Topic: GTX 1070 / 1060 mixed /w RX Vega 56 with super low efficient eth hashrate, why?! (Read 131 times)

Hi,

I am using a Asus Z270a and running 2 GTX 1060, 3 GTX 1070 and always when I add the RX Vega 56, the efficient hashhtrate is fluctuating like hell though the reported hashrate is always good. When I remove the RX Vega 56 my reported and efficient hashrate is nearly 1:1, but I absolutely cannot see where is the problem with mixing those cards?!  I am using a 1600W EVGA PSU with good risers, but as soon as I add the Vega 56. All drivers are up to date as well...

The Vega 56 is hashing without BIOS flash under 75% TDP with (in my opinion pretty good) 38 mh/s (reported). So, mostly out of the box. However, the fan from the VEGA 56 is also spinning as it wishes and does not "listen" to MSI afterburner. However, it seems that it overtakes the memory and core clocks adjusted with Afterburner (880 memory clock, lowest possible core cloclk 825 mHz or so). Prior to setting Virtual memory high, the cards did not work properly. Can the fluctuating hashrate solely be a physical memory issue?

All risers get powered as well, virtual memory set to 24gb to 32GB with 4gb physical RAM.

Temperatures of all cards between 55 to 62°C on average.

I overclocked the Nvidia GPU's mostly to its maximum and it seems that it influences the effective hashrate as well, but I don't really see why. One GTX 1070 with a Micron VRAM actually hashes with 65% TDP close to 34 mh/s and stable. Other 2 with 30.5 and 32.5, both Micron VRAM.

HOW can I figure out with card makes the problems in this configruation?

The rig does not crash and seems to be stable in the reported hashrate, but the effective hashrate with ethermine on Claymore is just a mess with the Vega 56. Someone had the same issue?

I added the graph below. You can see prior and after I added the RX Vega 56 (again).

HELP, highly appreciated as I tried already everything. https://imgur.com/a/Xj4Z3