Is this a possibility?

[Maximilian]

An overclocked GPU which is stable at 60 degrees becomes unstable at 70+? Anyone ever had something like that happen or is temperature an unlikely cause of instability?

 

[n0x1ous]

I think temperature can be a cause of instability but 70 degrees shouldn't be enough for it to be a factor. That's a great temp for a gpu under load....

 

[Termie]

What's the GPU?

And is it actually under load at 60C? Sounds like a very low temperature for an overclocked GPU unless it's under water.

My guess is that it's not a stable overclock. 70C would not cause instability in any GPU.

 

[nwo]

Temperature is an unlikely cause in my opinion. Either the video card is defective or unstable at those clock speeds, or it could be a power supply or wall outlet issue.

 

[JBT]

If its your 5850 thats pretty low. I think low 80's for an OCed card is pretty normal under load.

With that said If its an old card that has been OCed for quite some time who knows what they will start doing when its ready to die. You could have VRM's getting toasty due to old thermal compound etc.

 

[Maximilian]

Its a 7850, i whacked it up to 1050/1400 from 860/1200. Didnt touch the powertune +/-20% thing at all.

Seemed stable overclocked at ~60 degrees then i stuck a 7970 next to it, now it gets to 72 overclocked. Its for dogecoin mining. Im testing its stability at stock speeds now, see if i get another bluescreen. Could be it was never really stable or something else completely though.

Still waiting for pcie risers.

 

[JBT]

Honestly leave the RAM clocks at stock, they aren't going to help your performance.

 

[nwo]

I have my 7850 clocked at 1100/1300 using stock voltage and +20 powertune. I get around 370-375kH/s mining with it. It's an XFX Core 1GB 7850.

 

[Bubbleawsome]

It can. I ran my 770 at an insane 1400 bench and it was stable until it hit ~85c. If I turned the fan up it would be fine, but 85c was bad. I think it might have been TDP throttling, but temp target was 95c.

 

[KingFatty]

Perhaps it's the power supply, where adding the 2nd card put too much strain when you overclock both cards and the voltage dips? So while the temp also increases, maybe the temp is a red herring and really it's the voltage load and weakened PSU?

 

[nwo]

Yep, that's a good point. If the BSOD only occurs with the second card, the PSU may be the one to blame. I've had similar issues with my PSUs....

 

[Maximilian]

Well its been stable for around 12 hours so far with no overclock so thats promising. Ill not touch the memory clocks again then if it dosent make a big difference.

The PSU is a borrowed antec quattro from a mate. I dont want to plug my HCP 1200 in again because long story short a bolt fell out of it when I removed it from my main rig lol. Not good! So the PSU is a bit of an uncertain quantity, those quattros never reviewed all that well IIRC.

 

[taserbro]

Here's a review that describes a card that seems to be solid under 65degrees but dodgey when going above.
http://www.hardwareasylum.com/reviews/video/evga_gtx780ti_classified-kp/page13.aspx

 

[BrightCandle]

A colder transistor will run at a higher clockspeed given the same voltage. This is one of the reasons liquid nitrogen and liquid helium allow such impressive overclocks, that would not be possible without the transistors also being made very cold, its not just the cooling potential that they are doing.

Watercoolers often use chillers to take the temperature down another 10C to get exactly ambient as its often worth another 30Mhz or so.

Its absolutely possible that an increase in temperature of the environment could result in an overclock becoming unstable.