Source: Own illustration
Supercomputing (“SC”) is the largest international conference in the research field of high-performance computing. Here, researchers and industry partners from all over the world discuss current topics in high-performance computing. The conference takes place annually in the USA, this year under the motto “I Am HPC.” from 12 November to 17 November in Denver, Colorado. A total of 14,000 participants and 438 exhibitors came together this year. The HPC Group from the IT Centre was also there with six participants.
Paper Sessions and Workshops
The conference itself consists of a diverse programme. In particular, paper sessions and workshops make up the scientific part of the conference (“Technical Programme”): Here, researchers present their accepted publications in talks and answer questions from the audience. This year, the thematic focus was on the application of artificial intelligence to large simulation problems (such as climate and weather simulations) as well as the optimisation and necessary infrastructure for carrying out highly scalable training runs for neural networks, the application of which is now known to the general public through ChatGPT.
Other topics such as the performance and correctness analysis of parallel programmes also attracted a large audience: the HPC group was represented with contributions in the workshop on performance analysis tools and in the workshop on correctness analysis tools. Isa Thärigen, Simon Schwitanski and Prof Matthias Müller presented the main results of their accepted research work in talks.
Informal Dialogue and Exciting Presentations
Birds-of-a-Feather sessions are also part of the conference: Here, researchers can exchange ideas informally with short presentations on a predetermined topic area. Radita Liem was able to present her research work in the field of parallel I/O in one such session. Fabian Orland also presented his research work on coupling HPC simulations with AI as a “Sofatalk” at the ScaDS.AI (Centre for Scalable Data Analytics and Artificial Intelligence) stand.
Source: Own illustration
Radita Liem, Simon Schwitanski, Isa Thärigen, Fabian Orland
The audience’s feedback on the presentations was consistently positive. In addition to the exchange with long-standing research partners, the technical programme also resulted in new research collaborations that will be intensified in the coming weeks and months.
Tutorials, Technology Exhibition and Inspiring Presentations
Tutorials are another part of the programme: here, participants are introduced to a new subject area by leading experts in a half-day or full-day course. For many years, Dr. Christian Terboven, together with partners from other research institutions, has offered a tutorial on OpenMP tasking and the extended functionalities of OpenMP. This year’s tutorial was again very well received by the participants.
In addition to the technical programme, an exhibition of all the major companies and research institutions from the IT sector took place at the same time: Manufacturers of HPC systems, processors, graphics cards and network technology as well as cloud providers and suppliers of software solutions presented their products to a wide audience. Research institutions advertised their current and planned research projects. Here, too, the focus this year was on technologies and applications in the field of artificial intelligence.
Source: Own illustration
Radita Liem
The supporting programme of the conference offered various presentations in the large plenary session: Under the motto “I Am HPC.”, various well-known personalities in the HPC community presented their careers in research. They also discussed how inclusivity can be promoted in the HPC community and is already being practised. A final get-together during a visit to the Denver Art Museum allowed for an informal exchange between the participants over a buffet and drinks.
Successful Participation
This year’s participation in Supercomputing was a complete success. Presentations and discussions in the various sessions made the HPC Group’s research activities visible to a broad target audience. The strengthening and promotion of existing and new research and industry collaborations was another important aspect. After the conference is before the conference: next year’s Supercomputing will take place under the motto “hpc creates” in Atlanta, Georgia.
Source: Own illustration
Isa Thärigen
List of Publications
Interested parties can view the accepted publications from the HPC group at ACM:
An Event Model for Trace-Based Performance Analysis of MPI Partitioned Point-to-Point Communication
Mapping High-Level Concurrency from OpenMP and MPI to ThreadSanitizer Fibers
RMARaceBench: A Microbenchmark Suite to Evaluate Race Detection Tools for RMA Programs
Towards Correctness Checking of MPI Partitioned Communication in MUST
Adding Microbenchmarks with SIMD Data Races to DataRaceBench
Improve and Stabilize the Classification Results of DataRaceBench
Responsible for the content of this article are Simon Schwitanski and Fabian Orland.
