PC crashes

[Guest]

Hello good people of LTT forum. I am currently looking for help to diagnose a colleagues PC. Whenever he is gaming for too long, the PC without any clue what's wrong crashes.

 

Build:

• AMD ryzen 7 5800X
• Gigabyte X570 AORUS ULTRA
• 4x8GB HyperX ram sticks HX432C16PB3K2/16
• Gigabyte GeForce RTX 3060 EAGLE OC 12G
• NVME SSD
• EVGA 750Watt PSU
• Clear Windows 11 Pro

 

What I tried so far:

I managed to replicate the issue by running RDR2 benchmark on loop and sure as day it happens somewhere around the 3rd or 4th loop. The PC resets with no BSOD screen a simply reboots.

Event Viewer show only events about sudden reboot and no extra Errors.

Windows Minidump files show just ntoskrnl.exe and PSHED.dll, so those are not saying much. Files attached.

 

I tried swapping the PSU for a stronger one (1000Watt Seasonic), but no change. So power spikes won't be the issue.

I put in a spare Nvidia 2070 Super and the PC showed no symptoms (the RDR2 loop ran over an hour with no crash).

This leads me to believe that the 3060 is the issue, although it didn't crash when testing in my PC (12th gen i7, Seasonic 1000W PSU, 2x16GB Corsair Dominator RAM).

 

Stressing the GPU with Furmark had no effect on it. The stress test ran for about an hour and the GPU purred like a kitten.

I ran some test on the GPU and recorded the values, the hotspot temperature was somewhere around 105°C. I repasted it and changed all thermal pads, which lead to hotspots temperature dropping to around 97°C.

This also had no effect on the crashing. All recorded values are attached.

I tried two different Nvidia Drivers, the newest from Nvidia and the version from Gigabyte site https://www.gigabyte.com/Graphics-Card/GV-N3060EAGLE-OC-12GD-rev-20/support#dl

I ran a VRAM test tool (https://github.com/GpuZelenograd/memtest_vulkan) which reported zero issues.

Finally I tried lowering GPU boost in AORUS engine (by -15, then by -50) but no change.

 

I attached the minidump files, GPU-Z logs and OpenHardwareMonitor logs.

Please help. So far, the only conlusion I have is, that the GPU is bad. But it is very strange to say so, considering it works in other PC or under stress tests.

Minidump.zip OpenHardwareMonitorLog-2025-10-28.csv GPU-Z Sensor Log.csv

[strange13930]

29 minutes ago, Mathes58 said:

Hello good people of LTT forum. I am currently looking for help to diagnose a colleagues PC. Whenever he is gaming for too long, the PC without any clue what's wrong crashes.

 

Build:

• AMD ryzen 7 5800X
• Gigabyte X570 AORUS ULTRA
• 4x8GB HyperX ram sticks HX432C16PB3K2/16
• Gigabyte GeForce RTX 3060 EAGLE OC 12G
• NVME SSD
• EVGA 750Watt PSU
• Clear Windows 11 Pro

 

What I tried so far:

I managed to replicate the issue by running RDR2 benchmark on loop and sure as day it happens somewhere around the 3rd or 4th loop. The PC resets with no BSOD screen a simply reboots.

Event Viewer show only events about sudden reboot and no extra Errors.

Windows Minidump files show just ntoskrnl.exe and PSHED.dll, so those are not saying much. Files attached.

 

I tried swapping the PSU for a stronger one (1000Watt Seasonic), but no change. So power spikes won't be the issue.

I put in a spare Nvidia 2070 Super and the PC showed no symptoms (the RDR2 loop ran over an hour with no crash).

This leads me to believe that the 3060 is the issue, although it didn't crash when testing in my PC (12th gen i7, Seasonic 1000W PSU, 2x16GB Corsair Dominator RAM).

 

Stressing the GPU with Furmark had no effect on it. The stress test ran for about an hour and the GPU purred like a kitten.

