Opencv matrix multiplication

C++
,
1

Hi everybody,
i found that when i multiply vector in vector the run time is much more than when i multiply matrix in matrix in OpenCV 2.4.7
consider that both of the vector and matrices has the same number of multiplication.
for example:

 cv::Mat mat1(186, 186, CV_64FC1, cv::Scalar::all(20.));
  cv::Mat vec1(186, 1, CV_64FC1, cv::Scalar::all(20.));
  cv::Mat vec1_tr(1, 186, CV_64FC1, cv::Scalar::all(20.));
  int64_t matrix_time{ 0 };
  for (int i = 0; i < m_loop_count; ++i)
  {
    tick_tock mat_tick;
mat_tick.tick();
    cv::Mat out = (vec1_tr * mat1) * vec1;
   //  34782 multiplication
    matrix_time += mat_tick.tock();
  }
int mul_count = (186* 186) + 186; //34782 
  cv::Mat mat2(mul_count, 1, CV_64FC1, cv::Scalar::all(20.));
  cv::Mat mat2_tr(1, mul_count, CV_64FC1, cv::Scalar::all(20.));

  int64_t vec_time{ 0 };
  for (int i = 0; i < m_loop_count ; ++i)
  {
    tick_tock vec_tick;
			vec_tick.tick();
    cv::Mat out = mat2_tr * mat2;//34782 multiplication
    vec_time += vec_tick.tock();
  }

the matrix multiplication is much faster (2 times) than vector but both has 34782 multiplication.
can anybody explain why?
thank you.

2

that died years ago.
please throw it away, and update to recent 4.5.x (no discussion)