient infrastructure.AMD and Hewlett Packard Enterprise (HPE) are responding decisively, announcing a strategic expansion of their long-standing partnership focused on delivering cutting-edge solutions for large-scale AI and High-Performance Computing (HPC). This collaboration centers around two key initiatives: the AMD Helios rack-scale AI architecture and the Herder supercomputer for the High-Performance Computing Center Stuttgart (HLRS) in Germany. These developments signal a notable step towards democratizing access to advanced AI capabilities and accelerating innovation across research, cloud, and enterprise sectors.
Helios: A Cohesive, Open Platform for AI Scalability
AMD and HPE are jointly developing Helios, a purpose-built rack-scale AI platform designed to simplify the deployment and scaling of demanding AI workloads. This isn’t simply a collection of components; it’s a fully integrated system leveraging the strengths of both companies. Helios combines AMD’s powerful compute technologies – EPYC CPUs, Instinct GPUs, Pensando networking, and the open ROCm software stack – with HPE’s system innovation and a purpose-built, high-bandwidth networking solution.
A critical element of Helios is its integration of a scale-up switch, co-developed with Broadcom, utilizing the Ultra Accelerator Link over Ethernet (UALoE) standard. This commitment to open, standards-based technologies is paramount, ensuring interoperability and avoiding vendor lock-in – a key consideration for organizations investing in long-term AI infrastructure.
Key features and benefits of the Helios platform include:
* Exceptional Performance: Helios delivers up to 2.9 exaFLOPS of FP4 performance per rack, powered by AMD Instinct MI455X GPUs and next-generation AMD EPYC Venice CPUs.
* Simplified Deployment: Built on the OCP Open Rack Wide design, Helios streamlines deployment timelines and offers a scalable, flexible architecture.
* Open Ecosystem: The ROCm software ecosystem fosters flexibility and innovation, enabling developers to optimize AI and HPC workloads across a diverse range of applications.
* Enhanced Efficiency: The integrated design optimizes power consumption and resource utilization, reducing operational costs.
* Future-proofing: The commitment to open standards ensures compatibility with future hardware and software advancements.
“With Helios, we’re taking that collaboration further, bringing together the full stack of AMD compute technologies and HPE’s system innovation,” stated Lisa Su, AMD chair and CEO. “This delivers an open, rack-scale AI platform that drives new levels of efficiency, scalability and breakthrough performance for our customers in the AI era.”
Herder: Empowering European Scientific Revelation
Complementing the Helios platform, HPE and AMD are also powering Herder, the next-generation supercomputer for HLRS in Germany.built on the HPE Cray Supercomputing GX5000 platform and equipped with AMD Instinct MI430X GPUs and next-generation AMD EPYC Venice CPUs, Herder is poised to become a cornerstone of European scientific research and industrial innovation.
Herder’s architecture is specifically designed to support both traditional HPC workloads and the burgeoning field of AI, enabling researchers to develop and deploy hybrid HPC/AI workflows. This versatility is crucial for tackling complex challenges that require the combined strengths of both approaches.
Michael Resch, director of HLRS, emphasized the importance of this dual capability: “Our scientific user community requires that we continue to support traditional applications of HPC for numerical simulation. at the same time, we are seeing growing interest in machine learning and artificial intelligence.Herder’s system architecture will enable us to support both of these approaches.”
Strategic Implications and Future Outlook
These announcements underscore a critical trend in the AI infrastructure market: the move towards integrated,rack-scale solutions that prioritize performance,scalability,and openness. HPE and AMD are positioning themselves as leaders in this space, offering a compelling alternative to proprietary systems.
The availability of Helios is slated for 2026, while herder is scheduled for delivery in the second half of 2027, with full operational status expected by the end of that year. These timelines demonstrate a clear commitment to delivering tangible solutions to address the growing demand for AI computing power.
Why This Matters to You:
* For Cloud Service Providers: Helios offers a pathway to faster AI deployments,greater flexibility,and reduced risk in scaling AI computing infrastructure.
* For Researchers: Herder provides access to state-of-the-art computing resources for groundbreaking scientific discovery.
Keep reading