Parallel Computing

What it is about

You learn the basics and practices of parallel computing on multi-core CPUs, GPUs, and in distributed systems. In the end you will be able to assess performance bottlenecks, design and implement parallel algorithms (OpenMP, CUDA/HIP/SYCL, MPI/PGAS) and analyse performance using models such as Roofline.

What you will be able to do

• Gustafson’s Law, Amdahl’s Law and Dennard Scaling understand and apply
• create a Roofline model and evaluate the performance of an algorithm within it
• work with distributed and shared memory architectures
• apply speedup, efficiency as well as strong and weak scaling
• implement parallel algorithms with OpenMP
• implement parallel algorithms for GPUs (CUDA/HIP/SYCL)
• understand and apply distributed memory concepts such as MPI/PGAS

What the module consists of

• VorlesungPresentation of concepts and keynote lectures
• Interaktive ÜbungenApplication of concepts to practical examples
• Hausaufgabenfive extensive tasks for independent deepening
• Projektarbeitindividual project to apply and deepen what has been learned with written documentation and presentation

Teaching method

• ProfessorvortragConveying concepts and theory
• Hands‑on Übungenpractical application of concepts to real problems
• Projektbasiertes Lernenown projects for deepening and applying the course content
• Hausaufgabenindependent solving of challenging questions and subsequent presentation