Question GPU crashes within a few minutes of opening a game

[el3mentry]

UPDATE:

The answer is - Graphics Driver.​

I reinstalled a nearly 5 month old graphics driver for my video card and upon testing over 2 days, I believe I have fixed this issue.

However, it is not that simple. Only one of these things fixed my issue:
Removing the graphics driver completely using DDU in Safe Mode.
or
Installation of an older driver, which is more stable than the newest driver.

I did reinstall my graphics driver earlier to try to fix this issue, but the process of uninstallation was using DDU without Safe Mode. Also, I installed the newest driver which was offered. There is no way of knowing what made it right.

---

Over the past month, my video card has been crashing a few minutes into playing games.
I use other programs which are also resource intensive but nothing happens.

Everything works as usual, but as soon as I open a game, within a couple of minutes, the video card crashes.
First, the signal stops going to the monitor, then the sound stops coming from the game which indicates that the running applications has also crashed. Then it stays like that, following which I force restart.

My peripherals do work when the signal stops, as I am able to take screenshots when pressing the Win+prtsc, but when I view them after the restart, it's just black.

Here is the most confusing part, which makes me question if this is a PSU / Video Card issue:
When I reboot, and open the game again - it doesn’t crash, I play for hours but nothing happens.

Here, I have attached 2 images, 1st image is recorded when the crash happens & the 2nd image is post reboot, when the crash does not happen. They are recorded using MSI Afterburner.

In the following log, its visible when the crash happens. Everything including the power draw instantly drops to ZERO:


Here, in this log, I simply play the game for around 10-12 minutes and then exit the game normally:



So, the occurrence of this crash is almost simultaneous in nature.

After the reboot, in the event viewer, I open ‘System’ logs -
I see no errors logged before the crash.
Post boot, the only relevant errors are:

Kernel Power 41 (63)
TPM-WMI 1033

Also, there are no files in the MiniDump folder in C:/Windows.

These are the few things I tried:

• Tried reinstalling the Graphics Driver using DDU.
• Used memtest86 and waited for at least 3 passes. There were no errors.
• Ran DISM.exe /Online /Cleanup-image /Restorehealth
• Ran sfc /scannow
• Ran chkdsk /f /r
• Deleted TEMP folders

---

Description of every hardware component and if I believe if it's causing this crash or not :

CPU - Ryzen 5 3600 , age : ~2 years
I don't know if a bad CPU can cause this behaviour.
Core Performance Boost, Performance Boost Overdrive turned off in the BIOS
Average idle temperatures stays between 40-45 degree celsius, ingame temps are also normal.

Graphics Card - Zotac Nvidia GTX 1660 Ti , age: ~3.5 years
PSU - Antec VP550P Plus 550 Watt 80 Plus , age: ~2.5 years

As mentioned before, games work fine and does not crash after the reboot. So, if any of these were affected it would cause crashes on every instance and not in simultaneous instances. Still, I don’t know enough to completely count them out.

Motherboard - Gigabyte B550M DS3H , age: 2.5 years
BIOS is updated, no BSODs or random reboots, seems unlikely

Memory - G.Skill 8GB * 2 DDR4 2400Mhz , age: 4 years
No diminishing performance, no POST issue, no random crashes, no corrupt files
As mentioned before, testing done using memtest86 for 2 hours, no errors were detected (I know 2 hours might not be enough for 100% assurance)

SSD - XPG Spectrix S40G
Health @ 92% , latest firmware installed

Chipset Drivers installed are latest , from AMD[.]com

Operating System : Windows 11

 

[VirtualLarry]

PSU, SSD, or possibly CPU, if it had been run with CPB and PBO previously.

I've personally seen 3600 CPUs degrade a bit to the point of intermittant crashes / reboots, but I've also had Adata SSDs flake out, esp. nvme ones.

 

[Shmee]

I suspect PSU somewhat. I would do a CPU stress test next, so test the 3600. Should be able to rule it out fairly easily with this, or confirm it is unstable. I would focus next on the PSU and GPU. Possibly one thing to do that is a good idea, would be to get a new, quality PSU for testing, and try swapping that in. It is good to have a spare PSU anyway.

If a new PSU doesn't fix the issue, I would suspect the video card.

Alternatively, you can try turning down the power budget of the GPU first, to see if that helps with the issue. Though even if this helps, it doesn't necessarily indicate bad PSU, as it could just be the GPU is stable at lower power settings but not higher, just due to the frequency-voltage curve.

 

[el3mentry]

PSU, SSD, or possibly CPU, if it had been run with CPB and PBO previously.

I've personally seen 3600 CPUs degrade a bit to the point of intermittant crashes / reboots, but I've also had Adata SSDs flake out, esp. nvme ones.

