The European processor designer YesPearl has revealed, within the framework of the ISC 2022 High Performance conference, which is taking place in Hamburg, that it has reached two agreements with HPE and with Nvidiafocused on the development of high performance computing (HPC) and exascale systems in Europe.
The agreement it has with HPE involves the joint development of a supercomputing platform combining HPE technology and SiPearl’s Rhea processor, an ARM-based chip with RISC-V controllers, which will not take long to appear and which is expected to be integrated in the next exascale systems. This is not the only purpose of the agreement, since it has also been signed to expand heterogeneous computing options in supercomputing, apart from supporting and accelerating the adoption of exascale systems in Europe.
Among the contributions that HPE will make to the agreement are those related to high-performance networks with its interconnection slingshot, as well as a software compiler and programming stack based on the HPE Cray programming environment. Slingshot, which HPE acquired with the purchase of Cray, is a high-performance networking standard that is compatible with Ethernet and is present in many supercomputers already in operation today. Among them, Frontier, which has achieved the first place in the list of the 500 most powerful systems in the world.
As for the HPE Cray programming environment, it is a development software package that is made up of various compilers, programming languages, and input/output, mathematical, and scientific libraries. It also includes various performance optimization and profiling tools.
Another purpose of this agreement is the development of joint solutions between HPE and SiPearl for the HPE Center of Excellence in Grenoble, which has just deployed a new superior computer with Artificial Intelligence, which they have called Champolion.
As for the agreement signed with Nvidia, also for collaboration, it pursues technical and business development and aims to create joint hardware and software solutions. thus, both will develop a proxy platform for activities that will combine SiPearl processors and Nvidia GPUs. In addition, they will launch joint activities with several research institutions in Europe, on topics ranging from simulation functions in open source to modeling of research-oriented tools.
In addition, the deal is focused on getting Nvidia’s accelerated computing elements, as well as its network product portfolio, to operate smoothly and seamlessly with SiPearl’s Rhea processor, as well as other CPUs it may launch in the future. Both will also work on expanding ARM’s high-performance computing ecosystem.
SiPearl, born from the European Processor Initiative (EPI), has among its goals the development of a new generation of high-level European microprocessors. The Rhea Processor is designed to be your first fruit. Its launch is scheduled for before the end of this year 2022, and it will carry 72 Neoverse V1 64-bit ARM cores, in addition to being compatible with HBM2E and DDR5 memory technologies.