Commit 39b90ae9 authored by Alexander Alekhin's avatar Alexander Alekhin

core: dispatch convert

parent 33d477e0
......@@ -3,6 +3,7 @@ set(the_description "The Core Functionality")
ocv_add_dispatched_file(mathfuncs_core SSE2 AVX AVX2)
ocv_add_dispatched_file(stat SSE4_2 AVX2)
ocv_add_dispatched_file(arithm SSE2 SSE4_1 AVX2 VSX3)
ocv_add_dispatched_file(convert SSE2 AVX2)
# dispatching for accuracy tests
ocv_add_dispatched_file_force_all(test_intrin128 TEST SSE2 SSE3 SSSE3 SSE4_1 SSE4_2 AVX FP16 AVX2)
......
This diff is collapsed.
......@@ -3,10 +3,16 @@
// of this distribution and at http://opencv.org/license.html
#include "precomp.hpp"
#include "opencl_kernels_core.hpp"
#include "convert.hpp"
namespace cv {
CV_CPU_OPTIMIZATION_NAMESPACE_BEGIN
BinaryFunc getConvertFunc(int sdepth, int ddepth);
BinaryFunc get_cvt32f16f();
BinaryFunc get_cvt16f32f();
#ifndef CV_CPU_OPTIMIZATION_DECLARATIONS_ONLY
/*namespace hal {
......@@ -66,7 +72,7 @@ void addRNGBias64f( double* arr, const double* scaleBiasPairs, int len )
}*/
template<typename _Ts, typename _Td, typename _Twvec> inline void
template<typename _Ts, typename _Td, typename _Twvec> static inline void
cvt_( const _Ts* src, size_t sstep, _Td* dst, size_t dstep, Size size )
{
sstep /= sizeof(src[0]);
......@@ -97,7 +103,7 @@ cvt_( const _Ts* src, size_t sstep, _Td* dst, size_t dstep, Size size )
// in order to reduce the code size, for (16f <-> ...) conversions
// we add a conversion function without loop unrolling
template<typename _Ts, typename _Td, typename _Twvec> inline void
template<typename _Ts, typename _Td, typename _Twvec> static inline void
cvt1_( const _Ts* src, size_t sstep, _Td* dst, size_t dstep, Size size )
{
sstep /= sizeof(src[0]);
......@@ -140,7 +146,10 @@ static void cvtCopy( const uchar* src, size_t sstep,
#define DEF_CVT_FUNC(suffix, cvtfunc, _Ts, _Td, _Twvec) \
static void cvt##suffix(const _Ts* src, size_t sstep, uchar*, size_t, \
_Td* dst, size_t dstep, Size size, void*) \
{ cvtfunc<_Ts, _Td, _Twvec>(src, sstep, dst, dstep, size); }
{ \
CV_INSTRUMENT_REGION(); \
cvtfunc<_Ts, _Td, _Twvec>(src, sstep, dst, dstep, size); \
}
////////////////////// 8u -> ... ////////////////////////
......@@ -225,16 +234,16 @@ DEF_CVT_FUNC(16f32f, cvt1_, float16_t, float, v_float32)
///////////// "conversion" w/o conversion ///////////////
static void cvt8u(const uchar* src, size_t sstep, uchar*, size_t, uchar* dst, size_t dstep, Size size, void*)
{ cvtCopy(src, sstep, dst, dstep, size, 1); }
{ CV_INSTRUMENT_REGION(); cvtCopy(src, sstep, dst, dstep, size, 1); }
static void cvt16u(const ushort* src, size_t sstep, uchar*, size_t, ushort* dst, size_t dstep, Size size, void*)
{ cvtCopy((const uchar*)src, sstep, (uchar*)dst, dstep, size, 2); }
{ CV_INSTRUMENT_REGION(); cvtCopy((const uchar*)src, sstep, (uchar*)dst, dstep, size, 2); }
static void cvt32s(const int* src, size_t sstep, uchar*, size_t, int* dst, size_t dstep, Size size, void*)
{ cvtCopy((const uchar*)src, sstep, (uchar*)dst, dstep, size, 4); }
{ CV_INSTRUMENT_REGION(); cvtCopy((const uchar*)src, sstep, (uchar*)dst, dstep, size, 4); }
static void cvt64s(const int64* src, size_t sstep, uchar*, size_t, int64* dst, size_t dstep, Size size, void*)
{ cvtCopy((const uchar*)src, sstep, (uchar*)dst, dstep, size, 8); }
{ CV_INSTRUMENT_REGION(); cvtCopy((const uchar*)src, sstep, (uchar*)dst, dstep, size, 8); }
/* [TODO] Recover IPP calls
......@@ -380,148 +389,17 @@ BinaryFunc getConvertFunc(int sdepth, int ddepth)
return cvtTab[CV_MAT_DEPTH(ddepth)][CV_MAT_DEPTH(sdepth)];
}
