Help to diagnose PC crash during gaming?

Justnotsleeping · 5 Sep 2024 at 10:53

Hi,

I'm wondering if you can help diagnose an issue which is causing the computer to crash during graphically intensive games. I posted previously a while ago as I thought it was a GPU (6800XT) overheating issue but i'm not so sure now.

I'm playing graphically intensive games such as Nobody Wants To Die and Star Wars Outlaws, and I get random crashes where PC turns off. The computer does not switch off completely as motherboard and RAM lights are still on, but nothing on the screen (shows no input signal). I have to manually turn off and back on the computer, then loads into BIOS with an error about video card incompatible. This is what made me originally think it was a GPU crashing issue.

GPU temps have a 20-25% delta between GPU Temp & Hotspot (70-75C GPU/92-97C Hotspot). Apparently this is in specification for Radeon GPUs which have a hotspot limit of 100-105 degrees. I repasted the GPU with some Thermal Grizzly Kryonaut just in case and while the temps have come down a degree or two the issue is the same. Case airflow is good and my CPU is cooled by an NZXT Kraken AIO

I tried running Firmark 2.0 which gets temps to a similar level as about but with no issues, even after leaving on loop for over half an hour. Is there a 'more intensive' benchmark I can try?

I'm wondering if the issue might be something else, to do with motherboard, memory or PSU (Corsair RM 750x watt Gold). Could power spikes be causing the issue? I would have thought 750w is plenty for this card.

Any ideas would be appreciated.

Thanks!

mickyflinn · 5 Sep 2024 at 11:37

Test memory to rule it out.

Download and run memtest

fewfe3 · 5 Sep 2024 at 11:46

Had a similar ish issue a while ago that I thought could be the GPU. Changed PSU, memory, and graphics card, and it turned out to be something to do with the CPU and it boosting. had to disable some auto boosting in the bios and it has been fine ever since.

I need to go back and take another look at it, to see if a bios update or something else would fix it.

Justnotsleeping · 5 Sep 2024 at 20:10

mickyflinn said:
Test memory to rule it out.

Download and run memtest

Thanks for suggestion, I left Memtest running for a few hours, didnt pick any errors up

Justnotsleeping · 5 Sep 2024 at 20:12

fewfe3 said:
Had a similar ish issue a while ago that I thought could be the GPU. Changed PSU, memory, and graphics card, and it turned out to be something to do with the CPU and it boosting. had to disable some auto boosting in the bios and it has been fine ever since.

I need to go back and take another look at it, to see if a bios update or something else would fix it.

I've just got the standard auto overclock in Ryzen Master. I'll maybe revert back to base and uninstall Ryzen Master, as well as checking for new BIOS updates

tamzzy · 5 Sep 2024 at 22:36

First of all I would reset bios and run *everything* at stock and also disabled expo and see if it still happens.
Regarding furmark even if it passes does not mean that the GPU is stable. Furmark is a power virus and applies a constant, stable load on the GPU, whereas in games, the load is far more varied so you get power spikes. I don't think there's a benchmark that can replicate gaming...except gaming lol

However to me, the fact that you can pass furmark and still occasionally crash...my 2p would be on the PSU not able to keep up with the 6800xt transient spikes (this can be as high as 580w and can cause the PSU ocp to trip)
The only way to be certain is to use another higher wattage PSU.

Justnotsleeping · 6 Sep 2024 at 09:00

Perhaps its time to upgrade the PSU to a 1000w. I suppose the benefit will be it will be able to be used for my next upgrade. I had been thinking about upgrading from AM4 to AM5 but was going to hang on until the Zen 5 3D chips are released.

ED209 · 6 Sep 2024 at 09:16

I would try the following

- Make sure your gpu drivers are fully up to date
- Make sure your bios is fully updated
- Reset bios to factory and just enable EXPO or for testing leave this off and enable it if no crashes happen.

Is there anything in event viewer to indicate what is causing the crash? (could be a driver or DLL crash)

Do you have another PC you could test the GPU in or borrow another PSU from?

Justnotsleeping · 6 Sep 2024 at 10:19

ED209 said:
I would try the following

- Make sure your gpu drivers are fully up to date
- Make sure your bios is fully updated
- Reset bios to factory and just enable EXPO or for testing leave this off and enable it if no crashes happen.

Is there anything in event viewer to indicate what is causing the crash? (could be a driver or DLL crash)

Do you have another PC you could test the GPU in or borrow another PSU from?

I have the latest motherboard bios, chipset drivers and Adrenaline drivers.
Reset the bios to factory defaults. After doing this, I thought this looked promising. I had 40mins of Star Wars Outlaws...then boom....black screen!

Unfortunately I don't have another PC to get another PSU from or test GPU, as my other one is a mini PC I use for a plex server

ED209 · 6 Sep 2024 at 10:21

Justnotsleeping said:
I have the latest motherboard bios, chipset drivers and Adrenaline drivers.
Reset the bios to factory defaults. After doing this, I thought this looked promising. I had 40mins of Star Wars Outlaws...then boom....black screen!
Unfortunately I don't have another PC to get another PSU from or test GPU, as my other one is a mini PC I use for a plex server

Did Windows log anything in event viewer when the crash happened?

Was EXPO enabled on the ram when you did the reset or did you leave it off for testing?

