vertexprogramming

Vertex programming refers to writing programs that run on graphics processing units at the vertex processing stage of the graphics pipeline. These programs, typically implemented as vertex shaders, compute per-vertex attributes such as transformed positions, normals, and texture coordinates, and pass data to subsequent pipeline stages.

Historically, fixed-function vertex processing dominated early GPUs, but in the 1990s and 2000s programmable shading allowed developers to implement custom transformations. Early vertex programs used assembly-like languages such as NVIDIA's vertex program language and ARB_vertex_program. Modern pipelines use high-level shading languages such as GLSL, HLSL, or SPIR-V-based languages to write vertex shaders that are compiled and run on the GPU.

In a typical vertex program, input attributes include vertex position, normal, tangent, texture coordinates, and skinning weights. The program applies model-view-projection transformations, computes lighting at the vertex if needed, and can perform morph targets or skeletal animation (bone transforms). It outputs a clip-space position (often gl_Position) and a set of per-vertex varyings that are interpolated across the primitive for the fragment shader. Vertex programs aim to maximize performance and minimize bandwidth by doing as much work as possible per vertex, sometimes trading per-vertex accuracy for speed.

In modern graphics APIs, vertex programming remains a core part of the shader pipeline, though compute shaders and advanced shading techniques have expanded possibilities. Understanding vertex programming is foundational for 3D rendering, computer graphics education, and real-time graphics development.