Question RTX 2080 Super high temp underload and load fans

[Gunsmoke1]

Hello guys,
So I have a problem with my gigabyte RTX 2080 Super OC G8. The problem is that while gaming and is under 99% load it reaches around 85-86C and gpu fans starts to spin at 94-95% speed, it pretty loud and it starts driving me nuts.

Recently, while my gpu's warranty was still valid I took gpu to the service and it seems they didn't find any issue, they've stress tested it and the temp limits are pretty decent. I'm attaching some result screenshots of warranty service.
https://ibb.co/xfD8NxK
https://ibb.co/6WrctyP
https://ibb.co/ygthMJ9
In the past my gpu worked fine, other component works perfect, cpu temp is normal, except gpu. And it doesn't matter if its winter and room's temp is +20C or if it's summer and room's temp is +30C. My gpu still reaches this high temp after about 3mins underload while gaming and fans starts spin like crazy.
Airflow of my PC is decent too - 3 exhaust fans and two intake fans at the bottom plus AIO fans. Adding photo of chassis too:
https://ibb.co/1z80VnB
It's very odd that warranty service manages to get 71-75C of the same gpu while on my system it runs at 85C. Warranty service suggested me to try and reinstall windows on my PC but I'm having a hard time believing this would make any difference. Other than that, I'm out of ideas what can I try to do to normalize these temps.
If anyone has a clue what I should try, please tell, any help is appreciated. Thanks.

 

[Darkbreeze]

What are your full hardware specs?

Given the age of that card and if you have used it pretty regularly putting a good amount of miles on it, it may be time for the cooler to have new thermal paste applied.

I'd also make sure you have the most recent stable motherboard BIOS installed, the latest motherboard chipset drivers and the latest Nvidia drivers. Might even be a good idea to run the DDU to do a clean uninstall of all the Nvidia settings and configuration and then do a fresh install of the Nvidia drivers. And if you're not already you might also want to not rely on the card's own configuration for the cooling fan curve, and use either Gigabytes tweak program or MSI Afterburner to configure your own curve. Just keep in mind that you have to keep ANY graphics card tweak program open at all times, although it can be reduced to the system tray, or else it will not continue to control the fan curve and it will revert to the card's default routine.

 

[Gunsmoke1]

What are your full hardware specs?

Given the age of that card and if you have used it pretty regularly putting a good amount of miles on it, it may be time for the cooler to have new thermal paste applied.

I'd also make sure you have the most recent stable motherboard BIOS installed, the latest motherboard chipset drivers and the latest Nvidia drivers. Might even be a good idea to run the DDU to do a clean uninstall of all the Nvidia settings and configuration and then do a fresh install of the Nvidia drivers. And if you're not already you might also want to not rely on the card's own configuration for the cooling fan curve, and use either Gigabytes tweak program or MSI Afterburner to configure your own curve. Just keep in mind that you have to keep ANY graphics card tweak program open at all times, although it can be reduced to the system tray, or else it will not continue to control the fan curve and it will revert to the card's default routine.

My full specs are: ryzem 3700x, 32gb of ram 3000mhz, mb - aorus x570 pro, psu - Corsair RM850x, chassis - fractal design define r6.
At first, I also thought of replacing thermal paste, but since warranty service shows 71-75C of my GPU it doesn't make any sense to change thermal paste, right?

I'll try to do clean uninstall of nvidia drivers and then reinstalling it freshly. What do you think of reinstalling windows as warranty service employee suggested me? Can windows reinstall make any difference in temperature?

P.S. what's interesting, in the past my gpu temps were fine even under load, it was like 71-72C, even during the summer, when room was +28C.

 

[Darkbreeze]

If your temps used to be 72°C and now they are reaching more like 85°C, and nothing else has changed, that tells me that over time the cooling has diminished. Technically that card is fine up to about 89°C which is where it should start throttling, but you really don't want that to happen regularly either because then you'll start eventually causing electromigration to occur in small doses. Cumulatively that will eventually have an effect. Has this card ever been overclocked for a period of time?

Do you control the fan curve manually or just let it ride with the default configuration?

Doing a clean install of Windows can't hurt. It might not fix anything, but if it's been a while since one was done it also can't hurt and might actually be helpful. Be sure to also reinstall the latest motherboard chipset, network and audio drivers from the motherboard product page. Don't EVER rely on the Microsoft supplied drivers as a permanent solution. They are rarely full featured drivers like the manufacturer's drivers for any given hardware component are and are mainly only there in case no other driver is available or to get you by until you can install the proper drivers.

I think running the fan curves manually could certainly help if you are not already doing that. Also, you might want to look at increasing the fan curves for your case fans and AIO slightly as well to ensure there is enough cool ambient air reaching the card and that the exhaust fans are getting the heat out of the case fast enough.

 

[Phaaze88]

It's very odd that warranty service manages to get 71-75C of the same gpu while on my system it runs at 85C.

We don't know the testing environment. They could've been using an open bench.
It's not unusual for case ambient to be around 10C or higher than room ambient.

