Skip to content

Functions Overview

Functions in WGSL are named, reusable subroutines that group a series of statements to perform specific tasks. Because shaders operate in a highly parallel execution environment, WGSL functions are designed for high efficiency and predictable execution.

The GPU Execution Model

Shaders are executed on the GPU across thousands of concurrent threads (or invocations). To ensure maximum hardware performance, WGSL functions are built under specific rules:

  • Compiler Inlining: User-defined functions have virtually zero execution overhead. The GPU compiler almost always performs inline expansion (replacing function calls directly with the function's body code) to avoid the overhead of traditional call-stack allocations.
  • No Dynamic Dispatch: Function calls are resolved statically at compile time.
  • Predictable Performance: These compile-time decisions ensure that your shader execution paths remain highly optimized and uniform.

Section Contents

This section covers the core mechanics of writing and calling functions in WGSL:

  • Syntax: Declaring functions using the fn keyword, specifying input parameters, defining return types, and applying strict typing.
  • Function Calls & Recursion: Function invocation mechanics, copy-on-pass parameter behavior, and the structural hardware constraints that prohibit recursion.
  • Pipeline Entry Points: Special entry point stages annotated with @vertex, @fragment, or @compute that serve as interface boundaries with the WebGPU API.
  • Must Use Attributes: Safety-critical compiler guarantees using the @must_use attribute to prevent silently discarding evaluation results.