From 8efb48a6847c5cd166c561127ae6611150963ce3 Mon Sep 17 00:00:00 2001 From: Nikhil Raj Date: Fri, 19 May 2023 11:14:28 +0100 Subject: Update Doxygen docu for 23.05 Signed-off-by: Nikhil Raj Change-Id: I0a992286f14fa68fcc6e5eba31ac39fed003cbbe --- .../structarmnn_1_1_batch_mat_mul_descriptor.xhtml | 709 +++++++++++++++++++++ 1 file changed, 709 insertions(+) create mode 100644 23.05/structarmnn_1_1_batch_mat_mul_descriptor.xhtml (limited to '23.05/structarmnn_1_1_batch_mat_mul_descriptor.xhtml') diff --git a/23.05/structarmnn_1_1_batch_mat_mul_descriptor.xhtml b/23.05/structarmnn_1_1_batch_mat_mul_descriptor.xhtml new file mode 100644 index 0000000000..0f7986b4fb --- /dev/null +++ b/23.05/structarmnn_1_1_batch_mat_mul_descriptor.xhtml @@ -0,0 +1,709 @@ + + + + + + + + + + + + + +ArmNN: BatchMatMulDescriptor Struct Reference + + + + + + + + + + + + + + + + +
+
+ + + + ArmNN + + + +
+
+  23.05 +
+
+
+ + + + + + + +
+
+ +
+
+
+ +
+ +
+
+ + +
+ +
+ +
+
+Public Member Functions | +Static Public Member Functions | +Public Attributes | +List of all members
+
+
BatchMatMulDescriptor Struct Reference
+
+
+ +

A BatchMatMulDescriptor for the BatchMatMul operator. + More...

+ +

#include <Descriptors.hpp>

+
+Inheritance diagram for BatchMatMulDescriptor:
+
+
+ + +BaseDescriptor + +
+ + + + + + + + + + + +

+Public Member Functions

 BatchMatMulDescriptor (bool transposeX=false, bool transposeY=false, bool adjointX=false, bool adjointY=false, DataLayout dataLayoutX=DataLayout::NCHW, DataLayout dataLayoutY=DataLayout::NCHW)
 
bool operator== (const BatchMatMulDescriptor &rhs) const
 
- Public Member Functions inherited from BaseDescriptor
virtual bool IsNull () const
 
virtual ~BaseDescriptor ()=default
 
+ + + + + + + + + + + + + + +

+Static Public Member Functions

static std::pair< std::pair< unsigned int, unsigned int >, std::pair< unsigned int, unsigned int > > GetAxesToMul (const BatchMatMulDescriptor &desc, const TensorShape &tensorXShape, const TensorShape &tensorYShape)
 
static std::pair< std::vector< unsigned int >, std::vector< unsigned int > > GetAxesNotMul (const BatchMatMulDescriptor &desc, const TensorShape &inputXShape, const TensorShape &inputYShape)
 
static std::pair< unsigned int, unsigned int > GetAxesToMul (DataLayout dataLayout, const TensorShape &tensorShape)
 Static helper to get the two axes (for each input) for multiplication. More...
 
static std::vector< unsigned int > GetAxesNotMul (DataLayout dataLayout, const TensorShape &tensorShape)
 Static helper to get the axes (for each input) that will not be multiplied together. More...
 
static PermutationVector GetPermuteVec (DataLayout dataLayout, const TensorShape &tensorShape)
 Static helper to get the axes which will be transposed. More...
 
+ + + + + + + + + + + + + + + + +

+Public Attributes

bool m_TransposeX
 Transpose the slices of each input tensor Transpose and Adjoint can not both be set to true for the same tensor at the same time. More...
 
bool m_TransposeY
 
bool m_AdjointX
 Adjoint the slices of each input tensor Transpose and Adjoint can not both be set to true for the same tensor at the same time. More...
 
bool m_AdjointY
 
DataLayout m_DataLayoutX
 Data layout of each input tensor, such as NHWC/NDHWC (leave as default for arbitrary layout) More...
 
DataLayout m_DataLayoutY
 
+

Detailed Description

+

A BatchMatMulDescriptor for the BatchMatMul operator.

+ +

Definition at line 1551 of file Descriptors.hpp.

