cuda::CudaDriverFrontend::Context Class Reference

CUDA Driver API context. More...

#include <CudaDriverFrontend.h>

Collaboration diagram for cuda::CudaDriverFrontend::Context:

List of all members.

Public Types
enum	MemcpyKind {   HostToHost = 0, HostToDevice = 1, DeviceToHost = 2, DeviceToDevice = 3,   MemcpyKind_invalid }
Public Member Functions
	Context ()
	~Context ()
void	_memcpy (void *dst, const void *src, size_t count, MemcpyKind kind)
	performs a memcpy on selected device
void	_memoryError (const void *address, size_t count, const std::string &func="")
	report a memory error and throw an exception
void	_enumerateDevices ()
	create devices if they do not exist
executive::Device &	_getDevice ()
	gets the current device for the current thread
std::string	_formatError (const std::string &message)
	returns an Ocelot-formatted error message
void	_registerModule (ModuleMap::iterator module)
void	_registerModule (const std::string &name)
void	_registerAllModules ()
Public Attributes
ModuleMap	_modules
	Registered modules.
HostThreadContext	_hostThreadContext
	object storing context information needed while configuring calls
KernelLaunchConfiguration	_launchConfiguration
	configuration of next kernel
RegisteredKernelMap	_kernels
	maps kernel symbols to module-kernels
RegisteredTextureMap	_textures
	maps texture symbols to module-textures
RegisteredGlobalMap	_globals
	maps symbol pointers onto their device names
EventSet	_events
	set of live events
StreamSet	_streams
	set of live streams
DimensionMap	_dimensions
	The dimensions for multi-dimensional allocations.
GLBufferMap	_buffers
	Registered opengl buffers and mapping to graphics resources.
executive::Device *	_device
	device
int	_selectedDevice
	index of device bound to context
int	_nextSymbol
	the next symbol for dynamically registered kernels
unsigned int	_flags
	The device flags.
FatBinaryVector	_fatBinaries
	fatbinaries
translator::Translator::OptimizationLevel	_optimization
	optimization level
int	_referenceCount
	number of references to this context

---

Detailed Description

CUDA Driver API context.

---

Member Enumeration Documentation

enum cuda::CudaDriverFrontend::Context::MemcpyKind

Enumerator:

HostToHost
HostToDevice
DeviceToHost
DeviceToDevice
MemcpyKind_invalid

---

Constructor & Destructor Documentation

cuda::CudaDriverFrontend::Context::Context

(

)

cuda::CudaDriverFrontend::Context::~Context

(

)

---

Member Function Documentation

void cuda::CudaDriverFrontend::Context::_enumerateDevices

(

)

create devices if they do not exist

std::string cuda::CudaDriverFrontend::Context::_formatError

(

const std::string &

message

)

returns an Ocelot-formatted error message

executive::Device & cuda::CudaDriverFrontend::Context::_getDevice

(

)

gets the current device for the current thread

void cuda::CudaDriverFrontend::Context::_memcpy	(	void *	dst,
		const void *	src,
		size_t	count,
		MemcpyKind	kind
	)

performs a memcpy on selected device

void cuda::CudaDriverFrontend::Context::_memoryError	(	const void *	address,
		size_t	count,
		const std::string &	func = ""
	)

report a memory error and throw an exception

void cuda::CudaDriverFrontend::Context::_registerAllModules

(

)

void cuda::CudaDriverFrontend::Context::_registerModule

(

ModuleMap::iterator

module

)

void cuda::CudaDriverFrontend::Context::_registerModule

(

const std::string &

name

)

---

Member Data Documentation

GLBufferMap cuda::CudaDriverFrontend::Context::_buffers

Registered opengl buffers and mapping to graphics resources.

executive::Device* cuda::CudaDriverFrontend::Context::_device

device

DimensionMap cuda::CudaDriverFrontend::Context::_dimensions

The dimensions for multi-dimensional allocations.

EventSet cuda::CudaDriverFrontend::Context::_events

set of live events

FatBinaryVector cuda::CudaDriverFrontend::Context::_fatBinaries

fatbinaries

unsigned int cuda::CudaDriverFrontend::Context::_flags

The device flags.

RegisteredGlobalMap cuda::CudaDriverFrontend::Context::_globals

maps symbol pointers onto their device names

HostThreadContext cuda::CudaDriverFrontend::Context::_hostThreadContext

object storing context information needed while configuring calls

RegisteredKernelMap cuda::CudaDriverFrontend::Context::_kernels

maps kernel symbols to module-kernels

KernelLaunchConfiguration cuda::CudaDriverFrontend::Context::_launchConfiguration

configuration of next kernel

ModuleMap cuda::CudaDriverFrontend::Context::_modules

Registered modules.

int cuda::CudaDriverFrontend::Context::_nextSymbol

the next symbol for dynamically registered kernels

translator::Translator::OptimizationLevel cuda::CudaDriverFrontend::Context::_optimization

optimization level

int cuda::CudaDriverFrontend::Context::_referenceCount

number of references to this context

int cuda::CudaDriverFrontend::Context::_selectedDevice

index of device bound to context

StreamSet cuda::CudaDriverFrontend::Context::_streams

set of live streams

RegisteredTextureMap cuda::CudaDriverFrontend::Context::_textures

maps texture symbols to module-textures

---

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

• ocelot/cuda/interface/CudaDriverFrontend.h
• ocelot/cuda/implementation/CudaDriverFrontend.cpp