HPE has announced a new expansion of its family of Cray supercomputers with three new equipment from the HPE Cray EX and HPE Cray XD ranges. Both are built on exascale innovation from HPE, providing significant performance and AI capabilities at scale to tackle data-intensive workloads. Also to promote AI and machine learning initiatives, as well as to accelerate the development of products and services.
the new supercomputer HPE Cray EX2500 It features the same architecture as the HPE Cray EX4000, making it the first exascale-class system but 24% smaller in size, making it easy to integrate into an enterprise data center.
It incorporates a 100% direct liquid cooling system, which allows it to improve its energy efficiency, and represents a more cost-efficient solution for companies that want greater performance at scale but with a lower carbon footprint than other systems, and that have a small space. In addition, it will be compatible with 4th generation AMD APYC processors and 4th generation Intel Xeon Scalable processors.
The other two supercomputers presented, the HPE Cray XD2000 and the XD6500 they are high-density servers built on a design that integrates the HPE and Cray portfolios to deliver maximum performance for advanced workloads. Like modeling, simulation and Artificial Intelligence.
Both are compatible with traditional enterprise data centers, offering a wide range of options to customize the technologies. These include selection options for CPU, accelerators, storage, interconnect, and power and cooling options. All can be customized depending on the workloads that the systems have to face. The HPE Cray DX6500 will be compatible with 4th generation Intel Xeon Scalable processors and Nvidia H100 Tensor Core GPUs.
These three new supercomputers will have the technology HPE Slingshot, an Ethernet interconnection intended to ease the demands for high speed and improved congestion control, and to make complex applications run more smoothly. In addition, they incorporate Cray Clusterstore E1000to offer intelligently tiered storage that supports data-intensive workloads such as training AI models.
They also have the HPE Cray programming environment, which provides a set of integrated software with compilers and development tools that allow code portability. In this way, developers can execute code where they need it. It can be used to optimize modeling, simulation, analysis and Artificial Intelligence applications.
All new Cray supercomputers are ready to enable Artificial Intelligence at scale, and can be equipped with HPE Machine Learning Development Environment, a machine learning software platform that enables you to build and train more powerful AI models faster. In addition, they will be compatible with the latest CPUs, GPUs and accelerators, giving them advanced and accelerated computing capabilities to tackle demanding workloads.
HPE will also be the first to deploy a client system using fourth-generation Xeon Scalable chips for the US Department of Energy’s Los Alamos National Laboratory. will be the system cross roadsand will support the research and development tasks of the National Nuclear Security Administration (NNSA) of the United States, to address the critical modeling and simulation, in high resolution 3D, of nuclear weapons, with the aim of guaranteeing their reliability and security.
HPE Cray EX2500 Supercomputers are now available to select customers. Those with 4th generation Xeon Scalable processors will be available to all in Q1 2023, and those with 4th generation AMD EPYE chips will arrive for all customers in Q2 2023.
As for the HPE Cray XD2000, if they carry the fourth generation AMD EPYC processor with air cooling, they are already available. Those that carry this same chip, but have direct liquid cooling systems, will arrive from December 5. As for those with fourth generation Intel Xeon Scalable, they will arrive in the second quarter of 2023, and the HPE Cray XD6500 with said chips will arrive in the first quarter of next year. In that same quarter, the HPE Cray XD6500 with Nvidia H100 Tensor Core GPU will arrive.
