Speedup of parallel odd-even transposition sort

Hi all,
I understand that the worst case sort time for parallel odd-even
transposition sort using n processors sorting n numbers is O(n), but
I've read that the absolute speedup is O(log n), and I can't
understand how that came about. Can anyone explain it to me?

Thank you.

Regards,
Rayne

.