Executing algorithms on the GPU

Graphics processing units (GPUs) were originally designed and used for processing points and pixels for computer graphics rendering. Briefly, what the GPUs did was retrieve buffers of pixel data or vertex data, perform a simple operation on each buffer individually, and store the result in a new buffer (to eventually be displayed).

Here are some examples of simple, independent operations that could be executed on the GPU at an early stage:

• Transform a point from world coordinates to screen coordinates
• Perform a lighting calculation at a specific point (by lighting calculation, I am referring to calculating the color of a specific pixel in an image)

As these operations could be performed in parallel, the GPUs were designed for executing small operations in parallel. Later on, these graphics operations became programmable, although the programs were written in terms of computer graphics (that is, the memory reads were done...