Graphics Card Fan Won't Change/Increase

[izytang]

I've recently had problems while playing games where the whole system crashes. After touching the video card and realizing it was much too hot, I decided to monitor (with SIW) the temperatures while playing games and sure enough, they rise to over 100 C. At the same time, I'll check the fan speed and it'll be in the range of 80% (correct), but the actual speed is 1140 RPM (25% default). I've updated the drivers of my GeForce GTX 570 card (version 337.88) and no change.

I decided to disable nVidia's GeForce Experience software and manually attempt to change the fan speed. I installed both MSI Afterburner and EVGA Precision to manually control the fan speeds (using one at a time, so there are no conflicts). When doing so, I'll get temporary lapses of having control for 5 - 20 minutes, then it goes back to the 1140 RPM default. Note that the speed will be what I set it (~80%), but the fan speed doesn't adjust to the correct revolution count, and remains at 1140 RPM. I also tried setting fan speed curves, and there is no change.

The behavior of it all is interesting as well. Sometimes the fan speed does adjust without any problems, both in manual and auto (as well as curves). These periods last from a few minutes to about an hour. When the system does seemingly decide to set the GPU's fan to 1140 RPM (25%), it stays at that regardless of what settings I change (via Afterburner or Precision). That period lasts from 10 minutes to an hour, after which I have control again. The card's fan is capable of running at full speed and operates faithfully when it does. There are no airflow issues and it has been running in its current configuration and location for two years (ie. no system component changes). Because of this, I believe the issue is software-related. Whether that is in the operating system or the graphics card firmware, I'm not sure.

Has anyone else had similar problems? Afterburner and Precision are nice, but is there a deeper level to look at and change variables associated with the card's fan? Any help would be fully appreciated. For system specifications, see the following:

Hardware

Motherboard: ASRock M3A770DE / Bios: P1.80
CPU: AMD Athlon II X4 640
Memory: 20 GB DDR3 G.Skill F3-12800CL9
Graphics Card: nVidia GeForce GTX 570 (distributed by eVGA)
Audio: M-Audio Delta 1010LT Multichannel interface on PCI
Storage: 6 TB (3 TB Seagate, 2 TB Hitachi, 1 TB Seagate)

Software

Operating System: Windows 7 Professional 64-bit
Video Drivers: GeForce version 337.88
Video BIOS: 70.10.58.00.72
MSI Afterburner 3.0.1
EVGA Precision 4.2.1.2143

 

[i7Baby]

Try your card in another pc.

Try another card in your pc.

100C - your card is probably cooked.

 

[izytang]

Try your card in another pc.

Try another card in your pc.

100C - your card is probably cooked.

Hey, thanks for the response. I unfortunately don't have another system or card to test - can I make a determination without doing such? Also, I believe the card is still good. A lot of folks are saying that temperatures around 90 or 100 C are fine. (Source: http://www.tomshardware.com/forum/290865-33-safe-temp)

When playing a game, the GPU's temperature slowly rises to 100 C, at which point I exit the game so that it can cool. It appears to get up to around 110 C before the system crashes - but I don't want to test that limit too often. As stated before, when the fan does open it's control, it can operate at full speed no problem. Before looking at the hardware issues, are there any software related things that could be causing this?

 

[i7Baby]

The link says 70C OK. At 100C they get to shut down. Double check your temps.

Try Furmark or similar stress testing.

Is it overclocked?

What psu are you using - make and model?

 

[izytang]

The link has quotes such as:

"No worries friend,most GPUs are capable of reaching 100 celcius.Your gpu temps are fine."

and

"And in games around 70-90. Hit 100 when playing Bad Company 2"

The thread seems to indicate that temperatures approaching 100 are okay, but it isn't clear what the exact limit is. It would be nice to get a manufacturer specification on temperature ranges for GPUs. nVidia has the GTX570 spec'd as having a maximum operating temperature of 97 C. (Source: http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-570/specifications)

It isn't clear whether that number is the maximum safe operation or the max before failures occur. I'm actually an engineer by trade, and when we specify equipment, we provide the maximum temperature that can be sustained indefinitely before failure rates rise above 1% for a pre-determined duration. It's not clear if nVidia is using a similar definition.

Anyway, the notion of temperatures is a red herring. Whether it's okay at 90 - 100 C for any length of time doesn't matter if it still shuts down during gaming. The question is why isn't the fan speed being adjusted on command?

Answering your questions:
1. The graphics card is not overclocked. It is set to it's stock settings: 797 MHz cord clock, 1950 MHz memory clock.
2. The power supply is a 750 watt Corsair CX series CX750

I've attached a couple of screenshots to illustrate what is happening.

Exhibit A: When the GPU's fan is in "no control" mode, it doesn't matter what the fan speed is set to. It spins at the minimum 1140 RPM.

Exhibit B: When the GPU's fan is in "control" mode, I can adjust it at will and the real speed updates.

 

[i7Baby]

The command goes through the pcie slot. Try cleaning the slot out.

Also blow the card out.

It is showing your fan running at 4350rpm - surely thats 85%

 

[izytang]

Anyone else have some ideas on what might be causing this issue? Once I get the opportunity, I'll swap graphics cards and/or whole systems and see if the problem duplicates. I still feel the issue is software related.