BE VS LE data layout (&sldwi)

Say you have a vector int a = {1,2,3,4}, vector int b = {5,6,7,8}
The lay out of a in the memory on a BE machine is pretty simply, just as it is:
Because BE means high word is save in low address ( but it is natural for array access)
Addr0x   0 1 2 3
BE:     [1,2,3,4]
On LE it is the opposite (high word is save in high address):
Addr0x   0 1 2 3
LE:     [4,3,2,1]

say you have xxsldwi(a,b,3)
On BE: it is just 0 1 2 3   4 5 6 7
                 [1,2,3,4] [5,6,7,8]
        The Result RT = {4,5,6,7}
        if you want to use vec_shuffle to implement xxsldwi, you have to pass in
        shuffle_vector((vector int)a, (vector int)b, 3,4,5,6), you can simply image BE
        is acess from Left to right.
        if you do  for (int i = 0; i < 4; i++)
  {
    printf("RT[%d]=%d\n\n",i, RT[i]);
  }
    it is just :
   c[0]=4

   c[1]=5

   c[2]=6

   c[3]=7
    
But On LE, for the same array a and b,
           it is just: 3 2 1 0   7 6 5 4
                      [4,3,2,1] [8,7,6,5]

        The Result RT = [1,8,7,6] in the register
        So no matter what, when xxsldwi(a,b,3), the word[3] of A, followed by
        word[0], word[1] and word[2] will be put into the result vector register
        if you want to use vec_shuffle to implement xxsldwi, you have to pass in
        shuffle_vector((vector int)a, (vector int)b,5,6,7,0), you can simply image LE
        is acess from right to left.
        if you do  for (int i = 0; i < 4; i++)
  {
    printf("RT[%d]=%d\n\n",i, RT[i]);
  }
    it is just :
   c[0]=6

   c[1]=7

   c[2]=8

   c[3]=1