Course "Highly Parallel Programming of GPUs"
Responsible Lecturer:
Prof. Dr. Wolfgang E. Nagel
Contact:
Course description:
This course teaches students to develop and tune algorithms for manycore architectures. It focusses mainly on modern graphics cards, which are used in consumer market as well as in High Performance Computers.
Based on the Nvidia CUDA programming model, programming techniques for massively parallel environments are conveyed and trained during the exercises. Moreover, we demonstrate various GPU APIs for parallelization, computer graphics and deep learning and train you to work with remote GPU instances (Cloud, HPC).
The course consists of 14 lectures and 11 exercises, which alternate during the semester.
In the remaining exercises, students will process small group projects, which are offered in cooperation with the partners of the GPU Center of Excellence. These projects provide students with a concrete insight into scientific research while they also deepen what students learned during the lectures and exercises.
The lecture concludes with the presentation of these projects.
Scope:
4 SWS (2 SWS Lecture, 2 SWS Exercise)
Targeted Study Program:
- Diplom-Studiengänge Informatik, Informationssystemtechnik
- Master-Studiengänge Informatik, Medieninformatik
- Master-Studiengang Computational Science and Engineering
- Master-Studiengang Computational Modeling and Simulation
- weitere Interessenten der Studiengänge in Natur- und Ingenieurwissenschaften
Lecture held during:
Winter semester
Recommended Semester(s):
- Diplom-Studiengänge - Hauptstudium
- Master-Studiengänge - 1./3. Semester
Modules:
INF-B-510, INF-B-520, INF-B-530, INF-B-540, INF-BAS7, INF-PM-FOR, INF-VERT5, MA-CSE-35, CMS-VC-ELV1, CMS-VC-ELV2
Study Material and Possible Post-Exam Reviews:
Lecture- und exercises will be uploaded to OPAL. Post exam reviews are announced there as well. Furthermore, we provide exam information at the OPAL course.
