Games crash do not know why

[LaraJolien]

Hi, my games have been crashing loads lately while giving graphics errors. I have tried numerous things to fix the issue. My latest attempt to fixing the problem was a complete reinstall of windows on a different drive which did not help. This makes me think the problem is not software related. Could my GPU be dying? It has no artifacting problems just the crashes. My GPU is a Gigabyte RTX 2060 OC edition and I have had it for about 4 years. It currently reaches 77 C while at 99% usage at most but it used to be at 83 C at 99% usage all the time before I fixed that issue (for about 2 years). I had this problem in Valorant and Minecraft but it has not occurred since I set my max FPS to 160 and the GPU usage stays below 60%. I do still have the problem in Fortnite where the GPU usage is over 95%. The crash does not happen instantly it always happens about 15 minutes after starting the game and will then do it about every 3 minutes after the first crash when relaunching the game.

 

The rest of my specs:

CPU: Intel core I5-13600KF

Motherboard: MSI mpg b760i edge wifi

RAM: 2x Trident Z Royal 8gb DDR4 3200 mHz

SSD: WD Black SN850X 2tb

PSU: 500W from BE QUIET!

 

Let me know if you have any idea what could cause the issue and how to resolve it.

Thanks!

[Agall]

1 minute ago, LaraJolien said:

Hi, my games have been crashing loads lately while giving graphics errors. I have tried numerous things to fix the issue. My latest attempt to fixing the problem was a complete reinstall of windows on a different drive which did not help. This makes me think the problem is not software related. Could my GPU be dying? It has no artifacting problems just the crashes. My GPU is a Gigabyte RTX 2060 OC edition and I have had it for about 4 years. It currently reaches 77 C while at 99% usage at most but it used to be at 83 C at 99% usage all the time before I fixed that issue (for about 2 years). I had this problem in Valorant and Minecraft but it has not occurred since I set my max FPS to 160 and the GPU usage stays below 60%. I do still have the problem in Fortnite where the GPU usage is over 95%. The crash does not happen instantly it always happens about 15 minutes after starting the game and will then do it about every 3 minutes after the first crash when relaunching the game.

 

The rest of my specs:

CPU: Intel core I5-13600KF

Motherboard: MSI mpg b760i edge wifi

RAM: 2x Trident Z Royal 8gb DDR4 3200 mHz

SSD: WD Black SN850X 2tb

PSU: 500W from BE QUIET!

 

Let me know if you have any idea what could cause the issue and how to resolve it.

Thanks!

Try using a software like MSI Afterburner to limit the power% aka TDP to 85% and see if you're still getting the same issues.

 

500W is potentially fine for an RTX 2060, but with a 13600kf and if that PSU is a few years old, its quite possible that its not enough.

[LaraJolien]

47 minutes ago, Agall said:

Try using a software like MSI Afterburner to limit the power% aka TDP to 85% and see if you're still getting the same issues.

 

500W is potentially fine for an RTX 2060, but with a 13600kf and if that PSU is a few years old, its quite possible that its not enough.

Thanks for the reply, I set the power limit to 73% (the lowest it could go) in MSI afterburner and it reduces the amount of crashes slightly but they still happen frequently.

[Agall]

5 minutes ago, LaraJolien said:

Thanks for the reply, I set the power limit to 73% (the lowest it could go) in MSI afterburner and it reduces the amount of crashes slightly but they still happen frequently.

Another thing to try ontop of 73% would be to set a TDP limit in the UEFI for your 13600kf. I would set it around 105W-120W, whichever is an option.

 

Really, I would suggest a new power supply, something in the 750W-850W range since that's generally peak value in 2024. Ideally though, you can create a scenario to determine whether its a PSU overdraw scenario. Noting that the 13600k can spike in excess of 200W.

 

The fact that framerate limiting or TDP limiting helps gives credence to the issue being a power problem, whether that's an overdraw of the PSU or instability of the 12V rail at higher loads.

[LaraJolien]

51 minutes ago, Agall said:

Another thing to try ontop of 73% would be to set a TDP limit in the UEFI for your 13600kf. I would set it around 105W-120W, whichever is an option.

 

Really, I would suggest a new power supply, something in the 750W-850W range since that's generally peak value in 2024. Ideally though, you can create a scenario to determine whether its a PSU overdraw scenario. Noting that the 13600k can spike in excess of 200W.

 

The fact that framerate limiting or TDP limiting helps gives credence to the issue being a power problem, whether that's an overdraw of the PSU or instability of the 12V rail at higher loads.

