CudaDeviceBuffer Class Reference

Simple memory buffer using un-managed pinned host memory when compiled with NVCC. Obviously this class is making explicit used of CUDA so replace it with your own memory allocator if you are not using CUDA. More...

#include <nanovdb/util/CudaDeviceBuffer.h>

Public Member Functions
	CudaDeviceBuffer (uint64_t size=0)
	CudaDeviceBuffer (const CudaDeviceBuffer &)=delete
	Disallow copy-construction. More...
	CudaDeviceBuffer (CudaDeviceBuffer &&other) noexcept
	Move copy-constructor. More...
CudaDeviceBuffer &	operator= (const CudaDeviceBuffer &)=delete
	Disallow copy assignment operation. More...
CudaDeviceBuffer &	operator= (CudaDeviceBuffer &&other) noexcept
	Move copy assignment operation. More...
	~CudaDeviceBuffer ()
	Destructor frees memory on both the host and device. More...
void	init (uint64_t size)
uint8_t *	data () const
uint8_t *	deviceData () const
void	deviceUpload (void *stream=0, bool sync=true) const
	Copy grid from the CPU/host to the GPU/device. If sync is false the memory copy is asynchronous! More...
void	deviceDownload (void *stream=0, bool sync=true) const
	Copy grid from the GPU/device to the CPU/host. If sync is false the memory copy is asynchronous! More...
uint64_t	size () const
	Returns the size in bytes of the raw memory buffer managed by this allocator. More...
bool	empty () const
	Returns true if this allocator is empty, i.e. has no allocated memory. More...
void	clear ()
	De-allocate all memory managed by this allocator and set all pointer to NULL. More...

Static Public Member Functions
static CudaDeviceBuffer	create (uint64_t size, const CudaDeviceBuffer *context=nullptr)

Detailed Description

Simple memory buffer using un-managed pinned host memory when compiled with NVCC. Obviously this class is making explicit used of CUDA so replace it with your own memory allocator if you are not using CUDA.

Note

While CUDA's pinned host memory allows for asynchronous memory copy between host and device it is significantly slower then cached (un-pinned) memory on the host.

Constructor & Destructor Documentation

CudaDeviceBuffer ( uint64_t  size = 0 )

inline

CudaDeviceBuffer ( const CudaDeviceBuffer &  )

delete

Disallow copy-construction.

CudaDeviceBuffer ( CudaDeviceBuffer &&  other )

inlinenoexcept

Move copy-constructor.

~CudaDeviceBuffer ( )

inline

Destructor frees memory on both the host and device.

Member Function Documentation

void clear ( )

inline

De-allocate all memory managed by this allocator and set all pointer to NULL.

CudaDeviceBuffer create ( uint64_t  size, const CudaDeviceBuffer *  context = nullptr  )

inlinestatic

uint8_t* data ( ) const

inline

Warning

Note that the pointer can be NULL is the allocator was not initialized!

uint8_t* deviceData ( ) const

inline

void deviceDownload ( void *  stream = 0, bool  sync = true  ) const

inline

Copy grid from the GPU/device to the CPU/host. If sync is false the memory copy is asynchronous!

void deviceUpload ( void *  stream = 0, bool  sync = true  ) const

inline

Copy grid from the CPU/host to the GPU/device. If sync is false the memory copy is asynchronous!

Note

This will allocate memory on the GPU/device if it is not already allocated

bool empty ( ) const

inline

Returns true if this allocator is empty, i.e. has no allocated memory.

void init ( uint64_t  size )

inline

CudaDeviceBuffer& operator= ( const CudaDeviceBuffer &  )

delete

Disallow copy assignment operation.

CudaDeviceBuffer& operator= ( CudaDeviceBuffer &&  other )

inlinenoexcept

Move copy assignment operation.

uint64_t size ( ) const

inline

Returns the size in bytes of the raw memory buffer managed by this allocator.