+

Constructor & Destructor Documentation

+ +

◆ BatchMatMulDescriptor()

+ +
+
+ + + + + +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
BatchMatMulDescriptor (bool transposeX = false,
bool transposeY = false,
bool adjointX = false,
bool adjointY = false,
DataLayout dataLayoutX = DataLayout::NCHW,
DataLayout dataLayoutY = DataLayout::NCHW 
)
+ 		+inline
+
+ +

Definition at line 1553 of file Descriptors.hpp.

+
1559  : m_TransposeX(transposeX)
+
1560  , m_TransposeY(transposeY)
+
1561  , m_AdjointX(adjointX)
+
1562  , m_AdjointY(adjointY)
+
1563  , m_DataLayoutX(dataLayoutX)
+
1564  , m_DataLayoutY(dataLayoutY)
+
1565  {}
+
+
+
+

Member Function Documentation

+ +

◆ GetAxesNotMul() [1/2]

+ +
+
+ + + + + +
+ + + + + + + + + + + + + + + + + + + + + + + + +
std::pair< std::vector< unsigned int >, std::vector< unsigned int > > GetAxesNotMul (const BatchMatMulDescriptor & desc,
const TensorShape & inputXShape,
const TensorShape & inputYShape 
)
+ 		+static
+
+ +

Definition at line 467 of file Descriptors.cpp.

+
471 {
+
472  return { GetAxesNotMul(desc.m_DataLayoutX, inputXShape),
+
473  GetAxesNotMul(desc.m_DataLayoutY, inputYShape) };
+
474 }
+
+

References BatchMatMulDescriptor::m_DataLayoutX, and BatchMatMulDescriptor::m_DataLayoutY.

+ +

Referenced by BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ GetAxesNotMul() [2/2]

+ +
+
+ + + + + +
+ + + + + + + + + + + + + + + + + + +
std::vector< unsigned int > GetAxesNotMul (DataLayout dataLayout,
const TensorShape & tensorShape 
)
+ 		+static
+
+ +

Static helper to get the axes (for each input) that will not be multiplied together.

+ +

Definition at line 497 of file Descriptors.cpp.

+
500 {
+
501  auto axesToMul = BatchMatMulDescriptor::GetAxesToMul(dataLayout, tensorShape);
+
502  std::vector<unsigned int> axesNotMul;
+
503  for(unsigned int i = 0; i < tensorShape.GetNumDimensions(); i++)
+
504  {
+
505  if(i == axesToMul.first || i == axesToMul.second)
+
506  {
+
507  continue;
+
508  }
+
509  axesNotMul.push_back(i);
+
510  }
+
511  return axesNotMul;
+
512 }
+
+

References BatchMatMulDescriptor::GetAxesToMul(), and TensorShape::GetNumDimensions().

+ +
+
+ +

◆ GetAxesToMul() [1/2]

+ +
+
+ + + + + +
+ + + + + + + + + + + + + + + + + + + + + + + + +
std::pair< std::pair< unsigned int, unsigned int >, std::pair< unsigned int, unsigned int > > GetAxesToMul (const BatchMatMulDescriptor & desc,
const TensorShape & tensorXShape,
const TensorShape & tensorYShape 
)
+ 		+static
+
+ +

Definition at line 459 of file Descriptors.cpp.

+
463 {
+
464  return { GetAxesToMul(desc.m_DataLayoutX, tensorXShape),
+
465  GetAxesToMul(desc.m_DataLayoutY, tensorYShape) };
+
466 }
+
+

References BatchMatMulDescriptor::m_DataLayoutX, and BatchMatMulDescriptor::m_DataLayoutY.

+ +

Referenced by BatchMatMulDescriptor::GetAxesNotMul(), BatchMatMulDescriptor::GetPermuteVec(), BatchMatMulLayer::InferOutputShapes(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ GetAxesToMul() [2/2]

+ +
+
+ + + + + +
+ + + + + + + + + + + + + + + + + + +
std::pair< unsigned int, unsigned int > GetAxesToMul (DataLayout dataLayout,
const TensorShape & tensorShape 
)
+ 		+static
+
+ +

Static helper to get the two axes (for each input) for multiplication.

+ +

Definition at line 476 of file Descriptors.cpp.

+
479 {
+
480  auto numDims = tensorShape.GetNumDimensions();
+
481  std::pair<unsigned int, unsigned int> axes = { numDims-2, numDims-1 };
+
482  switch(dataLayout)
+
483  {
+
484  case DataLayout::NDHWC:
+
485  case DataLayout::NHWC:
+
486  axes.first -= 1;
+
487  axes.second -= 1;
+
488  break;
+
489  case DataLayout::NCDHW:
+
490  case DataLayout::NCHW:
+
491  default:
+
492  break;
+
493  }
+
494  return axes;
+
495 }
+
+

References TensorShape::GetNumDimensions(), armnn::NCDHW, armnn::NCHW, armnn::NDHWC, and armnn::NHWC.

+ +
+
+ +

◆ GetPermuteVec()

+ +
+
+ + + + + +
+ + + + + + + + + + + + + + + + + + +
PermutationVector GetPermuteVec (DataLayout dataLayout,
const TensorShape & tensorShape 
)
+ 		+static
+
+ +

Static helper to get the axes which will be transposed.

+ +

Definition at line 514 of file Descriptors.cpp.

+
517 {
+
518  std::vector<unsigned int> vec;
+
519  auto axesToMul = BatchMatMulDescriptor::GetAxesToMul(dataLayout, tensorShape);
+
520  for(unsigned int i = 0; i < tensorShape.GetNumDimensions(); i++)
+
521  {
+
522  if(i == axesToMul.first)
+
523  {
+
524  vec.push_back(i+1);
+
525  }
+
526  else if(i == axesToMul.second)
+
527  {
+
528  vec.push_back(i-1);
+
529  }
+
530  else
+
531  {
+
532  vec.push_back(i);
+
533  }
+
534  }
+
535  return PermutationVector(vec.data(),
+
536  static_cast<unsigned int>(vec.size()));
+
537 }
+
+

References BatchMatMulDescriptor::GetAxesToMul(), and TensorShape::GetNumDimensions().

+ +

Referenced by BatchMatMulLayer::InferOutputShapes(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ operator==()

+ +
+
+ + + + + +
+ + + + + + + + +
bool operator== (const BatchMatMulDescriptor & rhs) const
+ 		+inline
+
+ +

Definition at line 1567 of file Descriptors.hpp.

+
1568  {
+
1569  return m_TransposeX == rhs.m_TransposeX &&
+
1570  m_TransposeY == rhs.m_TransposeY &&
+
1571  m_AdjointX == rhs.m_AdjointX &&
+
1572  m_AdjointY == rhs.m_AdjointY &&
+
1573  m_DataLayoutX == rhs.m_DataLayoutX &&
+
1574  m_DataLayoutY == rhs.m_DataLayoutY;
+
1575  }
+
+

References BatchMatMulDescriptor::m_AdjointX, BatchMatMulDescriptor::m_AdjointY, BatchMatMulDescriptor::m_DataLayoutX, BatchMatMulDescriptor::m_DataLayoutY, BatchMatMulDescriptor::m_TransposeX, and BatchMatMulDescriptor::m_TransposeY.

+ +
+
+

Member Data Documentation

+ +

◆ m_AdjointX

+ +
+
+ + + + +
bool m_AdjointX
+
+ +

Adjoint the slices of each input tensor Transpose and Adjoint can not both be set to true for the same tensor at the same time.

+ +

Definition at line 1584 of file Descriptors.hpp.

+ +

Referenced by armnn::ClBatchMatMulValidate(), ClBatchMatMulWorkload::ClBatchMatMulWorkload(), BatchMatMulLayer::InferOutputShapes(), armnn::NeonBatchMatMulValidate(), NeonBatchMatMulWorkload::NeonBatchMatMulWorkload(), BatchMatMulDescriptor::operator==(), StringifyLayerParameters< BatchMatMulDescriptor >::Serialize(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ m_AdjointY

+ +
+
+ + + + +
bool m_AdjointY
+
+ +

Definition at line 1585 of file Descriptors.hpp.

