GPU overheating

[22350]

Hello,

I am having an issue with one of my Nvidia 2080ti overheating under load.

It's the bottom card.  

 

Strangely, it seems to run a bit colder until it doesn't

 

System

Huananzhi F8 motherboard

e5-2699 v3 xeon

128g 2133

dual Nvidia RTX 2080 ti FE

 

 

 

[Middcore]

83 C isn't a problem. 

[freeagent]

My top card used to sit at 100c just running Heaven 

[22350]

9 minutes ago, Middcore said:

83 C isn't a problem. 

It is when it gets up to 89 and the fans sound like they are taking off.  it is also sitting right nest to another gpu 20 degrees cooler

[22350]

1 minute ago, freeagent said:

My top card used to sit at 100c just running Heaven 

yea, but was that because it was a 2 space nvlink set up with no air gap? also i am talking about the bottom card

 

 

[freeagent]

Just now, 22350 said:

yea, but was that because it was a 2 space nvlink set up with no air gap?

 

There was about 1mm between the two cards.. it was ridiculous. It was so hot I broke the pair up because I felt like I was abusing them lol. I am hard on my hardware, but I do have a heart

 

83 isn't too bad, that space you have is luxurious  

 

You could add some stronger fans, but then you will have to deal with a little bit more noise.. I would wait till the summer

[Middcore]

8 minutes ago, 22350 said:

It is when it gets up to 89 and the fans sound like they are taking off.  it is also sitting right nest to another gpu 20 degrees cooler

Your video tops out at 83, if you want people to know it's getting up to 89, say so. 

 

This is what you're going to get with this kind of setup. One card or the other or both will be choked for air. Only way to correct it would probably be to do a full custom loop, which probably wouldn't be cost-effective on two-generation-old cards. For the amount of money invested in such a project you could get a faster single GPU. 

[22350]

2 minutes ago, Middcore said:

Your video tops out at 83, if you want people to know it's getting up to 89, say so. 

 

This is what you're going to get with this kind of setup. One card or the other or both will be choked for air. Only way to correct it would probably be to do a full custom loop, which probably wouldn't be cost-effective on two-generation-old cards. For the amount of money invested in such a project you could get a faster single GPU. 

Yes,  agreed.  The video ends at 83, but it keeps going.  strange thing is that it's the bottorm card that is heating up.  The top card is fine

[22350]

5 minutes ago, freeagent said:

There was about 1mm between the two cards.. it was ridiculous. It was so hot I broke the pair up because I felt like I was abusing them lol. I am hard on my hardware, but I do have a heart

 

83 isn't too bad, that space you have is luxurious  

 

You could add some stronger fans, but then you will have to deal with a little bit more noise.. I would wait till the summer

yea, like i mentioned to middcore, it keeps going up to 89.  The entire time the top card stays under 70

[freeagent]

1 minute ago, 22350 said:

yea, like i mentioned to middcore, it keeps going up to 89.  The entire time the top card stays under 70

Maybe try pulling your PCI slot covers, that might help a little. Have you tried switching the cards around to see if the temps change or stay the same?

[22350]

20 minutes ago, freeagent said:

Maybe try pulling your PCI slot covers, that might help a little. Have you tried switching the cards around to see if the temps change or stay the same?

That was my next thing to do 

[22350]

So peeling the onion back on this....

 

If I pull the cards and swap them, the nvidia driver gets destroyed and I loose my screen.

 

If I turn off sli and restart with just the primary card (originally the secondary card), the driver remains intact.  If I add the secondary card (originally the primary card) and boot, everything remains fine.  If I enable sli, it completely destroys the nvidia driver.

 

I have tried DDU in safe mode and same results with fresh driver install.

 

If I go throught the same steps with the cards in their original position, everything loads fine.

 

Very Strange

 

The take away is that I haven't been able to see if the bottom card overheats in the primary slot.

[22350]

So I figured out it was bad thermal paste on that card