Class Tensor<T>

Represents data in a multidimensional array-like structure.

Inheritance
object
Tensor
Tensor<T>
Implements
Inherited Members
Namespace: Unity.InferenceEngine
Assembly: Unity.InferenceEngine.dll
Syntax
[MovedFrom("Unity.Sentis")]
public class Tensor<T> : Tensor, IDisposable where T : unmanaged
Type Parameters
Name Description
T
Remarks

Tensors are the fundamental data structure used to represent multidimensional arrays of data, such as images and audio. Use them as inputs for models, and to download a copy of the backend output values (backendType).

Ownership and Lifetime
Tensors manage native memory resources. You must call Dispose() of a tensor when it is no longer needed. The ownership of the tensor's internal data (dataOnBackend) belongs to the Tensor object itself. Disposing the tensor also disposes its underlying data.

Data Representation
A Tensor's structure is defined by its shape (a TensorShape object) and its dataType. The actual data is held by an ITensorData implementation, which dictates the physical storage location BackendType. Data within the tensor is stored in a flattened, row-major format.

Asynchronous Operations and Data Access
Tensor data can be pending a device is performing computations, or if the data is stored on a non-readable device-specific type (for example GPU memory). To get a CPU readable copy of the data, use ReadbackAndClone(), or ReadbackAndCloneAsync() for an asynchronous operation. You can check the status of an asynchronous readback request with IsReadbackRequestDone().

Examples

The following example demonstrates how to interact with a Tensor object, ensuring its data is available on the CPU for reading, and then properly disposing of resources. 

 For a full workflow example, refer to [Workflow example](xref:sentis-workflow-example).

 </p>

  // Create a tensor
         m_Tensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });
         // Alternatively with `using` so you don't need to call `Dispose()` when you are done with the tensor
         using var m_OtherTensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });

         // Get a CPU-accessible clone of the tensor. This copy owns its own data.
         Tensor<float> cpuCopyTensor = m_Tensor.ReadbackAndClone() as Tensor<float>;

         // Release memory.
         cpuCopyTensor.Dispose();
         m_Tensor.Dispose();

Constructors

Tensor(TensorShape, bool)

Initializes and returns a tensor with the specified shape.

Declaration
public Tensor(TensorShape shape, bool clearOnInit = true)
Parameters
Type Name Description
TensorShape shape

The shape of the tensor.
bool clearOnInit

Whether to clear the tensor data to zeros.
See Also

Tensor(TensorShape, NativeArray<T>, int)

Initializes and returns a tensor with specified shape and a native T array of srcData data.

Declaration
public Tensor(TensorShape shape, NativeArray<T> srcData, int dataStartIndex = 0)
Parameters
Type Name Description
TensorShape shape

The shape of the tensor.
NativeArray<T> srcData

The data elements of the tensor.
int dataStartIndex

The index of the first tensor element in the srcData array.
See Also

Tensor(TensorShape, ITensorData)

Initializes and returns a Tensor with the specified shape, an ITensorData data.

Declaration
public Tensor(TensorShape shape, ITensorData data)
Parameters
Type Name Description
TensorShape shape

The shape of the tensor.
ITensorData data

The optional tensor data.
See Also

Tensor(TensorShape, T[], int)

Initializes and returns a tensor with specified shape and a T[] array of srcData data.

Declaration
public Tensor(TensorShape shape, T[] srcData, int dataStartIndex = 0)
Parameters
Type Name Description
TensorShape shape

The shape of the tensor.
T[] srcData

The data elements of the tensor.
int dataStartIndex

The index of the first tensor element in the srcData array.
See Also

Properties

this[int]

Returns the tensor element at offset d0.

Declaration
public T this[int d0] { get; set; }
Parameters
Type Name Description
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int]

Returns the tensor element at offset (d1, d0), which is position d1 * stride0 + d0.

Declaration
public T this[int d1, int d0] { get; set; }
Parameters
Type Name Description
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int, int]

Returns the tensor element at offset (d2, d1, d0), which is position d2 * stride1 + d1 * stride0 + d0.

Declaration
public T this[int d2, int d1, int d0] { get; set; }
Parameters
Type Name Description
int d2

Axis 2.
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int, int, int]

Returns the tensor element at offset (d3, d2, d1, d0), which is position d3 * stride2 + d2 * stride1 + d1 * stride0 + d0 in this tensor.

Declaration
public T this[int d3, int d2, int d1, int d0] { get; set; }
Parameters
Type Name Description
int d3

Axis 3.
int d2

Axis 2.
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int, int, int, int]

Returns the tensor element at offset (d4, d3, d2, d1, d0), which is position d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0.

