renamed gpu::cudev namespace -> cuda::device

cudev is used for new device layer
This commit is contained in:
Vladislav Vinogradov
2013-07-23 11:33:51 +04:00
parent e895b7455e
commit 8282f6ebc1
165 changed files with 539 additions and 539 deletions

View File

@@ -43,7 +43,7 @@
#ifndef __OPENCV_GPU_DEVICE_BLOCK_HPP__
#define __OPENCV_GPU_DEVICE_BLOCK_HPP__
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
struct Block
{

View File

@@ -47,7 +47,7 @@
#include "vec_traits.hpp"
#include "vec_math.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
//////////////////////////////////////////////////////////////
// BrdConstant

View File

@@ -45,7 +45,7 @@
#include "detail/color_detail.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
// All OPENCV_GPU_IMPLEMENT_*_TRAITS(ColorSpace1_to_ColorSpace2, ...) macros implements
// template <typename T> class ColorSpace1_to_ColorSpace2_traits

View File

@@ -87,7 +87,7 @@ namespace cv { namespace cuda
namespace cv { namespace cuda
{
namespace cudev
namespace device
{
__host__ __device__ __forceinline__ int divUp(int total, int grain)
{

View File

@@ -45,7 +45,7 @@
#include "common.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 200

View File

@@ -49,7 +49,7 @@
#include "../limits.hpp"
#include "../functional.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
#ifndef CV_DESCALE
#define CV_DESCALE(x, n) (((x) + (1 << ((n)-1))) >> (n))
@@ -146,7 +146,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2RGB_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2RGB<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2RGB<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -219,7 +219,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2RGB5x5_TRAITS(name, scn, bidx, green_bits) \
struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2RGB5x5<scn, bidx, green_bits> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2RGB5x5<scn, bidx, green_bits> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -300,7 +300,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB5x52RGB_TRAITS(name, dcn, bidx, green_bits) \
struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB5x52RGB<dcn, bidx, green_bits> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB5x52RGB<dcn, bidx, green_bits> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -346,7 +346,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_GRAY2RGB_TRAITS(name, dcn) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::Gray2RGB<T, dcn> functor_type; \
typedef ::cv::cuda::device::color_detail::Gray2RGB<T, dcn> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -388,7 +388,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_GRAY2RGB5x5_TRAITS(name, green_bits) \
struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::Gray2RGB5x5<green_bits> functor_type; \
typedef ::cv::cuda::device::color_detail::Gray2RGB5x5<green_bits> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -430,7 +430,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB5x52GRAY_TRAITS(name, green_bits) \
struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB5x52Gray<green_bits> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB5x52Gray<green_bits> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -481,7 +481,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2GRAY_TRAITS(name, scn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2Gray<T, scn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2Gray<T, scn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -532,7 +532,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2YUV_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2YUV<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2YUV<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -620,7 +620,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_YUV2RGB_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::YUV2RGB<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::YUV2RGB<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -699,7 +699,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2YCrCb_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2YCrCb<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2YCrCb<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -778,7 +778,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_YCrCb2RGB_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::YCrCb2RGB<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::YCrCb2RGB<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -854,7 +854,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2XYZ_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2XYZ<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2XYZ<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -929,7 +929,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_XYZ2RGB_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::XYZ2RGB<T, scn, dcn, bidx> functor_type; \
typedef ::cv::cuda::device::color_detail::XYZ2RGB<T, scn, dcn, bidx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1067,7 +1067,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2HSV_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HSV<T, scn, dcn, bidx, 180> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HSV<T, scn, dcn, bidx, 180> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1075,7 +1075,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <typename T> struct name ## _full_traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HSV<T, scn, dcn, bidx, 256> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HSV<T, scn, dcn, bidx, 256> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1083,7 +1083,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HSV<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HSV<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1091,7 +1091,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _full_traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HSV<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HSV<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1207,7 +1207,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_HSV2RGB_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::HSV2RGB<T, scn, dcn, bidx, 180> functor_type; \
