What is a GPU Triangle?

What is a GPU Triangle?

We all know that GPU is used for graphic applications. Explaining in simple terms,
A GPU has to synthesis a graphic image based on the information given.
A game provides some parameters to the GPU such as vertices, vertex shader, fragment shader etc to the GPU.

The vertices define objects in the game such as the gun, opponent, trees etc.
So now the GPU knows what appears where and size of the object based on the vertices.
Then begins the process of reconstruction of the image frame in a game.

Just like a living organism is composed of cells, a graphic image is composed of Triangles.

Thus an image frame in a game consisting of objects like guns or zombies, are made of innumerable number of triangles. Thus knowing the vertices represented in a co-ordinate system, the GPU constructs triangles in another co-ordinate system.

The size of triangle though, varies with the manufacturer of the GPU.

Image of a dolphin constructed using Triangles

The performance of a GPU is measured using Triangle Throughput (Triangles per second) although this is not accurate as the size varies according to the manufacturer.

As a rough estimate, the Power VR 6 series GPU in the iphone 5s has a Triangle throughput of 68.2 Million triangles per second in an off-screen test and perhaps the world's fastest GPU till date i.e Nvidia GTX 580 has a triangle throughput of 2 Billion Triangles per second.