By checking the source code of opencv, there are sub-functions and some sub-functions further in sub-functions, generally how can we get to know how many operations in total of does this function need and how many times of memory read and write we need in the function?
通过检查opencv的源代码，子函数中还有子函数和一些子函数，一般来说，我们怎样才能知道这个函数需要多少次操作，以及在函数中需要多少次的内存读写?
Trying calculating by hand looks quite a lot of effort, is somebody trying to count this by hand or any other smart way to know it ?
用手来计算看起来很费劲，有人想用手算一下还是用其他聪明的方法来计算?
cv::Ptr<cv::FilterEngine> cv::createGaussianFilter( int type, Size ksize,
double sigma1, double sigma2,
int borderType )
{
Mat kx, ky;
createGaussianKernels(kx, ky, type, ksize, sigma1, sigma2);
return createSeparableLinearFilter( type, type, kx, ky, Point(-1,-1), 0, borderType );
}
static void createGaussianKernels( Mat & kx, Mat & ky, int type, Size ksize,
double sigma1, double sigma2 )
{
int depth = CV_MAT_DEPTH(type);
if( sigma2 <= 0 )
sigma2 = sigma1;
// automatic detection of kernel size from sigma
if( ksize.width <= 0 && sigma1 > 0 )
ksize.width = cvRound(sigma1*(depth == CV_8U ? 3 : 4)*2 + 1)|1;
if( ksize.height <= 0 && sigma2 > 0 )
ksize.height = cvRound(sigma2*(depth == CV_8U ? 3 : 4)*2 + 1)|1;
CV_Assert( ksize.width > 0 && ksize.width % 2 == 1 &&
ksize.height > 0 && ksize.height % 2 == 1 );
sigma1 = std::max( sigma1, 0. );
sigma2 = std::max( sigma2, 0. );
kx = getGaussianKernel( ksize.width, sigma1, std::max(depth, CV_32F) );
if( ksize.height == ksize.width && std::abs(sigma1 - sigma2) < DBL_EPSILON )
ky = kx;
else
ky = getGaussianKernel( ksize.height, sigma2, std::max(depth, CV_32F) );
}
cv::Mat cv::getGaussianKernel( int n, double sigma, int ktype )
{
const int SMALL_GAUSSIAN_SIZE = 7;
static const float small_gaussian_tab[][SMALL_GAUSSIAN_SIZE] =
{
{1.f},
{0.25f, 0.5f, 0.25f},
{0.0625f, 0.25f, 0.375f, 0.25f, 0.0625f},
{0.03125f, 0.109375f, 0.21875f, 0.28125f, 0.21875f, 0.109375f, 0.03125f}
};
const float* fixed_kernel = n % 2 == 1 && n <= SMALL_GAUSSIAN_SIZE && sigma <= 0 ?
small_gaussian_tab[n>>1] : 0;
CV_Assert( ktype == CV_32F || ktype == CV_64F );
Mat kernel(n, 1, ktype);
float* cf = kernel.ptr<float>();
double* cd = kernel.ptr<double>();
double sigmaX = sigma > 0 ? sigma : ((n-1)*0.5 - 1)*0.3 + 0.8;
double scale2X = -0.5/(sigmaX*sigmaX);
double sum = 0;
int i;
for( i = 0; i < n; i++ )
{
double x = i - (n-1)*0.5;
double t = fixed_kernel ? (double)fixed_kernel[i] : std::exp(scale2X*x*x);
if( ktype == CV_32F )
{
cf[i] = (float)t;
sum += cf[i];
}
else
{
cd[i] = t;
sum += cd[i];
}
}
sum = 1./sum;
for( i = 0; i < n; i++ )
{
if( ktype == CV_32F )
cf[i] = (float)(cf[i]*sum);
else
cd[i] *= sum;
}
return kernel;
}
cv::Ptr<cv::FilterEngine> cv::createSeparableLinearFilter(
int _srcType, int _dstType,
InputArray __rowKernel, InputArray __columnKernel,
Point _anchor, double _delta,
int _rowBorderType, int _columnBorderType,
const Scalar& _borderValue )
{
Mat _rowKernel = __rowKernel.getMat(), _columnKernel = __columnKernel.getMat();
_srcType = CV_MAT_TYPE(_srcType);
_dstType = CV_MAT_TYPE(_dstType);
int sdepth = CV_MAT_DEPTH(_srcType), ddepth = CV_MAT_DEPTH(_dstType);
int cn = CV_MAT_CN(_srcType);
CV_Assert( cn == CV_MAT_CN(_dstType) );
int rsize = _rowKernel.rows + _rowKernel.cols - 1;
int csize = _columnKernel.rows + _columnKernel.cols - 1;
if( _anchor.x < 0 )
_anchor.x = rsize/2;
if( _anchor.y < 0 )