typedef ::cv::cuda::device::color_detail::HSV2RGB<T, scn, dcn, bidx, 180> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1215,7 +1215,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <typename T> struct name ## _full_traits \
{ \
typedef ::cv::cuda::cudev::color_detail::HSV2RGB<T, scn, dcn, bidx, 255> functor_type; \
typedef ::cv::cuda::device::color_detail::HSV2RGB<T, scn, dcn, bidx, 255> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1223,7 +1223,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::HSV2RGB<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::HSV2RGB<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1231,7 +1231,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _full_traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::HSV2RGB<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::HSV2RGB<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1340,7 +1340,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2HLS_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HLS<T, scn, dcn, bidx, 180> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HLS<T, scn, dcn, bidx, 180> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1348,7 +1348,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <typename T> struct name ## _full_traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HLS<T, scn, dcn, bidx, 256> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HLS<T, scn, dcn, bidx, 256> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1356,7 +1356,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HLS<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HLS<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1364,7 +1364,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _full_traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2HLS<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2HLS<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1480,7 +1480,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_HLS2RGB_TRAITS(name, scn, dcn, bidx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::HLS2RGB<T, scn, dcn, bidx, 180> functor_type; \
typedef ::cv::cuda::device::color_detail::HLS2RGB<T, scn, dcn, bidx, 180> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1488,7 +1488,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <typename T> struct name ## _full_traits \
{ \
typedef ::cv::cuda::cudev::color_detail::HLS2RGB<T, scn, dcn, bidx, 255> functor_type; \
typedef ::cv::cuda::device::color_detail::HLS2RGB<T, scn, dcn, bidx, 255> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1496,7 +1496,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::HLS2RGB<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::HLS2RGB<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1504,7 +1504,7 @@ namespace cv { namespace cuda { namespace cudev
}; \
template <> struct name ## _full_traits<float> \
{ \
typedef ::cv::cuda::cudev::color_detail::HLS2RGB<float, scn, dcn, bidx, 360> functor_type; \
typedef ::cv::cuda::device::color_detail::HLS2RGB<float, scn, dcn, bidx, 360> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1649,7 +1649,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2Lab_TRAITS(name, scn, dcn, srgb, blueIdx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2Lab<T, scn, dcn, srgb, blueIdx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2Lab<T, scn, dcn, srgb, blueIdx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1762,7 +1762,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_Lab2RGB_TRAITS(name, scn, dcn, srgb, blueIdx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::Lab2RGB<T, scn, dcn, srgb, blueIdx> functor_type; \
typedef ::cv::cuda::device::color_detail::Lab2RGB<T, scn, dcn, srgb, blueIdx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1861,7 +1861,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_RGB2Luv_TRAITS(name, scn, dcn, srgb, blueIdx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::RGB2Luv<T, scn, dcn, srgb, blueIdx> functor_type; \
typedef ::cv::cuda::device::color_detail::RGB2Luv<T, scn, dcn, srgb, blueIdx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \
@@ -1962,7 +1962,7 @@ namespace cv { namespace cuda { namespace cudev
#define OPENCV_GPU_IMPLEMENT_Luv2RGB_TRAITS(name, scn, dcn, srgb, blueIdx) \
template <typename T> struct name ## _traits \
{ \
typedef ::cv::cuda::cudev::color_detail::Luv2RGB<T, scn, dcn, srgb, blueIdx> functor_type; \
typedef ::cv::cuda::device::color_detail::Luv2RGB<T, scn, dcn, srgb, blueIdx> functor_type; \
static __host__ __device__ __forceinline__ functor_type create_functor() \
{ \
return functor_type(); \

View File

@@ -47,7 +47,7 @@
#include "../warp.hpp"
#include "../warp_shuffle.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
namespace reduce_detail
{

View File

@@ -47,7 +47,7 @@
#include "../warp.hpp"
#include "../warp_shuffle.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
namespace reduce_key_val_detail
{

View File

@@ -47,7 +47,7 @@
#include "../vec_traits.hpp"
#include "../functional.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
namespace transform_detail
{

View File

@@ -46,7 +46,7 @@
#include "../common.hpp"
#include "../vec_traits.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
namespace type_traits_detail
{

View File

@@ -45,7 +45,7 @@
#include "../datamov_utils.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
namespace vec_distance_detail
{

View File

@@ -43,7 +43,7 @@
#ifndef __OPENCV_GPU_DYNAMIC_SMEM_HPP__
#define __OPENCV_GPU_DYNAMIC_SMEM_HPP__
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template<class T> struct DynamicSharedMem
{

View File

@@ -46,7 +46,7 @@
#include "common.hpp"
#include "warp_reduce.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
struct Emulation
{

