Some guys are just mean here.. I know i'm a noob on this forums but cmon we gotta help each other out.. I asked if I should mod my bios and not one person answered me. Anyways I went ahead and did it glad I did and don't know why it took me a year lol.. most cards get me around 29-30 even one card does 31mhs... Now one set of cards gets to 31mhs but with tons of errors.. so i lowered clocks. overnight each card got around 30 errors is this bad? If I clock those cards lower I get MHS lowered then before modded.. Should I try another bios on those? Also if somebody could point me to a good modded vbios repository would be awesome thanks.
For my rig I am having a couple thousand errors in a day. I have 6 Sapphire Nitro+ RX580 total hash 185 Mh for solo mining with 800Watt from wall. System never crashes up to 48 hours. I always restarted system before that time, maybe it could run even further. So i think you can overclock your card a little bit further. Use HWinfo64, overclock each card seperately and if there are no huge number of error in couple minutes you can use that values safely.
Your system does not need to crash. It can look stable. But you have no idea how much shares you've lost due tu your too overclocked gpus/mem straps. Memory errors indicates, the card is not running comfortable in 100% condition. So you could definatelly have more shares with cards without mem errors. In early days I was alsi running my cards like that and I had no clew what memory errors are. But one day I learned everything needed, optimized my cards, my cards showed lower hashrate, but my overall accepted efective hashrate increased. And that's the point.
I dont think i am losing so much shares. Because my avarage hashrate and reported hashrates are almost identical on nanopool even %1.5 devfee