Discussion Gpu troubleshooting help

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
i have a zotac 1080 ti mini that appears to be throttling at around 40c in heaven or any game, causing the image to stutter. it will ramp up to about 45-50c then slam back to 37-40c. ive changed the paste and reinstalled drivers a few times, got it to work for a bit, then swapped a fan and it has been doing the same thing ever since. i tried a g12 setup a few days ago reinstalled the heatsink and am using two noctua a9x14s via a two fan splitter. im worried i may have damaged something, but the card has worked at normal temps/clocks on a couple occasions and the board seems fine. Is there anything I am missing, maybe not enough paste (contact seems fine) that might be causing this?
 
Last edited:

PhilippNikko

Efficiency Noob
Sep 23, 2020
6
0
Looks like GPU was damaged
Or temp. sensor do not works correctly
Try any other GPU and test it
OR better try to test Your 1080 on another PC, if it`s possible
 

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
Has this problem not solved here?
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
Has this problem not solved here?
it wasnt, still exhibiting the same issues after the heatsink was reinstalled. last thread i will post about it
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
Install Afterburner. Set power limit and memory clock at minimum. Click Apply.
Does this help?
Maybe VRM or VRAM are overheating?
I'll give this a shot, I don't suspect that the VRAM is overheating, thermal pads make contact and the thermal pads seem to be fine
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
ok so i made some changes to the power saving settings to prefer maximum performance and disabled the pcie link power setting on the power plan in windows which seems to have eliminated part of the problem. I repasted the card again and made sure mounting pressure was good. So now i've been able to break beyond the high 40c mark and get to a stable loop in heaven. the odd thing is that now the card will down clock from about 1800mhz every 10 degrees or so, hit like 100 to 300 mhz, and then climb up to the next 10c mark and then do it again until it gets to the 70s, where it is finally stable. memory clock and fan speed is fine and i can't seem to find the cause of the weird 10c throttle then climb to the next 10c mark thing.

also did a clean install of windows to see if it was a software thing and it doesnt seem to be. i've also noticed that setting a custom fan curve in afterburner seems to have netted some stability gains as well. Im really baffled by this and really want to get the card fixed. Any ideas on what I might be missing?
 

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
What exactly are the modifications (hardware and software) made to the card?

I suggest, bring the card back to its most stock/default condition and then test again.
Eg the Zotac 1080Ti mini has base clock of 1506 MHz and boost clock of 1620 MHz. How on earth you are getting 1800 MHz? Is the card heavily overclocked?
And, 300 MHz sounds like the standard idle clock.

Has the card ever been flashed to another VBIOS?
If so, maybe consider flashing back to its VBIOS (check TPU's database) when all else fail.
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
at this point, the only mod is an nf a9x14 (92mm) and a scythe kaze jyu slim 2000 rpm (100mm) fan attached to the heatsink and run off the card's fan header via a mini 4 pin to 2x pwm splitter. I haven't touched the vbios and the card is running at stock settings since the clean install. Like i said, i just applied the fan curve to it in AB, nothing else has been changed. that's the curiosity, i can't imagine that the fan splitter would cause all of the issues im having.

with regard to the clock, that's the peak at the high 70s during the heaven run. might have gotten a good chip to allow that clock, but nothing im doing to it, as far as i know
 
Last edited:

Bubba

Average Stuffer
Nov 4, 2019
63
24
smallformfactor.net
Sounds like VRM thermal throttling to me.

This site here says their highest stable overclock is 1745 MHz

If you want your 1080 Ti to have a long life, then go easy on the VRM's.
Or get better VRM cooling.
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
one last thing, the clock goes immediately from mid 1800s to 150 ish mhz when it thottles. could it be the vrm wacking out due to poor thermal pads? the pads on the card make contact, but have definitely seen better days. I'm thinking the vrm might at least be a contributing factor
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
Sounds like VRM thermal throttling to me.

This site here says their highest stable overclock is 1745 MHz

If you want your 1080 Ti to have a long life, then go easy on the VRM's.
Or get better VRM cooling.

I'm assuming new thermal pads are in order. is there any use in doing copper heatsinks on the memory or is that a waste of time? I'm assuming that would be a waste
 

Bubba

Average Stuffer
Nov 4, 2019
63
24
smallformfactor.net
The Zotac 1080 Ti mini has 5+1 VRM.
Not so good for a 250W TDP card.


For comparison, my 1660 Ti (120W TDP) has 4+2 VRM

I doubt new/better thermal pads will help much.
Heatsinks require airflow. Did you set the fan to 100% speed in afterburner?
You could try blowing air across the card's backside in order to cool the PCB.
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
i just ordered a new set of thicker thermal grizzly pads, so we'll see if that was the issue. im my research for doing the fan swap, other shave been able to run the card with dual nf a9's without much issue. definitely seems like the VRMs or some other voltage control issue is going on, not really any other thing other than maybe one of my power cables is messed up. i've got some pslate cables coming on monday, so i think thats about it in terms of remaining variables. if not, might just have to retire the card
 

Diamorif

Cable-Tie Ninja
Original poster
Jun 9, 2020
211
105
It was the VRM throttling. Reapplied some new 3mm pads from kryonaut to the whole card, switched the cables to the stock sf600 ones and now I'm up and running. Turns out the 1800mhz wasnt a fluke, I'm running at 1860-1880mhz at 76c. thank you all for the help,.