This five-day on-site block course provides a solid introduction into the architecture and efficient programming of modern parallel high-performance computers. The full range of hardware features is covered - from the single core to the highly parallel level. OpenMP, MPI, and CUDA are introduced as the dominant programming  models for shared and distributed-memory systems and GPUs. Code optimization strategies based on simple performance models will help participants understand the principal bottlenecks of parallel computing. Extensive hands-on exercises are provided to allow attendees to apply the theoretical concepts right away. The programming exercises will be available in Fortran and C.

Attendees will be given access to the HPC cluster at the University of Greifswald Computing Center.

Course prerequisites: Attendees should be familiar with the Linux command line and with handling a Linux system remotely (editing, compiling) via SSH. Basic programming skills are required.  

Lecturer: Dr. Georg Hager, Institute of Physics at the University of Greifswald and Erlangen National High Performance Computing Center (NHR@FAU)

Course dates and times: September 21-27, 2023 (Thu, Fri, Mon, Tue, Wed), 9.00 a.m. - 4:30 p.m.

Location: RZ Uni Greifswald, Felix-Hausdorff-Stra├če 18, 17489 Greifswald

Registration: If you want to attend, please send e-mail to georg.hager@uni-greifswald.de