Trait TensorDescriptor 

pub trait TensorDescriptor: MemoryDescriptor {
    // Required methods
    fn shape(&self) -> &[usize];
    fn stride(&self) -> &[usize];
    fn element_size(&self) -> usize;
}

A tensor is memory with shape, stride, and element size metadata.

This trait extends MemoryDescriptor with tensor-specific metadata. The underlying memory could be externally owned, self-owned, or a view.

Shape and Stride

• shape() returns the number of elements in each dimension
• stride() returns the number of elements to skip when incrementing each dimension
• element_size() returns the number of bytes per element

For a contiguous tensor with shape [2, 3, 4]:

• stride would be [12, 4, 1] (row-major/C order)
• total elements = 2 * 3 * 4 = 24
• total bytes = 24 * element_size()

Required Methods

fn shape(&self) -> &[usize]

Shape of the tensor (number of elements per dimension).

fn stride(&self) -> &[usize]

Stride of the tensor (elements to skip per dimension).

stride[i] indicates how many elements to skip when incrementing dimension i.

fn element_size(&self) -> usize

Number of bytes per element.

Dyn Compatibility

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementations on Foreign Types

impl TensorDescriptor for Arc<dyn TensorDescriptor + Send + Sync>

fn shape(&self) -> &[usize]

fn stride(&self) -> &[usize]

fn element_size(&self) -> usize

impl TensorDescriptor for Arc<dyn TensorDescriptor>

fn shape(&self) -> &[usize]

fn stride(&self) -> &[usize]

fn element_size(&self) -> usize

Implementors