Question GPU Crashes Unless Power Limit Is Put On It

Jamps

Reputable
Aug 27, 2016
16
1
4,515
0
I've been dealing with this issue for a while and assumed it was a PSU issue, after replacing my PSU with a new EVGA Supernova 1000 G2 and having the same issue I've decided to ask for help. The crash seems to occur whenever putting my GPU under load, through benchmarks or GPU intensive games (Any video games that are CPU intensive have no issue). My computer black screens (recently one of my three monitors is pink screening if that's a thing) and the GPU fans spin to maximum. I've already flashed my GPU BIOS and have previously flashed my motherboard BIOS around the time this issue began to occur. I've also tried re-installing drivers and have yet to check any physical thermal issues with the GPU as I have never taken one apart before.

My specs are as follows: Intel i7 9700k @ 4.7 Ghz, ASUS STRIX GTX 1080 8GB, EVGA Supernova 1000 G2 and Gigabyte Z390 I AORUS PRO

I have two GPU-Z log files one from the crash and one from running a benchmark quickly with 55% power but, I'm not sure of the best way to share them on this forum.

I know this isn't a lot of information to help so, if you need anything else please ask.
 

Jamps

Reputable
Aug 27, 2016
16
1
4,515
0
I think that if you replaced the PSU...and we can be confident it's not that.
I think it may be the GPU itself and I would try it in another system to confirm.
I tend to not think it's software related.
I won't be able to test the GPU in another PC until tomorrow but, I will try that when I can.
 

Jamps

Reputable
Aug 27, 2016
16
1
4,515
0
Sorry...that slipped my mind.
You could also try powering down and reseating the GPU itself....although I think this is less likely to be a solution it is something to try.
Sorry I should specify that since it's a mini-ITX build replacing the PSU means taking everything except the Motherboard CPU and CPU cooler out of the PC.
 
Jul 6, 2021
20
0
10
0
Have you monitored the temperature while you apply load? If its climbing way too fast and then blackscreens, taking everything else you've said into account, its a hardware issue. Since its a 1080, its probably like 4-5 years old?

Also, you said you've never taken a GPU apart. So, has anybody put new thermal paste and thermal pads on that GPU ever? It may need simple maintenance.

Watch a tutorial on how to change them, Its pretty simple. Also, clean everything including the heatsink and fans just to make sure.
 
Dec 22, 2020
13
0
10
0
I have a somewhat similar problem with my 1070 (laptop version). Mine crashes (BSOD and system reboots) if the voltage isn't locked at high number and the GPU temp is too low. Happens as soon as I run any game or application that uses the Nvidia card (even just launching Steam crashes it sometimes). But if I set a constant voltage (1093mV in my case, but can be set lower) and let the temp rise to at least 60C (takes a while when idle), it works just fine, as if nothing was wrong. It's really odd.

Sorry if my reply isn't providing useful advice, just wanted to share my similar experience.
 

vov4ik_il

Respectable
I have a somewhat similar problem with my 1070 (laptop version). Mine crashes (BSOD and system reboots) if the voltage isn't locked at high number and the GPU temp is too low.
Totally different issue. Probably the clock is set too high for the card or one of the power phases is bad.


Sorry I should specify that since it's a mini-ITX build replacing the PSU means taking everything except the Motherboard CPU and CPU cooler out of the PC.
ITX builds would usually have to be power limited and fine-tuned to keep thermals within envelope. Try running OCCT to find the right limits.
9700k is a fat chip too, see if you can undervolt it or power limit it to reduce heating.
 
Jun 27, 2021
55
6
35
0
Just sharing my two cents.

Recently had an AIO fail, and as soon as the coolant got into the GPU, it black screened and fans went to 100%. I know you haven't had a leak, but if you are confident with it, you could try taking the GPU apart and giving it a deep clean, and re-applying the paste on the chip. Just my two cents!
 

dorsai

Distinguished
How old is the 1080 ? You might need to remove the heatsink, clean everything, and apply some new thermal paste. I'd replace the thermal pads on the memory at the same time with some good quality replacements. The tim on GPUs tends to dry out, harden, and crack after a few years of use and no longer functions they way it should to transfer heat away from the GPU core or ram.
 

Jamps

Reputable
Aug 27, 2016
16
1
4,515
0
Taking what you guys have said into consideration I'm going to try replacing the thermal pads and putting some new thermal paste on the card. This seems like the most logical issue and is something that I considered myself but wanted more input to confirm. Will update with results!
 

Jamps

Reputable
Aug 27, 2016
16
1
4,515
0
Have you monitored the temperature while you apply load? If its climbing way too fast and then blackscreens, taking everything else you've said into account, its a hardware issue. Since its a 1080, its probably like 4-5 years old?

Also, you said you've never taken a GPU apart. So, has anybody put new thermal paste and thermal pads on that GPU ever? It may need simple maintenance.

Watch a tutorial on how to change them, Its pretty simple. Also, clean everything including the heatsink and fans just to make sure.
The temps don't seem to climb very fast however, the second it hits it's boost clock it crashes so I'm thinking something similar as you.
 
Jul 6, 2021
20
0
10
0
The temps don't seem to climb very fast however, the second it hits it's boost clock it crashes so I'm thinking something similar as you.
Maybe the overall temp is not rising too fast but since you say it crashes when it reaches boost block it may have a very high temp hotspot that makes it crash once it reaches that point. Report back once you've given it maintenance.
 

Schlachtwolf

Respectable
This is very possibly a VRM heat issue and changing the thermal pads may well help, the memory moduls in such a small build can degrade faster than in a more airflow orientated build due to the heat build-up that a smaller case will bring.

It is however VERY important to get the right thickness of pad as too thin and not enough contact is made to the cooler/backplate, and if to thick it can bend the PCB and cause an uneven pressure on the GPU chip itself when putting it back together.
 

Jamps

Reputable
Aug 27, 2016
16
1
4,515
0
I've been dealing with this issue for a while and assumed it was a PSU issue, after replacing my PSU with a new EVGA Supernova 1000 G2 and having the same issue I've decided to ask for help. The crash seems to occur whenever putting my GPU under load, through benchmarks or GPU intensive games (Any video games that are CPU intensive have no issue). My computer black screens (recently one of my three monitors is pink screening if that's a thing) and the GPU fans spin to maximum. I've already flashed my GPU BIOS and have previously flashed my motherboard BIOS around the time this issue began to occur. I've also tried re-installing drivers and have yet to check any physical thermal issues with the GPU as I have never taken one apart before.

My specs are as follows: Intel i7 9700k @ 4.7 Ghz, ASUS STRIX GTX 1080 8GB, EVGA Supernova 1000 G2 and Gigabyte Z390 I AORUS PRO

I have two GPU-Z log files one from the crash and one from running a benchmark quickly with 55% power but, I'm not sure of the best way to share them on this forum.

I know this isn't a lot of information to help so, if you need anything else please ask.
Just wanted to update everyone it was a thermal compound issue. I wish I took pictures it was as though 90% of the thermal paste had evaporated. Thank you to everyone who helped this thread, if you have a similar issue I hope this helps.
 
Reactions: Schlachtwolf

ASK THE COMMUNITY

TRENDING THREADS