Declaration
public T this[int d4, int d3, int d2, int d1, int d0] { get; set; }
Parameters
Type Name Description
int d4

Axis 4.
int d3

Axis 3.
int d2

Axis 2.
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int, int, int, int, int]

Returns the tensor element at offset (d5, d4, d3, d2, d1, d0), which is position d5 * stride4 + d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0.

Declaration
public T this[int d5, int d4, int d3, int d2, int d1, int d0] { get; set; }
Parameters
Type Name Description
int d5

Axis 5.
int d4

Axis 4.
int d3

Axis 3.
int d2

Axis 2.
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int, int, int, int, int, int]

Returns the tensor element at offset (d6, d5, d4, d3, d2, d1, d0), which is position d6 * stride5 + d5 * stride4 + d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0.

Declaration
public T this[int d6, int d5, int d4, int d3, int d2, int d1, int d0] { get; set; }
Parameters
Type Name Description
int d6

Axis 6.
int d5

Axis 5.
int d4

Axis 4.
int d3

Axis 3.
int d2

Axis 2.
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

this[int, int, int, int, int, int, int, int]

Returns the tensor element at offset (d7, d6, d5, d4, d3, d2, d1, d0), which is position d7 * stride6 + d6 * stride5 + d5 * stride4 + d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0.

Declaration
public T this[int d7, int d6, int d5, int d4, int d3, int d2, int d1, int d0] { get; set; }
Parameters
Type Name Description
int d7

Axis 7.
int d6

Axis 6.
int d5

Axis 5.
int d4

Axis 4.
int d3

Axis 3.
int d2

Axis 2.
int d1

Axis 1.
int d0

Axis 0.
Property Value
Type Description
T
See Also

Methods

AsReadOnlyNativeArray()

Exposes the linear memory data representation of this tensor as a readonly NativeArray.

Declaration
public NativeArray<T>.ReadOnly AsReadOnlyNativeArray()
Returns
Type Description
NativeArray<T>.ReadOnly

NativeArray of tensor data.
See Also

AsReadOnlySpan()

Exposes the linear memory data representation of this tensor as a ReadOnlySpan.

Declaration
public ReadOnlySpan<T> AsReadOnlySpan()
Returns
Type Description
ReadOnlySpan<T>

Span of tensor data.
See Also

DownloadToArray()

Blocking Call to return a copy of linear memory representation of the data in this tensor.

the returned array is a deepcopy of the tensor, the caller of this methods is now responsible for it. If you modify the contents of the returned array, it will not modify the underlying tensor

Declaration
public T[] DownloadToArray()
Returns
Type Description
T[]

T array copy of tensor data.
See Also

DownloadToNativeArray()

Blocking Call to return a copy of linear memory representation of the data in this tensor.

the returned native array is a deepcopy of the tensor, the caller of this methods is now responsible for it. If you modify the contents of the returned native array, it will not modify the underlying tensor

Declaration
public NativeArray<T> DownloadToNativeArray()
Returns
Type Description
NativeArray<T>

T native array copy of tensor data.
See Also

ReadbackAndClone()

Blocking download task of the internal data.

Declaration
public Tensor<T> ReadbackAndClone()
Returns
Type Description
Tensor<T>

returns cpu copy of the tensor.
See Also

ReadbackAndCloneAsync()

Schedules asynchronous download task of the internal data.

Declaration
public Awaitable<Tensor<T>> ReadbackAndCloneAsync()
Returns
Type Description
Awaitable<Tensor<T>>

awaitable tensor on the cpu.
See Also

Reshape(TensorShape)

Changes the logical shape of the tensor without changing the backing data's physical allocation.

Declaration
public override void Reshape(TensorShape shape)
Parameters
Type Name Description
TensorShape shape

The new shape for the tensor. The total number of elements in the new shape must fit within the currently allocated backend tensor data's capacity.
Overrides
Exceptions
Type Condition
AssertionException

Thrown if the new shape's total elements exceed the allocated capacity.
See Also

Upload(NativeArray<T>)

Uploads a contiguous block of tensor data to internal storage.

Declaration
public void Upload(NativeArray<T> srcData)
Parameters
Type Name Description
NativeArray<T> srcData

The data to upload.
See Also

Upload(T[])

Uploads a contiguous block of tensor data to internal storage.

Declaration
public void Upload(T[] srcData)
Parameters
Type Name Description
T[] srcData

The data to upload.
See Also

Implements

See Also