So the PSU is most certainly the problem? I have Asiahorse cable extensions could that have anything to do with the power issue? and I have a lot of usb devices plugged in does that make a big difference?

What exactly do you mean with instability of the 12V rail?

Thanks a lot already! My gaming experience is already a lot better than before you helped me!

[QuantumSingularity]

IF you system shits down completely, your PSU is almost certainly overheating. If it's just restarting it's probably some protection tripping like OCP or OPP. Better replace it because as mentioned above 500W are not really enough for a 13600ks with dedicated GPU.

Custom cables can be an issue like they were in the case were i was dealing with a PC for over a week. Usually they are cause restarts though, not system shutdowns. 

Also as mentioned earlier - use Afterburner or HWinfo64 to check the temps of the GPU while you run a stress test in FurMark. Had a couple of instances with Asus RTX 3000 series GPUs where the memory chips were only half covered by the thermal pads and were overheating.

[Agall]

24 minutes ago, LaraJolien said:

So the PSU is most certainly the problem? I have Asiahorse cable extensions could that have anything to do with the power issue? 

Its definitely the best theory so far, but its not a certainty until you have another power supply available. Outside of obvious thermal limitations/failures, its unlikely since I'm assuming you transplanted this RTX 2060 from another system/motherboard. I've assumed the most recent variable being a CPU+motherboard upgrade, which would be a sufficient increase in peak and steady state power draw to push a PSU problem over the edge.

 

Cable extensions can actually cause problems, whether that's the CPU EPS connectors, VGA aux power, or 24 pin (since technically the 24 pin does supply 12V power through the PCIe connector to the GPU). There's a loss of voltage when increasing cable length, and if your PSU is wearing to the point where the 12V rail isn't maintaining a stable 12V's, then cable extension will compound that.

 

Before you go buy a new PSU, try removing all the cable extensions you're using. Its a possibility, but less likely than just a power supply issue. At least its a $0 test comparably.

[LaraJolien]

15 minutes ago, QuantumSingularity said:

IF you system shits down completely, your PSU is almost certainly overheating. If it's just restarting it's probably some protection tripping like OCP or OPP. Better replace it because as mentioned above 500W are not really enough for a 13600ks with dedicated GPU.

Custom cables can be an issue like they were in the case were i was dealing with a PC for over a week. Usually they are cause restarts though, not system shutdowns. 

Also as mentioned earlier - use Afterburner or HWinfo64 to check the temps of the GPU while you run a stress test in FurMark. Had a couple of instances with Asus RTX 3000 series GPUs where the memory chips were only half covered by the thermal pads and were overheating.

The PC never shuts off or restarts just game crashes. I tried FurMark (which crashed once too) and the temps seem fine. (I put the TDP back to 100% for this test)

[LaraJolien]

4 minutes ago, Agall said:

Its definitely the best theory so far, but its not a certainty until you have another power supply available. Outside of obvious thermal limitations/failures, its unlikely since I'm assuming you transplanted this RTX 2060 from another system/motherboard. I've assumed the most recent variable being a CPU+motherboard upgrade, which would be a sufficient increase in peak and steady state power draw to push a PSU problem over the edge.

 

Cable extensions can actually cause problems, whether that's the CPU EPS connectors, VGA aux power, or 24 pin (since technically the 24 pin does supply 12V power through the PCIe connector to the GPU). There's a loss of voltage when increasing cable length, and if your PSU is wearing to the point where the 12V rail isn't maintaining a stable 12V's, then cable extension will compound that.

 

Before you go buy a new PSU, try removing all the cable extensions you're using. Its a possibility, but less likely than just a power supply issue. At least its a $0 test comparably.

Thanks, I will try removing the cable extensions. I did upgrade the mobo and cpu about a year ago and everything was perfectly fine up until about 2 months ago so that is weird. I have also had the cable extensions for the entire time I've had the new mobo and cpu. All the crashes just appeared all at once one day and did not gradually increase over time. One thing to note is that I have a custom PC case which had a prototype 3D printed motherboard bed that started to warp overtime due to heat and pressed up against the bottom of the mobo right underneath the pcie slot the gpu is in. I do not know if this could cause an issue though.

[Agall]

9 minutes ago, LaraJolien said:

Thanks, I will try removing the cable extensions. I did upgrade the mobo and cpu about a year ago and everything was perfectly fine up until about 2 months ago so that is weird. I have also had the cable extensions for the entire time I've had the new mobo and cpu. All the crashes just appeared all at once one day and did not gradually increase over time. One thing to note is that I have a custom PC case which had a prototype 3D printed motherboard bed that started to warp overtime due to heat and pressed up against the bottom of the mobo right underneath the pcie slot the gpu is in. I do not know if this could cause an issue though.

 

That could very easily be apart of the problem since part of a case's function is a proper ground as well. That can result in the buildup of impedance/capacitance in components that can affect input current/voltage. I think it would be worth normalizing your system in a standard steel case and without extensions to see if you get these issues still. Unless apart of that case's design was the addition of grounding wires between the motherboard and power supply.

 

17 minutes ago, LaraJolien said:

The PC never shuts off or restarts just game crashes. I tried FurMark (which crashed once too) and the temps seem fine. (I put the TDP back to 100% for this test)

In this, you're reaching the thermal limit at only 167W. If you expand out the 'GPU Performance Limiters' we can confirm what's the performance limitation too.

 

From HWinfo's website; "It's pretty common for the +12V rail to drop slightly under higher load. 11.85 should be still OK. under a benchmark with highest settings possible the PCIe 12v Input Voltage comes out between 11.80-11.85; a fair droop but it's not really what occurs under usual gaming load"

 

Yours is dropping to 11.671V which is low enough to be a problem, that could be because of the PSU and/or the extension cables, or some of the circuitry on the GPU itself. That should be the voltage the GPU is sensing from the 12V rail which should be higher than what the PSU is pumping out.

[LaraJolien]

58 minutes ago, Agall said:

 

That could very easily be apart of the problem since part of a case's function is a proper ground as well. That can result in the buildup of impedance/capacitance in components that can affect input current/voltage. I think it would be worth normalizing your system in a standard steel case and without extensions to see if you get these issues still. Unless apart of that case's design was the addition of grounding wires between the motherboard and power supply.

 

In this, you're reaching the thermal limit at only 167W. If you expand out the 'GPU Performance Limiters' we can confirm what's the performance limitation too.

 

From HWinfo's website; "It's pretty common for the +12V rail to drop slightly under higher load. 11.85 should be still OK. under a benchmark with highest settings possible the PCIe 12v Input Voltage comes out between 11.80-11.85; a fair droop but it's not really what occurs under usual gaming load"

 

Yours is dropping to 11.671V which is low enough to be a problem, that could be because of the PSU and/or the extension cables, or some of the circuitry on the GPU itself. That should be the voltage the GPU is sensing from the 12V rail which should be higher than what the PSU is pumping out.

The only performance limiter under load is Power according to HWinfo.

Would connecting the Mobo screws to the metal PSU chassis be a good way to ground the system or is that bad for the PSU? How is the system grounded in a normal case?

I'll see if I can improve the 12V rail Voltage.

Thanks for all the help so far!

[QuantumSingularity]

I am not sure a 2060 is supposed to reach a hot spot of 96°C when the package temps are reading 75°C. 21°C delta between average and hotspot seems a bit high. Probably dried up paste or improper contact or both. Could also be causing the crashes if the GPU suddenly drops to voltage too much to reduce the temperature. With performance limiters those temps should be under 60°C, especially with fans at 2400 rpm

[Agall]

7 minutes ago, LaraJolien said:

The only performance limiter under load is Power according to HWinfo.

Would connecting the Mobo screws to the metal PSU chassis be a good way to ground the system or is that bad for the PSU? How is the system grounded in a normal case?

I'll see if I can improve the 12V rail Voltage.

Thanks for all the help so far!

In a normal case, the grounds are done through the motherboard standoffs to the case. The power supply also being grounded via its metal screws into the housing. If you were designing your own non-metallic case though, I would suggest having a grounding wire between one of the motherboard standoff screws and the PSU. Normal DC circuit grounds are handled via the pins, so its not like its grounding the whole wattage of the PC through that.

 

Basically its a wire with two loops on each end that you put through the screw, so that each screw is grounded that way. For the PSU, you could have it behind the screw holes for the case, where it just needs to be able to touch the screw which grounds through its threads. Good policy to make sure there's no paint on the back end of it though, since those screws are designed to wear the paint on the case and self tap into bare steel on the PSU housing and case.

[LaraJolien]

I have removed the extension cable and now the 12v rail voltage seems fine. Sadly no change in the number of cashes so the 12v rail does not seem to cause the problem. I have also tried playing CS2 now and that crashes almost instantly every time I launch it even with the GPU being at around 60C and the TDP set to 80%. Might just be a CS2 thing though I'm not sure.