#ifdef HAVE_OPENCL
static bool ocl_convertFp16( InputArray _src, OutputArray _dst, int sdepth, int ddepth )
BinaryFunc get_cvt32f16f()
{
int type = _src.type(), cn = CV_MAT_CN(type);
_dst.createSameSize( _src, CV_MAKETYPE(ddepth, cn) );
int kercn = 1;
int rowsPerWI = 1;
String build_opt = format("-D HALF_SUPPORT -D srcT=%s -D dstT=%s -D rowsPerWI=%d%s",
sdepth == CV_32F ? "float" : "half",
sdepth == CV_32F ? "half" : "float",
rowsPerWI,
sdepth == CV_32F ? " -D FLOAT_TO_HALF " : "");
ocl::Kernel k("convertFp16", ocl::core::halfconvert_oclsrc, build_opt);
if (k.empty())
return false;
UMat src = _src.getUMat();
UMat dst = _dst.getUMat();
ocl::KernelArg srcarg = ocl::KernelArg::ReadOnlyNoSize(src),
dstarg = ocl::KernelArg::WriteOnly(dst, cn, kercn);
k.args(srcarg, dstarg);
size_t globalsize[2] = { (size_t)src.cols * cn / kercn, ((size_t)src.rows + rowsPerWI - 1) / rowsPerWI };
return k.run(2, globalsize, NULL, false);
return (BinaryFunc)cvt32f16f;
}
#endif
} // cv::
void cv::Mat::convertTo(OutputArray _dst, int _type, double alpha, double beta) const
BinaryFunc get_cvt16f32f()
{
CV_INSTRUMENT_REGION();
if( empty() )
{
_dst.release();
return;
}
bool noScale = fabs(alpha-1) < DBL_EPSILON && fabs(beta) < DBL_EPSILON;
if( _type < 0 )
_type = _dst.fixedType() ? _dst.type() : type();
else
_type = CV_MAKETYPE(CV_MAT_DEPTH(_type), channels());
int sdepth = depth(), ddepth = CV_MAT_DEPTH(_type);
if( sdepth == ddepth && noScale )
{
copyTo(_dst);
return;
}
Mat src = *this;
if( dims <= 2 )
_dst.create( size(), _type );
else
_dst.create( dims, size, _type );
Mat dst = _dst.getMat();
BinaryFunc func = noScale ? getConvertFunc(sdepth, ddepth) : getConvertScaleFunc(sdepth, ddepth);
double scale[] = {alpha, beta};
int cn = channels();
CV_Assert( func != 0 );
if( dims <= 2 )
{
Size sz = getContinuousSize2D(src, dst, cn);
func( src.data, src.step, 0, 0, dst.data, dst.step, sz, scale );
}
else
{
const Mat* arrays[] = {&src, &dst, 0};
uchar* ptrs[2] = {};
NAryMatIterator it(arrays, ptrs);
Size sz((int)(it.size*cn), 1);
for( size_t i = 0; i < it.nplanes; i++, ++it )
func(ptrs[0], 1, 0, 0, ptrs[1], 1, sz, scale);
}
return (BinaryFunc)cvt16f32f;
}
//==================================================================================================
void cv::convertFp16( InputArray _src, OutputArray _dst )
{
CV_INSTRUMENT_REGION();
int sdepth = _src.depth(), ddepth = 0;
BinaryFunc func = 0;
switch( sdepth )
{
case CV_32F:
if(_dst.fixedType())
{
ddepth = _dst.depth();
CV_Assert(ddepth == CV_16S /*|| ddepth == CV_16F*/);
CV_Assert(_dst.channels() == _src.channels());
}
else
ddepth = CV_16S;
func = (BinaryFunc)cvt32f16f;
break;
case CV_16S:
//case CV_16F:
ddepth = CV_32F;
func = (BinaryFunc)cvt16f32f;
break;
default:
CV_Error(Error::StsUnsupportedFormat, "Unsupported input depth");
return;
}
CV_OCL_RUN(_src.dims() <= 2 && _dst.isUMat(),
ocl_convertFp16(_src, _dst, sdepth, ddepth))
Mat src = _src.getMat();
int type = CV_MAKETYPE(ddepth, src.channels());
_dst.create( src.dims, src.size, type );
Mat dst = _dst.getMat();
int cn = src.channels();
CV_Assert( func != 0 );
if( src.dims <= 2 )
{
Size sz = getContinuousSize2D(src, dst, cn);
func( src.data, src.step, 0, 0, dst.data, dst.step, sz, 0);
}
else
{
const Mat* arrays[] = {&src, &dst, 0};
uchar* ptrs[2] = {};
NAryMatIterator it(arrays, ptrs);
Size sz((int)(it.size*cn), 1);
#endif
for( size_t i = 0; i < it.nplanes; i++, ++it )
func(ptrs[0], 0, 0, 0, ptrs[1], 0, sz, 0);
}
}
CV_CPU_OPTIMIZATION_NAMESPACE_END
} // namespace
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