Trait DctToWaveletStageAccelerator 

pub trait DctToWaveletStageAccelerator {
    // Provided methods
    fn supports_dwt97_batch(&self) -> bool { ... }
    fn supports_htj2k97_codeblock_batch(&self) -> bool { ... }
    fn supports_htj2k97_i16_preencoded_batch(&self) -> bool { ... }
    fn supports_htj2k97_compact_preencoded_batch(&self) -> bool { ... }
    fn dct_grid_to_reversible_dwt53(
        &mut self,
        _job: DctGridToReversibleDwt53Job<'_>,
    ) -> Result<Option<ReversibleDwt53FirstLevel>, TranscodeStageError> { ... }
    fn dct_grid_to_reversible_dwt53_batch(
        &mut self,
        _jobs: &[DctGridToReversibleDwt53Job<'_>],
    ) -> Result<Option<Vec<ReversibleDwt53FirstLevel>>, TranscodeStageError> { ... }
    fn dct_grid_to_dwt53(
        &mut self,
        _job: DctGridToDwt53Job<'_>,
    ) -> Result<Option<Dwt53TwoDimensional<f64>>, TranscodeStageError> { ... }
    fn dct_grid_to_dwt97(
        &mut self,
        _job: DctGridToDwt97Job<'_>,
    ) -> Result<Option<Dwt97TwoDimensional<f64>>, TranscodeStageError> { ... }
    fn dct_grid_to_dwt97_batch(
        &mut self,
        _jobs: &[DctGridToDwt97Job<'_>],
    ) -> Result<Option<Vec<Dwt97TwoDimensional<f64>>>, TranscodeStageError> { ... }
    fn dct_grid_to_htj2k97_codeblock_batch(
        &mut self,
        _jobs: &[DctGridToHtj2k97CodeBlockJob<'_>],
        _options: Htj2k97CodeBlockOptions,
    ) -> Result<Option<Vec<PrequantizedHtj2k97Component>>, TranscodeStageError> { ... }
    fn dct_grid_to_htj2k97_preencoded_batch(
        &mut self,
        _jobs: &[DctGridToHtj2k97CodeBlockJob<'_>],
        _options: Htj2k97CodeBlockOptions,
    ) -> Result<Option<Vec<PreencodedHtj2k97Component>>, TranscodeStageError> { ... }
    fn dct_grid_i16_to_htj2k97_preencoded_batch(
        &mut self,
        _jobs: &[DctGridI16ToHtj2k97CodeBlockJob<'_>],
        _options: Htj2k97CodeBlockOptions,
    ) -> Result<Option<Vec<PreencodedHtj2k97Component>>, TranscodeStageError> { ... }
    fn dct_grid_i16_to_htj2k97_compact_preencoded_batch(
        &mut self,
        _jobs: &[DctGridI16ToHtj2k97CodeBlockJob<'_>],
        _options: Htj2k97CodeBlockOptions,
    ) -> Result<Option<PreencodedHtj2k97CompactBatch>, TranscodeStageError> { ... }
    fn dct_grid_i16_to_htj2k97_preencoded_batch_groups(
        &mut self,
        _groups: &[DctGridI16ToHtj2k97CodeBlockBatch<'_, '_>],
        _options: Htj2k97CodeBlockOptions,
    ) -> Result<Option<Vec<Vec<PreencodedHtj2k97Component>>>, TranscodeStageError> { ... }
    fn dct_grid_i16_to_htj2k97_compact_preencoded_batch_groups(
        &mut self,
        _groups: &[DctGridI16ToHtj2k97CodeBlockBatch<'_, '_>],
        _options: Htj2k97CodeBlockOptions,
    ) -> Result<Option<PreencodedHtj2k97CompactBatchGroups>, TranscodeStageError> { ... }
    fn last_dwt97_batch_stage_timings(&self) -> Option<Dwt97BatchStageTimings> { ... }
}

Optional backend for SIMD, GPU, or other accelerated transform stages.

Provided Methods

fn supports_dwt97_batch(&self) -> bool

Whether this accelerator wants same-geometry 9/7 batch jobs offered.

The default is false so CPU-only fallback paths do not pay the memory cost of materializing batch-owned float DCT blocks before immediately falling back.

fn supports_htj2k97_codeblock_batch(&self) -> bool

Whether this accelerator wants same-geometry 9/7 batches offered as prequantized HTJ2K code-block jobs before the float-band hook.

fn supports_htj2k97_i16_preencoded_batch(&self) -> bool

Whether this accelerator wants same-geometry 9/7 preencoded HTJ2K batches offered with dequantized i16 DCT blocks before materializing the generic f64 block representation.

fn supports_htj2k97_compact_preencoded_batch(&self) -> bool

Whether this accelerator wants the compact i16 preencoded HTJ2K batch hook offered before the owned preencoded hook.

fn dct_grid_to_reversible_dwt53( &mut self, _job: DctGridToReversibleDwt53Job<'_>, ) -> Result<Option<ReversibleDwt53FirstLevel>, TranscodeStageError>

Optionally compute the direct DCT-grid to one-level reversible integer 5/3 projection.

