Furmark - very high GPU temps on stability

[gramboh]

Hi guys,

I recently downloaded Furmark as an artifact/stability checker for my video card overclock. Running in stability mode (1280x1024 2X AA) with the fan on auto quickly ramps the fan speed up to 100% and the card slowly creeps up to 103c after 200 seconds. I closed the benchmark at this point as I didn't want it to hit 105 and beyond.

Is this normal? In Crysis the highest temp I've seen (1920x1200 everything very High) after playing for an hour or two is 87c. In TF2 the temp rarely goes much over 75c.

Running in benchmark mode it gets up to 86c in 60 seconds at 1280x1024 no AA. The score is 4408 o3Marks if it matters (73 avg FPS).

Thanks

 

[Mem]

Originally posted by: gramboh
Hi guys,

I recently downloaded Furmark as an artifact/stability checker for my video card overclock. Running in stability mode (1280x1024 2X AA) with the fan on auto quickly ramps the fan speed up to 100% and the card slowly creeps up to 103c after 200 seconds. I closed the benchmark at this point as I didn't want it to hit 105 and beyond.

Is this normal? In Crysis the highest temp I've seen (1920x1200 everything very High) after playing for an hour or two is 87c. In TF2 the temp rarely goes much over 75c.

Running in benchmark mode it gets up to 86c in 60 seconds at 1280x1024 no AA. The score is 4408 o3Marks if it matters (73 avg FPS).

Thanks

FurMark does stress the video card more then any game out there that I know of,end of the day if its stable with FurMark then any game should be stable too.

My 4870 under load of a normal game goes to about 86-88c,FurMark can make it go to 92c (Max for my card) with no artifacts or stability problems.

I do hear the 280 can get quite hot,I'm sure 280 owners will chime in later about it.

 

[airhendrix13]

Sounds pretty hot to me. I think the GT200 cards are suppose to run cooler than the 8800 cards, and my 8800GTS maxed out at 85C. I could be wrong though.

 

[OCGuy]

RMA time

 

[AmberClad]

Wow, people are obsessed with HairyDonutMark...

Originally posted by: Ocguy31
RMA time

If it were one of the defective ones, it seems unlikely he'd be able to play Crysis at all, for several hours on end. Donutmark is pretty intensive, more so than any games, I've found.

 

[chizow]

Would help if we knew your case temps and fan set-up. 105C under any conditions seems hot to me, but as others stated, most of the problem cards will quickly rise to 105C even under moderate load or gaming.

 

[Mr Fox]

Originally posted by: gramboh
Hi guys,

I recently downloaded Furmark as an artifact/stability checker for my video card overclock. Running in stability mode (1280x1024 2X AA) with the fan on auto quickly ramps the fan speed up to 100% and the card slowly creeps up to 103c after 200 seconds. I closed the benchmark at this point as I didn't want it to hit 105 and beyond.

Is this normal? In Crysis the highest temp I've seen (1920x1200 everything very High) after playing for an hour or two is 87c. In TF2 the temp rarely goes much over 75c.

Running in benchmark mode it gets up to 86c in 60 seconds at 1280x1024 no AA. The score is 4408 o3Marks if it matters (73 avg FPS).

Thanks

You have a Bad Card.. they are prone to issues like this... 105C is too hot the VRAM will just error... 85C is the highest VRAM will run without major issues.



http://forums.anandtech.com/me...=2213906&enterthread=y

 

[Leon]

105C is throttling point, that's why the temp wont get any higher. There appears to be a production problem with some GTX280 - RMA it immediately.

 

[OCGuy]

Originally posted by: Mr Fox



You have a Bad Card.. they are prone to issues like this... 105C is too hot the VRAM will just error... 85C is the highest VRAM will run without major issues.

Thats odd...my old 8800GTS 640 would hit 92C when OC'd and I never had major issues. Got data?

But yes 105C is definatly across the line.

 

[Mr Fox]

Originally posted by: Ocguy31

Originally posted by: Mr Fox



You have a Bad Card.. they are prone to issues like this... 105C is too hot the VRAM will just error... 85C is the highest VRAM will run without major issues.

Thats odd...my old 8800GTS 640 would hit 92C when OC'd and I never had major issues. Got data?

But yes 105C is definatly across the line.

Look at The Technical Data for any of the GDDR4 Vendors... it will run up to 85C-90C before needing recalibration... the VRAM is removed from the GPU, but in the case of the Overheating Failure Mode it will saturate the Cooler... the one mode changes Power Consumption from 19 amps to 25+ Amps.

In the Xtreme thread they had noticed the power consumption issue, and that was what the data worked out to...

