Currently i have power 100, core 100, memory 580 and fan 80.
I would try power to 75-80, core 100, memory 350, fan set to user defined automatic fan control and set the value higher for the different temperatures.
then see if it run stable and memory you might can go up to 500 in smaller steps
I have to disagree. Core clock does nothing but cause problems.
set your power to 100 for troubleshooting. Memory to +500. Fans to 100%
Increase your memory by 50 for every hour of stability you get.
add to the end of your bat file -r 0 so that it auto restarts cards when they die.
You can also try -r 60 and have it restart the miner automatically every hour.
Once you figure out where the issue is, increase -r to 120, 240, etc until you get the kinks worked out.
Video card failures are almost always due to shit psu, too much overclock or too much heat. The back plates on these 1070's get really hot if you do not supplement with a fan so think about using one of those. Almost all of my crashing stopped when i got those things dialed in.
My settings are 75 power, 0 clock, +650 memory and I'm hashing at 32MH per card. I have a portable fan running on each rig, though ill replace these when i rebuild the frames and put 120mm case fans in to get the heat away from the cards. At 70C on the cards, there was nothing but issues with 8 rigs. Now that I'm down to 50C with dialed in mem clocks, I have 14 reliable rigs that dump coins into my wallet all day.