2012
EMSS 2012 Proceeding
Using graphic processors for highspeed simulations and other high performance computations
Authors: Thomas Wiedemann
Abstract
Current hardware development is characterized by an increasing number of multi-core processors. The performance advantages of dual and quad core processors have already been applied in high-speed calculations of video streams and other multimedia tasks. New options arise from the increasing power of new graphic processors. They include up to 1600 shading processors, which can also be used for universal computations at present. The paper discusses possible applications of graphic processors in simulation and other areas of high computation needs, like FEM or flow-analysis. The implementation of parallel threads on more than one core requires substantial changes in the software structure, which are only possible inside the source code. The paper also introduces feasible architectures and compares the CUDA and OpenCL approach.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
