OSSForks/opencv
1
0
Fork 0
mirror of https://github.com/zebrajr/opencv.git synced 2025-12-06 00:19:46 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge pull request #27734 from cudawarped:cuda_double4_dep

Browse Source 
[cuda] Add compatibility layer for vector types due for depreciation in CUDA 14.0
This commit is contained in:
Alexander Smorkalov 2025-09-09 14:49:48 +03:00 committed by GitHub
commit efea09120b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
5 changed files with 54 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
modules/core/include/opencv2/core/cuda/cuda_compat.hpp Normal file
View File

@ -0,0 +1,38 @@
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
#ifndef OPENCV_CUDA_CUDA_COMPAT_HPP
#define OPENCV_CUDA_CUDA_COMPAT_HPP
#include <cuda.h>
namespace cv { namespace cuda { namespace device { namespace compat
{
#if CUDA_VERSION >= 13000
using ulonglong4 = ::ulonglong4_16a;
using double4 = ::double4_16a;
__host__ __device__ __forceinline__
double4 make_double4(double x, double y, double z, double w)
{
return ::make_double4_16a(x, y, z, w);
}
#else
using ulonglong4 = ::ulonglong4;
using double4 = ::double4;
__host__ __device__ __forceinline__
double4 make_double4(double x, double y, double z, double w)
{
return ::make_double4(x, y, z, w);
}
#endif
using ulonglong4Compat = ulonglong4;
using double4Compat = double4;
__host__ __device__ __forceinline__
double4Compat make_double4_compat(double x, double y, double z, double w)
{
return make_double4(x, y, z, w);
}
}}}}
#endif // OPENCV_CUDA_CUDA_COMPAT_HPP

3
modules/core/include/opencv2/core/cuda/vec_math.hpp
View File

@ -45,6 +45,7 @@
#include "vec_traits.hpp"
#include "saturate_cast.hpp"
#include "cuda_compat.hpp"
/** @file
* @deprecated Use @ref cudev instead.
@ -54,6 +55,8 @@
namespace cv { namespace cuda { namespace device
{
using cv::cuda::device::compat::double4;
using cv::cuda::device::compat::make_double4;
// saturate_cast

4
modules/core/include/opencv2/core/cuda/vec_traits.hpp
View File

@ -44,6 +44,7 @@
#define OPENCV_CUDA_VEC_TRAITS_HPP
#include "common.hpp"
#include "cuda_compat.hpp"
/** @file
* @deprecated Use @ref cudev instead.
@ -53,6 +54,9 @@
namespace cv { namespace cuda { namespace device
{
using cv::cuda::device::compat::double4;
using cv::cuda::device::compat::make_double4;
template<typename T, int N> struct TypeVec;
struct __align__(8) uchar8

8
modules/core/src/cuda/gpu_mat.cu
View File

@ -51,10 +51,12 @@
#include "opencv2/core/cuda.hpp"
#include "opencv2/cudev.hpp"
#include "opencv2/core/cuda/utility.hpp"
#include "opencv2/core/cuda/cuda_compat.hpp"
using namespace cv;
using namespace cv::cuda;
using namespace cv::cudev;
using cv::cuda::device::compat::double4Compat;
device::ThrustAllocator::~ThrustAllocator()
{
@ -341,7 +343,7 @@ void cv::cuda::GpuMat::copyTo(OutputArray _dst, InputArray _mask, Stream& stream
{0,0,0,0},
{0,0,0,0},
{0,0,0,0},
{copyWithMask<double>, copyWithMask<double2>, copyWithMask<double3>, copyWithMask<double4>}
{copyWithMask<double>, copyWithMask<double2>, copyWithMask<double3>, copyWithMask<double4Compat>}
};
if (mask.channels() == channels())
@ -424,7 +426,7 @@ GpuMat& cv::cuda::GpuMat::setTo(Scalar value, Stream& stream)
{setToWithOutMask<short>,setToWithOutMask<short2>,setToWithOutMask<short3>,setToWithOutMask<short4>},
{setToWithOutMask<int>,setToWithOutMask<int2>,setToWithOutMask<int3>,setToWithOutMask<int4>},
{setToWithOutMask<float>,setToWithOutMask<float2>,setToWithOutMask<float3>,setToWithOutMask<float4>},
{setToWithOutMask<double>,setToWithOutMask<double2>,setToWithOutMask<double3>,setToWithOutMask<double4>}
{setToWithOutMask<double>,setToWithOutMask<double2>,setToWithOutMask<double3>,setToWithOutMask<double4Compat>}
};
funcs[depth()][channels() - 1](*this, value, stream);
@ -455,7 +457,7 @@ GpuMat& cv::cuda::GpuMat::setTo(Scalar value, InputArray _mask, Stream& stream)
{setToWithMask<short>,setToWithMask<short2>,setToWithMask<short3>,setToWithMask<short4>},
{setToWithMask<int>,setToWithMask<int2>,setToWithMask<int3>,setToWithMask<int4>},
{setToWithMask<float>,setToWithMask<float2>,setToWithMask<float3>,setToWithMask<float4>},
{setToWithMask<double>,setToWithMask<double2>,setToWithMask<double3>,setToWithMask<double4>}
{setToWithMask<double>,setToWithMask<double2>,setToWithMask<double3>,setToWithMask<double4Compat>}
};
funcs[depth()][channels() - 1](*this, mask, value, stream);

5
modules/dnn/src/cuda/vector_traits.hpp
View File

@ -11,6 +11,7 @@
#include "memory.hpp"
#include "../cuda4dnn/csl/pointer.hpp"
#include "opencv2/core/cuda/cuda_compat.hpp"
#include <type_traits>
@ -34,9 +35,11 @@ namespace cv { namespace dnn { namespace cuda4dnn { namespace csl { namespace de
* v_store(output_vPtr, vec);
*/
using cv::cuda::device::compat::ulonglong4Compat;
namespace detail {
template <size_type N> struct raw_type_ { };
template <> struct raw_type_<256> { typedef ulonglong4 type; };
template <> struct raw_type_<256> { typedef ulonglong4Compat type; };
template <> struct raw_type_<128> { typedef uint4 type; };
template <> struct raw_type_<64> { typedef uint2 type; };
template <> struct raw_type_<32> { typedef uint1 type; };