Numerical GPGPU


Introduction to Numerical General Purpose GPU Computing with NVIDIA CUDA

(compact course)

Course rate

Twice a year (summer and winter).

Course audience

Undergraduate (possible) and graduate (recommended).


This introductory compact course will be a hands-on tutorial on how to access the tremendous compute capabilities of modern Graphics Processing Units (GPUs) to accelerate scientific and especially numerical codes. The compact course adresses the general hardware- and programming models which serve as the foundation of using GPUs as floating point arithmetic accelerators especially using NVIDIA's Compute Unified Device Architecture (CUDA) and provides practical insight into first steps as well as basic and some advanced techniques to port numerical code to GPUs and optimise it for this powerful hardware architecture.


Good/advanced programming skills are mandatory. Coding examples throughout the lecture are based on C/C++ with Linux, so participants are expected to be able to read and understand C/C++ code snippets, and work with a Linux command line. Access to capable Linux GPU machines will be provided but private CUDA-capable Laptops can be used too (please make sure your CUDA installation and development environment is running). The lecture will be held in English language.

Current iteration

  • Date: Mo. Oct. 10, 2016 / Tue. Oct. 11, 2016
  • 10am - 12, 1pm - 4
  • Place: CIP Pool, M946 (Math-Tower Vogelpothsweg 87)
  • Lecturers Raphael Münster and Dirk Ribbrock
  • Organizers / Advisors: Markus Geveler, Stefan Turek
  • Language: German

Last iteration

Course date and location (Winter 2015/2016)

  • Date: Fr. 22.1.2016, 10am - 12, 1pm - 4pm
  • Place: CIP Pool, M946 (Math-Tower Vogelpothsweg 87)
  • Lecturer(s): Amin Safi and Raphael Münster
  • Organizers / Advisors: Markus Geveler, Stefan Turek

Course material

Please see the Files section.

Registration and contact

Please write an email containing your name and affiliation with subject REGISTER to: gpu-course( replace me with AT please )