[QST] how can I use copy swizzle to deal with non-consecutive elements？

Assume that there are 4x4 elements in the shared memory. I can use composition(Swizzle<2, 0, 2>{},...) to swizzle each element successfully，
But now I want to swizzle in unit of 2x2 elements, just like :

//        0     1      2      3    
//     +----+----+----+----+
//  0  |  a  |  a  |  b  |  b  | 
//     +----+----+----+----+
//  1  |  a  |  a  |  b  |  b  | 
//     +----+----+----+----+
//  2  |  c  |  c  |  d  |  d  | 
//     +----+----+----+----+
//  3  |  c  |  c  |  d  |  d  | 
//     +----+----+----+----+

I divide 4x4 into 2x2 pieces(a, b , c, d), each pieces is unit of 2*2 elements, I want to use some methods like composition(Swizzle<1, 0, 1>{},...) to swizzle it, but each pieces is not consecutive in smem, Is there any solution?

Start by reading the CUTLASS Swizzle and composition APIs referenced in the issue, focusing on how they map shared-memory coordinates and whether they support non-consecutive 2x2 groups. Confirm whether the requested 4x4-to-2x2 mapping can be expressed with the existing API; done means providing a tested usage pattern or documenting the limitation and needed direction.