At random intervals the GPU fans will begin to run at 100% (4300 or so rpm).  If left to continue doing so the displays will shut off and a "cold reboot" will be necessary.  A standard restart continues the same cycle. When it occurs the GPU monitoring (NV-Z 0.7.2) starts showing fan speed at "0%, 4300 RPM" and GPU temp jumps up and down erratically (by liks 30 degrees indicating some sort of read error).  This has happened on this machine with both the 660ti FTW and the GTX980SC I put in to try and resolve the problem.  I have seen many other posts where people have had the same issue with many different GeForce cards but have been unable to find the solution for it.
 
The 660ti runs fine in another machine as does the 980.
 
The 980 is running correctly in the machine right now and fan/temp monitoring is working fine but I never know when it is going to go haywire.
 
I have not installed any new programs except EVGA OC-x and EVGA PrecisionX to monitor this issue (installed after the issue started) and have no other programs for monitoring or affecting the GPU.
 
This problem started recently (3-4 weeks ago).  The computer has been running fine for more than a year.
 
MoBo bios is at the most recent release.
 
I reverted drivers to 344.11 for the 980GTX and 314.22 for the 660ti hoping to resolve the issue but it happened again today.
 
Any help would be appreciated.
 
 
ASUS M5A99X Pro rev2.0, AMD FX4170, 16GB Corsair, Thermaltake 850W, GTX980, 500GB WD Cavair, ASUS PB728Q and HP 2311xi

Thanks for the suggestions and for responding to the post.  The issue started prior to installing precision but I tried it with and without precision and still same result.  It also occurred on this machine with both the GTX660ti initially and then again today with the GTX980SC (brand new two weeks ago).  The GTX660ti does run fine on another machine.
 
I am running windows 8.1.  Has anyone had a Windows 8.1 driver issue cause this?
 
And Scarlett with all those SSD drives.  Any suggestions for me on one to replace my main spinner?

Thanks about the drives.
Not sure on the heat sensor but it seems that the temp reporting goes wonky since the GPU monitoring has the temp jumping all over the place.  If I get into Bios right away there seems to be some voltage issues going on but i'm not sure what's causing them and If I hit the memtest button (which causes a coldreboot and fresh driver load) the issue goes away!  Could it be power-supply related?  This Thermaltake 850 is about 2 years old so I wouldn't think it's bad but I don't know how to check.  It even ran the fans on the GTX980 in sleep mode!  I think this might be a Windows 8.1 issue but the boards there are soooo helpful.
 
Seems a very similar issue to one that I have seen posted about:
 
 
 
 
and here is even an old school one!
 
Edit: well apparently the forums won't allow me to post any links.
 
 

bsmegreg

Scarlet-Tech
I dug gest changing the thermal material on the core/cooler if possible. A friend went through this and it was a small bubble in the thermal material on his 780, causing one part of the core to overheat and shut off.

Don't touch the inside of the card, I would imagine any warranty is void if you down that road

Nope.  I pay EVGA for the extra warranty and RMA service so if the card goes bad I will RMA it.  No opening it up for me!

Maybe he has a virus on his PC? When the fan goes up, does the GPU usage go to max? Otherwise, the problem could be a bad a sensor like Sajin said. So maybe it wouldn't be a bad idea to RMA the card and cross your fingers.

The 660 has been running for the past week in the other machine without issues.  I tried the 980 in the other machine for about a day running all day on World of Tanks without issue.  Both cards had the issue in my main machine.
 
When I get into bios after the issue occurs the voltage for the card and for the CPU are running low.  The ASUS EZ setup shows the voltages in red like 1500 mv for the CPU and 600 mv for the GPU.
 
If I hit the MEM check button that Asus builds onto the Mobo or do a cold reboot (APU power off) then it seems to reset things and it works fine for a time.  I have to make sure I shut the computer down after using it because if it goes into sleep mode then it immediately has the issue on wake up (I can replicate the fault by letting it go into sleep mode).

Uh, it's a BIOS issue its not a GPU problem because MemOK resets the BIOS for a period of time allowing for a clean boot and it's also a CPU undervoltage. Is your BIOS upto date? Otherwise you could try to reset it or flash it and that would probably fix the issue. MemOK doesn't save the settings it changes so eventually they will revert which is why it only fixes the issue for a little bit. But clearly for BIOS/CMOS has a bad setting.

Other things you could try:
set power for everything to AUTO in BIOS

Reset BIOS

Make sure PCI settings are good in BIOS