i uploaded a screen shot on the imgur link above
thnx for the replies,
i know the rig seems expensive for the speed but i am expecting future driver developments by RADEON WINDOWS and mobo manufacturer that will uptick the currently and by intent unutilised potential of these GPUS.
there is a reason why these cards come with a BIOS switch
and fancy led colour switch, we will get into that later....
as PanzerSheriff guessed correctly,
2 MAIN PILLARS (imo) of MINING WITH VEGA64 are:
1) GPU Operating temperature (
especialy Memory temp as opposed to CPU)
and ofcourse
2) Kw/h energy cost
at present the VEGA's draw too much power for the Hashing speed they can attain in comparison to lesser cards which at present are more efficient.
The rig as is currently burns through almost 1700W total (8 GPU's mobo, screen) with memory overclocking and undervolting.
Using Crimson 17.11.1 you can avoid the Crossfire issues and other complexities that came with later drivers.
Avoid the mining driver like the plague unless you must use it as it is riddled with bugs and uncompatibilities that will probably test your patience.
HBM at this stage and for mining the way we are testing is NOT needed and only causes problems with more than 3 GPU's (we will be testing it on a separate 6 card rig for MONERO coming soon)
Unified settings on Wattman are:
Dynamic CPU clock set min state6 @ 998 and max state7 @1000 (setting the same value in both fields causes issues)
Auto Voltage
Dynamic Memory clock set min/max @ 1100
Auto Temperature
Voltage set @ -24% (Wattman has issues between -10 to -20 avoid this range where possible)
Dynamic Fan speed set min 2500 target 4000 although this depends HEAVILY on the ambient temp, we achieve the temps seen on the screenshot with +8C ambient room temp, outside +2C (i control this by allowing more or less outside air to enter the room, also i am noting temps creating a reference table for future use, will post it when complete)
Vitrual memory is set to min 16Gb to 20Gb max (very important for stability and long running times)
If you are using later drivers you must
switch off Crossfire (good luck keeping the setting stable)
These settings will give you 43,7MH/s stable on all cards and here is the kicker:
Not all GPU's are the same, some can take more Memory overclocking than others, in conjunction with PCIe BUS abilities of the mobo and mobo drivers you are using, the Asrock is the winner in this respect as we have 10 VEGA's on it whereas the ASUS can only take 8 VEGA's max at present. ASUS has promised a driver update in Q4 of 2017 to support more but with 22 days left in this year the update has not come yet.
I was able, after a lot of trial and error to get 3 cards running at 45.5MH/s stable by uping the memory clock to 1150 and the voltage to -17%.
Again if you attempt this assume that one GPU can do it whilst another cant on top of the mobo instability when trying this on more than half the cards on your rig. Trial and error yields fruits but the process can be a nut cracker.
i cannot stress enough how
important it is keeping memory temp below 70C 
