generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

implemented faster transpose with SEE auto vectorization

Browse Source
This commit is contained in:
hbristow 2013-07-03 23:34:02 -07:00
parent baa946c390
commit d126263983
3 changed files with 50 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
modules/matlab/compile.cmake
View File

@ -3,7 +3,7 @@ macro(listify OUT_LIST IN_STRING)
endmacro()
listify(MEX_INCLUDE_DIRS_LIST ${MEX_INCLUDE_DIRS})
set(MEX_CXXFLAGS "CXXFLAGS=\$CXXFLAGS -pedantic -Wall -Wextra -Weffc++ -Wno-unused -Wold-style-cast -Wshadow -Wmissing-declarations -Wmissing-include-dirs -Wnon-virtual-dtor -Wno-newline-eof")
set(MEX_CXXFLAGS "CXXFLAGS=\$CXXFLAGS -msse -msse2 -msse3 -msse4.1 -msse4.2 -pedantic -Wall -Wextra -Weffc++ -Wno-unused-parameter -Wold-style-cast -Wshadow -Wmissing-declarations -Wmissing-include-dirs -Wnon-virtual-dtor -Wno-newline-eof")
file(GLOB SOURCE_FILES "${CMAKE_CURRENT_BINARY_DIR}/src/*.cpp")
foreach(SOURCE_FILE ${SOURCE_FILES})
# strip out the filename

2
modules/matlab/include/bridge.hpp
View File

@ -122,7 +122,7 @@ public:
// --------------------------------------------------------------------------
// MATLAB TYPES
// --------------------------------------------------------------------------
Bridge& operator=(const mxArray*) { return *this; }
Bridge& operator=(const mxArray* obj) { ptr_ = obj; return *this; }
Bridge(const mxArray* obj) : ptr_(obj) {}
MxArray toMxArray() { return ptr_; }

86
modules/matlab/include/mxarray.hpp
View File

@ -21,8 +21,6 @@
#ifdef __cplusplus
extern "C" {
#endif
void mkl_somatcopy(char, char, size_t, size_t, const float, const float*, size_t, float*, size_t);
void mkl_domatcopy(char, char, size_t, size_t, const double, const double*, size_t, double*, size_t);
#ifdef __cplusplus
}
#endif
@ -224,6 +222,12 @@ public:
* just encapsulate it
*/
MxArray(const mxArray* ptr) : ptr_(const_cast<mxArray *>(ptr)), owns_(false) {}
MxArray& operator=(const mxArray* ptr) {
dealloc();
ptr_ = const_cast<mxArray *>(ptr);
owns_ = false;
return *this;
}
/*!
* @brief explicit typed constructor
@ -372,7 +376,7 @@ public:
template <typename Scalar>
cv::Mat toMat() const {
cv::Mat mat(cols(), rows(), CV_MAKETYPE(cv::DataType<Scalar>::type, channels()));
cv::Mat mat(rows(), cols(), CV_MAKETYPE(cv::DataType<Scalar>::type, channels()));
switch (ID()) {
case mxINT8_CLASS: deepCopyAndTranspose<int8_t, Scalar>(*this, mat); break;
case mxUINT8_CLASS: deepCopyAndTranspose<uint8_t, Scalar>(*this, mat); break;
@ -397,7 +401,7 @@ public:
Scalar* real() { return static_cast<Scalar *>(mxGetData(ptr_)); }
template <typename Scalar>
Scalar* imag() { return static_cast<Scalar *>(mxGetData(ptr_)); }
Scalar* imag() { return static_cast<Scalar *>(mxGetImagData(ptr_)); }
template <typename Scalar>
const Scalar* real() const { return static_cast<const Scalar *>(mxGetData(ptr_)); }
@ -413,6 +417,7 @@ public:
std::string str;
str.reserve(size()+1);
mxGetString(ptr_, const_cast<char *>(str.data()), str.size());
mexPrintf(str.c_str());
return str;
}
@ -491,21 +496,48 @@ cv::Mat MxArray::toMat<Matlab::InheritType>() const {
// MATRIX TRANSPOSE
// ----------------------------------------------------------------------------
template <typename InputScalar, typename OutputScalar>
void gemt(const char major, const size_t M, const size_t N, const InputScalar* a, size_t lda, OutputScalar* b, size_t ldb) {
switch (major) {
case 'R':
for (size_t m = 0; m < M; ++m) {
InputScalar const * arow = a + m*lda;
InputScalar const * const aend = arow + N;
OutputScalar * bcol = b + m;
while (arow < aend) {
*bcol = *arow;
arow++;
bcol+=ldb;
}
}
return;
case 'C':
for (size_t n = 0; n < N; ++n) {
InputScalar const * acol = a + n*lda;
InputScalar const * const aend = acol + M;
OutputScalar * brow = b + n;
while (acol < aend) {
*brow = *acol;
acol++;
brow+=ldb;
}
}
return;
default:
error(std::string("Unknown ordering given: ").append(std::string(1,major)));
}
}
template <typename InputScalar, typename OutputScalar>
void deepCopyAndTranspose(const cv::Mat& in, MxArray& out) {
conditionalError(static_cast<size_t>(in.rows) == out.rows(), "Matrices must have the same number of rows");
conditionalError(static_cast<size_t>(in.cols) == out.cols(), "Matrices must have the same number of cols");
conditionalError(static_cast<size_t>(in.channels()) == out.channels(), "Matrices must have the same number of channels");
const InputScalar* inp = in.ptr<InputScalar>(0);
OutputScalar* outp = out.real<OutputScalar>();
const size_t M = out.rows();
const size_t N = out.cols();
for (size_t m = 0; m < M; ++m) {
const InputScalar* inp = in.ptr<InputScalar>(m);
for (size_t n = 0; n < N; ++n) {
// copy and transpose
outp[m + n*M] = inp[n];
}
}
gemt('R', out.rows(), out.cols(), inp, in.step1(), outp, out.rows());
}
template <typename InputScalar, typename OutputScalar>
@ -514,34 +546,10 @@ void deepCopyAndTranspose(const MxArray& in, cv::Mat& out) {
conditionalError(in.cols() == static_cast<size_t>(out.cols), "Matrices must have the same number of cols");
conditionalError(in.channels() == static_cast<size_t>(out.channels()), "Matrices must have the same number of channels");
const InputScalar* inp = in.real<InputScalar>();
const size_t M = in.rows();
const size_t N = in.cols();
for (size_t m = 0; m < M; ++m) {
OutputScalar* outp = out.ptr<OutputScalar>(m);
for (size_t n = 0; n < N; ++n) {
// copy and transpose
outp[n] = inp[m + n*M];
}
}
OutputScalar* outp = out.ptr<OutputScalar>(0);
gemt('C', in.rows(), in.cols(), inp, in.rows(), outp, out.step1());
}
template <>
void deepCopyAndTranspose<float, float>(const cv::Mat&, MxArray&) {
}
template <>
void deepCopyAndTranspose<double, double>(const cv::Mat&, MxArray&) {
}
template <>
void deepCopyAndTranspose<float, float>(const MxArray&, cv::Mat&) {
// use mkl
}
template <>
void deepCopyAndTranspose<double, double>(const MxArray&, cv::Mat& ) {
// use mkl
}
#endif