As far as the layout in memory is concern, the difference between parallel arrays and an array of structs is essentially:

PA:
AAAAABBBBBCCCCCDDDDD

AoS:
ABCDABCDABCDABCDABCD

The latter is preferable for many reasons, but one obvious one is that you're probably going to use the third A close to the third C, so it would be nice if they were close together in memory so you'd get fewer cache misses.