HPE has introduced three new multi-partner and multi-workload blade systems with advanced direct liquid cooling technologies, which become part of its next-generation supercomputing platform. In addition, the new hardware comes with unified HPE Supercomputing Management software and modern high-performance interconnection to help customers achieve productivity at scale in the era of converged AI and HPC workloads.
As highlighted by Trish Damkroger, senior vice president and general manager of AI and HPC Infrastructure Solutions at HPE, “organizations around the world that depend on supercomputing are looking for better computing performance for all their workloads. Our new HPE Cray Supercomputing platform answers customers’ needs for greater performance density in a unified AI and HPC architecture designed to deliver revolutionary results,” concludes the executive.
Higher computing performance
This portfolio expansion continues the launch of the HPE Cray Supercomputing GX5000, a solution designed specifically to enable new advances in supercomputing in the converging era of AI and HPC. The platform is reinforced with the new HPE Cray Supercomputing Storage Systems K3000, the industry’s first factory-integrated system with integrated open source Distributed Asynchronous Object Storage (DAOS) software, to ensure exceptional performance in supercomputing environments.
The HPE Cray Supercomputing Storage Systems K3000 storage system is based on the HPE ProLiant Compute DL360 Gen12 server. This system offers industry-leading compute performance, high memory density with scalability, and ultra-high data transfer rates to run demanding applications, integrating optimally into modern supercomputing ecosystems.
The Center for High Performance Computing at the University of Stuttgart (HLRS) and the Leibniz Supercomputing Center (LRZ) at the Bavarian Academy of Sciences and Humanities have selected the HPE Cray Supercomputing GX5000 as the platform for their next-generation supercomputers. This next-generation supercomputing portfolio is already gaining traction among some of the world’s most powerful systems, including the future Herder supercomputer at HLRS and the Blue Lion supercomputer at LRZ, opening up new scientific horizons.
New multipartner and multiworkload processing blades
Based on 100% direct liquid cooling, these three new blades allow customers to deploy industry-leading performance density for GPUs, such as the next-generation NVIDIA Rubin platform and AMD Instinct MI430X, or top-of-the-range CPUs, such as the next-generation AMD EPYC processors, codenamed “Venice.” The combination of these advances reinforces HPE’s position as a benchmark in scalable supercomputing.
HPE Cray Supercomputing addresses customer needs for greater performance density in a unified AI and HPC architecture
All compute blades feature a choice of four or eight HPE Slingshot 400 gigabits per second (Gbps) endpoints per blade, and the option of two NVMe solid state drives (SSD) per blade. It is possible to combine the three types of processing blades in a single HPE Cray Supercomputing GX5000 compute rack so that users configure the optimal mix of blades for their workloads, even in configurations designed to maximize performance in distributed supercomputing.
HPE Cray Supercomputing GX440n Accelerated Blade
This blade offers a universal compute engine for mixed-precision computing with four NVIDIA Vera CPUs and eight NVIDIA Rubin GPUs. Up to 24 blades can be configured per compute rack.
HPE Cray Supercomputing GX350a Accelerated Blade
This blade model is specifically designed for customers who need a universal compute engine for mixed-precision workloads with AMD CPUs and GPUs. It incorporates a new generation AMD EPYC “Venice” CPU, CPU and four AMD Instinct MI430X GPUs, the new MI400 series accelerator designed for AI and HPC governance. Supports up to 28 blades per computing rack.
HPE Cray Supercomputing GX250 Compute Blade
Designed for next-generation supercomputers, this blade is recommended for customers who want to create a CPU-only partition for double-precision workloads, with eight next-generation AMD EPYC™ “Venice” CPUs. It is possible to configure up to 40 blades per computing rack.
Unified, multitenant and secure management
HPE Supercomputing Management software introduces new capabilities to support multi-tenant, virtualized and container-based environments, offering greater flexibility to deploy converged AI and HPC processing and isolate workloads and user groups when necessary. From application performance optimization to comprehensive deployment and 24/7 support, HPE supercomputing services help users maintain the maximum value from their investment.
The availability of this equipment varies depending on each model. The first to hit the market will be the HPE Cray Supercomputing Storage Systems K3000 with HPE ProLiant Compute servers, which will be available in early 2026. The rest of those mentioned, including the HPE Supercomputing Management software, would follow in early 2027, thus consolidating HPE’s commitment to continuous innovation in the field of supercomputing.
