class InferenceEngine::TensorDesc

Overview

This class defines Tensor description. More…

#include <ie_layouts.h>

class TensorDesc
{
public:
    // construction

    TensorDesc(, , );
    TensorDesc(, , );
    TensorDesc(, );
    TensorDesc();

    // methods

    voidconst SizeVector&Layout reshape(, );
    voidconst SizeVector&const BlockingDesc& reshape(, );
    SizeVector& getDims();
    const SizeVector& getDims() const;
    voidconst SizeVector& setDims();
    Layout getLayout() const;
    voidLayout setLayout();
    const Precision& getPrecision() const;
    voidconst Precision& setPrecision();
    const BlockingDesc& getBlockingDesc() const;
    boolconst TensorDesc& operator == () const;
    boolconst TensorDesc& operator != () const;
    size_tconst SizeVector& offset() const;
    size_tsize_t offset() const;
    static Layoutconst SizeVector& getLayoutByDims();
    static Layoutsize_t getLayoutByRank();
};

Detailed Documentation

This class defines Tensor description.

Construction

TensorDesc(, , )

The constructor creates the tensor descriptor using blocking descriptor.

Parameters:

precision	memory precision
dims	memory dimensions
blockDesc	blocking descriptor

TensorDesc(, , )

The constructor creates the tensor descriptor using standard layout.

Parameters:

precision	memory precision
dims	memory dimensions
layout	memory layout

TensorDesc(, )

The constructor creates the empty tensor descriptor with precision and layout.

Parameters:

precision	memory precision
layout	memory layout

TensorDesc()

The default constructor which creates empty tensor descriptor.

Methods

voidconst SizeVector&Layout reshape(, )

Reshapes the tensor descriptor.

Parameters:

dims	new dimensions
layout	new layout if it is necessary

voidconst SizeVector&const BlockingDesc& reshape(, )

Reshapes the tensor descriptor.

Parameters:

dims	new dimensions
blockDesc	new blocking descriptor

SizeVector& getDims()

Returns the vector of dimensions.

Returns:

dimensions

const SizeVector& getDims() const

Returns the constant vector of dimensions.

Returns:

dimensions

voidconst SizeVector& setDims()

Sets dimensions.

Parameters:

dims	new dimensions

Layout getLayout() const

Returns the memory layout.

Returns:

layout

voidLayout setLayout()

Sets the layout.

Parameters:

l	memory layout

const Precision& getPrecision() const

Returns the memory precision.

Returns:

precision

voidconst Precision& setPrecision()

Sets the memory precision.

Parameters:

p

precision

const BlockingDesc& getBlockingDesc() const

Returns the blocking descriptor.

Returns:

blocking descriptor

boolconst TensorDesc& operator == () const

The comparison operator for the TensorDesc.

Parameters:

rhs	object to compare

Returns:

true if objects are equal

boolconst TensorDesc& operator != () const

The comparison operator for the TensorDesc.

Parameters:

rhs	object to compare

Returns:

true if objects aren’t equal

size_tconst SizeVector& offset() const

Calculates offset for the vector of dimensions.

Parameters:

v	vector of dimensions

Returns:

offset

size_tsize_t offset() const

Calculates offset for the local offset.

Parameters:

l	local offset

Returns:

offset

static Layoutconst SizeVector& getLayoutByDims()

Returns the standard layout for dimensions.

Parameters:

dims	the vector of dimensions

Returns:

the standard memory layout

static Layoutsize_t getLayoutByRank()

Returns the standard layout for the specified tensor rank.

Parameters:

rank	of the requested layout

Returns:

the standard memory layout