| | |
Summary: General Purpose GPUs (GPGPUs)
Using a Graphics Processor Unit (GPU) for Feature
Extraction from Turbulent Flow datasets
Andrew Davidson
Frameworks Group, CCT
Introduction-Why GPUs?
· GPUs are very powerful, and highly cost-effective
· Due to demand for "top of the line" graphics, GPU
growth has surpassed CPU growth
· GPUs have shown promising co-processing abilities
· With the release of high level shading languages, the
use of GPUs for general purpose processing is
possible, and very fast when applied to certain
algorithms.
Cg and Fragment Programs
· Fragment Programs used to do the computation are
written in Cg
· Cg is a shading language used on GPUs, due to
different architectures direct C code conversions are
impossible
|
