Trait vrl::SIMDVector

source · 
pub trait SIMDVector<const N: usize>: SIMDBase<N> + Neg<Output = Self> + Arithmetic + ArithmeticAssign<Self> + Arithmetic<Self::Element> + ArithmeticAssign<Self::Element> + PartialEq + Into<Self::Underlying> + Into<[Self::Element; N]> + for<'a> From<&'a [Self::Element; N]> + SIMDPartialLoad<Self::Element> + SIMDPartialStore<Self::Element> + SIMDFusedCalc + Default + Copy + Clone + Debug + Index<usize> + IndexMut<usize> + Sum + Product
where
    [Self::Element; N]: Into<Self>,
    Self::Element: Arithmetic<Self, Self>,
    Self::Underlying: Into<Self>,
{ }
Expand description

Represents a packed vector containing N values of type Element.

Default::default initializes all elements of vector with zero.

All arithmetic operations (Neg, Add, etc) are applied vertically, i.e. “elementwise”.

§extract vs index

index (aka operator []) extracts index-th element of the vector. If value of the vector is expected to be in a register consider using extract. Use this function if only the vector is probably stored in memory.

In the following example the vector is stored in the heap. Using []-indexing in the case is as efficient as dereferencing the corresponding pointer.

let many_vectors = vec![Vec4f::new(1.0, 2.0, 3.0, 4.0); 128];
assert_eq!(many_vectors.index(64)[2], 3.0);

Here is an example of inefficient usage of index. The vector wouldn’t even reach memory and would stay in a register without that [1]. extract should be used instead.

let mut vec = Vec4f::new(1.0, 2.0, 3.0, 4.0);
vec *= 3.0;
vec -= 2.0;
let second_value = vec[1];
assert_eq!(second_value, 4.0);

Object Safety§

This trait is not object safe.

Implementors§