Do you get those Hs rates for a prolonged period of time, or just for a few minutes?
to state that a card Hs at XXX it should average the rate for at least 6 hours.   
Would you be so kind to share OverdriveNT settings and which power table are you using?

I also have 3 of Vega 56 with Bios 64 and a "real" Vega 64, but they get shy of 1950Hs each.
Sometime they shoot 1 or 2 lines at 2300 Hs but that does not count.

Thank you!

 The reference cards should all be the same, as the guts are entirely provided by AMD.

 Aftermarket cards will vary a LOT though - like that Sapphire Nitro+ model with the INSANE 3 power connector setup.


 My single STOCK Vega 56 gets 1940-1950 consistantly long-term, I've not bothered playing with the BIOS as I'm not sure how long it's going to be pointed at Monero mining and it doesn't NEED a modded BIOS for it's later "long-term" home.

Youre not the only one circusdad also got those error reports. I might have to try this for a week and see which miner is stable.

I think it is the other way around.
Cast is 1.5% donation and STAK is 2%.

At least on my case, --fastswitch may have lowered the reporting of Hs rate on the miner window, but it has actually increased the rate on the pool monitor, which is what matter at the end, since it is how we get paid.

Now I have been getting 9000Hs for the last 24 hours, on 4 Vega and one 580, at 830w at the wall. Before I was getting 8400Hs maximum.
It tend to lower the Hs rate if it runs without auto restarting for more than 2 hours, but if i put it on 60/90 minutes cycles, it keeps the average at 9000hs for 24 hours.
On the miner screen it shows 8650Hs yet at the pool I get 9000Hs.
I cannot complain....

For the Vegas these are the settings:

I have set up the rig using most of Circus Daddy's Vega Mining guide

Blockchain drivers august 23

1 vega 64
3 vega 56  bios 64

Regedit Power table 1100Mhz

P6 1212   890
P7 1408   890

P3 1100  870

Power 0%
Max temp and Target temp at 75
Fans at 2800, 4900


Cast has also lifted Hs rate on another PC on which I run 2 of RX580 and Andrenalin drivers. It run steady at 1900/2000Hs on both XMR and ETN, without any crashing. before I was only reaching 1750Hs for both 580 cards.
On the miner screen it shows a combined Hs rate of 1664Hs for both cards, yet at the pool I get 2000Hs.

I finally got it to run reliably.
I hope you get better luck.

confirmed 662 shares 0 errors (2 hours) at 11600H/s average on 6 Vega 56 cards with these OverdriveNTool settings:

[Profile_0]
Name=Vega56
GPU_P0=852;900
GPU_P1=991;900
GPU_P2=1084;900
GPU_P3=1138;900
GPU_P4=1150;900
GPU_P5=1202;900
GPU_P6=1212;905
GPU_P7=1407;935
Mem_P0=167;900
Mem_P1=500;900
Mem_P2=800;900
Mem_P3=935;900

nice numbers.. At which speed does XMR Kast mine?

I have no experience with the Vega 64, but for the looks of it, you might need to find the sweet spot for your Vega 64, MHz wise. Vega 56 sweet spot is around 930-935MHz, again, from personal experience, this might differ.

At anything above 940MHz, if I set a Vega as the main GPU, everything seems normal in your OS' desktop, but then after doing something as simple as using GPU-Z to check stats, screen artefacts start to show everywhere on the screen, like when you overclocked a card too much.

Using Cast XMR alone seems fine, though, your OS' desktop won't show artefacts and the cards will be running, BUT I got random hangs so I used a 10 minutes timer to restart Cast XMR and OnedriveNTool, and you get more errors. It's been 2 days the 6 Vega 56 rig is running stable and the difference is night and day, you can sleep well.

Also overclocking can lead to de-configuration of your card's settings, as if the card forget their own settings which you had set up at startup, and can get over 80º with the fan spinning at something like 1000rpm and be happy with it --this means the card can wear off and not last much.

Yes, it happens to me too. For what I read, Asus Z270-a is not the best mobo to run Vegas. I'm changing it tomorrow. Will let you know if it get more stable.

cast_xmr-vega -S pool.supportxmr.com:5555 -u yourwallet -p x --nonicehash -G 0,1,2,3,4

--nonicehash is optional, -G will depend on the amount of cards you have. In my case, I have 5 cards.

I don't think it is a mobo issue. Will check the job # in JSON output to see if that be used to monitor freezes.

Anyone else seen this or worked around it?