Wissensspritze on October 28, 2011 at 12.00 with Christian Federspiel and Michael Aspetsberger
The rise of graphic processors units (GPUs) in high performance computing started in the children’s room. Due to their extreme parallel design GPUs entered the field of scientific simulations and settle now down in real time applications.
CPU based systems have a performance increase of about 100 within one century due to Moore’s Law, yet hybrid CPU – GPU based systems face a performance increase of about 1.000 within one century due to energy- and cost-efficient scalability.
In this wispri we show you how to make use of multiple GPUs and many thousand processing cores to get on the fast lane.