“Supercomputing is on the cusp of a new era – the exascale era,” wrote Pete Ungaro, ex-CEO of Cray and now Senior Vice President and General Manager, HPC and Mission Critical Solutions at HPE, in a blog post . HPE acquired supercomputing specialist Cray in May 2019 for around $ 1.3 billion. Today we stand on the threshold of a new era, said Ungaro. Because the integration of Cray in HPE is practically complete for the employees, the critical business systems, the roadmaps and the brands.
“This is a milestone that we call Business Day One.” The manager speaks of a new era of computing that differs significantly from previous approaches. In future, the focus would no longer be on individual, gigantic supercomputers or specific, unique technologies. In view of the rapid growth in data, it is much more a question of the convergent use of a wide variety of techniques to address new requirements in terms of analytics, AI and machine learning in the context of different applications and workloads.
The new supercomputers combine and integrate a number of different technologies from both companies – from Cray and HPE. The manufacturer wants to offer high-performance computing for a broader range of applications in companies. From the point of view of those responsible for HPE, the requirements are currently primarily about gaining new insights into current business developments, customer relationships or the resilience of supply chains with the help of data analysis. The aim is to integrate different combinations of HPC, analysis and AI workloads – often in real time.
This combination of different workloads is computationally intensive. For this, HPE wants to bring its new supercomputers into play in the future. The heart of the new systems is a software infrastructure, on the basis of which various combinations of bare metal and containerized workloads could run. Brandon Draeger, who is responsible for marketing for the compute products at HPE, stated that this is not possible with the conventional super-computing architectures currently available. This requires a flexible new software infrastructure that enables developers to create a new generation of applications around data in a kind of cloud.
Read how Hewlett Packard Enterprise is repositioning itself
“HPE Cray supercomputers bridge the worlds of supercomputing, the cloud, and the data center by enabling aggregated workloads to run in an as-a-service delivery in a supercomputing environment,” said Draeger. According to HPE, the systems support different processor architectures as well as different accelerator options. The current models are available in two versions. The “HPE Cray EX Supercomputer” runs with liquid cooling, which includes all components in the system, and therefore also functions compactly in space-saving blade configurations. This HPC variant primarily addresses high-performance requirements. The air-cooled “HPE Cray Supercomputers” are based on Apollo’s own compute nodes, which are connected to one another via HPE’s “Slingshot” interconnect technology. In terms of functionality and software, there was no difference between the two model series.