Engineering Mechanics

Proceedings Vol. 17 (2011)

ENGINEERING MECHANICS 2011

Copyright © 2011 Institute of Thermomechanics, Academy of Sciences of the Czech Republic, v.v.i., Prague

 

Fischer C.
pages 135 - 138, full text

In many research fields the numerical problems demand extremely large computational power. As a consequence, researchers are constantly demanding new and more powerful computers. Recently, the graphics chip producer NVIDIA launched the CUDA parallel computing architecture, which enables scientists to utilize the extreme power available on modern and relatively cheap Graphical Processor Units (GPUs). Necessity to change the common approach together with significant difference in the GPU architecture prevents the wide usage of the graphics hardware. A flexible programming interface to GPU has been recently implemented in numerical packages like Mathematica™ or Matlab®. Such an interface does not hide the differences completely, but greatly simplifies the development of the optimized code. The contribution presents simple examples of GPU utilization in a real engineering environment.

Text and facts may be copied and used freely, but credit should be given to these Proceedings.

All papers were reviewed by members of the scientific committee.