Recently, AMD announced an expanded partnership with Hewlett Packard Enterprise (HPE), aiming to accelerate the development of the next generation of open and scalable artificial intelligence infrastructure. This collaboration will be based on AMD's "Helios" architecture, an open-stack platform designed for large-scale AI workloads. HPE will become one of the first system providers to adopt the "Helios" architecture, and will also integrate HPE's custom Juniper network switches developed in collaboration with Broadcom, ensuring high bandwidth and low latency connections between large-scale AI clusters.
Image source note: The image is AI-generated, and the image licensing service provider is Midjourney.
The "Helios" platform will integrate AMD's EPYC™ processors, Instinct™ GPUs, Pensando™ advanced networking technologies, and the ROCm™ open software stack, providing a unified platform that optimizes performance, efficiency, and scalability. This system is designed to simplify the deployment of large-scale AI clusters, helping users accelerate solution delivery in research, cloud computing, and enterprise environments, and improve infrastructure flexibility.
Dr. Lisa Su, Chairman and CEO of AMD, stated that HPE, as a long-term partner, has played an important role in high-performance computing. Through the "Helios" platform, both companies will further strengthen their collaboration, driving high performance and breakthrough achievements in the AI era.
Anthony Neri, CEO of HPE, pointed out that over the past decade, HPE and AMD have continuously pushed the boundaries of supercomputing, launching multiple supercomputing-level systems and advocating for open standards to accelerate innovation. The newly launched "Helios" architecture combined with HPE's customized network solutions will bring faster deployment, more flexibility, and lower risk for cloud service providers, helping them scale in the field of AI computing.
The "Helios" platform can provide up to 2.9 exaFLOPS of FP4 performance per rack, supporting AMD Instinct MI455X GPU, next-generation AMD EPYC "Venice" processors, and Pensando Vulcano network interface cards, suitable for AI and high-performance computing (HPC) workloads. This platform is designed based on the Open Compute Project (OCP), helping customers simplify the deployment process and offering scalable and flexible solutions.
In addition, HPE will launch a new supercomputer "Herder" for the High-Performance Computing Center at Stuttgart University (HLRS). It will be equipped with AMD Instinct™ MI430X GPUs and the next-generation AMD EPYC™ "Venice" processors, and is expected to be put into use by the end of 2027. "Herder" will become HLRS's flagship supercomputer, supporting scientific discovery and industrial innovation.
Key Points:
🌟 HPE will be one of the first system providers to adopt AMD's "Helios" architecture, promoting the development of an open AI platform.
⚡ The "Helios" platform can achieve up to 2.9 exaFLOPS of computing performance per rack, meeting the needs of large-scale AI.
🚀 The new supercomputer "Herder" will be put into use in 2027, supporting high-performance computing and AI innovation in Europe.
