Question GPU thermal throttles at a very low temp?

Aug 23, 2019
8
0
10
0
Just recently upgraded to a 2080ti from a 1070 and haven't been noticing much better performance. Running benchmarks and stress tests shows the GPU throttling at 60 Degrees consistently trying to overclock using software such as MSI afterburner and Asus GPU Tweak. In all games and all benchmarking programs like Kombuster and FurMark show the temps instantly spike up and hold very consistently to 60 Degrees, while the clocks throttle and FPS drops. Cant seem to find any reason why the GPU would thermal throttle at such a low temperature, unless there's some other bottleneck factor I haven't considered. My only thoughts are my power supply might not be enough for the GPU, or since I have a custom cooling loop on the GPU, it does not detect any fans (As MSI afterburner and GPU Tweak show 0% fan speed. The actual cooling fans are powered by a hub and the High Amp motherboard 4-pin) and throttles to protect itself. Any advice would be helpful.

My parts list:
https://pcpartpicker.com/list/LK7h6s
 
Aug 23, 2019
8
0
10
0
Which is the main monitor?
Why are you using 2 different gpu softwares? Only use one, uninstall the other.

What's the cooler for the cpu?

Motherboard drivers up to date?
I tried multiple software's to see if I could fix the throttle issue. I'm just using afterburner now. The cooler is my custom liquid cooling loop with a 360mm rad. The main monitor is the viotek

And all drivers are up to date
 
Aug 23, 2019
8
0
10
0
Windows power plan set to high performance?
Nvidia Control Panel set to max performance, and highest refresh available?

Go here^and post a link to your results.
Yes those settings are already set

Heres the link:

https://www.userbenchmark.com/UserRun/19550160

UserBenchmarks: Game 78%, Desk 135%, Work 102%
CPU: Intel Core i7-8700K - 87.5%
GPU: Nvidia RTX 2080 Ti - 70.2%
SSD: Samsung 970 Evo Plus NVMe PCIe M.2 500GB - 283.5%
SSD: Intel Raid 0 Volume 1TB - 148%
HDD: Intel Raid 0 Volume 4TB - 196.6%
RAM: Corsair CMW32GX4M4C3000C15 4x8GB - 65.4%
MBD: Asus ROG MAXIMUS X HERO (WI-FI AC)
 

Phaaze88

Admirable
Herald
You haven't enabled XMP in the bios? Ram doesn't run at the rated speed(3000mhz) out of the box.

I see you're on the latest Nvidia driver, but are your motherboard's drivers are up to date as well? https://www.asus.com/us/Motherboards/ROG-MAXIMUS-X-HERO-WI-FI-AC/HelpDesk_Download/

Double check all your power supply connections. Make sure the 2080Ti is fully seated in the motherboard and the 2x 8-pin cables are fully seated as well.

Also run System File Checker to make sure windows isn't missing any files. Link: Using System File Checker
 
Aug 23, 2019
8
0
10
0
You haven't enabled XMP in the bios? Ram doesn't run at the rated speed(3000mhz) out of the box.

I see you're on the latest Nvidia driver, but are your motherboard's drivers are up to date as well? https://www.asus.com/us/Motherboards/ROG-MAXIMUS-X-HERO-WI-FI-AC/HelpDesk_Download/

Double check all your power supply connections. Make sure the 2080Ti is fully seated in the motherboard and the 2x 8-pin cables are fully seated as well.

Also run System File Checker to make sure windows isn't missing any files. Link: Using System File Checker
Ive reinstalled all of my motherboard drivers. Everything installed expect the VGA driver which said "Your computer does not meet the minimum requirements for this software" and refused to run the installer. After moving some things around in the BIOS I ran User Benchmark again:

https://www.userbenchmark.com/UserRun/19565732

Everything is connected properly in the computer.

System file check said it had replaced some corrupt files. The GPU problem still persists
 

Phaaze88

Admirable
Herald
Ive reinstalled all of my motherboard drivers. Everything installed expect the VGA driver which said "Your computer does not meet the minimum requirements for this software" and refused to run the installer. After moving some things around in the BIOS I ran User Benchmark again:

https://www.userbenchmark.com/UserRun/19565732

Everything is connected properly in the computer.

System file check said it had replaced some corrupt files. The GPU problem still persists
Do you still have the old 1070 to test and see how it performs?

Have you tried the 2080Ti in the 2nd and 3rd pcie slots?


Something else that comes to mind about cooling: while Afterburner shows 60C, that's core temp. It doesn't count VRM throttling.
Download and heaven benchmark, open Afterburner(with at least core clock temperature, and power graphs visible).
Run Heaven and let it finish, then take a screenshot of those graphs and post it.
If there's no power throttling going on, then the core clock should be a straight line across.

While I don't believe the VRMs should be that much warmer than the gpu core, crazy stuff does happen.
What waterblock are you using on the gpu, by the way?
 
Aug 23, 2019
8
0
10
0
Do you still have the old 1070 to test and see how it performs?