Airflow of my PC is decent too - 3 exhaust fans and two intake fans at the bottom plus AIO fans.

Why? Did you test to confirm if they were actually beneficial to the gpu?

 

[Karadjgne]

Having more fans does not mean you have air Flow, just means you have air. Not necessarily that it actually goes anywhere or does anything of value.

View: https://youtu.be/ICMKUSff_6I

1 fan in the wrong orientation meant a cpu at 100°C instead of 30°C.

View: https://youtu.be/OupiPZybyBE

12 fans, 0 airflow.

Windows has had updates. Motherboard software has had updates, bios has had updates, games definitely have had updates, gpu drivers have had updates, there's any number of situations where just because You personally did not change anything, doesn't mean that a gpu is suddenly reset and is now using Raytracing, or windows defaulted back to using Xbox DVR, or game fixes turned on Ambient Occlusion etc.

You temps went up, and not to your liking. Something has changed. Figure out exactly what changed and you'll have an answer, until then, just saying you didn't personally change anything is moot. Could be nothing more than a piece of fluff fell off a fan blade and is blocking airflow in the gpu, or could be a setting got changed with an auto update, could be anything really.

 

[Darkbreeze]

He has airflow. If you take time to look at the pictures you'd see that. AIO in front as intake, two top exhaust fans and one rear exhaust fan, plus two bottom intakes. He has air FLOW with that configuration. Let's focus on real problems, not imaginary ones. LOL.

 

[Karadjgne]

I did. What I see is a pretty dusty case, and a big ol' chunk of dirt in the gpu heatsink, top row...

What I don't see

and two intake fans at the bottom

 

[Richj444]

He has airflow. If you take time to look at the pictures you'd see that. AIO in front as intake,

With the AIO set as intake, he's pumping hot air from the CPU right back into the case directly at the GPU. They need to exhaust the air, not blow it back in the case. Preferably up top and move the two top fans to the front as intake.

 

[Phaaze88]

With the AIO set as intake, he's pumping hot air from the CPU right back into the case directly at the GPU. They need to exhaust the air, not blow it back in the case. Preferably up top and move the two top fans to the front as intake.

What a cpu AIO/CLC does as an intake is a little fart compared to the heat an open air gpu cooler pumps into the case.
There aren't many cpus that pull the kind of power recent gpus do, and those that can require specific circumstances. Games are not one of them.
That 2080 Super can do up to 250w easily, of which that energy gets scattered inside the PC.
Ryzen 3700X is what, 70w? Nothing the coolers on these other devices can't deal with.

It all matters even less if the user has air conditioning.

 

[Richj444]

What a cpu AIO/CLC does as an intake is a little fart compared to the heat an open air gpu cooler pumps into the case.
There aren't many cpus that pull the kind of power recent gpus do, and those that can require specific circumstances. Games are not one of them.
That 2080 Super can do up to 250w easily, of which that energy gets scattered inside the PC.
Ryzen 3700X is what, 70w? Nothing the coolers on these other devices can't deal with.

It all matters even less if the user has air conditioning.

It matters a lot.

A Ryzen 3700X shows a TDP of 65 watts, which is the same as my I5-11400. I don't know what the Ryzen actually draws under heavy load, but my I5 draws around 175 watts and my AIO pumps a TON of heat out of the top of my case. I am very glad it's not going back into the case because that would cause some serious heating issues.

A simple but effective rule to follow when designing airflow is: Hot air out, cool air in.

 

[Karadjgne]

Here's a pretty simple but effective rule: Pump hot air out, bring cool air in.

No. Not that simple. Have to understand how aios work. The cpu has a certain amount of power converted into heat energy. That's picked up by the coolant. The coolant doesn't change temps, it's still 33°C or whatever ambient case temp is. Doesn't matter if the cpu is 70w at 70°C, the coolant is 33°C. It is just a medium, a way to move that energy from point A to point B. So when the coolant area reaches the rad, it's still 33°C but is holding 70w worth of energy. That energy is transfered to the metal fins, which do heat up, dissipating that 70w over the surface area where the fans pick up that radiated heat and blow it out of the rad.

So you get 70w dumped into the case. Not 70°C air, the temp of the cpu. Cpu could be 50°C or 100°C, it's still only 70w of energy. With a 120mm AIO, that 70w is concentrated into a 12cm x 12cm square, so is warmer, with a 360mm AIO it's dissipated over a 36cm x 12cm rectangle, so 'feels' cooler, but is still the exact same 70w worth of energy.

Cpu temp does not equal radiator temp, coolant temp or even the 'hot air' blown into the case when the aio is mounted as intake.

Exactly the same amount of energy as dumped into a case by an aircooler. Spread that 'hot air' over the entire volume of a case, it's Nothing even close to really affecting a gpu or the ambient temps used by the gpu.

What does affect the gpu is when the intake aio is in Push configuration as the radiator is a giant diffuser and kills any static pressure the fans would have created, so stops any direct airflow from the intakes from reaching the gpu. Put the aio as Pull, or use extra fans in Pull, and the rad has zero affect on static pressure putting airflow directly at the gpu.

