uni20::cuda::ReadAccess< T > Class Template Reference

Scoped read-only access to a CUDA buffer on one stream. More...

#include <buffer.hpp>

Public Types
using	element_type = T

Public Member Functions
T const *	data () const noexcept
	Return the typed device pointer for read-only CUDA work.
ReadAccess &	operator= (ReadAccess &&other) noexcept
ReadAccess &	operator= (ReadAccess const &)=delete
	ReadAccess (ReadAccess &&other) noexcept
	ReadAccess (ReadAccess const &)=delete
void	release () noexcept
	Publish this read completion and end the scoped access.
void	release_after_synchronization () noexcept
	End access after the caller has synchronized the operation stream.
void	release_with_completion (Completion completion) noexcept
	Publish a caller-recorded completion and end scoped access.
std::size_t	size () const noexcept
	Return the number of typed elements in the view.
std::size_t	size_bytes () const noexcept
	Return the number of bytes in the view.
CudaBuffer< T > const &	storage () const
	Return the CUDA buffer retained by this access object.
	~ReadAccess ()

Friends
class	CudaBuffer< T >

Detailed Description

template<typename T>
class uni20::cuda::ReadAccess< T >

Scoped read-only access to a CUDA buffer on one stream.

The constructor installs predecessor waits on the stream. The destructor records a reader completion at the current stream tail and publishes it back to the buffer.

Member Typedef Documentation

element_type

template<typename T >

using uni20::cuda::ReadAccess< T >::element_type = T

Constructor & Destructor Documentation

ReadAccess() [1/2]

template<typename T >

uni20::cuda::ReadAccess< T >::ReadAccess ( ReadAccess< T > const &  )

delete

ReadAccess() [2/2]

template<typename T >

uni20::cuda::ReadAccess< T >::ReadAccess ( ReadAccess< T > &&  other )

inlinenoexcept

~ReadAccess()

template<typename T >

uni20::cuda::ReadAccess< T >::~ReadAccess ( )

inline

Member Function Documentation

data()

template<typename T >

T const * uni20::cuda::ReadAccess< T >::data ( ) const

inlinenoexcept

Return the typed device pointer for read-only CUDA work.

operator=() [1/2]

template<typename T >

ReadAccess & uni20::cuda::ReadAccess< T >::operator= ( ReadAccess< T > &&  other )

inlinenoexcept

operator=() [2/2]

template<typename T >

ReadAccess & uni20::cuda::ReadAccess< T >::operator= ( ReadAccess< T > const &  )

delete

release()

template<typename T >

void uni20::cuda::ReadAccess< T >::release ( )

inlinenoexcept

Publish this read completion and end the scoped access.

release_after_synchronization()

template<typename T >

void uni20::cuda::ReadAccess< T >::release_after_synchronization ( )

inlinenoexcept

End access after the caller has synchronized the operation stream.

release_with_completion()

template<typename T >

void uni20::cuda::ReadAccess< T >::release_with_completion ( Completion  completion )

inlinenoexcept

Publish a caller-recorded completion and end scoped access.

size()

template<typename T >

std::size_t uni20::cuda::ReadAccess< T >::size ( ) const

inlinenoexcept

Return the number of typed elements in the view.

size_bytes()

template<typename T >

std::size_t uni20::cuda::ReadAccess< T >::size_bytes ( ) const

inlinenoexcept

Return the number of bytes in the view.

storage()

template<typename T >

CudaBuffer< T > const & uni20::cuda::ReadAccess< T >::storage ( ) const

inline

Return the CUDA buffer retained by this access object.

Friends And Related Symbol Documentation

CudaBuffer< T >

template<typename T >

friend class CudaBuffer< T >

friend

---

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

• buffer.hpp