Crazy gpu temps

[r3uz0r]

Hi everyone. I want to apologize in advance for the long post, but I really need your help.

I own 670sli (both with waterblocks) and last week one of the cards started heating a little more than usual so I decided to fix it this past weekend. Everything went well, both cards has great temps (around 26) until Sunday afternoon my screen goes black and I thought windows screwed up. So I rebooted and after that evga precision only detects one card and the nightmare begins. I noticed one of the cards had a little bit of water between the pcb and the backplate, so I unplugged everything, removed both backplate and waterblock, cleaned and dryed everything up and now I have placed the card with a lot of rice. I also noticed the power connector extensions were humid and they are now drying. Hopefully the card will survive. I will only test this in about 3/4 weeks. Do you think it will survive?

Now I have another problem (as if the other one wasnt enough). After all that mess I booted my pc with the other card (after I cheked if it had any water too, but it didnt) and now it has very hot temps.. I was in windows with a few programs opened and it was around 55C. After that I opened bf3 and it went to 102C. I immediately closed bf and then it went to 70C. It is very strange since on saturday it was around 26C. I already confirmed that thermal pads and tim are both correctly placed. I've done this several times before, I dont know why this is happening.

Can anyone please help me?
thank you very much and sorry for the big post.

 

[looz]

Does the system have enough water in circulation? oO

 

[r3uz0r]

Yeah, I think so. All the tubes are filled, reservoir too and my cpu is around 30C. I don't really know what is missing here. I cheked every inch of the gpublock and everything is placed according the instructions.. could the thermal pads be used too much? they were working perfectly fine on saturday tho

 

[Remmy]

Did you reuse the same thermal pads when you put it back together? New ones each time would be better

 

[r3uz0r]

Yeah, they are the same. Actually I never changed them, so now I guess it would be the best time to do it. Its weird that Saturday it was all good and suddenly it goes to hell

 

[Remmy]

It's worth a shot. And there was no problem with this GPU after the other got a leak?

 

[r3uz0r]

After the leak happened, I opened this "dry" card to see if there was any water on it, but it wasnt. Then I turned on my pc with this "dry" card and this temp problem started. Today I opened it again, changed the tim, placed the pads and the problem persists. I followed the instruction even tho I've done this quite few times now. I think I will buy new tim, pads and hope for the best.

 

[Remmy]

Yeah it's at least something cheap enough that it's worth a try, and if not, then you've eliminated that as the problem, and not lost too much time or money.

 

[looz]

I doubt they'll have a difference of 25c vs. 105c.

 

[Remmy]

It's just that nothing else changed is all

 

[r3uz0r]

Maybe it wont make a difference but I have to try something. Like remmy says, changing tim + pads wont be expensive. From what I see, it will be around 20 euros. I think its worth a try. But if you can come up with a reason why this is happening, or a fix for it please say so

 

[Surfie]

Ok my diagnosis steps involve a few questions at the outset.

1. The card which is working, is it the same card which was already heating up prior to the issue with the second card?

If so, I would take a closer look at the card itself.

2. Are we certain that the water is circling through the system?

3. Is the gpu block actually cycling water or could it be air locked?

Recognising that this is not your first outing, I am assuming the obvious things have already been checked. .. I.e. Now that you are running single card v you go in one side and out the other. .. etc

Once these questions have been answered I would start thinking about removing the water block and putting the air cooler on to run some basic card tests.

If you are still running into issues and you know that the air cooler is working (I.e. You can see and hear the fans spinning) Then, I'm sorry but the card is probably dying a slow death. If over clocked, wind it back to stock settings and check again. You might be able to extend its life this way a bit.

If the issue more or less goes away on air then it's a issue with either the block or how it's mounted.

Good luck!

 

[r3uz0r]

Ok my diagnosis steps involve a few questions at the outset.

1. The card which is working, is it the same card which was already heating up prior to the issue with the second card?

If so, I would take a closer look at the card itself.

2. Are we certain that the water is circling through the system?

3. Is the gpu block actually cycling water or could it be air locked?

Recognising that this is not your first outing, I am assuming the obvious things have already been checked. .. I.e. Now that you are running single card v you go in one side and out the other. .. etc

Once these questions have been answered I would start thinking about removing the water block and putting the air cooler on to run some basic card tests.

If you are still running into issues and you know that the air cooler is working (I.e. You can see and hear the fans spinning) Then, I'm sorry but the card is probably dying a slow death. If over clocked, wind it back to stock settings and check again. You might be able to extend its life this way a bit.

If the issue more or less goes away on air then it's a issue with either the block or how it's mounted.

Good luck!

1. No no, the card that is working had no problems before this happened

2. Yes. When I put the loop back together, in the beginning I could see bubbles going through it, so I guess now that its all filled, t he water is going through as well

3. Well I think so, but if it is air locked how can I fix it?

Yeah, I didnt think of that lol tks. I will try it out later today with the air cooler and see how it goes. Hopefully it is alright..

 

[Surfie]

1. No no, the card that is working had no problems before this happened

2. Yes. When I put the loop back together, in the beginning I could see bubbles going through it, so I guess now that its all filled, t he water is going through as well

3. Well I think so, but if it is air locked how can I fix it?

Yeah, I didnt think of that lol tks. I will try it out later today with the air cooler and see how it goes. Hopefully it is alright..

If it has liquid in the block, it's not airlocked.

Hmmm I just had a random thought. Were the card that was faulty before and this card plugged into the same pci express port when they started showing issues? If yes, It might be an issue with the motherboard not the gpu... For your sake I hope that it is, as they are cheaper to replace than gpus...

To test, plug this card into any other pci express port and test, if no issue, it's probably the motherboard port to blame.

 

[r3uz0r]

yeah, now that you said that, I'm using the same pci.. hmm worth a shot! Later today I will post the results ^^
thank you so much for your help and time mate, really appreciate it.

 

[Surfie]

No problem. Another (albeit unlikely) culprit is the psu. I've seen them do some really weird things when they were starting to die before.

Let us know how it goes!

 

[r3uz0r]

So I switched the pci and with the waterblock it was still showing very high temps. Now I got it on air and its working properly ^^ I'll also order some new thermal pads and in 3/4 weeks I'll try again wc both cards! Hopefully that will solve it and I won't need to buy new blocks

 

[Surfie]

So I switched the pci and with the waterblock it was still showing very high temps. Now I got it on air and its working properly ^^ I'll also order some new thermal pads and in 3/4 weeks I'll try again wc both cards! Hopefully that will solve it and I won't need to buy new blocks

Hmmm. Certainly sounds like either a mounting issue or a tim issue then.

Best of luck with it!

 

[Warchild]

I strongly think you did not have proper contact with the block and the GPU core. That can be the only explanation if it works fine on air. Either that or you maybe forgot to use TIM on one of the cards?

 

[d3rrial]

Or like it was previously suggested, that he used the water inlet as water outlet aswell so no water ran through the card, but bypassed it. (This can happen when you watercooled your cards in parallel before, not serially and simply forgot to put the outlet to the other side now)