Question Why does my 2080Ti crash in every game if I set the POWER TARGET above 50%?

[GDR]

I bought a few years ago a system with a 550W PSU and a GTX 1070 GPU that I was using with a 27 inch monitor. I sometimes had system crashes with a blue screen.​

Anyway, after a couple of years I sold the GPU and PSU and bought a Corsair HX850i PSU and upgraded my GPU to an ASUS dual OC 2080Ti and had no more PC blue screen crashes, but sometime, very rarely, games crashed when there were very intense scenes.
When I upgraded the hardware I did not reinstall windows.

After a year I upgraded my monitor to the 144Hz LG 34GK950F and since then the crashes have increased so much that every game crashes after a few seconds of gameplay.​

The system doesn't crash, only the game crashes to desktop.

​

First I thought it might be the high temperatures, but most games run under 70C. And the dual fans rarely run above 30% of capacity.
So I limited the temp to 70C and there have been no more crashes. If I go 71C limit all games crash after a few minutes.
After some time I played more with the settings and found out that the issue is the POWER TARGET. If I set it to 50% there are no more crashes. I can leave the temp to the 84C default limit.
What does this power target mean and how can I fix the problem?

For example in Divinity Original Sin 2 if I set the maximum FPS to 60 I get crashes after an hour or so of gameplay without any power targer limit. But if I crank up the FPS limit to 144Hz it crashes immediatly unless I set the GPU power target to 50%.
This happens in absolutely every game. From non demanding games like Two Point Hospital to Control and Gears 5.

Thank you very much for the support and I look forward to your reply!
My CPU is an i7-6700K

 

[rgd1101]

make and model of this 550w psu?

 

[GDR]

make and model of this 550w psu?

It doesn't matter, I sold it. I think it was Seasonic or something like that.
The current one is Corsair HX850i. I connected the GPU via 2x 8pin pci-e cables. I don't know if I should have used one with the extension/appendix or however it's called or if I did it right by using 2 separate cables.
Could this be the problem?

 

[vov4ik_il]

I connected the GPU via 2x 8pin pci-e cables. I don't know if I should have used one with the extension/appendix or however it's called or if I did it right by using 2 separate cables.
Could this be the problem?

2 Separate cables is better. Less voltage drop on the cabling itself. But it should not be critical in your case. The problem is elsewhere.

 

[vov4ik_il]

Is it crashing with BSOD? Try running OCCT on the GPU only (with 100% power target). And then on the CPU only. And then on everything.

 

[GDR]

I thought about different possible problems, from normal ones to the most outlandish.

• should I reinstall windows?
• could it be a faulty display port, though I've tested 3. now using a 1.4 DP
• is it because an incompatibility between NVIDIA GPU and Freesync monitor
• is the GPU faulty (I really hope not)
• is it because everything is connected to a single electrical outlet/socket via a strip extension cord
• is it the motherboards fault, it's a asrock z170 pro4s
• I have not changed anything in BIOS

 

[GDR]

Is it crashing with BSOD? Try running OCCT on the GPU only (with 100% power target). And then on the CPU only. And then on everything.

The dual ASUS GTX 1070 was giving me weekly blue screens
The dual ASUS RTX 2080Ti has never given me a blue screen.
I think the blue screens were because of the PSU

It's just that all games simply exit to desktop. Some give me errors, others don't.
For example if I set the power target to 50% in the ASUS program (I forgot the name) or in MSI Afterburner, a game can run for many hours with no issues.
But if I alt+tab and set the power target to anything higher the game suddenly closes.

 

[vov4ik_il]

But if I alt+tab and set the power target to anything higher the game suddenly closes.

You should not alt+tab to adjust settings while the game is running. There are many applications that will crash if you do it in this manner. The applications should be closed first and then things adjusted.

• should I reinstall windows? - Probably not
• could it be a faulty display port, though I've tested 3. now using a 1.4 DP - SO is it only faulty above 50%?
• is it because an incompatibility between NVIDIA GPU and Freesync monitor - SO is it only a problem above 50%?
• is the GPU faulty (I really hope not) - Maybe, hard to say at this point
• is it because everything is connected to a single electrical outlet/socket via a strip extension cord - It would trip a breaker if there is a problem
• is it the motherboards fault, it's a asrock z170 pro4s - Maybe, hard to say at this point, unlikely but possible.
• I have not changed anything in BIOS - No need to change anything there if it was working fine earlier.

 

[GDR]

Is it crashing with BSOD? Try running OCCT on the GPU only (with 100% power target). And then on the CPU only. And then on everything.

