An update. It seems (to my untrained eye) that something is wrong with the second SSD in the RAID configuration. Here's my question on serverfault related to what I saw with iostat - https://serverfault.com/questions/951096/difference-in-utilisation-reported-by-iostat-for-two-identical-disks-in-raid1
I've disabled RAID and rebooted the server to run the benchmarks with client=1,4,8,12 with shared_buffers=8MB (default) vs shared_buffers=2GB (optimised?) and will report back.
-- Saurabh.
