Hey!
Could someone tell me, why my efficiency is so low, and should I be worried about it? Mhs seems good.
You’ve got invalid shares on GPU0 reboot and it should clear the efficiency log, if it continues remove PL and set core to 1050 and reboot. Even if the card are exactly the same you have to except some uncertainties during manufacturing.
If you still get invalid shared try -1010 / -1020 in small increments until you get stable shares. Tring to Max out Hash rate can cause the opposite effect. 
1 Like
Just want to say a big thank you for all who contribute here it has helped me to find the best results for my rig. A question for the long term miners. What would be a safe MAX memory Clock?
My stable settings with a mix of MSI,ASUS, NVIDIA and RX 480 (Mixed with SAMSUNG, HYNIX and MICRON)
My Maxed out Settings:
I asked the question because the MSI SAMSUNG seems to be ok at 2800 MEM clock and the ASUS at 2400 - whilst testing at what point do each of the cards either give me a invalid share / crashes the miner and trying to maintain under 500W. This setting seems to be stable for both but worried I’m reducing shelf life. I’m changing it back to my stable settings for now but just wanted some Pro advice from those with more experience as I’m just a freshy round here.
Many thanks in advance and kind regards,
Nas
Guys. I would like to mine some RVN. What OC will be ok for my mixed cards ? On the picture are oc for ETH…
these OC for ETH obviously are not good. you can try with off sett values for memory -1004 as i have seen these are the best i think for those cards.
for raven coin you will need to up a bit the core clock. test and let us know.
first of all for the Hynix cards, use 930 core and -1004 mem and 75pl … for Samsung, i start with 1100 core and 2400 memory, and for cards that have rejects, i lower memory 100 at a time until the rejects stop … this will get you 32/33 mh …
for RVN, it’s very different, i use the +100 offset core and -1004 hynix or 1800 for samsung memory … but these cards are extremely inefficient at mining kawpow … be aware, you will barely make money on them, as they use more power, mine avg 14mh @100w for kawpow … i suggest sticking with ethash for these gpu’s for now and converting eth → rvn … i use 30xx for kawpow and 1660 for eth today … good luck!
Its not Zotac, its Hynix memory on all your cards. I do 378.5MH @ 836w on 12 x GTX 1660 Super’s
See attached, my Hynix memory cards also consume more wattage
I just picked up a new Asus TUF 1660 Super and mine has Samsung memory. Not sure if this is a new version or what but it seems to do a pretty stable 32-33MH/S. My Settings are as follows:
Algo: Ethash
Miner: Ethminer
Absolute Core: 1100
Mem: 2700
PL: 80
Drivers and miner version?
Thank you
Driver version:
N470.63.01
Miner Version:
ethminer-ethash v.0.19.0.2
I’ve read it, but I disagree… in practice I do absolut cc and if I change the PL I get a different result, don’t you?
Thank you MarcioS
I noticed the Samsung can clock quite a bit higher, I will run them for 24 hours and see what happens.
Thank you
One more thing to state , GPU drivers version is 460 the base which came when i installed the hiveos
The ones getting 35mh is impressive. Is that the ones with the Bios flash?
Hello Guys. Can i get better results for my Mixed cards? I wonder if i can change my oc for higher hash rate and less rejected shares? But in OCTOPUS algo for mining CFX better. Thanks for help 
This is original bios for palit Hynix memory
Hello, something curious, in my country have youtubers questioning the amount of MH reported for MH effective, and that for EXAMPLE a 1660s with MC 2100 doing its 32.30mh this would make LESS valid shares… than a 1660s with MC 1750 (which is natural for her if we look at techpowerup) doing her 31.22mh, look I’m doing the tests, measuring every 24 hours at exact times, it requires a few days, but I’m having better results with less mh, less effort on the memories .
Can anyone opine about it here?