Return Ok(Some(output)) when the backend handled the job bit-exactly relative to j2k’s scalar integer oracle. Return Ok(None) to use the scalar fallback.

fn dct_grid_to_reversible_dwt53_batch( &mut self, _jobs: &[DctGridToReversibleDwt53Job<'_>], ) -> Result<Option<Vec<ReversibleDwt53FirstLevel>>, TranscodeStageError>

Optionally compute a same-geometry batch of direct DCT-grid to one-level reversible integer 5/3 projections.

Backends should return outputs in the same order as jobs. Return Ok(None) to use the scalar per-component fallback.

fn dct_grid_to_dwt53( &mut self, _job: DctGridToDwt53Job<'_>, ) -> Result<Option<Dwt53TwoDimensional<f64>>, TranscodeStageError>

Optionally compute the direct DCT-grid to one-level 5/3 projection.

Return Ok(Some(output)) when the backend handled the job. Return Ok(None) to use the scalar fallback.

fn dct_grid_to_dwt97( &mut self, _job: DctGridToDwt97Job<'_>, ) -> Result<Option<Dwt97TwoDimensional<f64>>, TranscodeStageError>

Optionally compute the direct DCT-grid to one-level 9/7 transform.

Return Ok(Some(output)) when the backend handled the job. Return Ok(None) to use the scalar fallback.

fn dct_grid_to_dwt97_batch( &mut self, _jobs: &[DctGridToDwt97Job<'_>], ) -> Result<Option<Vec<Dwt97TwoDimensional<f64>>>, TranscodeStageError>

Optionally compute a same-geometry batch of direct DCT-grid to one-level 9/7 transforms.

Backends should return outputs in the same order as jobs. Return Ok(None) to use the scalar per-component fallback.

fn dct_grid_to_htj2k97_codeblock_batch( &mut self, _jobs: &[DctGridToHtj2k97CodeBlockJob<'_>], _options: Htj2k97CodeBlockOptions, ) -> Result<Option<Vec<PrequantizedHtj2k97Component>>, TranscodeStageError>

Optionally compute same-geometry DCT-grid 9/7 jobs directly into prequantized HTJ2K code-block components.

Backends should return one component per input job in the same order as jobs. Return Ok(None) to use the float-band path.

fn dct_grid_to_htj2k97_preencoded_batch( &mut self, _jobs: &[DctGridToHtj2k97CodeBlockJob<'_>], _options: Htj2k97CodeBlockOptions, ) -> Result<Option<Vec<PreencodedHtj2k97Component>>, TranscodeStageError>

Optionally compute same-geometry DCT-grid 9/7 jobs directly into preencoded HTJ2K code-block payloads.

Backends should return one component per input job in the same order as jobs. Return Ok(None) to use the prequantized or float-band path.

fn dct_grid_i16_to_htj2k97_preencoded_batch( &mut self, _jobs: &[DctGridI16ToHtj2k97CodeBlockJob<'_>], _options: Htj2k97CodeBlockOptions, ) -> Result<Option<Vec<PreencodedHtj2k97Component>>, TranscodeStageError>

Optionally compute same-geometry dequantized i16 DCT-grid 9/7 jobs directly into preencoded HTJ2K code-block payloads.

Backends should return one component per input job in the same order as jobs. Return Ok(None) to use the generic f64 preencoded path.

fn dct_grid_i16_to_htj2k97_compact_preencoded_batch( &mut self, _jobs: &[DctGridI16ToHtj2k97CodeBlockJob<'_>], _options: Htj2k97CodeBlockOptions, ) -> Result<Option<PreencodedHtj2k97CompactBatch>, TranscodeStageError>

Optionally compute same-geometry dequantized i16 DCT-grid 9/7 jobs into compact preencoded HTJ2K code-block payloads.

Backends should return one component per input job in the same order as jobs, with all component ranges pointing into the returned payload. Return Ok(None) to use the owned preencoded path.

fn dct_grid_i16_to_htj2k97_preencoded_batch_groups( &mut self, _groups: &[DctGridI16ToHtj2k97CodeBlockBatch<'_, '_>], _options: Htj2k97CodeBlockOptions, ) -> Result<Option<Vec<Vec<PreencodedHtj2k97Component>>>, TranscodeStageError>

Optionally compute multiple same-geometry dequantized i16 DCT-grid batches directly into preencoded HTJ2K code-block payloads.

Each input batch is internally same-geometry, but different batches may have different component dimensions. Backends should return one output vector per input batch, in order. Return Ok(None) to use the per-group fallback hooks.

fn dct_grid_i16_to_htj2k97_compact_preencoded_batch_groups( &mut self, _groups: &[DctGridI16ToHtj2k97CodeBlockBatch<'_, '_>], _options: Htj2k97CodeBlockOptions, ) -> Result<Option<PreencodedHtj2k97CompactBatchGroups>, TranscodeStageError>

Optionally compute multiple same-geometry dequantized i16 DCT-grid 9/7 batches into compact preencoded HTJ2K code-block payloads.

Each returned item corresponds to one input batch and contains one component per job in that batch. Return Ok(None) to use the owned preencoded grouped hook.

fn last_dwt97_batch_stage_timings(&self) -> Option<Dwt97BatchStageTimings>

Return backend stage timings for the most recent 9/7 batch dispatch.

Dyn Compatibility

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors

impl DctToWaveletStageAccelerator for CpuOnlyDctToWaveletStageAccelerator

impl DctToWaveletStageAccelerator for RayonReversibleDwt53Accelerator