Ok so I researched a bit OCCT. I downloaded it and started testing.
The OCCT test at 100% power target got 0 errors after 15 minutes.
The 3D test at 100% power target delivered less than 1 FPS and had over 100 errors.
The 3D test at 50% power target had 0 errors. - around 40 FPS
The 3D test at 70% power target went fine for about a minute, then after 80 seconds it had 133 errors. - around 60 FPS

The power test at 70% power target gets 100 FPS, I ran it for 30 seconds and every 10 seconds it droped to 0 FPS and generated 10 errors.
All tests were done at 3440x1440 resolution.

Do you need pictures?
I would also get a notification in my bottom right corner that said:
Application has been blocked from accessing Graphics hardware.

Do you have any clues what the problem might be?

 

[vov4ik_il]

Ok so I researched a bit OCCT. I downloaded it and started testing.
The OCCT test at 100% power target got 0 errors after 15 minutes.
The 3D test at 100% power target delivered less than 1 FPS and had over 100 errors.
The 3D test at 50% power target had 0 errors. - around 40 FPS
The 3D test at 70% power target went fine for about a minute, then after 80 seconds it had 133 errors. - around 60 FPS

The power test at 70% power target gets 100 FPS, I ran it for 30 seconds and every 10 seconds it droped to 0 FPS and generated 10 errors.
All tests were done at 3440x1440 resolution.

Do you need pictures?
I would also get a notification in my bottom right corner that said:
Application has been blocked from accessing Graphics hardware.

Do you have any clues what the problem might be?

The issue seems to be anywhere between PSU and GPU.
What do you see for +12V when you are doing those tests?

 

[GDR]

The issue seems to be anywhere between PSU and GPU.
What do you see for +12V when you are doing those tests?

OK, so I did the test again. I had to limit the power target at 70% otherwise the OCCT would instantly crash.
Even so, the test ran for 30 seconds before crashing. Just in time to get a screenshot.

This is idle:

This is with the test running:

 

[vov4ik_il]

Som Ting Wong...
Can you try switching the PSU to Single Rail and repeat the test? How are the CPU temps?

 

[GDR]

Som Ting Wong...
Can you try switching the PSU to Single Rail and repeat the test? How are the CPU temps?

Ok, so the test ran twice as long on single rail.
The CPU temps were always maximum, and the test stopped when I got the following error notification

This is the screen of the temps during the test

 

[GDR]

Yesterday I ran the first test on OCCT for 15 minutes.
The CPU usage was always at 100%.
The GPU usage never went above 25%
There were no errors detected.

 

[vov4ik_il]

I will have a better look on the screenshots later, temps are unreadable from my phone. Meanwhile, can you run Prime95 (AVX2) to torch the CPU (to completely make sure it is stable) like I did here? P.S note power use when you do it so we can see how much it takes w/o GPU being loaded.

 

[GDR]

I will have a better look on the screenshots later, temps are unreadable from my phone. Meanwhile, can you run Prime95 (AVX2) to torch the CPU (to completely make sure it is stable) like I did here? P.S note power use when you do it so we can see how much it takes w/o GPU being loaded.

Ok, just passed the 28k...
The cooler is running at max speed... temps sometimes reach 95C
How long is this test? It feels like it's gonna catch on fire...🧯🔥🔥🔥

 

[GDR]

P.S note power use when you do it so we can see how much it takes w/o GPU being loaded.

I don't know what you meant by this. I selected the small FFT's test, like you did in the image.
The test has been running for about 30 minutes. I just passed the 30k

 

[GDR]

I just passed the 50k
The test has been running for over an hour. I have to work tomorrow so I kinda have to go to bed now.
I took several screenshots.
This is the latest one>

I just passed 60k. I will stop the test now.

 

[vov4ik_il]

I just passed the 50k
The test has been running for over an hour. I have to work tomorrow so I kinda have to go to bed now.
I took several screenshots.
This is the latest one>

I just passed 60k. I will stop the test now.

It is good enough. Stability run is usually 24h but it would reveal prominent issues much faster than that.
The PSU power chart does not seem to exceed 200w for the system with CPU alone and there seem to be no problem with CPU stability or overclocking.
I am still at work and will be home in 3 hours or so. Will drop you a note then.
P.S 95C is a little high for the CPU, you would want to get the cooling to do a better job

 

[vov4ik_il]

I did not find anything abnormal in your screenshots (except the CPU cooling is on the weak side, but it does not seem to be the issue).
Is your GPU overclocked? If so, remove it and limit the frequency to the stock (remove any custom curves if you have them). Does your GPU have a warranty?
Final check:
Remove Asus Tweak utility and any other GPU related tweakers you might have.
DDU and reinstall the nVidia drivers (no utilities should be installed, just the drivers).
Use Single Rail on the PSU (removes ~40A overcurrent protection, you have a beefy card).
Rerun the OCCT test with a vanilla latest driver install.

