Have a got a temp problem???

[danceforfreedom]

I was reading through this other thread and noticed a shot of GPUz and saw how there wasn't much difference in temps between the different readings on the guys card. Here's mine...

I saw how theres about 18c difference in the main temps and the GPUz Temp #2 reading with my card.

This got me worried, I always have afterburner running and have checked after a few hours of gameplay before, and it never usually goes above 70. Made me think though, the temp afterburner shows is the main temp, and from the other guys card and shots there's usually not that much difference in the different temps. But with 18c difference in mine idle I wanted to see the difference when being stressed!

This is with Furmark in a 1280x1024 window with extreme burning and post effect on and fans at 70%...

I know the furmark settings are more of a stress than some games will be but im seriously thinking this can't be right. 120c which is a difference of 43c compared with the other reading, thats over 50% hotter than the other sections of the card.

The card is a 1 month old msi twin frozr (with twin fans), case has more than enough air flow and a low ambient temp inside. This cant be right surely, is it actually possible for there to be that much difference across the one card, I know its not safe. I don't know what part of the card 'Temp #2' is but is this likely a dodgy reading or if its a real reading a dodgy card!?

I seriously don't know what to think, any idea's anyone?

 

[JTrickle]

At idle there is only 5c difference between that sensor and the main one on my reference 5850. Under load I've seen the difference get to about 9c.

I'm not sure there is a lot you can do about it. You are going to have to hope its a faulty sensor. Unless you fancy overvolting and trying to 'force the issue'. If you can get a 130c+ reading out of it I rekon it will definitely be the sensor. Either that, or that part of the chip will be feeling like the space shuttle on re-entry and having trouble getting life insurance.

I've forgotten to turn my watercooler on a few times and auto shutdown is when the main chip hits 106c. Rumour is that the chips are certified for 125c even if auto is at a lower temp. But at what volts? And is it rated to survive 125c at idle, or prolonged exposure under load? Thats why those rated rumours from last year were no use.

If you were not dead set on overclocking I would just ignore it. Games will never get that chip as high as you did with furmark.

 

[danceforfreedom]

Either that, or that part of the chip will be feeling like the space shuttle on re-entry and having trouble getting life insurance.

LOL

The readings above are with the overclock: 900core, 1200memory at 1.275volts.

I've just reduced the volts to 1.25v to see it I can get it stable at the 900/1200 overclock and im about to run the furmark again.

The 43c difference in the readings is not normal. I know there's always a small gap and the 'Temp #2' is always the highest but 43c is definately way more than normal.

I hope its a dodgy sensor, like mentioned, ive got the volts down to 1.25 volts, im about to run furmark burn mode + post effect and ive also got the side of my case open.

Thing is really dont want to risk touching it if it is 120+ probably will cook my finger, think I should notice by running my palm just over the top....

 

[danceforfreedom]

Ive got the test running now, 15 minutes in. Can run my hand over the top of the card, even touch each place slightly for a second. I notice slight changes in heat along the card, but not to big.

Main temp is 75c along with the others but temp #2 says 115.5c. Its really hot all along the card but I can still touch for a second all along (with a sting). Im thinking that 75c is enough for that sting and if 115c was true I would really notice it?

Im touching not just the topside of the plastic or fibre or whatever it is of the card but also the metal that comes through the top!

 

[danceforfreedom]

After 45mins of extreme burn test in furmark It said it got to 120c on temp #2 but I could still touch it topside both card and metal with just a hot sting and no real difference all along the card to touch.

Does anyone have any idea whether the reading of 120c is real or more like a dodgy sensor?

Im quite sure if any part of the card was 120c I would have really noticed a big different compared to touching the rest, as the other readings where all 80c. Im quite sure its a dodgy reading but then don't have experience of judging by touch?