Cbuffer

Cbuffer, in the context of shader programming, refers to a constant or uniform buffer used to group read-only shader constants together for efficient access. In HLSL (High-Level Shading Language) and Direct3D, the keyword cbuffer declares such a block, which is bound to a binding slot and accessed as a single structured set of values within the shader.

A cbuffer is defined as a block with a name, an optional register binding, and a list of members, for example: cbuffer PerFrame : register(b0) { float4x4 viewProj; float3 cameraPos; float floatPadding; }; The register(bN) clause binds the buffer to a specific slot, allowing the CPU to update the constants once per frame or per scene as needed. The data within a cbuffer is read by the shader, and the GPU can stream updates efficiently when the entire buffer changes.

Layout and packing rules are central to using cbuffers effectively. Members are laid out according to 16-byte alignment rules: the memory is organized into 4-component vectors, so a float4 occupies a single slot, a float3 may cause padding to the next 16-byte boundary, and matrices or arrays are laid out accordingly. Packoffset can be used to control exact offsets within the buffer to optimize memory usage and keep layout predictable across shader stages.

Benefits include reduced CPU-GPU traffic, fewer resource bindings, and improved performance when constants are updated in bulk. Cbuffers are commonly used for per-frame or per-object constants such as transformation matrices, lighting parameters, and camera data. Limitations are hardware- and API-specific, including maximum buffer size, binding slot counts, and alignment constraints that influence how data should be organized. See also uniform blocks in other shading languages for cross-API context.