 

[GDR]

I did not find anything abnormal in your screenshots (except the CPU cooling is on the weak side, but it does not seem to be the issue).
Is your GPU overclocked? If so, remove it and limit the frequency to the stock (remove any custom curves if you have them). Does your GPU have a warranty?
Final check:
Remove Asus Tweak utility and any other GPU related tweakers you might have.
DDU and reinstall the nVidia drivers (no utilities should be installed, just the drivers).
Use Single Rail on the PSU (removes ~40A overcurrent protection, you have a beefy card).
Rerun the OCCT test with a vanilla latest driver install.

Ok, so I rebooted in safe mode.
I uninstalled the GPU drivers with DDU.
I restarted in normal mode.
I uninstalled Asus Tweak utility.
I did a custom clean install of the latest NVIDIA drivers.
I rebooted again.
I selected single rail from corsair link.
The power test and 3D test from OCCT run at less tha 1 FPS at 1920*1080 resolution.
At 3440*1440 resolution the app just crashes and I cannot run the test again unless I restart the computer.
Strangely, OCCT says no errors detected.
I even uninstalled Zone Alarm firewall.
On single rail I had a spike at 450W.
Should I leave it on single rail or multi rail?

When I had an Acer 1440p 27 inch 60hz monitor I don't remember having this issues.
Could it be related to my 34 inch ultrawide 144 hz freesync monitor? This is when the issues started to appear. I selected FreeSync - Extended from the monitor settings. The other options are basic and Off.
Or is it just that this monitor requires a higher resolution, therefore the GPU draws more than 50% power and that's why it crashes?

I just disabled fast motion and freesync from the monitor settings. OCCT test runs better, I see over 140FPS but after a few second it gets back to bellow 1 FPS

I tweaked the settings more, now I'm getting more than 150 FPS for 2 seconds before it drops to less than 1.
I tried the 3D test with 30 FPS Limit. The test goes well. It does not crash. After a minute or so there was a drop at >1FPS, but it went back at 30. The graph showed around 40% GPU usage. The fans were mostly turned off because the temp was under 60C.
As always OCCT detected no errors.😕

I honestly have no clue what the problem might be.
In the meantime I'm going to reinstall ASUS Tweak to use it as I've used it before, at 50% power target.

 

[vov4ik_il]

I doubt it is the monitor. To check it I would try a smaller resolution with FreeSync off, also try with hdmi cable just to make sure it is not monitor/cable related. It should still yield a problem if it is not related to the monitor or cable.
DO not use Asus Tweak for now. I want to exclude it from possible issue list.
For tests, leave the PSU on single rail. Multi-rail protects against high current (and possible fire) if one of the lines gets shorted out. On the GPU there is usually a fuse that burns earlier (not user replaceable). Not your case anyway.

You did not answer my warranty question.

 

[GDR]

I doubt it is the monitor. To check it I would try a smaller resolution with FreeSync off, also try with hdmi cable just to make sure it is not monitor/cable related. It should still yield a problem if it is not related to the monitor or cable.
DO not use Asus Tweak for now. I want to exclude it from possible issue list.
For tests, leave the PSU on single rail. Multi-rail protects against high current (and possible fire) if one of the lines gets shorted out. On the GPU there is usually a fuse that burns earlier (not user replaceable). Not your case anyway.

You did not answer my warranty question.

I bought it from a retailer that was providing pc parts to one of the largest shops in the country.
The shop sells them at overprice and offers 2-3 years warranty.
The retailer only gave me a 6 month warranty.

 

[GDR]

Ok, so I've searched a little bit on google if my monitor might cause games to crash.
I've found people, some with almost exactly my specs, same CPU, same resolution ultrawide monitor, describing exactly the problem I am facing.
It's uncanny how accurate his description matches mine.
Too bad nobody offered a solution.
This other person too.

 

[vov4ik_il]

Ok, so I've searched a little bit on google if my monitor might cause games to crash.
I've found people, some with almost exactly my specs, same CPU, same resolution ultrawide monitor, describing exactly the problem I am facing.
It's uncanny how accurate his description matches mine.
Too bad nobody offered a solution.
This other person too.

If you are positive about monitor being the issue, try the stuff I posted earlier (turn freesync off, lower resolution, or change the interface from dp to hdmi).
The FPS will surely go up but we are testing if the card can sustain 100% load at 100% power target regardless of resolution right?
I am skeptical about the monitor being the issue, but it is possible.