AMD and HPE Expand Collaboration to Advance Open Rack-Scale AI Infra

News Highlights

• HPE might be one of many first OEMs to undertake the AMD “Helios” structure, an open, full-stack AI platform engineered for large-scale AI workloads.
• HPE will leverage purpose-built HPE Juniper Networking switches in “Helios” in collaboration with Broadcom, to ship high-bandwidth, low-latency connectivity throughout large AI clusters.
• AMD and HPE to energy “Herder,” supercomputer constructed on the HPE Cray GX5000 platform with AMD Instinct MI430X GPUs and subsequent era AMD EPYC “Venice” CPUs, advancing HPC and Sovereign AI analysis throughout Europe.

Today, AMD (NASDAQ: AMD) introduced an expanded collaboration with HPE to speed up the subsequent era of open, scalable AI infrastructure constructed on AMD management compute applied sciences. HPE will grow to be one of many first system suppliers to undertake the AMD “Helios” rack-scale AI structure, which can combine a purpose-built HPE Juniper Networking scale-up swap – in collaboration with Broadcom – and software program for seamless, high-bandwidth connectivity over Ethernet.

“Helios” combines AMD EPYC CPUs, AMD Instinct GPUs, AMD Pensando superior networking and the AMD ROCm open software program stack to ship a cohesive platform optimized for efficiency, effectivity, and scalability. The system is engineered to simplify deployment of large-scale AI clusters, enabling sooner time to answer and larger infrastructure flexibility throughout analysis, cloud, and enterprise environments.

“HPE has been an distinctive long-term associate to AMD, working with us to redefine what is feasible in high-performance computing,” stated Dr. Lisa Su, Chair and CEO, AMD. “With ‘Helios’, we’re taking that collaboration additional, bringing collectively the complete stack of AMD compute applied sciences and HPE’s system innovation to ship an open, rack-scale AI platform that drives new ranges of effectivity, scalability, and breakthrough efficiency for our prospects within the AI period.”

“For greater than a decade, HPE and AMD have pushed the boundaries of supercomputing, delivering a number of exascale-class programs and championing open requirements that speed up innovation,” stated Antonio Neri, president and CEO at HPE. “With the introduction of the brand new AMD ‘Helios’ and our purpose-built HPE scale-up networking answer, we’re offering our cloud service supplier prospects with sooner deployments, larger flexibility, and decreased threat in how they scale AI computing of their companies.”

Advancing the Next Era of Industry Standard AI Infrastructure
The AMD “Helios” rack-scale AI platform delivers up to 2.9 exaFLOPS of FP4 efficiency per rack utilizing AMD Instinct MI455X GPUs, next-generation AMD EPYC “Venice” CPUs and AMD Pensando Vulcano NICs for scale-out networking, all unified by way of the open ROCm software program ecosystem that permits flexibility and innovation throughout AI and HPC workloads.

Built on the OCP Open Rack Wide design, “Helios” might help prospects and companions streamline deployment timelines and ship a scalable, versatile answer for demanding AI workloads.

This has enabled HPE to combine differentiated applied sciences for his or her prospects, particularly a scale-up Ethernet swap and software program designed for “Helios.” Developed in collaboration with Broadcom, the swap delivers optimized efficiency for AI workloads utilizing the Ultra Accelerator Link over Ethernet (UALoE) customary, reinforcing the AMD dedication to open, standards-based applied sciences.

HPE will provide the AMD “Helios” AI Rack-Scale Architecture worldwide in 2026.

Powering Europe’s Next Generation of HPC and AI Innovation

Herder, a brand new supercomputer for the High-Performance Computing Center Stuttgart (HLRS) in Germany, is powered by AMD Instinct MI430X GPUs and next-generation AMD EPYC “Venice” CPUs. Built on the HPE Cray Supercomputing GX5000 platform, Herder will provide world-class efficiency and effectivity for HPC and AI workloads at scale. The mixture of the AMD management compute portfolio with HPE’s confirmed system design will create a robust new device for sovereign scientific discovery and industrial innovation for European researchers and enterprises.

“The pairing of AMD Instinct MI430X GPUs and EPYC processors inside HPE’s GX5000 platform is an ideal answer for us at HLRS,” stated Prof. Dr. Michael Resch, Director of HLRS. “Our scientific person group requires that we proceed to assist conventional purposes of HPC for numerical simulation. At the identical time, we’re seeing rising curiosity in machine studying and synthetic intelligence. Herder’s system structure will allow us to assist each of those approaches, whereas additionally giving our customers the power to develop and profit from new sorts of hybrid HPC/AI workflows. This platform won’t solely make it doable for our customers to run bigger, extra highly effective simulations that lead to thrilling scientific discoveries, but additionally to develop extra environment friendly computational strategies which might be solely possible with the capabilities that such next-generation {hardware} provides.”

Delivery of Herder is scheduled for the second half of 2027 and it’s anticipated to go into service by the top of 2027. Herder will substitute HLRS’s present flagship supercomputer, known as Hunter.

The publish AMD and HPE Expand Collaboration to Advance Open Rack-Scale AI Infra first appeared on AI-Tech Park.