Editorial

With this second newsletter for 2022, it is time to look back at what has been achieved during the year. While working conditions have improved significantly, thanks to the changing pandemic situation, it has been a challenging year for PDC on various levels.

The ongoing installation of new hardware resources has been a major focus this year. In fact, the last hardware components for the Dardel high-performance computing (HPC) system are arriving at the time of writing. In recent months, there have been many changes in the system during which the team at PDC worked (often at very late hours) to keep the already deployed CPU partition of Dardel running and as stable as possible, despite various issues arising that were outside PDC’s control. Although it has not been possible to continuously maintain the target level of availability as a result of these disruptions, we are confident of resuming this level of service once the full upgraded system has passed all the acceptance tests during the first quarter of 2023. A very notable addition to the system was the graphics processing unit (GPU) partition, which pushed Dardel to 68th position on the Top500 list, a list that ranks the systems according to the throughput of floating-point operations while running the high-performance Linpack benchmark (see Dardel Fastest in Sweden ). Even more notable is that Dardel, with the new GPU partition, is now at position 5 in the Green500 list, which measures power efficiency by dividing the throughput of floating-point operations measured for the Top500 list by the amount of power that is consumed. The significance and importance of operating Dardel as one of the most power-efficient systems in the world are obvious, particularly given the significant increase in electricity prices. In addition, progress is being made on establishing Dardel’s little sister, namely the Dardel Cloud, which will be available for users soon (see Building the Dardel Cloud ). This resource will open up new usage models with researchers being able to deploy virtual servers with a lot of freedom regarding how to configure them. Also, data sharing according to FAIR principles (that is, with data being findable, accessible, interoperable and reusable) will become easier with a new object store.

With all of this new infrastructure which is based on the most recent technologies, support for researchers and engineers who use these infrastructures has to grow to keep pace with the technological hardware developments. The upcoming changes in the Swedish HPC landscape – due to the transition from the Swedish National Infrastructure for Computing (SNIC) to the National Academic Infrastructure for Supercomputing in Sweden (NAISS) at the start of next year – provide a good opportunity to think about leaping forward in respect to how support is provided for developing and using research software for HPC (see Supporting Research Software Development ). Such support will be provided to specific research communities through several new EuroHPC Centres of Excellence (see Successful Applications for EuroHPC CoEs ), and support is also available through ENCCS and SNIC/NAISS. Throughout this year, PDC continued both developing software for researchers, for example, in the area of brain research (see Describing Connectivity in Neuronal Network Models ), as well as supporting research communities through community training workshops and summer schools (see UppASD for Simulating Atomistic Spin-Dynamics  and BioExcel Summer School Live Again! ). A particular priority for next year will be to help researchers leverage the power of the new Dardel GPU partition.

PDC will also continue to strongly engage in broader training and education efforts. For instance, the CodeRefinery training events (see CodeRefinery Continues ), the series of PDC Summer Schools (see PDC Summer School 2022 ), and the series of introductory workshops (see Introduction to PDC Systems Workshop ) are being continued. We are happy to be maintaining close cooperation with the EuroHPC-funded Competence Centre ENCCS (see ENCCS Enters Its Second Phase ), for instance, with a joint GPU hackathon in March 2023. With ENCCS, PDC also shares the intention and ambition to encourage and facilitate wider business use of HPC (see Presenting Dardel for Businesses ).

This editorial began by describing the significant world-class upgrades to the infrastructure at PDC. These were made possible by venturing to take advantage of the most modern cutting-edge technology. To stay at the forefront of technological developments, it is crucial for KTH to be involved in HPC technology projects funded by EuroHPC (see Programming Models and I/O with DEEP-SEA and IO-SEA ), as well as being involved in the realisation of new HPC usage models like digital twins (see Paving the Path for Digital Twins in HPC ).

Let me finish with big thanks to the PDC team, which can be proud of all its achievements. PDC is currently undergoing a major review of its internal structures, and the interviews with all staff members confirmed that PDC is home to a diverse set of highly talented people. This is a strength of the organisation, on which we can build with confidence when we reconvene in 2023.

For the break at the turn of the year, I wish you all a well-deserved peaceful vacation, and I hope all of you will enjoy a good start to the new year and success in 2023.

Dirk Pleiter, Director PDC