opencv/modules/core/src/convert.dispatch.cpp
Yuantao Feng 3afe8ddaf8
core: Rename cv::float16_t to cv::hfloat (#25217)
* rename cv::float16_t to cv::fp16_t

* add typedef fp16_t float16_t

* remove zero(), bits() from fp16_t class

* fp16_t -> hfloat

* remove cv::float16_t::fromBits; add hfloatFromBits

* undo changes in conv_winograd_f63.simd.hpp and conv_block.simd.hpp

* undo some changes in dnn
2024-03-21 23:44:19 +03:00

405 lines
12 KiB
C++

// 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
#include "precomp.hpp"
#include "opencl_kernels_core.hpp"
#include "convert.simd.hpp"
#include "convert.simd_declarations.hpp" // defines CV_CPU_DISPATCH_MODES_ALL=AVX2,...,BASELINE based on CMakeLists.txt content
namespace cv {
namespace hal {
void cvt16f32f(const hfloat* src, float* dst, int len)
{
CV_INSTRUMENT_REGION();
CV_CPU_DISPATCH(cvt16f32f, (src, dst, len),
CV_CPU_DISPATCH_MODES_ALL);
}
void cvt32f16f(const float* src, hfloat* dst, int len)
{
CV_INSTRUMENT_REGION();
CV_CPU_DISPATCH(cvt32f16f, (src, dst, len),
CV_CPU_DISPATCH_MODES_ALL);
}
void addRNGBias32f(float* arr, const float* scaleBiasPairs, int len)
{
CV_INSTRUMENT_REGION();
CV_CPU_DISPATCH(addRNGBias32f, (arr, scaleBiasPairs, len),
CV_CPU_DISPATCH_MODES_ALL);
}
void addRNGBias64f(double* arr, const double* scaleBiasPairs, int len)
{
CV_INSTRUMENT_REGION();
CV_CPU_DISPATCH(addRNGBias64f, (arr, scaleBiasPairs, len),
CV_CPU_DISPATCH_MODES_ALL);
}
} // namespace
/* [TODO] Recover IPP calls
#if defined(HAVE_IPP)
#define DEF_CVT_FUNC_F(suffix, stype, dtype, ippFavor) \
static void cvt##suffix( const stype* src, size_t sstep, const uchar*, size_t, \
dtype* dst, size_t dstep, Size size, double*) \
{ \
CV_IPP_RUN(src && dst, CV_INSTRUMENT_FUN_IPP(ippiConvert_##ippFavor, src, (int)sstep, dst, (int)dstep, ippiSize(size.width, size.height)) >= 0) \
cvt_(src, sstep, dst, dstep, size); \
}
#define DEF_CVT_FUNC_F2(suffix, stype, dtype, ippFavor) \
static void cvt##suffix( const stype* src, size_t sstep, const uchar*, size_t, \
dtype* dst, size_t dstep, Size size, double*) \
{ \
CV_IPP_RUN(src && dst, CV_INSTRUMENT_FUN_IPP(ippiConvert_##ippFavor, src, (int)sstep, dst, (int)dstep, ippiSize(size.width, size.height), ippRndFinancial, 0) >= 0) \
cvt_(src, sstep, dst, dstep, size); \
}
#else
#define DEF_CVT_FUNC_F(suffix, stype, dtype, ippFavor) \
static void cvt##suffix( const stype* src, size_t sstep, const uchar*, size_t, \
dtype* dst, size_t dstep, Size size, double*) \
{ \
cvt_(src, sstep, dst, dstep, size); \
}
#define DEF_CVT_FUNC_F2 DEF_CVT_FUNC_F
#endif
#define DEF_CVT_FUNC(suffix, stype, dtype) \
static void cvt##suffix( const stype* src, size_t sstep, const uchar*, size_t, \
dtype* dst, size_t dstep, Size size, double*) \
{ \
cvt_(src, sstep, dst, dstep, size); \
}
#define DEF_CPY_FUNC(suffix, stype) \
static void cvt##suffix( const stype* src, size_t sstep, const uchar*, size_t, \
stype* dst, size_t dstep, Size size, double*) \
{ \
cpy_(src, sstep, dst, dstep, size); \
}
DEF_CPY_FUNC(8u, uchar)
DEF_CVT_FUNC_F(8s8u, schar, uchar, 8s8u_C1Rs)
DEF_CVT_FUNC_F(16u8u, ushort, uchar, 16u8u_C1R)
DEF_CVT_FUNC_F(16s8u, short, uchar, 16s8u_C1R)
DEF_CVT_FUNC_F(32s8u, int, uchar, 32s8u_C1R)
