Here is a tidbit of wisdom we gained by using a FLIR handheld Thermal Spectograph on the rig and wiring and why, as mentioned above you need a spare 12v computer fan. (i haven't found this info anywhere on the net so if you are reading this consider yourself ahead in the game)
on the imgur link above i have included pics of the ASUS and Asrock mobos, the red circle you see is the component that can OVERHEAT and can cause long run mining problems. We found it to go over 70C in relative low ambient temp causing the PCIe slot BUSing to malfunction and freeze/hang the system.
Because we are utililsing the cold outside temp to regulate ambient room temp a fan pointed on that component is enough but you might consider installing a small heat sink to go with the fan for better results if you live in warmer environments.
pics of the 10GPU rig and more info coming later...
on the imgur link above i have included pics of the ASUS and Asrock mobos, the red circle you see is the component that can OVERHEAT and can cause long run mining problems. We found it to go over 70C in relative low ambient temp causing the PCIe slot BUSing to malfunction and freeze/hang the system.
Because we are utililsing the cold outside temp to regulate ambient room temp a fan pointed on that component is enough but you might consider installing a small heat sink to go with the fan for better results if you live in warmer environments.
pics of the 10GPU rig and more info coming later...
Have you investigated the temperature dependence on the number of cards? Do you think that specific component that you pictured influences PCIe slots to stop working by transferring heat to them, or you suggest that motherboard has built in feature to stop when some of its elements pass predefined threshold temp. Have you solved the problem with introducing some kind of heating that you mentioned?
The fact that you reported lead either to the conclusion that Asrock , ASUS motherboard is not designed properly: it always overheats when there are many gpus (>10) or that you have motherboards that are not working properly or that high power VEGA consumption specifically heats the board.
In my case I render the picture from the integrated GPU on CPU(i3) but still the picture freezes so I might conclude it's not due to PCIe.
Have you tried to read some motherboard temperatures from HWInfo?
Here is my previous post for other thread so you can know my conf:
''
Hello,
I had the problem that system freezes at random time intervals while mining when using 10 Nvidia cards on H110 motherboard running on windows. I was aware of the limit on windows: 8 nvidia and 5 amd, or other way round so I got an information that windows solved that problem in Fall creators update. Unfortunately, my system was running two days and after that it started to freeze again on random time intervals when mining: approximately 2-3 hours. I tested both EWBF and DSTM miner. Has anyone reported that this win update can fix the problem?
This is a description my system conf:
Rig:
10x Zotac 1070 AMP Extreme
Asrock H110 Pro BTC
Intel i3
8GB RAM
500 HDD
2x corsair 1000W + Kolink 750W : 2750W
OC:
Power Limit: 65
GPU Core: 70
Memory Core: 270
Target Temp: 70
OS:
Windows 10, Fall creators Update
EWBF: 4600 sol/s Power usage: 160W/card
DSTM: 4750 sol/s
''
Hoping for a long and COLD winter so I can justify the power bill for my wife.... 