View File

@@ -48,7 +48,7 @@
#include "vec_math.hpp"
#include "type_traits.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <typename Ptr2D> struct PointFilter
{

View File

@@ -45,7 +45,7 @@
#include <cstdio>
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template<class Func>
void printFuncAttrib(Func& func)

View File

@@ -49,7 +49,7 @@
#include "type_traits.hpp"
#include "device_functions.h"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
// Function Objects
template<typename Argument, typename Result> struct unary_function : public std::unary_function<Argument, Result> {};

View File

@@ -47,7 +47,7 @@
#include <float.h>
#include "common.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <class T> struct numeric_limits;

View File

@@ -47,7 +47,7 @@
#include "detail/reduce.hpp"
#include "detail/reduce_key_val.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <int N, typename T, class Op>
__device__ __forceinline__ void reduce(volatile T* smem, T& val, unsigned int tid, const Op& op)

View File

@@ -45,7 +45,7 @@
#include "common.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(uchar v) { return _Tp(v); }
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(schar v) { return _Tp(v); }

View File

@@ -48,7 +48,7 @@
#include "opencv2/core/cuda/warp.hpp"
#include "opencv2/core/cuda/warp_shuffle.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
enum ScanKind { EXCLUSIVE = 0, INCLUSIVE = 1 };
@@ -174,13 +174,13 @@ namespace cv { namespace cuda { namespace cudev
__device__ T warpScanInclusive(T idata, volatile T* s_Data, unsigned int tid)
{
#if __CUDA_ARCH__ >= 300
const unsigned int laneId = cv::cuda::cudev::Warp::laneId();
const unsigned int laneId = cv::cuda::device::Warp::laneId();
// scan on shuffl functions
#pragma unroll
for (int i = 1; i <= (OPENCV_GPU_WARP_SIZE / 2); i *= 2)
{
const T n = cv::cuda::cudev::shfl_up(idata, i);
const T n = cv::cuda::device::shfl_up(idata, i);
if (laneId >= i)
idata += n;
}

View File

@@ -123,7 +123,7 @@
vmin4(a,b) per-byte unsigned minimum: min(a, b)
*/
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
// 2

View File

@@ -47,7 +47,7 @@
#include "utility.hpp"
#include "detail/transform_detail.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <typename T, typename D, typename UnOp, typename Mask>
static inline void transform(PtrStepSz<T> src, PtrStepSz<D> dst, UnOp op, const Mask& mask, cudaStream_t stream)

View File

@@ -45,7 +45,7 @@
#include "detail/type_traits_detail.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <typename T> struct IsSimpleParameter
{

View File

@@ -46,7 +46,7 @@
#include "saturate_cast.hpp"
#include "datamov_utils.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
#define OPENCV_GPU_LOG_WARP_SIZE (5)
#define OPENCV_GPU_WARP_SIZE (1 << OPENCV_GPU_LOG_WARP_SIZE)

View File

@@ -47,7 +47,7 @@
#include "functional.hpp"
#include "detail/vec_distance_detail.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <typename T> struct L1Dist
{

View File

@@ -46,7 +46,7 @@
#include "vec_traits.hpp"
#include "saturate_cast.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
// saturate_cast

View File

@@ -45,7 +45,7 @@
#include "common.hpp"
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template<typename T, int N> struct TypeVec;

View File

@@ -43,7 +43,7 @@
#ifndef __OPENCV_GPU_DEVICE_WARP_HPP__
#define __OPENCV_GPU_DEVICE_WARP_HPP__
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
struct Warp
{

View File

@@ -43,7 +43,7 @@
#ifndef OPENCV_GPU_WARP_REDUCE_HPP__
#define OPENCV_GPU_WARP_REDUCE_HPP__
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <class T>
__device__ __forceinline__ T warp_reduce(volatile T *ptr , const unsigned int tid = threadIdx.x)

View File

@@ -43,7 +43,7 @@
#ifndef __OPENCV_GPU_WARP_SHUFFLE_HPP__
#define __OPENCV_GPU_WARP_SHUFFLE_HPP__
namespace cv { namespace cuda { namespace cudev
namespace cv { namespace cuda { namespace device
{
template <typename T>
__device__ __forceinline__ T shfl(T val, int srcLane, int width = warpSize)