Struct TensorSpec 

pub struct TensorSpec {
    pub uid: i64,
    pub dtype: DtypeTag,
    pub dims: Vec<i64>,
    pub strides: Vec<i64>,
    pub alignment: i64,
    pub is_virtual: bool,
}

One tensor in a backend graph: unique id, dtype, dims, strides, alignment.

Fields

uid: i64

dtype: DtypeTag

dims: Vec<i64>

strides: Vec<i64>

alignment: i64

Byte alignment of the data pointer. cuDNN requires ≥ 4.

is_virtual: bool

Whether the tensor is virtual (intermediate result, no device-pointer binding required).

Implementations

impl TensorSpec

pub fn new( uid: i64, dtype: DtypeTag, dims: Vec<i64>, layout: TensorLayout, ) -> Self

Build a TensorSpec for dims under layout. For Strided the caller must call with_strides afterwards.

pub fn with_strides(self, strides: Vec<i64>) -> Self

pub fn with_alignment(self, alignment: i64) -> Self

pub fn virtualized(self) -> Self

pub fn rank(&self) -> usize

Trait Implementations

impl Clone for TensorSpec

fn clone(&self) -> TensorSpec

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TensorSpec

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

Formats the value using the given formatter.

impl Hash for TensorSpec

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for TensorSpec

fn eq(&self, other: &TensorSpec) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Eq for TensorSpec

impl StructuralPartialEq for TensorSpec