Hewlett Packard Enterprise (NYSE: HPE) has introduced a suite of advanced high-performance computing (HPC) and artificial intelligence (AI) products designed to accelerate scientific research and large-scale AI initiatives.
The new HPE Cray Supercomputing EX solutions and HPE ProLiant Compute XD server family provide cutting-edge compute, networking, and storage capabilities tailored for AI training, natural language processing (NLP), and multi-modal model training.
Transformative Cooling Technology
HPE’s latest offerings include a choice between traditional air cooling and HPE’s 100% fanless direct liquid cooling, an industry first, allowing for efficient and sustainable supercomputing environments.
Leveraging compute and accelerator technology from AMD, Intel, and NVIDIA, HPE's expanded portfolio provides customers with versatile solutions to match both performance demands and budgetary requirements.
<div className='custom-template-1'> </div>Trish Damkroger, HPE’s Senior Vice President and General Manager of HPC & AI Infrastructure Solutions, emphasized the growing demand for high-performance computing to support AI advancements: “Customers rely on us to rapidly deploy high-performance AI systems that drive value and innovation,” she noted.
HPE’s expertise in delivering integrated HPC systems aims to fast-track global initiatives in sovereign AI, scientific discovery, and industrial R&D.
HPE Cray Supercomputing EX
The centerpiece of HPE’s new HPC portfolio, the HPE Cray Supercomputing EX series, sets a new standard in speed, efficiency, and scalability.
Designed for research institutions and government agencies with sovereign AI goals, the HPE Cray EX systems integrate compute, networking, and storage solutions across every layer of supercomputing.
The lineup includes:
-
HPE Cray Supercomputing EX4252 Gen 2 Compute Blade: Featuring up to 98,304 cores in a single cabinet, this one-rack powerhouse is built with eight 5th Gen AMD EPYC processors, maximizing CPU density and performance in compact spaces. The EX4252 Gen 2 is expected in Spring 2025.
-
HPE Cray Supercomputing EX154n Accelerator Blade: Designed to expedite supercomputing workloads, this model can support up to 224 NVIDIA Blackwell GPUs per cabinet, each linked via NVIDIA’s NVLink technology for seamless data flow. The EX154n is anticipated by late 2025.
-
HPE Slingshot Interconnect 400: The latest in HPE’s high-speed interconnect technology, the Slingshot 400 delivers 400-gigabit-per-second speeds, doubling the performance of its predecessor while offering low latency and adaptive routing. The Slingshot 400 will be available on HPE Cray EX systems starting Fall 2025.
-
HPE Cray Supercomputing Storage Systems E2000: Built for high-volume data handling, the E2000 storage system more than doubles input/output (I/O) speeds over its previous generation. Powered by Lustre, an open-source file system, this storage solution minimizes idle time and enhances performance for CPU and GPU workloads. The E2000 will debut in early 2025.
-
HPE Cray Supercomputing User Services Software: Now available, this software helps optimize supercomputing workloads by managing system efficiency and power consumption, and enables flexible deployment of diverse workloads on HPE’s HPC infrastructure.
New HPE ProLiant Compute XD Series
HPE’s ProLiant Compute XD servers extend HPE’s capabilities in AI model training by offering purpose-built systems for rapid AI cluster deployment.
These servers, tailored for both service providers and large organizations, include options for in-depth customization, validation, and testing.
The lineup includes:
HPE ProLiant Compute XD680: With a focus on cost-efficiency, this air-cooled model features eight Intel Gaudi 3 AI accelerators, optimized for intensive AI training and inferencing tasks. The XD680 will be available in December 2024.
- HPE ProLiant Compute XD685: Designed for high performance and energy efficiency
- GPU Support: Supports up to 8 NVIDIA H200 SXM Tensor Core GPUs or Blackwell GPUs
- Cooling Technology: Uses HPE’s advanced liquid-cooling technology
- Launch Timeline: H200 GPU variant will launch in early 2025
- Additional Configurations: Will feature AMD Instinct MI325X accelerators and AMD EPYC CPUs, introduced earlier in October
AI and Supercomputing
By expanding its HPC and AI infrastructure portfolio, HPE aims to meet the evolving needs of customers who leverage AI models to drive scientific discoveries.
The HPE ProLiant Compute XD servers are part of HPE’s comprehensive AI portfolio, which also includes HPE Private Cloud AI and HPE ProLiant Compute DL servers. Attendees at the upcoming SC24 conference in Atlanta can see HPE's innovations firsthand.
