A Brief History of the Evolution of Graphics Processing Units A graphics processing unit (GPU) is usually a dedicated processor whose main purpose is to do vast amounts of computations in order to build images to be displayed. In order to render the 3D graphics one sees in modern day games, the GPU must perform countless floating-point calculations. GPUs can be found on virtually all commonly used electronics today such as phones, computers, video game consoles, etc. The term “GPU” didn’t actually exist until 1999 when the company NVIDIA used the term while marketing their new graphics card called the GeForce 256 . From the 1980’s until today, GPUs have gone through an exceptional amount of evolution with early graphics cards only being able to display simple vectors on screen to modern ones that can create full lush worlds that are hard to differentiate from our own world.
The process in which the GPU creates graphics is based off what is called a pipeline. A pipeline is essentially an assembly line in which the GPU creates vertices, and then turns them into primitives, then into pixel fragments (known as rasterization), which are displayed onto the screen . Early GPUs weren’t capable of handling the entire process and left the early parts to be calculated by the CPU. The whole process can be basically broken down into two parts. The first being the creation of the objects, while the other is just the rendering of the textures for said objects . Figure 1 shows a visual of this. In the mid 1980’s the company IBM created the first processor based video card for the computer. This allowed for the graphics card to take over all video tasks and thus freed up the CPU to be able to compute more calculations. It wasn’t a great success financially since it cost so much money, but it is important to note since it started a trend of having a separate processor on the video card to do the graphics computations instead of leaving it up to the CPU. In 1989 a company called Silicon Graphics Inc. (SGI) released a graphics API known as OpenGL. An API is known as “Application Programming Interface” and is just there to help programmers communicate with the hardware. This new API allowed programmers to really use all the horsepower available and has become one of the most widely used graphics APIs available .
Figure 1: Graphics Pipeline 
Then in the early 90’s SGI released what is known as the RealityEngine board. These boards had specific hardware dedicated to certain parts of the graphics pipeline, but still required the CPU to do most of the calculations. It wasn’t until the mid 90’s when companies started producing 3D graphics boards for the consumer . This came about because of the rise of computer games like Quake which required much more power to display the 3D graphics. There had been games before like Wolfenstein 3D, but it wasn’t completely 3D with most of the models being 2D. It was because of these games that many people started buying GPUs which were also evolving.
In 1996 the company 3DFX released their Voodoo1 card that sent 3D video cards flying into the mainstream. The first card to ever focus on solely 3D graphics, it was incapable of rendering 2D graphics and required a second card to do that task. It was so powerful though that people put up with it just so they could play games like Quake and Quake 2. It also supported Microsoft’s DirectX graphics API that was created for DOS in 1995. However even with all this power it still required the CPU to do the vertex transformations, while it took care of everything else in the pipeline .
It wasn’t until 1999 that a graphics card was created that took on all the responsibility of the pipeline as Figure 2 shows. This included new lighting and transform calculations. NVIDIA released their GeForce 256 and not only was it the first card to take on all the responsibility of the graphics pipeline; it also coined the term GPU. You could say that...
Please join StudyMode to read the full document