Pixel Pipelines

Pixel pipelines are graphics card components that process pixel information and are dedicated to accelerate image processing tasks. They have a reprogrammable processing core plus two independent frame buffers that are used to temporarily store image data and can operate on pixel data up to 200MB/s rates.

Pixel pipelines are comprised of pixel shaders and texture management units (TMU). If a graphics card has 24 pixel shaders and 24 TMUs, then that card is said to have 24 pixel pipelines. But it is not always a one-to-one ratio as some cards have more TMUs than shaders.

Pixel pipelines are also known as pixel processors.

The architecture of a pixel pipeline is now obsolete, replaced by unified shaders. In the previous architecture, there were pixel shaders and vertex shaders in the pipeline, where pixel shaders work on individual pixels and vertex shaders work on vertices to draw polygons faster. The downside to this is that sometimes, only one kind of shader does the bulk of the work, while the other is idle. This has been changed by using unified shaders which do different tasks, depending on the need. This is cheaper to manufacture, easier to program and more efficient, as all shaders are used for a task at any given moment.

Pixel pipelines are akin to manufacturing lines, where different processes are completed before getting to the final product. First, the pipelines receive data from the PCI bus or Accelerated Graphics Port (AGP) interface. Processes on the data are then completed sequentially before data is shown on the screen. These include clipping or removing the pixels not seen on the screen, generate more pixels, rasterization and then blending all the image elemnts before displaying on the monitor screen.

Post a Comment

0 Comments