Class providing functionality for querying the specified GPU properties. [詳解]

#include <cuda.hpp>

公開型
enum	ComputeMode { ComputeModeDefault , ComputeModeExclusive , ComputeModeProhibited , ComputeModeExclusiveProcess }

公開メンバ関数
CV_WRAP	DeviceInfo ()
	creates DeviceInfo object for the current GPU

CV_WRAP	DeviceInfo (int device_id)
	The constructors. [詳解]

CV_WRAP int	deviceID () const
	Returns system index of the CUDA device starting with 0.

const char *	name () const
	ASCII string identifying device

CV_WRAP size_t	totalGlobalMem () const
	global memory available on device in bytes

CV_WRAP size_t	sharedMemPerBlock () const
	shared memory available per block in bytes

CV_WRAP int	regsPerBlock () const
	32-bit registers available per block

CV_WRAP int	warpSize () const
	warp size in threads

CV_WRAP size_t	memPitch () const
	maximum pitch in bytes allowed by memory copies

CV_WRAP int	maxThreadsPerBlock () const
	maximum number of threads per block

CV_WRAP Vec3i	maxThreadsDim () const
	maximum size of each dimension of a block

CV_WRAP Vec3i	maxGridSize () const
	maximum size of each dimension of a grid

CV_WRAP int	clockRate () const
	clock frequency in kilohertz

CV_WRAP size_t	totalConstMem () const
	constant memory available on device in bytes

CV_WRAP int	majorVersion () const
	major compute capability

CV_WRAP int	minorVersion () const
	minor compute capability

CV_WRAP size_t	textureAlignment () const
	alignment requirement for textures

CV_WRAP size_t	texturePitchAlignment () const
	pitch alignment requirement for texture references bound to pitched memory

CV_WRAP int	multiProcessorCount () const
	number of multiprocessors on device

CV_WRAP bool	kernelExecTimeoutEnabled () const
	specified whether there is a run time limit on kernels

CV_WRAP bool	integrated () const
	device is integrated as opposed to discrete

CV_WRAP bool	canMapHostMemory () const
	device can map host memory with cudaHostAlloc/cudaHostGetDevicePointer

CV_WRAP DeviceInfo::ComputeMode	computeMode () const
	compute mode

CV_WRAP int	maxTexture1D () const
	maximum 1D texture size

CV_WRAP int	maxTexture1DMipmap () const
	maximum 1D mipmapped texture size

CV_WRAP int	maxTexture1DLinear () const
	maximum size for 1D textures bound to linear memory

CV_WRAP Vec2i	maxTexture2D () const
	maximum 2D texture dimensions

CV_WRAP Vec2i	maxTexture2DMipmap () const
	maximum 2D mipmapped texture dimensions

CV_WRAP Vec3i	maxTexture2DLinear () const
	maximum dimensions (width, height, pitch) for 2D textures bound to pitched memory

CV_WRAP Vec2i	maxTexture2DGather () const
	maximum 2D texture dimensions if texture gather operations have to be performed

CV_WRAP Vec3i	maxTexture3D () const
	maximum 3D texture dimensions

CV_WRAP int	maxTextureCubemap () const
	maximum Cubemap texture dimensions

CV_WRAP Vec2i	maxTexture1DLayered () const
	maximum 1D layered texture dimensions

CV_WRAP Vec3i	maxTexture2DLayered () const
	maximum 2D layered texture dimensions

CV_WRAP Vec2i	maxTextureCubemapLayered () const
	maximum Cubemap layered texture dimensions

CV_WRAP int	maxSurface1D () const
	maximum 1D surface size

CV_WRAP Vec2i	maxSurface2D () const
	maximum 2D surface dimensions

CV_WRAP Vec3i	maxSurface3D () const
	maximum 3D surface dimensions

CV_WRAP Vec2i	maxSurface1DLayered () const
	maximum 1D layered surface dimensions

CV_WRAP Vec3i	maxSurface2DLayered () const
	maximum 2D layered surface dimensions

CV_WRAP int	maxSurfaceCubemap () const
	maximum Cubemap surface dimensions

CV_WRAP Vec2i	maxSurfaceCubemapLayered () const
	maximum Cubemap layered surface dimensions

CV_WRAP size_t	surfaceAlignment () const
	alignment requirements for surfaces

CV_WRAP bool	concurrentKernels () const
	device can possibly execute multiple kernels concurrently

CV_WRAP bool	ECCEnabled () const
	device has ECC support enabled

CV_WRAP int	pciBusID () const
	PCI bus ID of the device

CV_WRAP int	pciDeviceID () const
	PCI device ID of the device

CV_WRAP int	pciDomainID () const
	PCI domain ID of the device

CV_WRAP bool	tccDriver () const
	true if device is a Tesla device using TCC driver, false otherwise

CV_WRAP int	asyncEngineCount () const
	number of asynchronous engines

CV_WRAP bool	unifiedAddressing () const
	device shares a unified address space with the host

CV_WRAP int	memoryClockRate () const
	peak memory clock frequency in kilohertz

CV_WRAP int	memoryBusWidth () const
	global memory bus width in bits

CV_WRAP int	l2CacheSize () const
	size of L2 cache in bytes

CV_WRAP int	maxThreadsPerMultiProcessor () const
	maximum resident threads per multiprocessor

CV_WRAP void	queryMemory (size_t &totalMemory, size_t &freeMemory) const
	gets free and total device memory

CV_WRAP size_t	freeMemory () const

CV_WRAP size_t	totalMemory () const

bool	supports (FeatureSet feature_set) const
	Provides information on CUDA feature support. [詳解]

CV_WRAP bool	isCompatible () const
	Checks the CUDA module and device compatibility. [詳解]

詳解

Class providing functionality for querying the specified GPU properties.

列挙型メンバ詳解

◆ ComputeMode

enum cv::cuda::DeviceInfo::ComputeMode

列挙値
ComputeModeDefault	default compute mode (Multiple threads can use cudaSetDevice with this device)
ComputeModeExclusive	compute-exclusive-thread mode (Only one thread in one process will be able to use cudaSetDevice with this device)
ComputeModeProhibited	compute-prohibited mode (No threads can use cudaSetDevice with this device)
ComputeModeExclusiveProcess	compute-exclusive-process mode (Many threads in one process will be able to use cudaSetDevice with this device)

構築子と解体子

◆ DeviceInfo()

CV_WRAP cv::cuda::DeviceInfo::DeviceInfo ( int device_id )

The constructors.

引数

device_id System index of the CUDA device starting with 0.

Constructs the DeviceInfo object for the specified device. If device_id parameter is missed, it constructs an object for the current device.

関数詳解

◆ isCompatible()

CV_WRAP bool cv::cuda::DeviceInfo::isCompatible ( ) const

Checks the CUDA module and device compatibility.

This function returns true if the CUDA module can be run on the specified device. Otherwise, it returns false .

◆ supports()

bool cv::cuda::DeviceInfo::supports ( FeatureSet feature_set ) const

Provides information on CUDA feature support.

引数

feature_set Features to be checked. See cuda::FeatureSet.

This function returns true if the device has the specified CUDA feature. Otherwise, it returns false

このクラス詳解は次のファイルから抽出されました:

core/cuda.hpp

公開型

公開メンバ関数