@tytanik - i experience the same issue issue like most of users with NV and more cards.
H110+ 13 x P106 setup, found everything from the start, have the following problems:
1. Overclocking works pretty strange, to clock the memory of the P106 from 4000 to 4900 i needed to put 1600 as parameter. Steps that increase the MHz are really not relevant. Putting 900 clock to 4400 something.
2. Finally stabilized the machine to 325MH and runs well but it reboots without reason 14 times for 10 hours. Temperatures are under 65C, target is 76C.
So how to fix those nonsense reboots ?
Not all the cards will overclock as well as others, so 1 badly working card can cause a reboot.
That being said, SMOS has some weird behavior with NVIDIA cards.
Until they are fixed, you're better off mining ETH with (H110+ 13 x P106) with the following overclock settings:
GPU Core: -200
Memory MHz: 500
Power Limit (watt): 70
Target Temperature ℃: 75
Minimum Fan speed %: 20
I have about 12 rigs, all of them with those settings.. no reboots, less noise and less heat.
thanks for the settings, what hashrate you have with those ? btw if I put 500 on memory clock my cards don't overclock with 500mhz but something like 200 only, that's why if I want to reach +900 i put 1600... very strange