I suspect PSU somewhat. I would do a CPU stress test next, so test the 3600. Should be able to rule it out fairly easily with this, or confirm it is unstable. I would focus next on the PSU and GPU. Possibly one thing to do that is a good idea, would be to get a new, quality PSU for testing, and try swapping that in. It is good to have a spare PSU anyway.

If a new PSU doesn't fix the issue, I would suspect the video card.

Alternatively, you can try turning down the power budget of the GPU first, to see if that helps with the issue. Though even if this helps, it doesn't necessarily indicate bad PSU, as it could just be the GPU is stable at lower power settings but not higher, just due to the frequency-voltage curve.

I just stress-tested my CPU using Cinebench. Nothing happend. Screenshot is attached.
As far as I know, if there was something wrong with the CPU, the stress test would have crashed, rebooted or BSOD'ed the PC.

 

[mikeymikec]

When the game crashes in this manner, can you do Ctrl + Alt + del, Task Manager, then kill the game process? - edit - no video signal, never mind

Have a look in Control Panel > Security & Reliability > Maintenance > view reliability history I believe the options are called.

It's a bit of a cheapo PSU you've got there, so it's possible that it's the problem, but IMO the graphics card is the obvious suspect. Any graphics corruptions at all?

I would get rid of AfterBurner for a while just to rule it out as a suspect.

 

[el3mentry]

I reinstalled a nearly 5-month-old graphics driver for my video card tested for over 2 days.

Removed the graphics driver completely using DDU in Safe Mode.
and
Installed an older driver, which is more stable than the newest driver.

It did not fix the issue.

 

[mikeymikec]

You already DDU'd the driver. In my experience not even that is necessary, let alone safe mode shenanigans. I personally would only turn to DDU if, despite say updating drivers, the driver control panel was still a mess with bizarre error messages etc.

I'd say the older driver is the more likely bet here, but is two days worth of testing definitely enough in your opinion to consider the situation resolved? It seems like the issue was intermittent to begin with.

 

[el3mentry]

You already DDU'd the driver. In my experience not even that is necessary, let alone safe mode shenanigans. I personally would only turn to DDU if, despite say updating drivers, the driver control panel was still a mess with bizarre error messages etc.

I'd say the older driver is the more likely bet here, but is two days worth of testing definitely enough in your opinion to consider the situation resolved? It seems like the issue was intermittent to begin with.

I can tell you with confidence that the video card crash was not intermittent. As I mentioned, the pattern was that on a cold/normal boot of the PC, opening a game would crash the video card. This happened every single time for the last 5 weeks. However, when I forced a restart and then opened a game again, the crash did not occur.

I read in a lot of solutions across some forums that DDU w/ Safe Mode was recommended. I had avoided the Safe Mode step before, as it also seemed a bit pointless to me.
However, this time, I wanted to make sure I was completely removing the old driver, so I used Safe Mode.

Following that, I started testing. I cold & warm booted my PC several times, running different games on each instance to see if the crash was happening. It did not happen, which is why I became confident that I had fixed it.
Also, you are right that 2 full days worth of testing might not be enough but the tests were quite strenuous & exhausting, so I wasn't left with doubt.

 

[mikeymikec]

"cold boot problems" screams hardware, just saying.

I'd be curious to know what would have happened if you cold booted, then immediately restart, then test for the problem.

If you have further problems, I'd suggest:

Do you have Fast Startup enabled? If so, disable it:
Control Panel > Power Options > Change what the buttons do (I think the option is labelled that, near top-left corner of the window, then you have to select an option to ungrey the tick boxes, then you can untick fast startup).

Fast startup would at least explain the "cold boot" aspect as fast startup only affects the shutdown/startup routine, whereas restart is a more thorough boot routine. Another way to test the "cold" aspect would be to leave the computer in sleep mode overnight and test as quickly as possible after waking it up.

Drivers and fast startup seems like an improbable combination given how long the feature has been around for, but perhaps it's fast startup + display driver + third factor fairly unique to your setup.

The last time I remember dealing with a literal cold boot scenario I had to wait until the hardware had cooled down overnight to reproduce the problem consistently. It was due to bad caps. That was over ten years ago.

Re fast startup - I've never trusted it, and it does very little for machines booting off SSDs. It's a corner-cutting measure that MS employs to make Windows look faster / more impressive.

 

[igor_kavinski]

No video signal?

Restart the graphics driver: https://www.makeuseof.com/reset-graphics-driver-windows/

If that works, you should get the video signal back.

 

[el3mentry]

it was the GPU. Had to RMA it. After RMA, am not facing the issue.

 

[Heartbreaker]

You could RMA a 3.5 year old card?