Struct CudaBatchGroup 

pub struct CudaBatchGroup { /* private fields */ }

One successful homogeneous CUDA-resident output group.

Implementations

impl CudaBatchGroup

pub const fn info(&self) -> &BatchGroupInfo

Shared decoded dimensions, type, color, transform, route, and layout.

pub fn source_indices(&self) -> &[usize]

Original input indices in dense batch order.

pub fn decoded_rects(&self) -> &[Rect]

Actual decoded rectangle for each image.

pub fn warnings(&self) -> &[Vec<J2kDecodeWarning>]

Non-fatal codec warnings for each image.

pub fn surfaces(&self) -> &[Surface]

CUDA-resident image views in dense batch order.

Grayscale groups expose one view per image. NHWC RGB/RGBA groups also expose compatible interleaved views over their dense group allocation. No decoded host staging is used.

pub const fn dense_output(&self) -> &CudaResidentBatchBuffer

Dense codec-owned allocation for exact-native grayscale or color output.

NCHW color groups must be consumed through this owner because a Surface describes interleaved pixels. Grayscale and NHWC RGB/RGBA groups return both this owner and compatible surface views.

pub fn into_parts( self, ) -> (BatchGroupInfo, Vec<usize>, Vec<Rect>, Vec<Vec<J2kDecodeWarning>>, Vec<Surface>, CudaResidentBatchBuffer)

Consume the group into metadata, compatible surfaces, and its required dense exact-native allocation.

Trait Implementations

impl Debug for CudaBatchGroup

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

Formats the value using the given formatter.