How to align cv :: Mat for AVX-512 / AVX2 architectures?

Disclamer: I'm new to simd so if this dirty peasant asks some bad questions.

In my opinion, AVX-512 architectures can handle up to 16 floating point variables, while AVX2 "only" 4.

To take advantage of this, the data must be aligned. As I found out here , it can be done with:

For AVX-512:

alignas(32) float a[8];

      

        
        
        
      

    

For AVX2:

alignas(16) float a[8];

      

        
        
        
      

    

Ok, so my first question is , since 16 is a factor of 32, why don't we always use alignas(32)

also for AVX2 architectures? Maybe (maybe) I'm missing something.

Then I have this function:

bool interpolate(const Mat &im, Mat &res, /*...*/){/*...*/}

      

        
        
        
      

    

which are allocated with

cv::Mat im(r, c, CV_32FC1); //similarly res

      

        
        
        
      

    

The Intel compiler report tells me that the two matrices are not aligned. So my second question is , how can I highlight them so that they match 16/32 ? I could allocate the aligned pointer and pass it to the constructor with cv::Mat

something like:

 float *aligned_ptr = /*allocate r*c 16/32 aligned floating points*/
 cv::Mat m (r, c, CV_32FC1, /* use aligned_ptr somehow*/);