container::TensorBuffer Class Reference

Interface to access the raw ref-counted data buffer. More...

#include <tensor_buffer.h>

Inheritance diagram for container::TensorBuffer:

Collaboration diagram for container::TensorBuffer:

Public Member Functions
	TensorBuffer (base::core::Allocator *alloc, void *data_ptr)
	Construct a new TensorBuffer object.
	TensorBuffer (base::core::Allocator *alloc, size_t size)
	TensorBuffer (void *data_ptr)
	Construct a new TensorBuffer object.
	TensorBuffer (TensorBuffer &&other) noexcept
	Move constructor for the TensorBuffer class.
	~TensorBuffer ()
	Destroy the TensorBuffer object.
void *	data () const
	Get the raw data pointer.
void	resize (size_t size)
	resize the tensor buffer
size_t	GetAllocatedBytes () const
	Get the size of the buffer.
TensorBuffer *	root_buffer ()
	Get the root TensorBuffer object.
base::core::Allocator *	allocator () const
	Get the Allocator object used in this class.
template<typename T >
T *	base () const
	Reinterpret the buffer as an array of type T.
virtual bool	OwnsMemory () const
	Check whether this TensorBuffer owns the underlying memory.
DeviceType	GetDeviceType () const
	Get the type of device used by the TensorBuffer.
TensorBuffer &	operator= (const TensorBuffer &other)
	Assignment operator overload for the TensorBuffer class.
TensorBuffer &	operator= (TensorBuffer &&other) noexcept
	Move assignment operator overload for the TensorBuffer class.
Public Member Functions inherited from base::core::counted_base
	counted_base ()
	Default constructor. Initializes the reference count to one.
void	ref () const
	Increases the reference count by one.
bool	unref () const
	Decreases the reference count by one.
int_fast32_t	ref_count () const
	Gets the current reference count.
bool	ref_count_is_one () const
	Checks if the reference count is one.

Private Attributes
base::core::Allocator *	alloc_ = nullptr
	Pointer to the allocator used for memory allocation.
void *	data_ = nullptr
	Pointer to the underlying data buffer.
bool	owns_memory_ = false
	Bool to indicate whether this tensor owns it's memory.
int64_t	allocated_bytes_ = 0
	The number of bytes allocated for this buffer.

Additional Inherited Members
Protected Member Functions inherited from base::core::counted_base
virtual	~counted_base ()
	Virtual destructor.

Detailed Description

Interface to access the raw ref-counted data buffer.

Constructor & Destructor Documentation

TensorBuffer() [1/4]

container::TensorBuffer::TensorBuffer	(	base::core::Allocator *	alloc,
		void *	data_ptr
	)

Construct a new TensorBuffer object.

Parameters

alloc	Pointer to the allocator to use for memory allocation.
data_ptr	Pointer to the underlying data buffer.

TensorBuffer() [2/4]

container::TensorBuffer::TensorBuffer	(	base::core::Allocator *	alloc,
		size_t	size
	)

Here is the call graph for this function:

TensorBuffer() [3/4]

container::TensorBuffer::TensorBuffer ( void *  data_ptr )

explicit

Construct a new TensorBuffer object.

This is a reference TensorBuffer, does not owns memory itself.

Parameters

data_ptr

Pointer to the given data buffer.

TensorBuffer() [4/4]

container::TensorBuffer::TensorBuffer ( TensorBuffer &&  other )

noexcept

Move constructor for the TensorBuffer class.

This constructor is used to move the contents and ownership of another TensorBuffer object into the newly created object using move semantics. The source TensorBuffer's resources will be taken over, and the source object will be left in a valid but unspecified state.

Parameters

other

The rvalue reference to the source TensorBuffer object to be moved.

Note

This function is declared as noexcept, indicating that it does not throw exceptions.

~TensorBuffer()

container::TensorBuffer::~TensorBuffer

(

)

Destroy the TensorBuffer object.

