New theory. I think the Toshiba 5TB (extracted from a retail external drive package), which is the Parity disk in this server, has a neutered firmware that may be interacting in a strange way with the AOC-SASLP-MV8 attached drives (mobo attached drives have normal write speed). I spotted what might be a clue with all the repeated troubleshooting reboots... the Asus M4A785-M BIOS is reporting the Toshiba as UDMA 5 and all the older WD 2TBs as UMDA 6. UDMA detection was set to auto and trying to set it manually to 6 was not possible as it was not an available option in the drop down; 5 was the highest for the Toshiba. Found the latest Asus BIOS and applied it. No change to detection of the Toshiba, write speeds still in the dumper.
I had a newish WD 2TB sitting here and I have swapped it in place of the Toshiba. Parity synch in progress and fingers crossed that it is the solution.
PS. mobo reports the replacement WD as UDMA 6.
QUESTION: is it safe to try and do some writes to the affected disks to test speed? Is it even useful to try unless the parity synch is done (speed degraded due to that process)?