I have 6 1070's on a rig with h81 pro btc and g1820. After 15-30min the cpu goes to 100% and the rig becomes unresponsive. Some of the cards start mining with 4-5mh and I start to get rejected shares. I'm using MSI AB to overclock my cards. Task manager reports System, System interrupts, EthDcrMiner64.exe or MSI AB maxing out the cpu.
A few people have been having similar issues - myself included. Haven't managed to find a solution yet. It might be hardware related, like insufficient ram for that many GPU's to maintain being run at max capacity from straps? Only speculating.
In short though, I'm in the same position as you and if I find a solution I'll post here or PM you.
Tried it on a different Win10 installation and it's the same.
Now I removed AB and I'm using the built-in -mclock and -powlim commands. I think there is an improvement, as it takes more time before this thing happens.
It's not the virtual memory, because hwinfo reports no more than 50-60% use. It's not the Defender, because it's disabled.
I tried HiveOS, but I saw that it doesn't support -strap and I removed it without testing. I can't get even close hash speeds so there is no point of trying in the first place.
This happens only on my Nvidia rig where I'm using claymore.
I'm using Win10 1709 build 16299.371 with Nvidia 430.86 drivers.
I have this exact motherboard and exact CPU as you except I had it with AMD gpus instead of Nvidia's. I can almost gaurantee your issue is a Windows 10 update.
Even with updates disabled, for some reason they kept installing. What would happen is either the rig would mine slower, speeds would jump up and down, Claymore remote manager would keep losing signal, or the miner would just crash.
Doing a cold reboot fixed the issue only to return a day or 2 later. Basically it was the update again trying to install. What I did was just leave the rig for 12 hours, let update install and you should be good for a couple more weeks.
Its a very old CPU and not meant for Win10.
