HPE has launched an updated Cray Supercomputing portfolio and new high-performance computing solutions within the ProLiant family.
HPE has launched an expanded portfolio of high-performance computing (HPC) and artificial intelligence (AI) infrastructure solutions, including its HPE Cray Supercomputing EX solutions and new systems optimized for large language model (LLM) training, natural language processing (NLP) and multi-modal model training. These solutions aim to accelerate scientific research and enable innovative breakthroughs for global customers.
HPE Cray Supercomputing EX: Industry-Leading HPC Solutions
HPE continues to lead in providing some of the world’s fastest, most energy-efficient supercomputers, built on the HPE Cray Supercomputing EX platform. This updated portfolio addresses the needs of research institutions and governments undertaking significant AI initiatives. With the industry’s first 100% fanless direct liquid cooling system, HPE’s portfolio supports compute nodes, networking, and storage, complemented by a new software solution.
- HPE Cray Supercomputing EX4252 Gen 2 Compute Blade: Featuring eight 5th Gen AMD EPYC™ processors, this compute blade supports up to 98,304 cores in a single cabinet, providing an exceptional CPU density for high-performing compute needs. It will be available in Spring 2025.
- HPE Cray Supercomputing EX154n Accelerator Blade: This system integrates up to 224 NVIDIA Blackwell GPUs in one cabinet, using the NVIDIA GB200 Grace Blackwell NVL4 Superchip, with four NVIDIA NVLink™-connected Blackwell GPUs and two NVIDIA Grace CPUs. Expected availability is by the end of 2025.
- HPE Slingshot Interconnect 400: Delivering speeds of 400 Gbps, this interconnect system doubles its predecessor’s line speed, featuring automated congestion management and adaptive routing for lower latency, ideal for extensive workload management. Availability for HPE Cray systems begins Fall 2025.
- HPE Cray Supercomputing Storage Systems E2000: This high-performance storage system, based on the Lustre file system, more than doubles I/O performance over previous versions, reducing idle times for both CPU and GPU nodes. It will be available in early 2025.
- HPE Cray Supercomputing User Services Software: Available now, this software enhances the user experience in managing compute-intensive workloads, optimizing system efficiency, regulating power, and supporting diverse workloads on supercomputing infrastructure.
HPE ProLiant Compute XD: Optimized for AI Model Training
HPE’s new ProLiant Compute XD servers are purpose-built to streamline the deployment of large AI clusters, catering to service providers and enterprises training proprietary AI models. HPE ProLiant Compute XD servers benefit from HPE’s robust experience deploying high-performance AI systems, and optional HPE Services support solution customization, integration, and full testing in HPE’s manufacturing facilities.
- HPE ProLiant Compute XD680 Server: Air-cooled and cost-optimized, this server is designed for demanding AI training, featuring a compact chassis with eight Intel® Gaudi® 3 AI accelerators. It will be available in December 2024.
- HPE ProLiant Compute XD685 Server: The XD685 server supports eight NVIDIA H200 SXM Tensor Core or Blackwell GPUs within a five-rack-unit chassis. Liquid cooling capabilities enable efficient thermal management for both CPUs and GPUs. The NVIDIA HGX H200 8-GPU version will be available in early 2025, with versions featuring AMD Instinct™ MI325X accelerators also available.
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | TikTok | RSS Feed