 

[OCGuy]

Originally posted by: Mr Fox

Originally posted by: Ocguy31

Originally posted by: Mr Fox



You have a Bad Card.. they are prone to issues like this... 105C is too hot the VRAM will just error... 85C is the highest VRAM will run without major issues.

Thats odd...my old 8800GTS 640 would hit 92C when OC'd and I never had major issues. Got data?

But yes 105C is definatly across the line.

Look at The Technical Data for any of the GDDR4 Vendors... it will run up to 85C-90C before needing recalibration... the VRAM is removed from the GPU, but in the case of the Overheating Failure Mode it will saturate the Cooler... the one mode changes Power Consumption from 19 amps to 25+ Amps.

In the Xtreme thread they had noticed the power consumption issue, and that was what the data worked out to...

Hmmm...I wasnt aware that Precision Tuner had a read out for VRAM tempurature...isnt it the GPU core?

 

[Mr Fox]

Originally posted by: Ocguy31

Originally posted by: Mr Fox

Originally posted by: Ocguy31

Originally posted by: Mr Fox



You have a Bad Card.. they are prone to issues like this... 105C is too hot the VRAM will just error... 85C is the highest VRAM will run without major issues.

Thats odd...my old 8800GTS 640 would hit 92C when OC'd and I never had major issues. Got data?

But yes 105C is definatly across the line.

Look at The Technical Data for any of the GDDR4 Vendors... it will run up to 85C-90C before needing recalibration... the VRAM is removed from the GPU, but in the case of the Overheating Failure Mode it will saturate the Cooler... the one mode changes Power Consumption from 19 amps to 25+ Amps.

In the Xtreme thread they had noticed the power consumption issue, and that was what the data worked out to...

Hmmm...I wasnt aware that Precision Tuner had a read out for VRAM tempurature...isnt it the GPU core?

I'm sorry you mis-understand...

The Stock Cooler was designed to dissipate 220-240 watts IE;18.33 - 20 amps
If the overheating is due to an over volt the observed difference is a 6.66+ Amp Over volt...

The Cards that are effected have Memory Error written all over the symptomology.
but it is due to the overvolt failure mode it is due to the heat being too much for the cooler... The Heat saturates the card, and the VRAM starts to error after getting too hot.

 

[BFG10K]

As others have said 105C is RMA time as it signifies you've got one of those faulty cards.

 

[Leon]

By the way, Furmark got Red Team worried as well

Linky

ATi's driver team obviously decided to axe the notorious "heater" application, by detecting the executable at startup, halving down the performance and thus the heat output, in the latest Catalyst 8.8 beta leak.

I can confirm this issue myself...when running Furmark with no AA on (4870) it black screens, with fan spinning at 100% (thats 4500rpm) and requiring a reboot. Watch out, this little app is evil :laugh:

 

[gramboh]

I don't think I have a faulty card since it doesn't get much over 85c in Crysis after a few hours. Reading posts from n7 and others on the EVGA forum about defecting cards (I read a lot while waiting for mine), I don't think I have one. The defective cards will hit 105c instantly in Crysis, or even in the menu for GRID. Actual gaming and 3DMark06/Vantage is fine, it's just Furmark that sends it soaring.

I never actually let it throttle as I closed Furmark once it hit 102/103.

I don't think I'll RMA unless I see actual artifacts in a game.

 

[Qbah]

My HD4870 hits 105C on one of the sensors during stress testing. Core is running at 91-92. Was running for an hour with no problems. As long as you don't have problems in games you're good to go imo.

 

[Keysplayr]

If you're playing Crysis for hours and only hitting 85C, then I would say you are fine. Just keep an eye out for any artifacting or lock ups/chugging/card throttling.

 

[Mem]

Originally posted by: gramboh
I don't think I have a faulty card since it doesn't get much over 85c in Crysis after a few hours. Reading posts from n7 and others on the EVGA forum about defecting cards (I read a lot while waiting for mine), I don't think I have one. The defective cards will hit 105c instantly in Crysis, or even in the menu for GRID. Actual gaming and 3DMark06/Vantage is fine, it's just Furmark that sends it soaring.

I never actually let it throttle as I closed Furmark once it hit 102/103.

I don't think I'll RMA unless I see actual artifacts in a game.

I love FurMark for the reason it stresses the video card like nothing else out there,if I was you I would just throw some more games at it for longer periods of time and see if it remains stable under load at normal gaming temps etc..if it does then you should be safe,its quite normal for me to test a new video card in a game all day,anyway end of the day you got the warranty covering you,remember no game will get as hot as FurMark does in testing.