DEF_CVT_FUNC_F2(32f8u, float, uchar, 32f8u_C1RSfs)
DEF_CVT_FUNC(64f8u, double, uchar)
DEF_CVT_FUNC_F2(8u8s, uchar, schar, 8u8s_C1RSfs)
DEF_CVT_FUNC_F2(16u8s, ushort, schar, 16u8s_C1RSfs)
DEF_CVT_FUNC_F2(16s8s, short, schar, 16s8s_C1RSfs)
DEF_CVT_FUNC_F(32s8s, int, schar, 32s8s_C1R)
DEF_CVT_FUNC_F2(32f8s, float, schar, 32f8s_C1RSfs)
DEF_CVT_FUNC(64f8s, double, schar)
DEF_CVT_FUNC_F(8u16u, uchar, ushort, 8u16u_C1R)
DEF_CVT_FUNC_F(8s16u, schar, ushort, 8s16u_C1Rs)
DEF_CPY_FUNC(16u, ushort)
DEF_CVT_FUNC_F(16s16u, short, ushort, 16s16u_C1Rs)
DEF_CVT_FUNC_F2(32s16u, int, ushort, 32s16u_C1RSfs)
DEF_CVT_FUNC_F2(32f16u, float, ushort, 32f16u_C1RSfs)
DEF_CVT_FUNC(64f16u, double, ushort)
DEF_CVT_FUNC_F(8u16s, uchar, short, 8u16s_C1R)
DEF_CVT_FUNC_F(8s16s, schar, short, 8s16s_C1R)
DEF_CVT_FUNC_F2(16u16s, ushort, short, 16u16s_C1RSfs)
DEF_CVT_FUNC_F2(32s16s, int, short, 32s16s_C1RSfs)
DEF_CVT_FUNC(32f16s, float, short)
DEF_CVT_FUNC(64f16s, double, short)
DEF_CVT_FUNC_F(8u32s, uchar, int, 8u32s_C1R)
DEF_CVT_FUNC_F(8s32s, schar, int, 8s32s_C1R)
DEF_CVT_FUNC_F(16u32s, ushort, int, 16u32s_C1R)
DEF_CVT_FUNC_F(16s32s, short, int, 16s32s_C1R)
DEF_CPY_FUNC(32s, int)
DEF_CVT_FUNC_F2(32f32s, float, int, 32f32s_C1RSfs)
DEF_CVT_FUNC(64f32s, double, int)
DEF_CVT_FUNC_F(8u32f, uchar, float, 8u32f_C1R)
DEF_CVT_FUNC_F(8s32f, schar, float, 8s32f_C1R)
DEF_CVT_FUNC_F(16u32f, ushort, float, 16u32f_C1R)
DEF_CVT_FUNC_F(16s32f, short, float, 16s32f_C1R)
DEF_CVT_FUNC_F(32s32f, int, float, 32s32f_C1R)
DEF_CVT_FUNC(64f32f, double, float)
DEF_CVT_FUNC(8u64f, uchar, double)
DEF_CVT_FUNC(8s64f, schar, double)
DEF_CVT_FUNC(16u64f, ushort, double)
DEF_CVT_FUNC(16s64f, short, double)
DEF_CVT_FUNC(32s64f, int, double)
DEF_CVT_FUNC(32f64f, float, double)
DEF_CPY_FUNC(64s, int64)
*/
BinaryFunc getConvertFunc(int sdepth, int ddepth)
{
CV_INSTRUMENT_REGION();
CV_CPU_DISPATCH(getConvertFunc, (sdepth, ddepth),
CV_CPU_DISPATCH_MODES_ALL);
}
#ifdef HAVE_OPENCL
static bool ocl_convertFp16( InputArray _src, OutputArray _dst, int sdepth, int ddepth )
{
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(sdepth == CV_32F ? "convertFp16_FP32_to_FP16" : "convertFp16_FP16_to_FP32", 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);
}
static bool ocl_convertTo(InputArray src_, OutputArray dst_, int ddepth, bool noScale, double alpha, double beta)
{
CV_INSTRUMENT_REGION();
CV_Assert(ddepth >= 0);
int stype = src_.type();
int sdepth = CV_MAT_DEPTH(stype);
int cn = CV_MAT_CN(stype);
int dtype = CV_MAKETYPE(ddepth, cn);
int wdepth = (sdepth == CV_64F) ? CV_64F : CV_32F;
bool needDouble = sdepth == CV_64F || ddepth == CV_64F;
bool doubleCheck = true;
if (needDouble)
{
doubleCheck = ocl::Device::getDefault().hasFP64();
}
bool halfCheck = true;
bool needHalf = sdepth == CV_16F || ddepth == CV_16F;
if (needHalf)
{
halfCheck = ocl::Device::getDefault().hasFP16();
}
if (!doubleCheck)
return false;
if (!halfCheck)
return false;
const int rowsPerWI = 4;
char cvt[2][50];
ocl::Kernel k("convertTo", ocl::core::convert_oclsrc,
format("-D srcT=%s -D WT=%s -D dstT=%s -D convertToWT=%s -D convertToDT=%s -D rowsPerWI=%d%s%s%s",
ocl::typeToStr(sdepth), ocl::typeToStr(wdepth), ocl::typeToStr(ddepth),
ocl::convertTypeStr(sdepth, wdepth, 1, cvt[0], sizeof(cvt[0])),
ocl::convertTypeStr(wdepth, ddepth, 1, cvt[1], sizeof(cvt[1])),
rowsPerWI,
needDouble ? " -D DOUBLE_SUPPORT" : "",
needHalf ? " -D HALF_SUPPORT" : "",
noScale ? " -D NO_SCALE" : ""
)
);
if (k.empty())
return false;
UMat src = src_.getUMat();
dst_.createSameSize(src_, dtype);
UMat dst = dst_.getUMat();
float alphaf = (float)alpha, betaf = (float)beta;
if (noScale)
k.args(ocl::KernelArg::ReadOnlyNoSize(src), ocl::KernelArg::WriteOnly(dst, cn));
else if (wdepth == CV_32F)
k.args(ocl::KernelArg::ReadOnlyNoSize(src), ocl::KernelArg::WriteOnly(dst, cn), alphaf, betaf);
else
k.args(ocl::KernelArg::ReadOnlyNoSize(src), ocl::KernelArg::WriteOnly(dst, cn), alpha, beta);
size_t globalsize[2] = {
(size_t)dst.cols * cn,
divUp((size_t)dst.rows, rowsPerWI)
};
if (!k.run(2, globalsize, NULL, false))
return false;
CV_IMPL_ADD(CV_IMPL_OCL);
return true;
}
#endif
void Mat::convertTo(OutputArray dst, int type_, double alpha, double beta) const
{
CV_INSTRUMENT_REGION();
if (empty())
{
dst.release();
return;
}
int stype = type();
int sdepth = CV_MAT_DEPTH(stype);
int ddepth = sdepth;
if (type_ >= 0)
ddepth = CV_MAT_DEPTH(type_);
else
ddepth = dst.fixedType() ? dst.depth() : sdepth;
bool noScale = std::fabs(alpha - 1) < DBL_EPSILON && std::fabs(beta) < DBL_EPSILON;
if (sdepth == ddepth && noScale)
{
copyTo(dst);
return;
}
CV_OCL_RUN(dims <= 2 && dst.isUMat(),
ocl_convertTo(*this, dst, ddepth, noScale, alpha, beta))
int cn = channels();
int dtype = CV_MAKETYPE(ddepth, cn);
Mat src = *this;
dst.create(dims, size, dtype);
Mat dstMat = dst.getMat();
BinaryFunc func = noScale ? getConvertFunc(sdepth, ddepth) : getConvertScaleFunc(sdepth, ddepth);
double scale[] = {alpha, beta};
CV_Assert( func != 0 );
if( dims <= 2 )
{
Size sz = getContinuousSize2D(src, dstMat, cn);
func(src.data, src.step, 0, 0, dstMat.data, dstMat.step, sz, scale);
}
else
{
const Mat* arrays[] = {&src, &dstMat, 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);
}
}
void UMat::convertTo(OutputArray dst, int type_, double alpha, double beta) const
{
CV_INSTRUMENT_REGION();
if (empty())
{
dst.release();
return;
}
#ifdef HAVE_OPENCL
int stype = type();
int sdepth = CV_MAT_DEPTH(stype);
int ddepth = sdepth;
if (type_ >= 0)
ddepth = CV_MAT_DEPTH(type_);
else
ddepth = dst.fixedType() ? dst.depth() : sdepth;
bool noScale = std::fabs(alpha - 1) < DBL_EPSILON && std::fabs(beta) < DBL_EPSILON;
if (sdepth == ddepth && noScale)
{
copyTo(dst);
return;
}
CV_OCL_RUN(dims <= 2,
ocl_convertTo(*this, dst, ddepth, noScale, alpha, beta))
#endif // HAVE_OPENCL
UMat src = *this; // Fake reference to itself.
// Resolves issue 8693 in case of src == dst.
Mat m = getMat(ACCESS_READ);
m.convertTo(dst, type_, alpha, beta);
(void)src;
}
//==================================================================================================
void 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 = getConvertFunc(CV_32F, CV_16F);
break;
case CV_16S:
case CV_16F:
ddepth = CV_32F;
func = getConvertFunc(CV_16F, CV_32F);
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);
for( size_t i = 0; i < it.nplanes; i++, ++it )
func(ptrs[0], 0, 0, 0, ptrs[1], 0, sz, 0);
}
}
} // namespace cv