Struct rand::prng::isaac::IsaacCore

[+] Show declaration

pub struct IsaacCore { /* fields omitted */ }

[−] Expand description

The core of IsaacRng, used with BlockRng.

Trait Implementations

impl Clone for IsaacCore

[+]

fn clone(&self) -> IsaacCore

[−]

Returns a copy of the value.

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

[−]

Performs copy-assignment from source.

impl Debug for IsaacCore

[+]

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

[−]

Formats the value using the given formatter.

impl SeedableRng for IsaacCore

[+]

type Seed = [u8; 32]

Seed type, which is restricted to types mutably-dereferencable as u8 arrays (we recommend [u8; N] for some N).

fn from_seed(seed: Self::Seed) -> Self

[−]

Create a new PRNG using the given seed.

fn from_rng<R: RngCore>(rng: R) -> Result<Self, Error>

[−]

Create a new PRNG seeded from another Rng.

fn seed_from_u64(state: u64) -> Self

[−]

Create a new PRNG using a u64 seed.

impl BlockRngCore for IsaacCore

[+]

type Item = u32

Results element type, e.g. u32.

type Results = IsaacArray<Self::Item>

Results type. This is the 'block' an RNG implementing BlockRngCore generates, which will usually be an array like [u32; 16].

fn generate(&mut self, results: &mut IsaacArray<Self::Item>)

[−]

Refills the output buffer, results. See also the pseudocode desciption of the algorithm in the IsaacRng documentation.

Optimisations used (similar to the reference implementation):

• The loop is unrolled 4 times, once for every constant of mix().
• The contents of the main loop are moved to a function rngstep, to reduce code duplication.
• We use local variables for a and b, which helps with optimisations.
• We split the main loop in two, one that operates over 0..128 and one over 128..256. This way we can optimise out the addition and modulus from s[i+128 mod 256].
• We maintain one index i and add m or m2 as base (m2 for the s[i+128 mod 256]), relying on the optimizer to turn it into pointer arithmetic.
• We fill results backwards. The reference implementation reads values from results in reverse. We read them in the normal direction, to make fill_bytes a memcopy. To maintain compatibility we fill in reverse.