_anchor.y = csize/2;
int rtype = getKernelType(_rowKernel,
_rowKernel.rows == 1 ? Point(_anchor.x, 0) : Point(0, _anchor.x));
int ctype = getKernelType(_columnKernel,
_columnKernel.rows == 1 ? Point(_anchor.y, 0) : Point(0, _anchor.y));
Mat rowKernel, columnKernel;
int bdepth = std::max(CV_32F,std::max(sdepth, ddepth));
int bits = 0;
if( sdepth == CV_8U &&
((rtype == KERNEL_SMOOTH+KERNEL_SYMMETRICAL &&
ctype == KERNEL_SMOOTH+KERNEL_SYMMETRICAL &&
ddepth == CV_8U) ||
((rtype & (KERNEL_SYMMETRICAL+KERNEL_ASYMMETRICAL)) &&
(ctype & (KERNEL_SYMMETRICAL+KERNEL_ASYMMETRICAL)) &&
(rtype & ctype & KERNEL_INTEGER) &&
ddepth == CV_16S)) )
{
bdepth = CV_32S;
bits = ddepth == CV_8U ? 8 : 0;
_rowKernel.convertTo( rowKernel, CV_32S, 1 << bits );
_columnKernel.convertTo( columnKernel, CV_32S, 1 << bits );
bits *= 2;
_delta *= (1 << bits);
}
else
{
if( _rowKernel.type() != bdepth )
_rowKernel.convertTo( rowKernel, bdepth );
else
rowKernel = _rowKernel;
if( _columnKernel.type() != bdepth )
_columnKernel.convertTo( columnKernel, bdepth );
else
columnKernel = _columnKernel;
}
int _bufType = CV_MAKETYPE(bdepth, cn);
Ptr<BaseRowFilter> _rowFilter = getLinearRowFilter(
_srcType, _bufType, rowKernel, _anchor.x, rtype);
Ptr<BaseColumnFilter> _columnFilter = getLinearColumnFilter(
_bufType, _dstType, columnKernel, _anchor.y, ctype, _delta, bits );
return Ptr<FilterEngine>( new FilterEngine(Ptr<BaseFilter>(), _rowFilter, _columnFilter,
_srcType, _dstType, _bufType, _rowBorderType, _columnBorderType, _borderValue ));
}
cv::Ptr<cv::BaseColumnFilter> cv::getLinearColumnFilter( int bufType, int dstType,
InputArray _kernel, int anchor,
int symmetryType, double delta,
int bits )
{
Mat kernel = _kernel.getMat();
int sdepth = CV_MAT_DEPTH(bufType), ddepth = CV_MAT_DEPTH(dstType);
int cn = CV_MAT_CN(dstType);
CV_Assert( cn == CV_MAT_CN(bufType) &&
sdepth >= std::max(ddepth, CV_32S) &&
kernel.type() == sdepth );
if( !(symmetryType & (KERNEL_SYMMETRICAL|KERNEL_ASYMMETRICAL)) )
{
if( ddepth == CV_8U && sdepth == CV_32S )
return makePtr<ColumnFilter<FixedPtCastEx<int, uchar>, ColumnNoVec> >
(kernel, anchor, delta, FixedPtCastEx<int, uchar>(bits));
if( ddepth == CV_8U && sdepth == CV_32F )
return makePtr<ColumnFilter<Cast<float, uchar>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_8U && sdepth == CV_64F )
return makePtr<ColumnFilter<Cast<double, uchar>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_16U && sdepth == CV_32F )
return makePtr<ColumnFilter<Cast<float, ushort>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_16U && sdepth == CV_64F )
return makePtr<ColumnFilter<Cast<double, ushort>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_16S && sdepth == CV_32F )
return makePtr<ColumnFilter<Cast<float, short>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_16S && sdepth == CV_64F )
return makePtr<ColumnFilter<Cast<double, short>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_32F && sdepth == CV_32F )
return makePtr<ColumnFilter<Cast<float, float>, ColumnNoVec> >(kernel, anchor, delta);
if( ddepth == CV_64F && sdepth == CV_64F )
return makePtr<ColumnFilter<Cast<double, double>, ColumnNoVec> >(kernel, anchor, delta);
}
else
{
int ksize = kernel.rows + kernel.cols - 1;
if( ksize == 3 )
{
if( ddepth == CV_8U && sdepth == CV_32S )
return makePtr<SymmColumnSmallFilter<
FixedPtCastEx<int, uchar>, SymmColumnVec_32s8u> >
