Deep Neural Network Library (DNNL)  1.3.0
Performance library for Deep Learning
Public Member Functions | Public Attributes | List of all members
dnnl::memory::desc Struct Reference

A memory descriptor. More...

#include <dnnl.hpp>

Collaboration diagram for dnnl::memory::desc:
Collaboration graph
[legend]

Public Member Functions

 desc ()
 Constructs a zero (empty) memory descriptor. More...
 
 desc (const memory::dims &dims, data_type data_type, format_tag format_tag, bool allow_empty=false)
 Constructs a memory descriptor. More...
 
 desc (const memory::dims &dims, data_type data_type, const memory::dims &strides, bool allow_empty=false)
 Constructs a memory descriptor by strides. More...
 
 desc (const dnnl_memory_desc_t &data)
 Constructs a memory descriptor from a C API data structure. More...
 
desc submemory_desc (const memory::dims &dims, const memory::dims &offsets, bool allow_empty=false) const
 Constructs a memory descriptor for a region inside an area described by this memory descriptor. More...
 
desc reshape (const memory::dims &dims, bool allow_empty=false) const
 Constructs a memory descriptor by reshaping an existing one. More...
 
desc permute_axes (const std::vector< int > &permutation, bool allow_empty=false) const
 Constructs a memory descriptor by permuting axes in an existing one. More...
 
memory::dims dims () const
 Returns dimensions of the memory descriptor. More...
 
memory::data_type data_type () const
 Returns the data type of the memory descriptor. More...
 
size_t get_size () const
 Returns size of the memory descriptor in bytes. More...
 
bool is_zero () const
 Checks whether the memory descriptor is zero (empty). More...
 
bool operator== (const desc &other) const
 An equality operator. More...
 
bool operator!= (const desc &other) const
 An inequality operator. More...
 

Public Attributes

dnnl_memory_desc_t data
 The underlying C API data structure.
 

Detailed Description

A memory descriptor.

Examples:
cnn_inference_f32.cpp, cnn_inference_int8.cpp, cnn_training_f32.cpp, cpu_cnn_training_bf16.cpp, cpu_matmul_quantization.cpp, cpu_rnn_inference_f32.cpp, cpu_rnn_inference_int8.cpp, cpu_sgemm_and_matmul.cpp, getting_started.cpp, gpu_opencl_interop.cpp, inference_int8_matmul.cpp, memory_format_propagation.cpp, performance_profiling.cpp, and rnn_training_f32.cpp.

Constructor & Destructor Documentation

◆ desc() [1/4]

dnnl::memory::desc::desc ( )
inline

Constructs a zero (empty) memory descriptor.

Such a memory descriptor can be used to indicate absence of an argument.

◆ desc() [2/4]

dnnl::memory::desc::desc ( const memory::dims dims,
data_type  data_type,
format_tag  format_tag,
bool  allow_empty = false 
)
inline

Constructs a memory descriptor.

Note
The logical order of dimensions corresponds to the abc... format tag, and the interpretation of the dimensions depends on the primitive that consumes the memory.
Parameters
dimsTensor dimensions.
data_typeData precision/type.
format_tagMemory format tag.
allow_emptyA flag signifying whether construction is allowed to fail without throwing an exception. In this case a zero memory descriptor will be constructed. This flag is optional and defaults to false.

◆ desc() [3/4]

dnnl::memory::desc::desc ( const memory::dims dims,
data_type  data_type,
const memory::dims strides,
bool  allow_empty = false 
)
inline

Constructs a memory descriptor by strides.

Note
The logical order of dimensions corresponds to the abc... format tag, and the interpretation of the dimensions depends on the primitive that consumes the memory.
Parameters
dimsTensor dimensions.
data_typeData precision/type.
stridesStrides for each dimension.
allow_emptyA flag signifying whether construction is allowed to fail without throwing an exception. In this case a zero memory descriptor will be constructed. This flag is optional and defaults to false.

◆ desc() [4/4]

dnnl::memory::desc::desc ( const dnnl_memory_desc_t data)
inline

Constructs a memory descriptor from a C API data structure.

Parameters
dataA C API dnnl_memory_desc_t structure.

Member Function Documentation

◆ submemory_desc()

desc dnnl::memory::desc::submemory_desc ( const memory::dims dims,
const memory::dims offsets,
bool  allow_empty = false 
) const
inline

Constructs a memory descriptor for a region inside an area described by this memory descriptor.

Parameters
dimsSizes of the region.
offsetsOffsets to the region from the encompassing memory object in each dimension.
allow_emptyA flag signifying whether construction is allowed to fail without throwing an exception. In this case a zero memory descriptor will be returned. This flag is optional and defaults to false.
Returns
A memory descriptor for the region.
Examples:
cpu_rnn_inference_f32.cpp, cpu_rnn_inference_int8.cpp, and rnn_training_f32.cpp.