+ +

Referenced by armnn::ClBatchMatMulValidate(), ClBatchMatMulWorkload::ClBatchMatMulWorkload(), BatchMatMulLayer::InferOutputShapes(), armnn::NeonBatchMatMulValidate(), NeonBatchMatMulWorkload::NeonBatchMatMulWorkload(), BatchMatMulDescriptor::operator==(), StringifyLayerParameters< BatchMatMulDescriptor >::Serialize(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ m_DataLayoutX

+ +
+
+ + + + +
DataLayout m_DataLayoutX
+
+ +

Data layout of each input tensor, such as NHWC/NDHWC (leave as default for arbitrary layout)

+ +

Definition at line 1588 of file Descriptors.hpp.

+ +

Referenced by armnn::ClBatchMatMulValidate(), ClBatchMatMulWorkload::ClBatchMatMulWorkload(), BatchMatMulDescriptor::GetAxesNotMul(), BatchMatMulDescriptor::GetAxesToMul(), BatchMatMulLayer::InferOutputShapes(), armnn::NeonBatchMatMulValidate(), NeonBatchMatMulWorkload::NeonBatchMatMulWorkload(), BatchMatMulDescriptor::operator==(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ m_DataLayoutY

+ +
+
+ + + + +
DataLayout m_DataLayoutY
+
+ +

Definition at line 1589 of file Descriptors.hpp.

+ +

Referenced by armnn::ClBatchMatMulValidate(), ClBatchMatMulWorkload::ClBatchMatMulWorkload(), BatchMatMulDescriptor::GetAxesNotMul(), BatchMatMulDescriptor::GetAxesToMul(), BatchMatMulLayer::InferOutputShapes(), armnn::NeonBatchMatMulValidate(), NeonBatchMatMulWorkload::NeonBatchMatMulWorkload(), BatchMatMulDescriptor::operator==(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ m_TransposeX

+ +
+
+ + + + +
bool m_TransposeX
+
+ +

Transpose the slices of each input tensor Transpose and Adjoint can not both be set to true for the same tensor at the same time.

+ +

Definition at line 1579 of file Descriptors.hpp.

+ +

Referenced by ClBatchMatMulWorkload::ClBatchMatMulWorkload(), BatchMatMulLayer::InferOutputShapes(), NeonBatchMatMulWorkload::NeonBatchMatMulWorkload(), BatchMatMulDescriptor::operator==(), StringifyLayerParameters< BatchMatMulDescriptor >::Serialize(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+ +

◆ m_TransposeY

+ +
+
+ + + + +
bool m_TransposeY
+
+ +

Definition at line 1580 of file Descriptors.hpp.

+ +

Referenced by ClBatchMatMulWorkload::ClBatchMatMulWorkload(), BatchMatMulLayer::InferOutputShapes(), NeonBatchMatMulWorkload::NeonBatchMatMulWorkload(), BatchMatMulDescriptor::operator==(), StringifyLayerParameters< BatchMatMulDescriptor >::Serialize(), and BatchMatMulQueueDescriptor::Validate().

+ +
+
+
The documentation for this struct was generated from the following files: +
+
+
armnn::BatchMatMulDescriptor::m_TransposeX
bool m_TransposeX
Transpose the slices of each input tensor Transpose and Adjoint can not both be set to true for the s...
Definition: Descriptors.hpp:1579
+ +
armnn::BatchMatMulDescriptor::m_DataLayoutX
DataLayout m_DataLayoutX
Data layout of each input tensor, such as NHWC/NDHWC (leave as default for arbitrary layout)
Definition: Descriptors.hpp:1588
+ + + + + + +
armnn::BatchMatMulDescriptor::m_AdjointX
bool m_AdjointX
Adjoint the slices of each input tensor Transpose and Adjoint can not both be set to true for the sam...
Definition: Descriptors.hpp:1584
+
armnn::BatchMatMulDescriptor::GetAxesToMul
static std::pair< std::pair< unsigned int, unsigned int >, std::pair< unsigned int, unsigned int > > GetAxesToMul(const BatchMatMulDescriptor &desc, const TensorShape &tensorXShape, const TensorShape &tensorYShape)
Definition: Descriptors.cpp:459
+
armnn::BatchMatMulDescriptor::GetAxesNotMul
static std::pair< std::vector< unsigned int >, std::vector< unsigned int > > GetAxesNotMul(const BatchMatMulDescriptor &desc, const TensorShape &inputXShape, const TensorShape &inputYShape)
Definition: Descriptors.cpp:467
+ + + + -- cgit v1.2.1