 

[Phaaze88]

It matters a lot.

A Ryzen 3700X shows a TDP of 65 watts, which is the same as my I5-11400. I don't know what the Ryzen actually draws under heavy load, but my I5 draws around 175 watts and my AIO pumps a TON of heat out of the top of my case. I am very glad it's not going back into the case because that would cause some serious heating issues.

A simple but effective rule to follow when designing airflow is: Hot air out, cool air in.

It doesn't. It's been overblown.
I've played around with it using my hardware as well, but the case was a CM H500P Mesh at the time, which I DIY'ed a mesh window in the top panel.
Since then, I've been baffled at all the cpu AIO/CLC intake disapproval, but seemingly no one so much as bats an eye at open air gpu coolers doing the exact same thing, yet worse...

AMD's TDP ratings for Zen have been much closer to their chips' actual power draw than Intel's have been.
Your 11400 draws around 175w regularly... in GAMES? What the-... my 7820X does like half that.
The heat coming off the top AIO is from the cpu, and EVERYTHING else below it. You are aware of this, right?
The statement about heat not going back into the case isn't accurate. It eventually makes its way back to the source: energy used by the PC is released into the room, increasing room ambient, and some of that is drawn back into the PC, increasing case ambient. It's all one big loop, but air conditioning trivializes it. If AC is not available to the user, then making use of lower gpu power limits and fans(appliance) to guide the air from the PC out of the room helps.

 

[Darkbreeze]

It matters a lot.

No, it really doesn't. LOTS and LOTS of testing has been done, and even with a high TDP processor the overall average internal case temp when using an intake configuration AIO very rarely exceeds a few degrees, which is not enough to have any realistic effect on the graphics card temperatures, besides which, the elephant in the room that you are intentionally trying to ignore is the fact that this configuration already existed in the exact same fashion prior to the change in graphics card temperatures which completes eliminates the AIO from having any relevance to the issue. It had the same AIO in front before, but has a different thermal profile now, so the problem is something that has either changed or aged, neither of which is going to be CPU or AIO related. If anything, an aged AIO would put LESS heat into the case and you'd see CPU thermal problems, not the other way around. Not unless you suddenly started overclocking the CPU and had not been before.

So again, we're looking at issues that can't have any relevance.

I did. What I see is a pretty dusty case, and a big ol' chunk of dirt in the gpu heatsink, top row...

No idea what image you're looking at, but the picture of the case internals that I'm seeing has none of that that I can see. It might not be clinically, eat off the surface clean, but it's clean enough in there. I certainly would be on board with blowing out the heatsink on the graphics card and repasting it, like I said before, but I don't see a lot of dust in there and I definitely can't see any big 'ol chunk of dirt unless I'm entirely blind. But regardless, by all means, blow out the fins on the heatsink. It can only help. Then if that doesn't help, consider replacing the paste. Bottom line though, cards tend to increase their power consumption over time to do the same amount of work they did before. This happens to pretty much all graphics cards and to some extent, CPUs as well. When components use more power to do the same amount of work, they produce more heat.

Blow out the fins and manually configure a fan curve for the graphics card using Afterburner or another utility and if that doesn't cure the problem, redo the paste. Pretty sure that will fix your issue, not that one actually exists since you're still below specifications but I do understand wanting to see similar temperatures as what you are used to seeing.

 

[Richj444]

No. Not that simple. Have to understand how aios work. The cpu has a certain amount of power converted into heat energy. That's picked up by the coolant. The coolant doesn't change temps, it's still 33°C or whatever ambient case temp is

Saying that the coolant do not increase in heat at all is totally and completely wrong. The CPU transfers its heat to the AIO surface. That heat is picked up by the coolant and circulated through the radiator fins. The fans dissipate the heat from the radiator. Depending on which way the fans blow, the heat either goes into the room or into the case.

It works exactly as the radiator in a car works, or the old radiators that were used to heat buildings. Pass a liquid over a heat source, then move that liquid to a radiator in a different location, where it is released.

Darkbreeze, I'm not intentionally ignoring anything. I realize that this doesn't account for a sudden increase in the OP's temps, and my guess matches yours, thermal paste - if nothing else changed. But it's still counter productive to use preheated air as part of a cooling solution, that's simple physics. But I am interested in reading some of those tests you mentioned regarding the intake configuration. Got any links handy?

And Phaaze88, I intended to add "in heavy usage" when talking about power draw for my cpu, and thought I had until I read your comment. Depending on the game it's definitely lower, probably no higher than 100 - 125 watts.

 

[Darkbreeze]

I'll see if I can dig them up. But again, it has no relevance to the situation because it's not a new addition to the equation.

 

[Richj444]

I'll see if I can dig them up. But again, it has no relevance to the situation because it's not a new addition to the equation.

As I said, I agree with that. Thanks.

Edit: I found this, interesting reading. If I read the chart correctly, front mounted with fans as intake and tubes at the top of the radiator, shows the GPU 7 degrees C higher than top mounted with fans as exhaust. Seems fairly significant to me. Of course it's just one source.

How to place your liquid cooler (msi.com)