Question Overheating issue after changing the thermal paste ?

[Rxzy1x]

i bought a used inno3d rtx 2080ti oc 3x edition and decided to change the thermal paste as the temps were hovering around 83-84c while under heavy loads. Nonetheless, i replaced the paste and i was expecting some improvement but the problem got worse. the gpu was hitting 86-88 C now and that too on stock settings with a pretty aggressive fan curve. idk what did i do wrong or maybe i just moved the thermal pads or something but i tried changing the paste again and still no improvement . can somebody help me out with this issue please as i don't really know what i am dealing with here.

 

[Aeacus]

i don't really know what i am dealing with here.

Haul your GPU to PC repair shop and pay them for proper repaste. This is easiest and fastest solution for your issue.

Btw, it looks like the thermal pads/paste you used, is inferior to the factory default one. That would explain the temp diff. Either that, or you screwed something up when reassembling the heatsink on GPU.

 

[Lutfij]

Welcome to the forums, newcomer!

idk what did i do wrong or maybe i just moved the thermal pads or something but i tried changing the paste again and still no improvement .
You're the one who disassembled the card, I'd try and retrace your steps and perhaps even inspect all placements for thermal pads and even thermal paste. Speaking of thermal paste, what past did you use? I'm assuming you used isopropyl alcohol to remove the thermal paste that was on the card?

You also need to make sure that the thermal pads are placed as they should be since any misalignments or height differences can prevent the cooling plate from making full contact with parts that need to transfer heat into the cooling assembly.

You should also mention a couple of more things.
1| Make and model of your case?
2| Number of fans in your case and their orientation?
3| Ambient room air temps?
4| What GPU did you have prior to the RTX2060?
5| Can you prove that the GPU wasn't used for mining?

 

[Rxzy1x]

Welcome to the forums, newcomer!

idk what did i do wrong or maybe i just moved the thermal pads or something but i tried changing the paste again and still no improvement .
You're the one who disassembled the card, I'd try and retrace your steps and perhaps even inspect all placements for thermal pads and even thermal paste. Speaking of thermal paste, what past did you use? I'm assuming you used isopropyl alcohol to remove the thermal paste that was on the card?

You also need to make sure that the thermal pads are placed as they should be since any misalignments or height differences can prevent the cooling plate from making full contact with parts that need to transfer heat into the cooling assembly.

You should also mention a couple of more things.
1| Make and model of your case?
2| Number of fans in your case and their orientation?
3| Ambient room air temps?
4| What GPU did you have prior to the RTX2060?
5| Can you prove that the GPU wasn't used for mining?

1.) antec nx series 410
2.) 5 fans and 2 for intake, 2 are for the cpu radiator and 1 for exhaust
3.) 25 c is the room ambient temp.
4.) I used a rx 5600xt before this card.
5.) I can't really prove that but the guy i bought this card from used it for rendering and also showed me some of his work, so i thought it was enough of a proof.

 

[Rxzy1x]

Haul your GPU to PC repair shop and pay them for proper repaste. This is easiest and fastest solution for your issue.

Btw, it looks like the thermal pads/paste you used, is inferior to the factory default one. That would explain the temp diff. Either that, or you screwed something up when reassembling the heatsink on GPU.

i used corsair tm30 as thermal paste. i am ordering thermal pads to replace the previous ones but i am not really sure of the thickness.

 

[Aeacus]

i used corsair tm30 as thermal paste

In thermal paste roundup, Corsair TM30 is 2nd worst,
link: https://web.archive.org/web/2021030...ages/guru3d_thermal_paste_roundup_2019,9.html

The diff between that and the best is ~6C, which is about what you're seeing.

 

[Rxzy1x]

yeah i understand that but still i think it should be better than stock considering the fact it was all dried up when i changed the paste, at least should have seen similar performance instead of reduction in performance.

In thermal paste roundup, Corsair TM30 is 2nd worst,
link: https://web.archive.org/web/2021030...ages/guru3d_thermal_paste_roundup_2019,9.html

The diff between that and the best is ~6C, which is about what you're seeing.

 

[Aeacus]

but still i think it should be better than stock

Not all stock/default parts/products are bad. There are actually a lot more sub-par/poor products out there than what the factory offers from stock.

Regarding thermal paste; Corsair is big name brand and they can afford to cheap out on thing or two. So, their thermal paste being sub-par, isn't that much of a surprise. Since after all, name sells more than actual quality of thermal paste. And Corsair tries to sell each and every product there is to be sold regarding PCs (including furniture).

For comparison, Arctic Cooling only focuses on coolers and fans. Mostly CPU air coolers, but also AIOs. There have been some GPU heatsinks as well. Due to that, they focus far more on good quality thermal paste, like the MX-4, which fares good on testing. Also MX-4 is better than the predecessor MX-2. Latest is MX-5 though. (I'm using MX-4 under my CPU.)

Thermal Grizzly only specializes in thermal compounds, so, their products being the best is essentially given.

 

[Phaaze88]

corsair tm30

How's the consistency of this paste? The direct die application of gpu dies(smooth surface) is different from cpu IHSes(rougher surface), so Aeacus' link doesn't really apply here. The type of surface affects which pastes would work.
Direct die prefers stickier pastes. Runny(er) pastes are more likely to leave areas of the die uncovered.

I'd suggest double checking what the die looks like underneath that cooler.

 

[Aeacus]

How's the consistency of this paste? The direct die application of gpu dies(smooth surface) is different from cpu IHSes(rougher surface), so Aeacus' link doesn't really apply here. The type of surface affects which pastes would work.

Direct die prefers stickier pastes. Runny(er) pastes are more likely to leave areas of the die uncovered.

Corsair decided to use a simple, easy-to-open cardboard packaging for the TM30 thermal compound. On the front, you can see the actual syringe through a plastic window in the middle. Just like most thermal compounds on the market, the TM30 is stored inside a syringe, which also acts as the applicator. Only a 3g version is available for purchase, which should be enough for around 10 to 15 applications on mainstream processors using the Intel LGA 115x or AMD AM4 sockets. There are markings on the tube to show how much thermal compound is left inside.

The TM30 has a dark-grey colour, similar to most thermal compounds we tested. It is more liquid than the others, which may contribute to the claimed longer lifespan with no drying or cracking.

Thermal conductivity: 3.8 W/mK
Density: 2.50 g/cm3

Source: https://web.archive.org/web/2021021...ages/guru3d_thermal_paste_roundup_2019,4.html

Also, haven't seen thermal paste roundup regarding GPUs. Hence why i used the one which tests CPUs.

 

[Eximo]

I have used MX-4 on a few GPUs, no complaints. I find it to be quite universal. A really chunky paste would be pretty bad for direct die.

I suspect more that the GPU had an underlying problem to begin with, or there is simply a pressure issue due to thermal pads or some other reason. Check the cooler for flatness, I have seen a few bent GPU coolers causing the board to bend and making poor contact with the GPU die.

 

[Phaaze88]

Thermal conductivity: 3.8 W/mK
Density: 2.50 g/cm3

Source: https://web.archive.org/web/2021021...ages/guru3d_thermal_paste_roundup_2019,4.html

Also, haven't seen thermal paste roundup regarding GPUs. Hence why i used the one which tests CPUs.

You can't rely on those numbers. Darn companies don't appear to have a universal test method.

View: https://www.youtube.com/watch?v=_kzGTOyCYrY


That TM30 does look to be on the runny side in your new link. It probably left some spots bare when OP mounted the cooler.

Paste roundup on gpus is due to how few actually repaste their cards. It's a niche, just like how folks have so much trouble finding info on gpu pad replacement. It may sound like it's a common thing here, but in the wide world, it isn't.

 

[Aeacus]

Paste roundup on gpus is due to how few actually repaste their cards. It's a niche, just like how folks have so much trouble finding info on gpu pad replacement. It may sound like it's a common thing here, but in the wide world, it isn't.

Application issues aside, the thermal paste roundup i linked before still has some value in it, since it shows the cooling capacity to be different between different pastes.

And i know that GPU repaste isn't that common as it is applying it to CPU. Though, based on topics what i've seen here in the forums, about half of all topics is actually niche.

 

[Rxzy1x]

Hii, an update guys, i opened up my gpu today and turns out the thermal pads at the back off the pcb were not applied properly so i did reapply and it kinda fixed my problem like idk the temps are still kinda high for default settings, but this card is one of the worst 2080ti so i think the problem is fixed. i also tried to apply equal pressure while mounting the cooler and screwed the cooler real tight and it seems to have fixed the problem. Maybe i can get better temps by replacing all the pads but i think i am just gonna leave it to a professional to that for me as i might break something again.

 

[Phaaze88]

since it shows the cooling capacity to be different between different pastes.

Cooling capacity??? That's for the actual cooler to deal with. The purpose of paste is a bridge, to aid in heat energy transfer from one surface to another. Said bridge just needs to have a proper connection, i.e., good coverage in those pits its filling. That's it, that's its job. What it does for operating thermals is minute and perhaps a little icing on the cupcake compared to one's cooler selection, which has far greater impact.


Let's just agree to disagree, please?

Hii, an update guys, i opened up my gpu today and turns out the thermal pads at the back off the pcb were not applied properly so i did reapply and it kinda fixed my problem like idk the temps are still kinda high for default settings, but this card is one of the worst 2080ti so i think the problem is fixed. i also tried to apply equal pressure while mounting the cooler and screwed the cooler real tight and it seems to have fixed the problem. Maybe i can get better temps by replacing all the pads but i think i am just gonna leave it to a professional to that for me as i might break something again.

Yeah, thermal pads should be the absolute last resort, 'cause it's so easy to go down a rabbit hole on that one.
-Some find out the thickness of the original pads, and pick based on that, but it isn't the only factor.
-Pads have a certain degree of hardness; soft, medium, firm, etc.
-There isn't a universal test condition for pads, so the advertised numbers need to be taken with some salt. For example, the temperature at which one set of pads starts to flex or mold is different from another brand's.

By the way, what is the gpu hot spot looking like? [Use Gpu-Z or hwinfo to see it.]
I'd like to ask about memory junction too, but it couldn't be seen on many 20 series cards via software - EVGA's FTWs were some of the only ones that did it.

 

[Rxzy1x]

Cooling capacity??? That's for the actual cooler to deal with. The purpose of paste is a bridge, to aid in heat energy transfer from one surface to another. Said bridge just needs to have a proper connection, i.e., good coverage in those pits its filling. That's it, that's its job. What it does for operating thermals is minute and perhaps a little icing on the cupcake compared to one's cooler selection, which has far greater impact.


Let's just agree to disagree, please?


Yeah, thermal pads should be the absolute last resort, 'cause it's so easy to go down a rabbit hole on that one.
-Some find out the thickness of the original pads, and pick based on that, but it isn't the only factor.
-Pads have a certain degree of hardness; soft, medium, firm, etc.
-There isn't a universal test condition for pads, so the advertised numbers need to be taken with some salt. For example, the temperature at which one set of pads starts to flex or mold is different from another brand's.

By the way, what is the gpu hot spot looking like? [Use Gpu-Z or hwinfo to see it.]
I'd like to ask about memory junction too, but it couldn't be seen on many 20 series cards via software - EVGA's FTWs were some of the only ones that did it.

hotspot was fine before , i mean considering i was getting 89c temps , the hotspot was hovering around 103 . But i haven't checked the hotspot now since i just used msi afterburner to stress test the card and with 100% fan and some oc , the card was running around 78c max. I'll look into that, thanks for reminding me.

 

[Rxzy1x]

Cooling capacity??? That's for the actual cooler to deal with. The purpose of paste is a bridge, to aid in heat energy transfer from one surface to another. Said bridge just needs to have a proper connection, i.e., good coverage in those pits its filling. That's it, that's its job. What it does for operating thermals is minute and perhaps a little icing on the cupcake compared to one's cooler selection, which has far greater impact.


Let's just agree to disagree, please?


Yeah, thermal pads should be the absolute last resort, 'cause it's so easy to go down a rabbit hole on that one.
-Some find out the thickness of the original pads, and pick based on that, but it isn't the only factor.
-Pads have a certain degree of hardness; soft, medium, firm, etc.
-There isn't a universal test condition for pads, so the advertised numbers need to be taken with some salt. For example, the temperature at which one set of pads starts to flex or mold is different from another brand's.

By the way, what is the gpu hot spot looking like? [Use Gpu-Z or hwinfo to see it.]
I'd like to ask about memory junction too, but it couldn't be seen on many 20 series cards via software - EVGA's FTWs were some of the only ones that did it.

hii i did some testing, the avg temp under 100 percent usage is 81 max and hotspots are hovering around 107 - 106 c under heavy load
should i be worried about that?

 

[Aeacus]

Cooling capacity??? That's for the actual cooler to deal with. The purpose of paste is a bridge, to aid in heat energy transfer from one surface to another. Said bridge just needs to have a proper connection, i.e., good coverage in those pits its filling. That's it, that's its job. What it does for operating thermals is minute and perhaps a little icing on the cupcake compared to one's cooler selection, which has far greater impact.

Well, how else can you explain the temperature difference found out by Guru3D, where only factor that changed, was different thermal paste?

Guru3D isn't the only one who has tested thermal pastes. TH did it too with far more different thermal pastes,
review: https://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616-18.html

Or are you dismissing Guru3D and Tom's Hardware findings, whereby it matters nothing at all, which thermal paste are you using? And the numbers in the findings are a lie.

 

[Phaaze88]

hii i did some testing, the avg temp under 100 percent usage is 81 max and hotspots are hovering around 107 - 106 c under heavy load
should i be worried about that?

Mmmk, so a typical range for the gap between the gpu core average and the hot spot is around 20C or lower.
What you've shared is a bit up there at 25C - BUT, whether it has always been around that or not is unknown... it's entirely possible this gap is the norm for the cooler, and IF - big IF - that were the case, then the cooler is barely adequate for the power the gpu is drawing under load. To then get around this, I would suggest setting a lower max power limit on the card. Instead of the 100 or 112%(max), opt for something like 90 or 85%.