Here is the call graph for this function:

Member Function Documentation

allocator()

base::core::Allocator * container::TensorBuffer::allocator

(

)

const

Get the Allocator object used in this class.

Returns

Allocator* Pointer to the Allocator object.

Here is the caller graph for this function:

base()

template<typename T >

T * container::TensorBuffer::base ( ) const

inline

Reinterpret the buffer as an array of type T.

Template Parameters

T	The type to reinterpret the buffer as.

Returns

T* Pointer to the underlying buffer reinterpreted as type T.

Here is the call graph for this function:

Here is the caller graph for this function:

data()

void * container::TensorBuffer::data

(

)

const

Get the raw data pointer.

Returns

void* Pointer to the underlying data buffer.

Here is the caller graph for this function:

GetAllocatedBytes()

size_t container::TensorBuffer::GetAllocatedBytes

(

)

const

Get the size of the buffer.

Returns

size_t The size of the buffer in bytes.

Here is the caller graph for this function:

GetDeviceType()

DeviceType container::TensorBuffer::GetDeviceType

(

)

const

Get the type of device used by the TensorBuffer.

Returns

MemoryType The type of memory used by the TensorBuffer.

Here is the call graph for this function:

Here is the caller graph for this function:

operator=() [1/2]

TensorBuffer & container::TensorBuffer::operator=

(

const TensorBuffer &

other

)

Assignment operator overload for the TensorBuffer class.

This operator is used to assign the values of another TensorBuffer object to the current object. It performs a deep copy of the data from the source TensorBuffer to the destination TensorBuffer.

Parameters

other

The source TensorBuffer object whose values will be assigned.

Returns

A reference to the current TensorBuffer object after the assignment.

Here is the call graph for this function:

operator=() [2/2]

TensorBuffer & container::TensorBuffer::operator= ( TensorBuffer &&  other )

noexcept

Move assignment operator overload for the TensorBuffer class.

This operator is used to move the contents and ownership of another TensorBuffer object into the current object using move semantics. The source TensorBuffer's resources will be taken over, and the source object will be left in a valid but unspecified state.

Parameters

other

The rvalue reference to the source TensorBuffer object to be moved.

Returns

A reference to the current TensorBuffer object after the move assignment.

Note

This function is declared as noexcept, indicating that it does not throw exceptions.

OwnsMemory()

bool container::TensorBuffer::OwnsMemory ( ) const

virtual

Check whether this TensorBuffer owns the underlying memory.

Returns

true If the TensorBuffer owns the underlying memory.

false If the TensorBuffer does not own the underlying memory.

Here is the caller graph for this function:

resize()

void container::TensorBuffer::resize

(

size_t

size

)

resize the tensor buffer

Here is the call graph for this function:

Here is the caller graph for this function:

root_buffer()

TensorBuffer * container::TensorBuffer::root_buffer

(

)

Get the root TensorBuffer object.

If this TensorBuffer is a sub-buffer of another TensorBuffer, returns that TensorBuffer. Otherwise, returns this.

Returns

TensorBuffer* Pointer to the root TensorBuffer object.

Here is the caller graph for this function:

Member Data Documentation

alloc_

base::core::Allocator* container::TensorBuffer::alloc_ = nullptr

private

Pointer to the allocator used for memory allocation.

allocated_bytes_

int64_t container::TensorBuffer::allocated_bytes_ = 0

private

The number of bytes allocated for this buffer.

data_

void* container::TensorBuffer::data_ = nullptr

private

Pointer to the underlying data buffer.

owns_memory_

bool container::TensorBuffer::owns_memory_ = false

private

Bool to indicate whether this tensor owns it's memory.

---

The documentation for this class was generated from the following files:

• /home/runner/work/abacus-develop/abacus-develop/source/source_base/module_container/ATen/core/tensor_buffer.h
• /home/runner/work/abacus-develop/abacus-develop/source/source_base/module_container/ATen/core/tensor_buffer.cpp