reorder of values in 128 bit vector in arm neon assembly code

Question

There are four 32 bit elements in a Neon register - say, Q0 - which is of size 128 bit.

1 2 3 4

I want the final data to be in order as shown below: 4 3 2 1

What Neon instruction can achieve the desired data order?

Answer 1

I don't think you can manage 4 words in a single instruction, but it can certainly be done in two:

vswp      d0, d1  ; exchange the two halves of q0, giving 3,4,1,2
vrev64.32 q0, q0  ; word-swap each doubleword of q0, giving 4,3,2,1

Note that the end result doesn't actually depend on which order you do the two operations in.