Editorial

Swedish high-performance computing (HPC) is in a transition phase at the moment, and centre stage is taken by the National Academic Infrastructure for Supercomputing in Sweden (NAISS). At the start of this year, NAISS took over responsibility for academic HPC, storage, and data services when the Swedish National Infrastructure for Computing came to a close. This change involves much more than a mere rebranding of the organisation and a relocation of the head office (see Nascent NAISS ). A well-founded sense of optimism has spread throughout Swedish academic institutions and industry alike, and this national optimism was tangible at the EuroHPC Summit 2023 conference, which was organised by the European High Performance Computing Joint Undertaking (EuroHPC JU) and held in Gothenburg in March (see EuroHPC Summit 2023 ).

The Swedish membership of the EuroHPC JU programme is bringing exascale computing to the doorsteps of Swedish researchers, and there is no alternative but for all of us to embrace this paradigm change and leverage it to the best of our abilities. This is where the concept of the HPC ecosystem for research becomes significant: knowledge, competencies, and applications are coming to the fore as being equally as important as hardware for utilising the full potential of the coming exascale HPC research infrastructure as we strive to advance sciences and engineering.

On the European level, a call was launched last year for centres of excellence (CoEs) to be formed and address the software side of the exascale ecosystem. Out of the nine CoEs that were awarded funding, no less than three are led by PDC (see Great Success for KTH & PDC in Innovations Towards Exascale ). This rather overwhelming success is quite obviously rooted in strong scientific research environments, and here a key role is being played by the Swedish e-Science Research Centre (SeRC, e-science.se ), which is led by the KTH Royal Institute of Technology. (The Karolinska Institute, Stockholm University and Linköping University are additional members.) But somewhat less obviously, perhaps, it is important to also acknowledge the long-term and continuous support that PDC has received directly from KTH to build up the competencies and knowledge at PDC that is needed now more than ever before.

In the new NAISS era of Swedish HPC, this strong commitment from KTH will continue, and it will arguably be no less than a decisive factor in realising the necessary component of applications within the Swedish national ecosystem that enables forefront research on our systems. There is now no longer a question of whether accelerators will reach into and affect work in your favourite e-science domain. However, that is not the same as saying we must all learn, or even care, about the often daunting technical details involved with modern parallel programming, some of which can be read about in this newsletter. PDC will take a leading national (and in some fields also international) role in the development of HPC software applications in sciences and engineering. We will do so with the ambition of providing user application interfaces that are less complex and that make it easier for less experienced users to access our increasingly complex hardware resources.

On the topic of NAISS hardware resources, PDC can proudly announce that the Dardel system has now reached its full intended capacity with the GPU and CPU partitions at positions 77 and 153, respectively, on the most recent TOP500 list (see Dardel Updates ). Bringing this heterogeneous system into use has involved – and continues to involve – hard work for the systems and application experts at PDC, but it represents a step that must be taken in order for our community of researchers to remain competitive. In addition, the experiences gained in this process will be invaluable for whatever the future has in store for Swedish HPC.

Patrick Norman, Director PDC