Supercomputers are expensive, but Hewlett Packard Enterprise has announced plans to make supercomputing accessible to more businesses by offering scaled-down, more affordable versions of its Cray supercomputers.
The new product portfolio includes HPE Cray EX and HPE Cray XD supercomputers, which are based on Oak Ridge National Laboratory’s Frontier exascale supercomputer. The servers come with a full range of hardware including compute, accelerated computing, interconnect, storage, software and flexible power and cooling options.
The first system is the HPE Cray EX2500, which has the same architecture as the HPE Cray EX4000 supercomputer used to build Frontier, but is 24 percent smaller to fit into enterprise data centers. Gerald Kleyn, HPE’s vice president of HPC/MCS/Edge hardware systems and silicon engineering, said the notable reduction is the reduction in cooling systems, not by using less hardware. The EX2500 uses the same switches and blades as the EX4000 used in Frontier.
“HPC customers have a specific budget that they’re trying to fit into, and they’re trying to get the most performance within that budget. So this allows us to go into spaces that have budget constraints that we couldn’t go into before with supercomputers,” Kleyn said.
The EX2500 is water-cooled rather than air-cooled, and HPE integrated the coolant distribution unit (CDU) into the server cabinet instead of having a separate cabinet like the EX4000 model. Not only does it save space, it also saves money. Kleyn said the HPE Cray EX2500 supercomputer costs only a fraction of the HPE Cray EX4000 supercomputer.
The next system is the HPE Cray XD2000, a 2U platform that fits into a standard 19-inch data center rack and is a descendant of HPE’s Apollo 2000 series. This server is either air-cooled or liquid-cooled, and the main reason you would use the XD2000 instead of the EX2500 is if your data center has the infrastructure for water cooling. Since the EX2500 only has water cooling, companies that insist on air cooling will use the XD2000.
HPE says it will offer compute blades for the XD2000 and EX2500, with both Intel’s fourth-generation Xeon Scalable (aka Sapphire Rapids) and AMD’s Epyc 4 Genoa CPUs.
The XD6500 is a 5U platform with up to eight Nvidia H100 Hopper GPUs, making it similar to Nvidia’s DGX servers. The XD6500 is suitable for advanced workloads including modeling, simulation and AI. Like the XD2000, it’s a rebrand of an older Apollo product.
With eight Hopper processors, it’s no surprise that the XD6500 is aimed at the AI market and machine learning developer environments to help data scientists build these AI workloads.
In addition to the hardware, the supercomputers feature HPE Slingshot, a high-speed Ethernet interconnect, the Cray Clusterstor E1000, which provides expanded storage through intelligent tiering, and the HPE Cray programming environment, which provides a fully integrated software suite with compilers and development tools. .
Kleyn said Cray servers are a step above newly released ProLiant servers because of liquid cooling. Liquid cooling allows for a denser server footprint, “which means you don’t have as many fiber optic cables running through the data center to connect all of them together. So you can bring them closer together, at a lower cost.” Get higher performance,” he said.
Cray XD and EX servers powered by AMD Epyc processors are now available, and Intel Xeon Scalable is slowly entering the market as Intel ramps up production. They can be purchased directly or through the GreenLake consumption model.