Struct StencilEvalPipeline 

pub struct StencilEvalPipeline { /* private fields */ }

Available on crate feature wgpu only.

Compute pipeline + bind-group layout for stencil evaluation.

Implementations

impl StencilEvalPipeline

pub fn new(device: &Device) -> Self

Build the pipeline with the default workgroup size (64).

pub fn with_workgroup_size(device: &Device, workgroup_size: u32) -> Self

Build the pipeline with a specific workgroup size.

pub fn encode( &self, device: &Device, encoder: &mut CommandEncoder, gpu_table: &StencilTableGpu, src_buffer: &Buffer, dst_buffer: &Buffer, src_desc: BufferDescriptor, dst_desc: BufferDescriptor, batch_range: Range<u32>, ) -> Result<(), KernelError>

Encode a stencil-evaluation dispatch for the output rows in batch_range into encoder.

pub fn encode_indexed( &self, device: &Device, encoder: &mut CommandEncoder, gpu_table: &StencilTableGpu, src_buffer: &Buffer, dst_buffer: &Buffer, indirection: &Buffer, src_desc: BufferDescriptor, dst_desc: BufferDescriptor, count: u32, ) -> Result<(), KernelError>

Encode a sparse (indexed) dispatch: recompute only the output rows named by indirection (e.g. from affected_outputs), leaving the other rows of dst_buffer untouched. count is the number of indirection entries.

Trait Implementations

impl Debug for StencilEvalPipeline

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.