Trait half::vec::HalfFloatVecExt

source · [−]
pub trait HalfFloatVecExt: SealedHalfFloatVec {
    fn reinterpret_into(self) -> Vec<u16>;
    fn from_f32_slice(slice: &[f32]) -> Self;
    fn from_f64_slice(slice: &[f64]) -> Self;
}
Expand description

Extensions to Vec<f16> and Vec<bf16> to support reinterpret operations.

This trait is sealed and cannot be implemented outside of this crate.

Required Methods

Reinterprets a vector of f16or bf16 numbers as a vector of u16 bits.

This is a zero-copy operation. The reinterpreted vector has the same memory location as self.

Examples
let float_buffer = vec![f16::from_f32(1.), f16::from_f32(2.), f16::from_f32(3.)];
let int_buffer = float_buffer.reinterpret_into();

assert_eq!(int_buffer, [f16::from_f32(1.).to_bits(), f16::from_f32(2.).to_bits(), f16::from_f32(3.).to_bits()]);

Converts all of the elements of a [f32] slice into a new f16 or bf16 vector.

The conversion operation is vectorized over the slice, meaning the conversion may be more efficient than converting individual elements on some hardware that supports SIMD conversions. See crate documentation for more information on hardware conversion support.

Examples
let float_values = [1., 2., 3., 4.];
let vec: Vec<f16> = Vec::from_f32_slice(&float_values);

assert_eq!(vec, vec![f16::from_f32(1.), f16::from_f32(2.), f16::from_f32(3.), f16::from_f32(4.)]);

Converts all of the elements of a [f64] slice into a new f16 or bf16 vector.

The conversion operation is vectorized over the slice, meaning the conversion may be more efficient than converting individual elements on some hardware that supports SIMD conversions. See crate documentation for more information on hardware conversion support.

Examples
let float_values = [1., 2., 3., 4.];
let vec: Vec<f16> = Vec::from_f64_slice(&float_values);

assert_eq!(vec, vec![f16::from_f64(1.), f16::from_f64(2.), f16::from_f64(3.), f16::from_f64(4.)]);

Implementations on Foreign Types

Implementors