Regensburg 2013 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

TUT: Tutorials

TUT 3: GPU Computing (SKM)

TUT 3.1: Tutorial

Sunday, March 10, 2013, 16:00–17:00, H4

High-performance computational physics on graphics processing units — •Tobias Kramer — Universität Regensburg, Germany

Single core computers are not gaining significant speed anymore and the future compute techniques require to program massively parallel processors. Graphics processing units (GPUs) consist of hundreds of processors and have the potential to speed-up computational physics codes tremendously.

To take advantage of GPUs requires to adopt the parallel programming paradigm from the beginning. The GPU tutorial gives first an overview over the GPU architecture, programming environments, basic algorithms. The second part focuses on applications to physical models, including coherent transport, interaction effects, and excitonic energy transfer in light-harvesting complexes.

Source code for some examples is available at

http://quantumdynamics.wordpress.com