Editorial

As we move towards the end of 2023, the transformation of Swedish high-performance computing (HPC) continues with the implementation of the National Academic Infrastructure for Supercomputing in Sweden (NAISS) and the different university branches within NAISS are finding their roles in this new landscape.

The establishment of NAISS has given a clear mandate to a single organisation and university to take full future control of Swedish hardware investments and operations. The previous model of tearing resources apart and spreading national investments between half a dozen different physical locations was not sustainable. Once the decision was made to change the approach, the NAISS proposal was widely embraced as is evidenced by the large number of universities that have declared themselves willing to pay NAISS membership fees. In this new NAISS era, it appears that the right balance has been found between localisation and delocalisation such that the new organisation embodies a national infrastructure with engaged universities. To implement centralisation for equipment but not application expertise may well turn out to be a potent recipe for HPC success.

Every Swedish national infrastructure, including that of HPC, is expected to be co-financed by the stakeholders to guarantee its necessity when gauged against other interests. The important decision made by NAISS here was to also take a step back from centralisation and let the lion’s share of the universities’ co-funding reside within the control of each of the universities. This is a bold and highly constructive move by NAISS that gives up organisational control of funding in favour of university participation in building an infrastructure that best meets the needs of its users. An adversary might say that was foolish because there is, in principle, no guarantee that the universities will continue funding Swedish HPC at the same levels as before. But I would argue that it is a risk worth taking since we need the universities to be invested and engaged in the question of how their co-funding is spent. Freedom is now given to each university to strategically link its e-science investments to those being made in other research areas at that university and not least in the experimental sciences

KTH is leading the way by expressing its full support for NAISS and continuing its strong financial commitment to the field of HPC. While the investments that are made will (in a sense “selfishly”) reflect the priorities of other research activities at KTH, they will also, for the most part, benefit the national HPC infrastructure. Notably, KTH will promote the notion of the NAISS ecosystem being an entity that, in addition to incorporating hardware and servers, also recognises knowledge, competencies, and software applications as being decisive factors for the advancement of natural sciences and engineering. Scientific software is in itself to be seen as part of the infrastructure, including the development and implementation of novel methods and algorithms. This view represents a paradigm change in Swedish HPC by blurring the boundary between research and infrastructure, and it can be implemented because the funding comes directly from KTH. This enables PDC and other research groups at KTH to participate in long-term efforts in the development of internationally leading scientific domain software that is anchored in strong research environments existing at KTH. It also promotes good software engineering practices and balances the work of Ph.D. students and postdocs, who typically have more short-term goals.

In return, for KTH, the development activities at PDC bring international recognition, industry contacts, and successful research environments that are able to compete for external funding in both the national and international arenas. KTH holds the leading position in the Swedish e-Science Research Centre (SeRC, e-science.se ), which also gathers together the Karolinska Institute, Stockholm University, and Linköping University. SeRC is highlighting the importance of the Research Software Engineers who are the backbone of these types of long-term scientific software development activities (see Research Software Engineer Teams: Organising the Most Advanced Level of User Support at the SeRC Universities ) and it is not a distant thought to use SeRC as an advisory body for the scientific software developments at PDC in the future. On the European level, PDC leads three of the EuroHPC centres of excellence (CoEs) and is a member of a fourth. Such an achievement is not only a token of success for KTH but also brings an international dimension into NAISS.

PDC has three flagship programs: GROMACS, VeloxChem, and Neko in the fields of molecular dynamics, quantum chemistry, and computational fluid dynamics, respectively. The uncompromisable goal for each of these software projects is to develop an open-source program that delivers an HPC performance beyond the state-of-the-art in the respective scientific domains. This exemplifies the national benefit of KTH’s strategic investment in HPC that also formed a cornerstone in the recent Swedish application to become a hosting nation within the EuroHPC JU. This application has been granted approval and, under the lead of Gert Svensson, preparations are being made for the procurement of a system (which will be named after Carl Axel Arrhenius) that will be far more powerful than anything presented before in Swedish HPC.

Patrick Norman, Director PDC