I ran some test on the GPU and recorded the values, the hotspot temperature was somewhere around 105°C. I repasted it and changed all thermal pads, which lead to hotspots temperature dropping to around 97°C.

This also had no effect on the crashing. All recorded values are attached.

I tried two different Nvidia Drivers, the newest from Nvidia and the version from Gigabyte site https://www.gigabyte.com/Graphics-Card/GV-N3060EAGLE-OC-12GD-rev-20/support#dl

I ran a VRAM test tool (https://github.com/GpuZelenograd/memtest_vulkan) which reported zero issues.

Finally I tried lowering GPU boost in AORUS engine (by -15, then by -50) but no change.

 

I attached the minidump files, GPU-Z logs and OpenHardwareMonitor logs.

Please help. So far, the only conlusion I have is, that the GPU is bad. But it is very strange to say so, considering it works in other PC or under stress tests.

Minidump.zip 3.32 MB · 0 downloads OpenHardwareMonitorLog-2025-10-28.csv 711.7 kB · 0 downloads GPU-Z Sensor Log.csv 1.25 MB · 0 downloads

try using only 2 ram sticks to see if it's a stability issue with the 3060

[Mumintroll]

Those hotspot temperatures are concerning.

Are there any difference in temperatures when the 3060 is in your pc case, compared to his?

Perhaps your case have a better airflow.

[Mark Kaine]

3 hours ago, Mathes58 said:

4x8GB HyperX

That's why. Most likely!

 

But 2x16 Trident Z 3200mhz (cause I know they should work)

and see if it gets fixed. Be sure there are no overclocks etc too, except XMP "3200"...

[Mark Kaine]

3 hours ago, Mumintroll said:

Those hotspot temperatures are concerning.

Are there any difference in temperatures when the 3060 is in your pc case, compared to his?

Perhaps your case have a better airflow.

Agreed it's way hot for a 3060?

 

So either that or ram imo (remember Ryzen hates 4 sticks of ram)

[191x7]

7 hours ago, Mathes58 said:

Hello good people of LTT forum. I am currently looking for help to diagnose a colleagues PC. Whenever he is gaming for too long, the PC without any clue what's wrong crashes.

 

Build:

• AMD ryzen 7 5800X
• Gigabyte X570 AORUS ULTRA
• 4x8GB HyperX ram sticks HX432C16PB3K2/16
• Gigabyte GeForce RTX 3060 EAGLE OC 12G
• NVME SSD
• EVGA 750Watt PSU
• Clear Windows 11 Pro

 

What I tried so far:

I managed to replicate the issue by running RDR2 benchmark on loop and sure as day it happens somewhere around the 3rd or 4th loop. The PC resets with no BSOD screen a simply reboots.

Event Viewer show only events about sudden reboot and no extra Errors.

Windows Minidump files show just ntoskrnl.exe and PSHED.dll, so those are not saying much. Files attached.

...

Latest motherboard Bios?

Latest SSD firmware?

Latest GPU vBios?

[Bjoolz]

15 hours ago, Mathes58 said:

Hello good people of LTT forum. I am currently looking for help to diagnose a colleagues PC. Whenever he is gaming for too long, the PC without any clue what's wrong crashes.

 

Build:

• AMD ryzen 7 5800X
• Gigabyte X570 AORUS ULTRA
• 4x8GB HyperX ram sticks HX432C16PB3K2/16
• Gigabyte GeForce RTX 3060 EAGLE OC 12G
• NVME SSD
• EVGA 750Watt PSU
• Clear Windows 11 Pro

 

What I tried so far:

I managed to replicate the issue by running RDR2 benchmark on loop and sure as day it happens somewhere around the 3rd or 4th loop. The PC resets with no BSOD screen a simply reboots.

Event Viewer show only events about sudden reboot and no extra Errors.

Windows Minidump files show just ntoskrnl.exe and PSHED.dll, so those are not saying much. Files attached.

 

I tried swapping the PSU for a stronger one (1000Watt Seasonic), but no change. So power spikes won't be the issue.

