Driving future computing hand in hand

idee

Teams generate more and better ideas and innovations: That's why the LRZ's next supercomputer is being developed together with technology providers in an innovation partnership. Photo: Nejc Soklic/Unsplash


The new system shall contain state-of-the-art technology, components that are not (yet) sold and that offer research new possibilities: For procuring its next supercomputer, the Leibniz Supercomputing Centre (LRZ) is relying on a new process called Innovationspartnerschaft (partnership for innovation). The architecture of the new high-performance computer (HPC) will be developed in a co-design process together with technology partners until 2024. Prototypes will be constructed; existing components adapted to specific needs. "It has been clear since the 2010s that the next generation of HPC systems, the exascale computers, will be co-designed; in the US we’ve seen it for a while, that scientific institutions rely on the joint development of hardware and software," explains Prof. Dr. Dieter Kranzlmüller, director of the LRZ. "We wanted to improve the procurement of our systems, and the partnership for innovation lent itself to this." The funding agencies, the Federal Ministry of Education and Research (BMBF) and the Bavarian State Ministry of Science and the Arts, also gave the green light. And so, the next supercomputer can be planned as the first in Europe via this new process. "Suited to our needs," Kranzlmüller expects, "we can research and evaluate innovative computing approaches together with manufacturers and finally implement them in the ExaMUC."

A vehicle for modernisation and economic promotion

In Germany, partnerships for innovation have been possible since 2016. The state and Länder support this procurement method because it supports the development of new offerings and innovative companies, and also helps modernise the administration and public institutions. "The public sector spends around 500 billion euros per year in Germany," notes Matthias Berg, head of the Competence Centre for Innovative Procurement, KOINNO. "If just one per cent of that goes to new products and services, that's a huge boost to the economy." Berg accompanies partnerships for innovation, most of which are set up for several years and proceed in several phases. At the LRZ, about ten specialists from different areas are supporting the procurement process, which began with the opening of the tender at the beginning of 2022: In several rounds and until mid-April, hardware companies competed on how they would realise the system ideas and technical specifications for the ExaMUC system. By mid-July, the bidding companies are requested to submit official offers with detailed co-design steps to the LRZ. End of July, the contract will finally be in place and the practical development work on the new system will start with the companies. And by 2024, it will be clear which company will build and install the ExaMUC.

"The partnership for innovation enables us to develop and optimise technologies for the next system with several suppliers at the same time," Herbert Huber cites advantages. He has been the head of the High Performance Systems department at the LRZ for ten years and helped already plan and procure several supercomputers. "This allows us to get to know the companies and their way of working well before the supercomputer is delivered." The LRZ's partnership for innovation also incorporates experience gained vir the LRZ Future Computing programme and with the Bavarian Energy, Architecture and Software Testbed (BEAST). The teams for Big Data and AI as well as the Computational X Support (CXS), which supports researchers in optimising algorithms and implementing them in HPC systems, also collaborate in the process. Hopes are high that the interaction of the supercomputer's components can be tested and improved in advance. In addition, the previous start-up difficulties should decrease when the system begins to operate. "When the first technical steps and contracts have been clarified," Huber plans, "we will also involve friendly users in the development phase.

Effort, discussions and risks

Partnerships for innovation are more elaborate than other procurement procedures: "They are worthwhile for products with high investment requirements and for research and development," observes Berg. "However, there is usually only a vague idea of the technology or service to be developed, and the contracts are concluded for research and development work and prototype construction." These touch on copyrights, patent rights and also marketing rights and are therefore drawn up more individually and in greater detail. There are also many precautions to be taken. Among other things, the LRZ must ensure that all companies work with the same information and that no one can learn about the technical innovations of the others. The prototypes of processors, accelerators and other components are therefore installed in different and closed racks. Within the task force, team members have to be cautious handling established personal relationships: "In the more family like HPC community, we are dealing with people and companies that we all know well," Huber says. "In personal conversations, it's easier to casually acquire information that may bring advantages."

supermucng

The SuperMUC-NG is now entering its second phase and will soon need a successor. Photo: V.Hohenegger/LRZ

Innovation expert Berg also recommends not underestimating the time required. "For the individual development steps, the design specifications and especially the test criteria should be carefully defined and elaborated," he says. "As many people as possible should be involved in the discussions so that different perspectives and needs are taken into account." Also, when several companies work on a problem and are rewarded for it, innovation partnerships definitely help to save money. According to experience, savings of up to 30 percent can be made in the IT sector because hardware or software is developed more specifically to meet real needs. Participating companies do run the risk that solutions will not work or will not be selected. In return, however, they can improve and market the products created during the cooperation. It is also not certain whether the plans of the tendering organisations can be realised: "It's quite possible that ideas will fail, but the experience is valuable for everyone involved. Failure is part and parcel of innovation, but the public sector has a hard time with it because it has to account for its spending”, Berg explains and encourages: "I have not yet seen an innovation partnership fail, although the procedure is complex and the risks are high." (vs)