Struct VisionHasherMultiDigest 

pub struct VisionHasherMultiDigest<const N: usize, const MN: usize> { /* private fields */ }

A Vision hasher with state size M=6 suited for parallelization.

Without using packed fields, there is only one advantage of an explicit parallelized Vision hasher over invoking the Vision hasher multiple times in parallel: we can amortize the cost of inversion in the sbox using Montogery’s trick. We slightly modify Montogery’s trick to use a binary tree structure, maximizing independence of multiplications for better instruction pipelining.

Trait Implementations

impl<const N: usize, const MN: usize> Clone for VisionHasherMultiDigest<N, MN>

fn clone(&self) -> VisionHasherMultiDigest<N, MN>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<const N: usize, const MN: usize> Default for VisionHasherMultiDigest<N, MN>

fn default() -> Self

Returns the “default value” for a type.

impl<const N: usize, const MN: usize> MultiDigest<N> for VisionHasherMultiDigest<N, MN>

type Digest = VisionHasherDigest

The corresponding non-parallelized hash function.

fn new() -> Self

Create new hasher instance with empty state.

fn update(&mut self, data: [&[u8]; N])

Process data, updating the internal state. The number of rows in data must be equal to parallel_instances().

fn finalize_into(self, out: &mut [MaybeUninit<Output<Self::Digest>>; N])

Write result into provided array and consume the hasher instance.

fn finalize_into_reset( &mut self, out: &mut [MaybeUninit<Output<Self::Digest>>; N], )

Write result into provided array and reset the hasher instance.

fn reset(&mut self)

Reset hasher instance to its initial state.

fn digest(data: [&[u8]; N], out: &mut [MaybeUninit<Output<Self::Digest>>; N])

Compute hash of data. All slices in the data must have the same length.

fn new_with_prefix(data: impl AsRef<[u8]>) -> Self

Create new hasher instance which has processed the provided data.

fn chain_update(self, data: [&[u8]; N]) -> Self

Process input data in a chained manner.