I put in a spare Nvidia 2070 Super and the PC showed no symptoms (the RDR2 loop ran over an hour with no crash).

This leads me to believe that the 3060 is the issue, although it didn't crash when testing in my PC (12th gen i7, Seasonic 1000W PSU, 2x16GB Corsair Dominator RAM).

 

Stressing the GPU with Furmark had no effect on it. The stress test ran for about an hour and the GPU purred like a kitten.

I ran some test on the GPU and recorded the values, the hotspot temperature was somewhere around 105°C. I repasted it and changed all thermal pads, which lead to hotspots temperature dropping to around 97°C.

This also had no effect on the crashing. All recorded values are attached.

I tried two different Nvidia Drivers, the newest from Nvidia and the version from Gigabyte site https://www.gigabyte.com/Graphics-Card/GV-N3060EAGLE-OC-12GD-rev-20/support#dl

I ran a VRAM test tool (https://github.com/GpuZelenograd/memtest_vulkan) which reported zero issues.

Finally I tried lowering GPU boost in AORUS engine (by -15, then by -50) but no change.

 

I attached the minidump files, GPU-Z logs and OpenHardwareMonitor logs.

Please help. So far, the only conlusion I have is, that the GPU is bad. But it is very strange to say so, considering it works in other PC or under stress tests.

Minidump.zip 3.32 MB · 0 downloads OpenHardwareMonitorLog-2025-10-28.csv 711.7 kB · 0 downloads GPU-Z Sensor Log.csv 1.25 MB · 0 downloads

You only get dump files from BSODs so you are 100% BSODing. You might not see it, but it's still happening. 

 

All of the dump files show a hardware error with the CPU. If anything is overclocked or undervolted, remove it. That includes the DOCP/XMP profile on the RAM and making sure that Precision Boost Overdrive is set as disabled in the BIOS. 3200 is usually not an overclock, but with four sticks it usually has a lower officially supported speed. AMD never gave any info on this, but ASRock kind of leaked it in one of their manuals saying 2933 (Run it at stock 2133 while testing). You are also on a beta BIOS so I would strongly suggest updating to F39 (You are on F39i, versions that end in a letter are beta versions). The beta version you are on has also been removed from the downloads list which usually means it had issues.

 

The CPU is giving two different errors, though they are very similar. One is a cache error when reading data from the L1 cache and the other is a bus error when fetching an instructions from the L1 cache. 

 

If you still crash a faulty CPU is the main suspect, but 5000 series has a ton of voltage issues so we can try some tweaks to see if we can get it to behave. 

• The first is if your motherboard has a setting for a voltage offset. If it does, set the CPU Core and SoC voltage offsets to +0.050v (Please read this number twice. Not 0.5v, but 0.05v).
• The second is setting a static voltage for the Core and SoC. We set a static voltage of 1.3v to the Core and 1.1v to the SoC.

If your board uses increments for the voltage instead of inputting a number, just get as close as you can. You can't use both at the same time so try one at a time.

The first one is more general 5000 series related when you get errors from the CPU memory controller. The second is something we've found helpful with mostly the higher end 5000 series chips like the 5800x, 5900x and 5950x across a wide range of crashes. So the second one is the most likely to help here, but trying the first one if that one doesn't help is completely fine.

[Guest]

Thank you everyone, you've been a great help and I figured out the issue.

 

As I suspected the GPU was overheating and it was due to poor airflow in the case as @Mumintroll suggested.

It really is a bad case. BeQuiet much like this one https://www.amazon.co.uk/dp/B07GT4G988 and the front intake just isn't big enough.

Removing the front panel lead to temperature drop of about 20°C

 

Also big thanks to @Bjoolz that analysis is amazing and as you suggested I updated BIOS to F39. Can't believe what you managed to figure out from BSOD files only.

 

It wasn't RAMs, as I've tried 1 stick, 2 sticks, XMP on and off.

 

Thank you again.