Struct SadOps 

pub struct SadOps<S: SimdOps> { /* private fields */ }

SAD operations using SIMD.

Implementations

impl<S: SimdOps> SadOps<S>

pub const fn new(simd: S) -> Self

Create a new SAD operations instance.

pub const fn simd(&self) -> &S

Get the underlying SIMD implementation.

pub fn sad_4x4( &self, src: &[u8], src_stride: usize, ref_block: &[u8], ref_stride: usize, ) -> u32

Calculate SAD for a 4x4 block.

Arguments

• src - Source block data (row-major, stride = src_stride)
• src_stride - Stride between source rows
• ref_block - Reference block data (row-major, stride = ref_stride)
• ref_stride - Stride between reference rows

Returns

Sum of absolute differences for all 16 pixels.

pub fn sad_8x8( &self, src: &[u8], src_stride: usize, ref_block: &[u8], ref_stride: usize, ) -> u32

Calculate SAD for an 8x8 block.

pub fn sad_16x16( &self, src: &[u8], src_stride: usize, ref_block: &[u8], ref_stride: usize, ) -> u32

Calculate SAD for a 16x16 block.

pub fn sad_32x32( &self, src: &[u8], src_stride: usize, ref_block: &[u8], ref_stride: usize, ) -> u32

Calculate SAD for a 32x32 block.

pub fn sad_nxn( &self, src: &[u8], src_stride: usize, ref_block: &[u8], ref_stride: usize, width: usize, height: usize, ) -> u32

Calculate SAD for an arbitrary block size.

Less efficient than size-specific functions but more flexible.

pub fn satd_4x4( &self, src: &[u8], src_stride: usize, ref_block: &[u8], ref_stride: usize, ) -> u32

Calculate SATD (Sum of Absolute Transformed Differences) for 4x4 block.

SATD applies a Hadamard transform before summing, providing a better cost metric for rate-distortion optimization.

Trait Implementations

impl<S: SimdOps + Default> Default for SadOps<S>

fn default() -> Self

Returns the “default value” for a type.