Struct Quantize

pub struct Quantize<'s, 'd, To> { /* private fields */ }

Quantize a slice of f32 elements to 8-bit integers using the formula:

y = saturate(round(x * inv_scale) + zero_point)

Where round rounds to the nearest i32 value with ties to even and saturate converts i32 to the small integer type To with saturation.

Implementations

impl<'s, 'd, To> Quantize<'s, 'd, To>

pub fn new( src: &'s [f32], dest: &'d mut [MaybeUninit<To>], inv_scale: f32, zero_point: To, ) -> Self

Trait Implementations

impl<'d> SimdOp for Quantize<'_, 'd, u8>

type Output = &'d mut [u8]

The type of the operation’s result.

fn eval<I: Isa>(self, isa: I) -> Self::Output

Evaluate the operation using the given instruction set.

fn dispatch(self) -> Self::Output

where Self: Sized,

Dispatch this operation using the preferred ISA for the current platform.