GPU Issue

[Spartan]

Before I start to seriously think about parting ways with more money to buy a better GPU, I need some advice on a temperature issue I'm having with my current one.

 

Now, before you go on reading, I know what you're thinking: "Wtf? Spartan, a TA, needs help with a temperature issue? :lul:".

 

Well, yes. This issue is not your usual one however. I'm essentially looking for someone that may know hardware a bit more in-depth to explain this, or at the least has experienced it and knows a solution.

 

Quick specs:

 

GPU - NVIDIA GeForce GTX 660 Twin Frozr

CPU - Intel i7 4790K 4GHz (water cooled)

RAM - 8GB

OS - Win 7 64 bit

 

My current GPU temperature is around 40-60°C depending on what I'm running on one of my two monitors. Certain applications like running a Twitch stream pushes it up to 60°, and it stays here with no problems whatsoever.

 

Previously when I would go into games, the temperature gets pushed into the 70s and would be fine there too. Recently it's started doing something strange. Let's say I run Rocket League. It's not a hard game to run graphically, needless to say I have no problem running it. However, slowly over time, the temperature rises from 60, to 70, to 80, and at around 80-90°C it will eventually shit the bed and raise the fan RPM to 3500 instantly, causing the temperature to dramatically fall back down to 60° within 5 - 10 seconds. At this point the temperature remains around 60 for a short while, and repeats the process over and over.

 

Things I've done to try and remedy the situation:

 

- Cleaned the dust out of all the fans inside the case.

- Replaced the thermal paste on the GPU.

- Ensured the PCI cable is properly attached, as well properly attached to the motherboard.

- Visually checked the outer and inner circuit board for any damage to the pins and capacitors (all seems ok).

- Set the MSI AfterBurner software to run a custom fan increase curve.

- Set the auto fan speed on/off.

- Benchmarked it with Heaven Benchmark.

- Looked for any anomalies with clock speeds.

- Updated NVIDIA drivers to the latest.

 

Well none of this has fixed or shown anything, the problem persists. My best guess is some faulty sensor chip causing a false reading, or something else similar. I can't think of much else to try to see if the issue still happens. Google search has shown some similar problems related to this but usually ends with "Replace it / RMA it". I don't think I have warranty on it any longer, its over 3 years old.

 

So, any thoughts?

[Rye]

Have you tried turning it off and back on?

[Cam]

Do you have another open slot that the GPU could be put into? If so, try that. And where did you replace thermal paste on the GPU...?

[Spartan]

Do you have another open slot that the GPU could be put into? If so, try that. And where did you replace thermal paste on the GPU...?

 

I haven't tried another slot yet, will attempt shortly. I replaced the thermal paste where it was previously lol (heatsink, core).

[Ham]

How's the actual temperature in your case overall? Since your CPU is watercooled it may not be suffering as much as your GPU if there is another component causing the heat. Maybe your PSU? I'm just throwing wild guesses here.

[Cam]

I haven't tried another slot yet, will attempt shortly. I replaced the thermal paste where it was previously lol (heatsink, core).

 

Ohh lol. I thought you had thermal paste around the slots or something.. LOL

 

Yea, just try it in a different slot and see if that helps.

[Spartan]

PCI slot #2 didn't change anything. I also tried not installing any Nvidia drivers and used the Windows one. Still no change.

 

The temperatures of everything inside my case is 100% fine. My GPU is fine at idle or with a bunch of broswer tabs open. Hell, even in-game it's fine but will randomly start increasing after a random amount of time.

[Drakanen]

It probably just decided it's done being your slave

 

Could just be a bug that the manufactures have to fix

Have you tried watching your task manager for anything that pops up when it starts to die?

[Zaraki]

And is this causing any problems? I mean, does it affect the performance?

It's obviously not normal, but if it doesn't start making sounds like a vaccuum cleaner or cause any crashes, I'd keep it like it is.

And is your desktop overclocked or not?

[Spartan]

Everything's set to default right now. When the RPM rockets up it makes a lot of noise as you may guess from the whirring motion. I wouldn't say it makes my performance suffer, but I don't want it to keep doing this cycle and eventually burn itself out from simply 'ignoring it'. Definitely don't want to damage anything else in there.