I've got a 5 GPU mining rig running built with EVGA GTX 1070 SC's. The miner has been running flawlessly for a good year now. The OS is windows 7 and the mining software is the latest phoenix miner. Power supply is an EVGA 1000 watt supernova G2. I've had this system moderately overclocked since I put it together. My most stable and efficient settings are running about .700 mv, core clocked at +100 over default, and memory clocked at +383 over default. Using nvidia inspector for the clock settings. P2 state is forced to off. Temps have never been over 55C on the cores. APC UPS backed up, so it has never seen a dropout or a hit to the power supply. I would say this is a well taken care of mining rig.
So the problem is this rig ran stable at 31.x MH/sec per card since I put it together. But in the last few months, the hashrates have been slowly creeping down to a point where all five cards are now running in the 28 to 29 MH/sec range. And I've had to work my voltage up to .840 mv just to maintain that. It seems like every time this machine gets a reboot, which is maybe once or twice per month, I have to creep up my voltage slightly just to maintain the hashrate I had before the reboot. As I mentioned, all five GPU's are dropping evenly. And I also have two other computers which also mine 24/7. They also have 2 EVGA GTX 1070 SC's each of roughly the same age, usage, and temps. Same OS, mining software, and the power supplies are both EVGA 650 watt supernova G2's. These rigs are not experiencing this drop in hashrate. For over a year, they're still running right where I initially set them.
What might be causing this even drop in hashrate? I don't think it's the GPU's themselves because of the even decline. I suspect either the power supply or possibly the riser cards used to mount 5 GPU's. From what I have said, what do you guys think? Has anyone else experienced this?