forge-vertex-pulling

To implement vertex pulling in an SDL3 GPU graphics pipeline, replace fixed-function vertex inputs with storage buffer reads. You do this by binding a storage buffer and configuring a pipeline with zero vertex attributes, allowing the vertex shader to fetch data.

Programmable vertex fetch is the process of reading vertex data from a storage buffer via SV_VertexID. You need it in SDL3 for real-time rendering pipelines with varying vertex formats or GPU-driven vertex assembly workflows.

You read vertex data from a storage buffer using SV_VertexID by writing a vertex shader that indexes into a StructuredBuffer. This eliminates per-format vertex attribute declarations by dynamically fetching vertex data on the GPU.

Yes, you can render meshes with different vertex formats through a single SDL3 GPU pipeline. By using programmable vertex fetch via SV_VertexID, the pipeline ignores fixed vertex attributes and dynamically reads varying layouts from a storage buffer.

Yes, vertex pulling supports compute-generated vertex data for indirect drawing. Because the vertex shader reads directly from a storage buffer, it fully enables flexible GPU-driven workflows and complex rendering pipelines.

Replacing fixed-function vertex input with storage buffer reads enables flexible vertex formats and GPU-driven workflows. It eliminates the need to declare per-format vertex attributes, streamlining pipelines for complex rendering scenarios.