Robert Strzodka and PPAM

Robert Strzodka is a researcher specialising in parallel computing and the use of graphics processing units (GPUs). He has been involved in the PPAM conference for many years, particularly in developing its GPU computing programme. What has this work involved, and what else is worth knowing about his scientific career?

Who is Robert Strzodka?

GPU-based scientific computing has been a central theme of Robert Strzodka’s research. Among the early researchers to explore scientific applications of graphics processors, he investigates how mathematical methods, algorithms and computer architectures can be designed together to improve the performance of parallel computation. Since 2015, he has held the Chair of Application Specific Computing at Heidelberg University in Germany.

His earlier career included a visiting assistant professorship at Stanford University in the United States, leadership of the Integrative Scientific Computing group at the Max Planck Institute for Informatics, and responsibility for the AmgX team at NVIDIA. He obtained his PhD in numerical mathematics from the University of Duisburg-Essen in 2004.

Strzodka and PPAM: the early years

At the 2009 International Conference on Parallel Processing and Applied Mathematics (PPAM), Strzodka joined Dominik Behr of AMD and Dominik Göddeke, then at the University of Dortmund, in leading a full-day, four-part GPU computing tutorial. The sessions covered GPU architecture, scientific applications, GPU clusters and OpenCL programming, including practical demonstrations.

At the same conference, Strzodka co-organised the GPU Computing minisymposium with José R. Herrero and Enrique S. Quintana-Ortí. It invited contributions exploring the use of graphics processors as additional computing devices in general-purpose applications.

Further tutorials on GPU computing

Strzodka continued to help deliver GPU tutorials at later PPAM editions. At PPAM 2015, he was one of four presenters of the six-hour tutorial “Scientific Computing with GPUs”. The programme covered CUDA programming, performance and memory optimisation, followed by practical exercises in which participants tested different ways to improve GPU performance.

Two years later, Strzodka and Göddeke delivered another edition of “Scientific Computing with GPUs”. Topics included asynchronous execution between the host and GPU, memory optimisation, multi-GPU computing and combining GPU processing with MPI. Strzodka also co-chaired the conference’s GPU Computing track.

In 2019, he worked with Göddeke and Manuel Ujaldón on the tutorial “Modern GPU Computing”. Its programme covered the Pascal, Volta and Turing GPU architectures, unified memory and tensor operations.

Between 2007 and 2019, GPU computing tutorials featured at seven consecutive PPAM conferences.

Invited talks and scientific sessions

Strzodka’s involvement extended beyond tutorials. At PPAM 2011, he delivered an invited lecture entitled “Accurate High Performance Multigrid Solvers on GPUs”. He also co-chaired sessions of the GPU Computing minisymposium, at which researchers presented work on various applications of the technology.

At PPAM 2022, Strzodka co-organised the “Special Session on GPU Computing” with José R. Herrero and Enrique S. Quintana-Ortí. The session examined practical GPU programming, including the trade-offs between higher-level interfaces and hand-written implementations, as well as tools and libraries for adapting applications from single-GPU to multi-GPU systems.

Strzodka was also a member of the PPAM Programme Committee for the 2022 edition, a role he continued to hold in 2024 and 2026.

Robert Strzodka’s contribution to PPAM

Across successive PPAM editions, Robert Strzodka delivered tutorials, helped organise minisymposia, chaired scientific sessions and participated in the work of the Programme Committee. Of these activities, his most sustained contribution was to the development of the conference programme devoted to GPU computing.