GpuDispatch

1 min read

Import: import system.gpu;

This reference lists the public declarations for this part of the library. See Reading API signatures for parameter and result notation. Shader-only declarations can be used only inside @compute, @vertex, @fragment, inferred helpers and ray tracing shader functions; using them elsewhere is a compile error.

GpuDispatch

How many threads a compute dispatch runs. The threads_* constructors take the number of threads you need; the dispatch rounds up to whole workgroups of the shader's @compute size, so a shader should check its global_id() against the real count. groups takes workgroup counts directly.

public struct GpuDispatch

threads_1d

Runs at least count threads along x.

public static fun threads_1d(count: int): GpuDispatch

threads_2d

Runs at least width by height threads, for example one per pixel.

public static fun threads_2d(width: int, height: int): GpuDispatch

threads_3d

Runs at least x by y by z threads.

public static fun threads_3d(x: int, y: int, z: int): GpuDispatch

groups

Runs exactly x by y by z workgroups.

public static fun groups(x: int, y: int, z: int): GpuDispatch

group_counts

The workgroup counts for a shader whose @compute size is workgroup. Thread counts round up to whole workgroups; zero or negative counts give zero workgroups, which runs nothing.

public fun group_counts(workgroup: GpuIVec3): GpuIVec3