◆ reshape()

desc dnnl::memory::desc::reshape ( const memory::dims dims,
bool  allow_empty = false 
) const
inline

Constructs a memory descriptor by reshaping an existing one.

The new memory descriptor inherits the data type. This operation is valid only for memory descriptors that have format_kind set to dnnl::memory::format_kind::blocked or dnnl::memory::format_kind::any.

The operation ensures that the transformation of the physical memory format corresponds to the transformation of the logical dimensions. If such transformation is impossible, the function either throws an exception (default) or returns a zero memory descriptor depending on the allow_empty flag.

The reshape operation can be described as a combination of the following basic operations:

  1. Add a dimension of size 1. This is always possible.
  2. Remove a dimension of size 1. This is possible only if the dimension has no padding (i.e. padded_dims[dim] == dims[dim] && dims[dim] == 1).
  3. Split a dimension into multiple ones. This is possible only if the size of the dimension is exactly equal to the product of the split ones and the dimension does not have padding (i.e. padded_dims[dim] = dims[dim]).
  4. Joining multiple consecutive dimensions into a single one. As in the cases above, this requires that the dimensions do not have padding and that the memory format is such that in physical memory these dimensions are dense and have the same order as their logical counterparts. This also assumes that these dimensions are not blocked.
    • Here, dense means: stride for dim[i] == (stride for dim[i + 1]) * dim[i + 1];
    • And same order means: i < j <=> stride for dim[i] < stride for dim[j].
Warning
Some combinations of physical memory layout and/or offsets or dimensions may result in a failure to make a reshape.
Parameters
dimsNew dimensions. The product of dimensions must remain constant.
allow_emptyA flag signifying whether construction is allowed to fail without throwing an exception. In this case a zero memory descriptor will be returned. This flag is optional and defaults to false.
Returns
A new memory descriptor with new dimensions.

◆ permute_axes()

desc dnnl::memory::desc::permute_axes ( const std::vector< int > &  permutation,
bool  allow_empty = false 
) const
inline

Constructs a memory descriptor by permuting axes in an existing one.

The physical memory layout representation is adjusted accordingly to maintain the consistency between the logical and physical parts of the memory descriptor.

The new memory descriptor inherits the data type. This operation is valid only for memory descriptors that have format_kind set to dnnl::memory::format_kind::blocked or dnnl::memory::format_kind::any.

The logical axes will be permuted in the following manner:

for (i: 0 .. ndims())
new_desc.dims()[permutation[i]] = dims()[i];

Example:

std::vector<int> permutation = {1, 0}; // swap the first and
// the second axes
dnnl::memory::desc expect_out_md(
assert(in_md.permute_axes(permutation) == expect_out_md);
Parameters
permutationAxes permutation.
allow_emptyA flag signifying whether construction is allowed to fail without throwing an exception. In this case a zero memory descriptor will be returned. This flag is optional and defaults to false.
Returns
A new memory descriptor with new dimensions.

◆ dims()

memory::dims dnnl::memory::desc::dims ( ) const
inline

Returns dimensions of the memory descriptor.

Potentially expensive due to the data copy involved.

Returns
A copy of the dimensions vector.
Examples:
inference_int8_matmul.cpp.

◆ data_type()

memory::data_type dnnl::memory::desc::data_type ( ) const
inline

Returns the data type of the memory descriptor.

Returns
The data type.

◆ get_size()

size_t dnnl::memory::desc::get_size ( ) const
inline

Returns size of the memory descriptor in bytes.

Returns
The number of bytes required to allocate a memory buffer for the memory object described by this memory descriptor including the padding area.
Examples:
cpu_rnn_inference_f32.cpp, cpu_rnn_inference_int8.cpp, and performance_profiling.cpp.

◆ is_zero()

bool dnnl::memory::desc::is_zero ( ) const
inline

Checks whether the memory descriptor is zero (empty).

Returns
true if the memory descriptor describes an empty memory and false otherwise.

◆ operator==()

bool dnnl::memory::desc::operator== ( const desc other) const
inline

An equality operator.

Parameters
otherAnother memory descriptor.
Returns
Whether this and the other memory descriptors have the same format tag, dimensions, strides, blocking, etc.

◆ operator!=()

bool dnnl::memory::desc::operator!= ( const desc other) const
inline

An inequality operator.

Parameters
otherAnother memory descriptor.
Returns
Whether this and the other memory descriptors describe different memory.

The documentation for this struct was generated from the following file: