Why is my GPU thermal throttling? - (SOLVED)

setrixhd · 26 Nov 2019 at 22:07

So ive been playing the new call of duty and i had noticed my GPU clock speed kept lowering itself as my FPS seemed awfully low than what it usually is.

The GPU in question is a watercooled Aorus RTX 2080 super. It hits 77 degrees Celsius when playing a game and does not budge no matter how fast my radiator fans are spinning.
I noticed in GPU-Z that the "Perfcap reason" said "Thrm" so im guessing this means my GPU is thermal throttling?

I also have the GPU at a target temp of 88 and a power target of 140% but this doesn't seem to have any effect as it doesn't even hit anywhere near the 88 degree target. (so why is it thermal throttling?)
The card stays at 77 while gaming. As soon as i start a game it jumps from the idle 30 degrees to the 77 degrees.

I had got the GPU from OverclockersUK around 4 months ago, they assembled the waterblock to the card themselves and it worked fine.

I ran a benchmark when i got it, the temp never went past 60 degrees and scored in the 10k mark leaderboard. when i run the same benchmark now it hits the 77 degrees and lowers the clockspeeds and scores in the 4k mark leaderboard.

So something is making the card run hot but what/why?. I cant think of what could be causing it as it HAD ran fine when i originally brought the card.

Im really lost as to what to do, is the card faulty? do i send it back? Why is it so hot? what could be causing the issue?

CPU idle temps = 25-30
CPU under load temps = 42 at around 55% load
GPU idle temps = 30
GPU under load temps = 77

EDIT!
Final update! YES REALLY! Thank you for all your help and advising me what to do. I drained the system and decided to take apart the gpu. I was a bit worried at first, thats one of the reasons i brought it already assembled by OverclockersUK as i didnt feel comfortable doing it myself. Anyway i took it apart and found the thermal paste didnt even cover the whole die. I am quite disappointed that a company like Overclockersuk did a poor job of applying thermal paste, however my GPU is in the upright orientation so perhaps this effects the thermal paste due to gravity? just a random thought.

I have since cleaned off the old thermal paste and reapplied my own. And now my GPU is working perfectly again!!! i now get 200FPS instead of the 80FPS i was getting.
i also noticed they didnt include the washers behind the screws for some reason. I did also put these on.

Once again thank you! Im glad this has been resolved

https://imgur.com/gallery/WJIy70R picture of thermal paste

gpuerrilla · 26 Nov 2019 at 22:14

Is the pump working?

3t3P · 26 Nov 2019 at 22:26

Wayyyyy too hot for a WCd card.

Wonder what paste application is looking like.

Could any microchannels be blocked? Are you running dye or pastel?

setrixhd · 26 Nov 2019 at 22:43

Th0nt said:
Is the pump working?

Pump is working as CPU temps are fine

setrixhd · 26 Nov 2019 at 22:53

3t3P said:
Wayyyyy too hot for a WCd card.

Wonder what paste application is looking like.

Could any microchannels be blocked? Are you running dye or pastel?

Im using the white EK-Cryofuel so im guessing its a dye? cant see the microchannels due to the colour but surely if they were blocked the temps would be high even when the card isn't underload?

3t3P · 26 Nov 2019 at 23:30

I had problems with blue cryofuel albeit not temperature ones.

Take the GPU out and remove the block to inspect it?

Dirk Diggler · 27 Nov 2019 at 08:57

What is the idle temperature? If it's a flow/fluid issue then I'd expect to see high idle temperature.

gpuerrilla · 27 Nov 2019 at 16:22

String said:
What is the idle temperature? If it's a flow/fluid issue then I'd expect to see high idle temperature.

setrixhd said:
GPU idle temps = 30
GPU under load temps = 77

Dirk Diggler · 27 Nov 2019 at 16:59

Could be paste it idle temperature is staying steady at 30c.

setrixhd · 27 Nov 2019 at 17:27

3t3P said:
I had problems with blue cryofuel albeit not temperature ones.

Take the GPU out and remove the block to inspect it?

I dont see how it could be a block issue though, the temps are fine at idle. Even when i put the card at "prefer maximum performance" in the nvidia control panel and the card runs at the full 1650MHz clock speed the temp goes up to 50 degrees.
Its only when i run a game and the GPU goes under heavy load which causes the high temps and low clock speeds.

I also dont know if i can take the block off and still be under warranty, being as OverclockersUK put the block on themselves.

setrixhd · 27 Nov 2019 at 17:30

String said:
Could be paste it idle temperature is staying steady at 30c.

That would mean it has dried out in 3 months? is that a likely scenario?

tyler_jrb · 27 Nov 2019 at 17:57

setrixhd said:
That would mean it has dried out in 3 months? is that a likely scenario?

Paste is possible but it’s unlikely, especially in that time frame unless the card wasn’t assembled correctly.

There may be enough flow to keep it cool at idle but not enough under load. I’ve seen that before with my own system.

If it goes from idle 30c and jumps up when you put it in something like furmark then it’s a flow/paste issue.

Should be hitting maybe 40-50c maximum after prolonged usage at max speed.

To determine which then you can just drain and refill the loop and visually see the flow. Or by how far the res drops down with the pump at low and full speed.

It’s also possible ocuk used the wrong screws to screw the block down, very easily done. That way it’s possibly separated and been held this long by the paste. Remove the backplate if it has one and see if the pcb moves about. Possibly remove a screw from around the core and measure against the manual on ek’s website.

setrixhd · 27 Nov 2019 at 19:05

tyler_jrb said:
Paste is possible but it’s unlikely, especially in that time frame unless the card wasn’t assembled correctly.

There may be enough flow to keep it cool at idle but not enough under load. I’ve seen that before with my own system.

If it goes from idle 30c and jumps up when you put it in something like furmark then it’s a flow/paste issue.

Should be hitting maybe 40-50c maximum after prolonged usage at max speed.

To determine which then you can just drain and refill the loop and visually see the flow. Or by how far the res drops down with the pump at low and full speed.

It’s also possible ocuk used the wrong screws to screw the block down, very easily done. That way it’s possibly separated and been held this long by the paste. Remove the backplate if it has one and see if the pcb moves about. Possibly remove a screw from around the core and measure against the manual on ek’s website.

If it was a flow rate issue wouldn't the CPU also be performing badly? also if it was a flow rate issue what would cause it to perform differently than it did 3 months ago? like why/how would the flow rate have changed?

Five.Stars · 27 Nov 2019 at 19:35

No chance it should be getting anywhere close to that temp.

Five.Stars · 27 Nov 2019 at 19:36

It's probably at least 30-40 degrees above what it should be operating at under load.

Drumroll · 27 Nov 2019 at 20:03

If using EK waterblock, make sure you're using the correct screws. The longest ones may give you the impression that they're tight enough, but they're not. Using the medium screws and sorted. Mine stays at nice 25C idle and 45C max 50C under load.

tyler_jrb · 27 Nov 2019 at 20:16

setrixhd said:
If it was a flow rate issue wouldn't the CPU also be performing badly? also if it was a flow rate issue what would cause it to perform differently than it did 3 months ago? like why/how would the flow rate have changed?

Ive had parts block up before. Radiators block up with crap from the coolant, or just crap from manufacture. To the point where there was a trickle coming into the reservoir even at full speed. Strip down cleaned out and back perfect again.

If the CPU is as it should be under max load then i'd say its likely a block/contact issue. As per my original post check the screws holding the pcb to the block. That they are the correct length as per ekwb's manual.

setrixhd · 27 Nov 2019 at 22:42

MrPotato said:
If using EK waterblock, make sure you're using the correct screws. The longest ones may give you the impression that they're tight enough, but they're not. Using the medium screws and sorted. Mine stays at nice 25C idle and 45C max 50C under load.

tyler_jrb said:
Ive had parts block up before. Radiators block up with crap from the coolant, or just crap from manufacture. To the point where there was a trickle coming into the reservoir even at full speed. Strip down cleaned out and back perfect again.

If the CPU is as it should be under max load then i'd say its likely a block/contact issue. As per my original post check the screws holding the pcb to the block. That they are the correct length as per ekwb's manual.

Im using a Phanteks block and looking at the manual the screws were all the same size, that is of course if OverclockersUK used those screws to begin with when they assembled the card. Ill have to drain the system soon and try to see if i can see anything wrong

Drumroll · 28 Nov 2019 at 19:10

Let's hope they've removed the protective film from the block.

tyler_jrb · 28 Nov 2019 at 19:24

setrixhd said:
Im using a Phanteks block and looking at the manual the screws were all the same size, that is of course if OverclockersUK used those screws to begin with when they assembled the card. Ill have to drain the system soon and try to see if i can see anything wrong

unlikely to be screws then if it is a phanteks block. EKWB just throws in a bag of assorted screws so its easy to mistake the wrong ones.

Any update? seen anything wrong with the card/loop?