Justnotsleeping · 6 Sep 2024 at 11:44

ED209 said:
Did Windows log anything in event viewer when the crash happened?

Was EXPO enabled on the ram when you did the reset or did you leave it off for testing?

I left EXPO/DCOP off.

Event manager seems to have logged a lot of errors looking at the summary but I don't know what I'm looking at really,

There was a critical Error with Kernel-Power which had an explanation of system rebooting without shutting down correctly. This makes sense as I had to power off and restart after the crash.

The most errors come under DeviceSetupManager with 34 errors in last hour, The description under there are similar to:-

Metadata staging failed, result=0x80070490 for container '{82953452-FE92-11EE-8FB1-BC542F625E8F}'

Am afraid I have no idea what that all means!

ED209 · 6 Sep 2024 at 12:04

Justnotsleeping said:
I left EXPO/DCOP off.

Event manager seems to have logged a lot of errors looking at the summary but I don't know what I'm looking at really,

There was a critical Error with Kernel-Power which had an explanation of system rebooting without shutting down correctly. This makes sense as I had to power off and restart after the crash.

The most errors come under DeviceSetupManager with 34 errors in last hour, The description under there are similar to:-

Metadata staging failed, result=0x80070490 for container '{82953452-FE92-11EE-8FB1-BC542F625E8F}'

Am afraid I have no idea what that all means!

Is Windows all up to date?
Are you using Windows 10 or 11?
Are you using more than 1 monitor?

To be honest I have no idea either but I tend to look online for issues etc.

Might be a pain but do you have a spare SSD\HDD you could do a clean install of windows on to see if you still get issues?

Justnotsleeping · 6 Sep 2024 at 13:06

ED209 said:
Is Windows all up to date?
Are you using Windows 10 or 11?
Are you using more than 1 monitor?

To be honest I have no idea either but I tend to look online for issues etc.

Might be a pain but do you have a spare SSD\HDD you could do a clean install of windows on to see if you still get issues?

Windows 11, all with latest updates. I have one 3440 x 1440 monitor.

I had been thinking of using a spare drive to do a windows clean install. If that doesn't work I might need to look at getting another PSU

ED209 · 6 Sep 2024 at 13:22

Justnotsleeping said:
I had been thinking of using a spare drive to do a windows clean install. If that doesn't work I might need to look at getting another PSU

I would do this as a next step, will then remove an OS\software\driver issue

henson0115 · 6 Sep 2024 at 13:32

1 test to rule out, that your cpu can handle your memory running at 3200mhz, although you'd have to have bad luck these days with the infinity fabric and 3200, it can still happen that the cpu can go that high. try downclocking the memory to 3000 and test to see if the IF can handle it. if it crashes its likely not the IF if it is still crashing check memory voltage, ive had a kit recently that had an xmp profile of 1.35 for 3600 but the memory just wasn't stable at 1.35 and had to up it to .1.38 to get it stable.

Justnotsleeping · 6 Sep 2024 at 14:44

henson0115 said:
1 test to rule out, that your cpu can handle your memory running at 3200mhz, although you'd have to have bad luck these days with the infinity fabric and 3200, it can still happen that the cpu can go that high. try downclocking the memory to 3000 and test to see if the IF can handle it. if it crashes its likely not the IF if it is still crashing check memory voltage, ive had a kit recently that had an xmp profile of 1.35 for 3600 but the memory just wasn't stable at 1.35 and had to up it to .1.38 to get it stable.

My memory as actually rated for 3600 but running at the non XMP profile default speed since i reser the bios to factory defaults. The only odd thing about the RAM is that one of the sticks LED lights does not come on always, although thats probably because i use Open RGB rather than installing Corsair software RGB software just for the RAM

henson0115 · 7 Sep 2024 at 09:59

Justnotsleeping said:
My memory as actually rated for 3600 but running at the non XMP profile default speed since i reser the bios to factory defaults. The only odd thing about the RAM is that one of the sticks LED lights does not come on always, although thats probably because i use Open RGB rather than installing Corsair software RGB software just for the RAM

im also using openrgb with rgb memory and had no issues like that, maybe try swapping the sticks around and see if you get the same issue with the lighting or a`lternatrively in openrgb have you got both stick enables in the lighting profile on the rigth hand side, you know where you can enable or disable specific components under the effects plugin?. in terms of the crashing if its still happening at stock speeds its likely not the memory then. next thing i would try if not already done, is disable PBO and all curves does it crash at stock cpu settings? what are board temps like when running under load, vrms etc

Justnotsleeping · 7 Sep 2024 at 12:40

henson0115 said:
im also using openrgb with rgb memory and had no issues like that, maybe try swapping the sticks around and see if you get the same issue with the lighting or a`lternatrively in openrgb have you got both stick enables in the lighting profile on the rigth hand side, you know where you can enable or disable specific components under the effects plugin?. in terms of the crashing if its still happening at stock speeds its likely not the memory then. next thing i would try if not already done, is disable PBO and all curves does it crash at stock cpu settings? what are board temps like when running under load, vrms etc

I have just done a new Windows installation so I can see if it was drivers, etc. Just running a few Cinebench tests and motherboard temps seem fine. Fingers crossed it will fix the issue, although im not convinced it will