N % blocksize != 0
If the blocksize isn’t a multiple of the number of elements in the matrix or the blocksize is larger than the number of elements in the matrix the kernel will fail. I know why they fail but i don’t know how to find a solution for this problem. The second one won’t occur as much as the first one so that won’t be a problem in my case. but the first one will. because my dataset isn’t an even number all the time. I have to look for a GCD and that GCD needs to be smaller than 22 otherwise I need to take 1 as blocksize.