Struct DataDrivenDispatch 

pub struct DataDrivenDispatch { /* private fields */ }

Computes and stores dispatch parameters that depend on the number of data elements only known at dispatch-preparation time (e.g., after a GPU readback or a CPU-side counter).

In a real GPU pipeline this feeds an indirect dispatch buffer; here we compute the DispatchGrid on the CPU side for portability and testing.

Implementations

impl DataDrivenDispatch

pub fn new(wg_x: u32, wg_y: u32, strategy: DataDispatchStrategy) -> Self

Create a new data-driven dispatch helper.

• wg_x / wg_y — workgroup size dimensions (must be ≥ 1).
• strategy — how to map element counts to workgroup grids.

pub fn linear(wg_size: u32) -> Self

Convenience constructor for a 1D strip with wg_size threads per workgroup.

pub fn square(wg_x: u32, wg_y: u32) -> Self

Convenience constructor for a 2D square grid with wg_x × wg_y threads per workgroup.

pub fn prepare(&mut self, element_count: u64) -> DispatchGrid

Prepare the dispatch grid for element_count data elements.

Returns the resulting DispatchGrid; the value is also stored internally and accessible via Self::grid.

pub fn grid(&self) -> Option<DispatchGrid>

The grid computed by the last Self::prepare call, or None if Self::prepare has not yet been called.

pub fn last_element_count(&self) -> u64

The element count supplied to the last Self::prepare call.

pub fn covered_elements(&self) -> u64

Minimum elements coverable by the last computed grid.

Returns 0 if Self::prepare has not been called.