Specific purpose hardware

From Green software

Revision as of 14:10, 15 September 2010 by Green (Talk | contribs)
(diff) ←Older revision | Current revision (diff) | Newer revision→ (diff)
Jump to: navigation, search

In most computer systems all computations are done on a CPU. CPUs are made to be generic, which sometimes costs efficiency. CPU's are now more often equipped with small specific purpose processing units. Examples of this are a Floating Point Unit (FPU) or Graphical Processing Unit (GPU).

Contents

Drawbacks

It is often impossible to rely on specific purpose hardware for use in generic software. Using a GPU for computation is often best done only in controlled environments.

Implementations

On some systems that have a GPU graphical computations can be handed of to the GPU instead. The GPU can do a lot more graphical computations per time, but also per Watt. With the invention of CUDA by NVIDIA, which stands for Compute Unified Device Architecture, the GPU can be used to compute generic problems. Especially in problems involving matrices or multiple nested for-loops this can be improve performance and energy-efficiency significantly.

The technique of using a GPU for compuations that can be done on a CPU as well is called General-purpose computing on graphics processing units (GPGPU).

Supercomputers have been constructed out of Playstation 3 systems linked together, because of their efficient specific purpose capabilities.

See Also

http://en.wikipedia.org/wiki/GPGPU

Sources

This best practice was recognized as such by IT professionals, described in Energy efficient software.

Personal tools