Skip to content

O
opencv
Commit 766d950f authored by Vladislav Vinogradov's avatar Vladislav Vinogradov
Browse files
Options

used new device layer in math per element operations

parent 7628e57f
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 0 additions and 264 deletions
modules/cudaarithm/src/cuda/math.cu
View file @ 766d950f
This diff is collapsed.
modules/cudaarithm/src/element_operations.cpp
View file @ 766d950f
......@@ -451,270 +451,6 @@ void cv::cuda::absdiff(InputArray src1, InputArray src2, OutputArray dst, Stream
arithm_op(src1, src2, dst, noArray(), 1.0, -1, stream, absDiffMat, absDiffScalar);
}
//////////////////////////////////////////////////////////////////////////////
// abs
namespace arithm
{
template <typename T>
void absMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
}
void cv::cuda::abs(InputArray _src, OutputArray _dst, Stream& stream)
{
using namespace arithm;
typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
static const func_t funcs[] =
{
absMat<unsigned char>,
absMat<signed char>,
absMat<unsigned short>,
absMat<short>,
absMat<int>,
absMat<float>,
absMat<double>
};
GpuMat src = _src.getGpuMat();
const int depth = src.depth();
CV_Assert( depth <= CV_64F );
CV_Assert( src.channels() == 1 );
if (depth == CV_64F)
{
if (!deviceSupports(NATIVE_DOUBLE))
CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
}
_dst.create(src.size(), src.type());
GpuMat dst = _dst.getGpuMat();
funcs[depth](src, dst, StreamAccessor::getStream(stream));
}
//////////////////////////////////////////////////////////////////////////////
// sqr
namespace arithm
{
template <typename T>
void sqrMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
}
void cv::cuda::sqr(InputArray _src, OutputArray _dst, Stream& stream)
{
using namespace arithm;
typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
static const func_t funcs[] =
{
sqrMat<unsigned char>,
sqrMat<signed char>,
sqrMat<unsigned short>,
sqrMat<short>,
sqrMat<int>,
sqrMat<float>,
sqrMat<double>
};
GpuMat src = _src.getGpuMat();
const int depth = src.depth();
CV_Assert( depth <= CV_64F );
CV_Assert( src.channels() == 1 );
if (depth == CV_64F)
{
if (!deviceSupports(NATIVE_DOUBLE))
CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
}
_dst.create(src.size(), src.type());
GpuMat dst = _dst.getGpuMat();
funcs[depth](src, dst, StreamAccessor::getStream(stream));
}
//////////////////////////////////////////////////////////////////////////////
// sqrt
namespace arithm
{
template <typename T>
void sqrtMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
}
void cv::cuda::sqrt(InputArray _src, OutputArray _dst, Stream& stream)
{
using namespace arithm;
typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
static const func_t funcs[] =
{
sqrtMat<unsigned char>,
sqrtMat<signed char>,
sqrtMat<unsigned short>,
sqrtMat<short>,
sqrtMat<int>,
sqrtMat<float>,
sqrtMat<double>
};
GpuMat src = _src.getGpuMat();
const int depth = src.depth();
CV_Assert( depth <= CV_64F );
CV_Assert( src.channels() == 1 );
if (depth == CV_64F)
{
if (!deviceSupports(NATIVE_DOUBLE))
CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
}
_dst.create(src.size(), src.type());
GpuMat dst = _dst.getGpuMat();
funcs[depth](src, dst, StreamAccessor::getStream(stream));
}
////////////////////////////////////////////////////////////////////////
// exp
namespace arithm
{
template <typename T>
void expMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
}
void cv::cuda::exp(InputArray _src, OutputArray _dst, Stream& stream)
{
using namespace arithm;
typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
static const func_t funcs[] =
{
expMat<unsigned char>,
expMat<signed char>,
expMat<unsigned short>,
expMat<short>,
expMat<int>,
expMat<float>,
expMat<double>
};
GpuMat src = _src.getGpuMat();
const int depth = src.depth();
CV_Assert( depth <= CV_64F );
CV_Assert( src.channels() == 1 );
if (depth == CV_64F)
{
if (!deviceSupports(NATIVE_DOUBLE))
CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
}
_dst.create(src.size(), src.type());
GpuMat dst = _dst.getGpuMat();
funcs[depth](src, dst, StreamAccessor::getStream(stream));
}
////////////////////////////////////////////////////////////////////////
// log
namespace arithm
{
template <typename T>
void logMat(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
}
void cv::cuda::log(InputArray _src, OutputArray _dst, Stream& stream)
{
using namespace arithm;
typedef void (*func_t)(PtrStepSzb src, PtrStepSzb dst, cudaStream_t stream);
static const func_t funcs[] =
{
logMat<unsigned char>,
logMat<signed char>,
logMat<unsigned short>,
logMat<short>,
logMat<int>,
logMat<float>,
logMat<double>
};
GpuMat src = _src.getGpuMat();
const int depth = src.depth();
CV_Assert( depth <= CV_64F );
CV_Assert( src.channels() == 1 );
if (depth == CV_64F)
{
if (!deviceSupports(NATIVE_DOUBLE))
CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
}
_dst.create(src.size(), src.type());
GpuMat dst = _dst.getGpuMat();
funcs[depth](src, dst, StreamAccessor::getStream(stream));
}
////////////////////////////////////////////////////////////////////////
// pow
namespace arithm
{
template<typename T> void pow(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
}
void cv::cuda::pow(InputArray _src, double power, OutputArray _dst, Stream& stream)
{
typedef void (*func_t)(PtrStepSzb src, double power, PtrStepSzb dst, cudaStream_t stream);
static const func_t funcs[] =
{
arithm::pow<unsigned char>,
arithm::pow<signed char>,
arithm::pow<unsigned short>,
arithm::pow<short>,
arithm::pow<int>,
arithm::pow<float>,
arithm::pow<double>
};
GpuMat src = _src.getGpuMat();
const int depth = src.depth();
const int cn = src.channels();
CV_Assert(depth <= CV_64F);
if (depth == CV_64F)
{
if (!deviceSupports(NATIVE_DOUBLE))
CV_Error(cv::Error::StsUnsupportedFormat, "The device doesn't support double");
}
_dst.create(src.size(), src.type());
GpuMat dst = _dst.getGpuMat();
PtrStepSzb src_(src.rows, src.cols * cn, src.data, src.step);
PtrStepSzb dst_(src.rows, src.cols * cn, dst.data, dst.step);
funcs[depth](src_, power, dst_, StreamAccessor::getStream(stream));
}
//////////////////////////////////////////////////////////////////////////////
// compare
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment