Event
Grafikprozessoren für andere Berechnungen
In November 2006 NVIDIA introduced CUDA, an architecture and programming environment for parallel computation on supported graphics processors. Instead of using a GPU only for displaying graphics, developers could also address other computational tasks. A contemporary technical brief describes tools with C support. NVIDIA documents
Parallel processing of many data items is not equally suitable for every task. NVIDIA announced a public beta of the SDK and C compiler in February 2007; that release is distinct from the 2006 architecture introduction and earlier access for registered developers. Sources
Sources
- NVIDIA-Technikpapier, Programmierhandbuch und Beta-Mitteilung: Architektur, Programmiermodell und Datierungen.
Offen
Unabhängige zeitgenössische Anwendungen und Grenzen der ersten Implementierung ergänzen. / Add independent contemporary applications and limitations of the initial implementation.