(kernel, anchor, delta, symmetryType, FixedPtCastEx<int, uchar>(bits),
SymmColumnVec_32s8u(kernel, symmetryType, bits, delta));
if( ddepth == CV_16S && sdepth == CV_32S && bits == 0 )
return makePtr<SymmColumnSmallFilter<Cast<int, short>,
SymmColumnSmallVec_32s16s> >(kernel, anchor, delta, symmetryType,
Cast<int, short>(), SymmColumnSmallVec_32s16s(kernel, symmetryType, bits, delta));
if( ddepth == CV_32F && sdepth == CV_32F )
return makePtr<SymmColumnSmallFilter<
Cast<float, float>,SymmColumnSmallVec_32f> >
(kernel, anchor, delta, symmetryType, Cast<float, float>(),
SymmColumnSmallVec_32f(kernel, symmetryType, 0, delta));
}
if( ddepth == CV_8U && sdepth == CV_32S )
return makePtr<SymmColumnFilter<FixedPtCastEx<int, uchar>, SymmColumnVec_32s8u> >
(kernel, anchor, delta, symmetryType, FixedPtCastEx<int, uchar>(bits),
SymmColumnVec_32s8u(kernel, symmetryType, bits, delta));
if( ddepth == CV_8U && sdepth == CV_32F )
return makePtr<SymmColumnFilter<Cast<float, uchar>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
if( ddepth == CV_8U && sdepth == CV_64F )
return makePtr<SymmColumnFilter<Cast<double, uchar>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
if( ddepth == CV_16U && sdepth == CV_32F )
return makePtr<SymmColumnFilter<Cast<float, ushort>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
if( ddepth == CV_16U && sdepth == CV_64F )
return makePtr<SymmColumnFilter<Cast<double, ushort>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
if( ddepth == CV_16S && sdepth == CV_32S )
return makePtr<SymmColumnFilter<Cast<int, short>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
if( ddepth == CV_16S && sdepth == CV_32F )
return makePtr<SymmColumnFilter<Cast<float, short>, SymmColumnVec_32f16s> >
(kernel, anchor, delta, symmetryType, Cast<float, short>(),
SymmColumnVec_32f16s(kernel, symmetryType, 0, delta));
if( ddepth == CV_16S && sdepth == CV_64F )
return makePtr<SymmColumnFilter<Cast<double, short>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
if( ddepth == CV_32F && sdepth == CV_32F )
return makePtr<SymmColumnFilter<Cast<float, float>, SymmColumnVec_32f> >
(kernel, anchor, delta, symmetryType, Cast<float, float>(),
SymmColumnVec_32f(kernel, symmetryType, 0, delta));
if( ddepth == CV_64F && sdepth == CV_64F )
return makePtr<SymmColumnFilter<Cast<double, double>, ColumnNoVec> >
(kernel, anchor, delta, symmetryType);
}
CV_Error_( CV_StsNotImplemented,
("Unsupported combination of buffer format (=%d), and destination format (=%d)",
bufType, dstType));
return Ptr<BaseColumnFilter>();
}
1 个解决方案
#1
0
Of course, you could disassemble the program, that will give you an instruction count and you can also easily count reads and writes with e.g. grep. I suggest using a cross-compiler to disassemble into some nice RISC assembler like MIPS - reading arcane, bloated X86 assembler is not fun.
当然，您可以分解程序，这将给您一个指令计数，并且您也可以很容易地计数读和写，例如grep。我建议使用一个交叉编译器来分解成一些像MIPS这样的优秀的RISC汇编器——阅读晦涩的、臃肿的X86汇编程序并不是一件有趣的事情。
#1
0
Of course, you could disassemble the program, that will give you an instruction count and you can also easily count reads and writes with e.g. grep. I suggest using a cross-compiler to disassemble into some nice RISC assembler like MIPS - reading arcane, bloated X86 assembler is not fun.
当然，您可以分解程序，这将给您一个指令计数，并且您也可以很容易地计数读和写，例如grep。我建议使用一个交叉编译器来分解成一些像MIPS这样的优秀的RISC汇编器——阅读晦涩的、臃肿的X86汇编程序并不是一件有趣的事情。