When parts of the graphic pipeline become programmable news programming languages appeared to provide access to the welcome features. The graphics architecture is known by its many execution units, that’s why it’s called the new massively parallel processing architecture.
The first languages are well suited to implement advanced algorithms inside the graphics pipeline that provide more realism and nice effects to the scene. To use those languages the programmer has to know some architecture details and basic OpenGL or DirectX commands to control the pipeline.

C for graphics

Developed by Nvidia in collaboration with Microsoft to provide vertex and fragment processors programmability. It is compatible with OpenGL and DirectX.
With CG you can develop your programs in a C-like portable language, running with a host CPU program that starts and control the graphic pipeline.


OpenGL Shading Language, also called GLslang, created by OpenGL ARM.
High Level Shading Language, created by Microsoft to work with the DirectX API.
Both are C-based, well suited to develop shader programs and need a CPU host to run.

CUDA architecture

Developed by Nvidia, CUDA provide a fast learn environment that is API-independent, C-like language and well suited for generic programming on the GPU. The CUDA language is based in thread hierarchy, shared memory and barrier synchronization.
The CUDA main advantage is its independence from the graphics APIs, it manage the pipeline for you, with some kind of high level commands. However, so as the other languages, it’s also need a CPU host to run.

Last edited Aug 16, 2009 at 3:44 PM by Lenna, version 2


No comments yet.