Have you tried the 2080Ti in the 2nd and 3rd pcie slots?


Something else that comes to mind about cooling: while Afterburner shows 60C, that's core temp. It doesn't count VRM throttling.
Download and heaven benchmark, open Afterburner(with at least core clock temperature, and power graphs visible).
Run Heaven and let it finish, then take a screenshot of those graphs and post it.
If there's no power throttling going on, then the core clock should be a straight line across.

While I don't believe the VRMs should be that much warmer than the gpu core, crazy stuff does happen.
What waterblock are you using on the gpu, by the way?
I do have the old 1070, but it would be a bitch to reinstall in my loop.

I ran heaven benchmark and have the graphs attached. My next guess is that the VRM's are thermal throttling. Could be poor thermal pad installation. My next move might be to reinstall the block. The block is a EK vector RTX 2080ti RGB block.

Yes thats the graph from GPU tweak. I only used those because it was easier to expand, I dont use that software for overclocking

View: https://imgur.com/a/9eF3dDe
 
Aug 23, 2019
8
0
10
0
Quite a number of drops.
If the WB is sitting level on the PCB, then the pads over the VRMs are making poor contact, and you will likely need some slightly bigger ones.
So i replaced the thermal pads on the waterblock. You were correct in that the VRM's were not making sufficient contact (In fact, no contact at all). After installing thicker pads the GPU will now reach thermal limit, but continues to throttle and i see exactly the same performance and issues as before, just at 88 degrees this time instead of 60. It was my mistake to assume the previous owner had installed the block correctly.

Even with sufficient thermal pad contact (And thermal paste for the core), I can't see any reason why the card would thermal throttle so quickly in a liquid cooling loop. In fact, it idles at 40-42 degrees where my previous 1070 never reached above 38-40 under full load in the same loop (With idle temperatures at or just above ambient, consistently).

Ill most likely be returning the GPU and purchasing a new one at this point. The card may have been faulty from the start.
 

Phaaze88

Admirable
Herald
So i replaced the thermal pads on the waterblock. You were correct in that the VRM's were not making sufficient contact (In fact, no contact at all). After installing thicker pads the GPU will now reach thermal limit, but continues to throttle and i see exactly the same performance and issues as before, just at 88 degrees this time instead of 60. It was my mistake to assume the previous owner had installed the block correctly.

Even with sufficient thermal pad contact (And thermal paste for the core), I can't see any reason why the card would thermal throttle so quickly in a liquid cooling loop. In fact, it idles at 40-42 degrees where my previous 1070 never reached above 38-40 under full load in the same loop (With idle temperatures at or just above ambient, consistently).

Ill most likely be returning the GPU and purchasing a new one at this point. The card may have been faulty from the start.
Now it looks like you alleviated the VRM issue, but now the gpu core isn't making solid contact because the new pads are too thick.
https://www.techpowerup.com/vgabios/204087/asus-rtx2080ti-11264-180910
88C is the thermal limit, too.

I'm left wondering if the previous owner made any other modifications to this card. If this is the PCB of an Asus 2080Ti Turbo, then it should've been compatible with this WB: https://www.ekwb.com/shop/ek-vector-rtx-2080-ti-rgb-nickel-plexi
 
Aug 23, 2019
8
0
10
0
Now it looks like you alleviated the VRM issue, but now the gpu core isn't making solid contact because the new pads are too thick.
https://www.techpowerup.com/vgabios/204087/asus-rtx2080ti-11264-180910
88C is the thermal limit, too.

I'm left wondering if the previous owner made any other modifications to this card. If this is the PCB of an Asus 2080Ti Turbo, then it should've been compatible with this WB: https://www.ekwb.com/shop/ek-vector-rtx-2080-ti-rgb-nickel-plexi
ill be trying one more thermal graphite pad to see if I can have everything making sufficient contact with the block and get those temps down. Other than that, is there anything else we haven't considered that might be worth a look?
 

99StefanRO

Commendable
Oct 25, 2016
98
15
1,545
1
Have you tried looking for hidden bitcoin miners in your computer?

This is what I encountered: I have a 1060. Everytime I went into a game, it performed well for some minutes then started dropping and lagging like never before. I tried everything and one day I went and downloaded some Advanced Task Manager and Security Task Manager to see what's going on. I waited until the GPU did it and seen that the CPU also went up by 15-20% usage. It turned out it was a miner called updatedg.exe somewhere in Local Disk C. I had to find it and delete it manually, now I am doing fine and my GPU is going great again.
 

Phaaze88

Admirable
Herald
ill be trying one more thermal graphite pad to see if I can have everything making sufficient contact with the block and get those temps down. Other than that, is there anything else we haven't considered that might be worth a look?

I asked for help on the matter, and one answer I got back was that it's possible that the VRMs may have already been damaged by heat, IF nothing else was altered on the card.
Returning it isn't such a bad idea at this point.
 

ASK THE COMMUNITY

TRENDING THREADS