Hey guys.
I have two rigs and issues with fan speeds.
0) 2x HIS IceQ X2 OC R9 390X
1) 1x ASUS DirectCU II R9 270X and 1x Sapphire Dual-X R9 270X
The 390X-rig runs quite stable. But sometimes I notice the temperature is rising above 90°C without a notable increase in fan speed. I sometimes set the fan speed to 50% on both cards to keep the temperature below 80°C, but I think that's not good for the fans to keep them running at high speeds.
The 270X-rig also runs stable, but the temperatures are going crazy. The cards are at temperatures of above 100°C after several minutes of mining. And yet again the fan speeds do not auto-adjust (it seems). I have to set the fan speeds to at least 60% to keep the cards running below 80°C.
I am wondering: Why are the cards not increasing fan speed automatically? 90 or even 100°C as far away from healthy I guess!
Any ideas how to automatically increase fan speeds without manually monitoring the devices all day long?
The rigs are open and well ventilated. It's just the fans not kicking in hard enough. Any way to adjust the auto-settings?
Cheers
ordoe
Edit: Ubuntu 15.04, AMD proprietary drivers 15.9
0 ·
Comments
60% fan speed is not a problem. I had many of my cards running with fan speeds at 80% - 95% before I built custom cases. Now they all run around 45% - 55% and stay a lot cooler, as well.
But my main question remains, why aren't the fans kicking in even at 100° ?
I think some cards are locked to max 70%.
As for the 100 degree issue, that sounds like a driver issue as you say I was always under the impression that the cards would simply shut off at like 90ish degrees.
Also from what i've read, gamers are wrong. Games have fluctuating resource use causing fans to spin up and down as needed.
Mining is a consistent work load, fans will stabilize at one speed.
High speeds are less of a concern for the long term health of the fans.
Constantly going from low to high speeds all of the time is MUCH harder on the longevity of a fan.
Chip temps are your biggest concern, having a GPU at 90degrees over a week is harder on the chip than 70degrees.
So if you are ok with the sound, try and find a stable speed that keeps your temps lower. If you trust the build quality of the cards, then let the fans spin higher. But they should still be spinning higher than you think overall anyway.
I created a shell script which keeps track of gpu temperatures and sets fan speeds accordingly. It also creates a html dump so I can monitor them from remote any time. And it logs the temperatures to CSV. That why I see if there were any issues during the day.
Now my cards run between 60 and 80 C with fan speeds of 40-55%. Stable setup.
Does it work on Ubuntu?
https://gist.github.com/anonymous/2fb7bf2aa2eee0d929fd
I also thought about adding highcharts which reads the csv, but I couldnt wrap my head around the highcharts API yet :-)