Question System crashes randomly and I can´t figure it out ?

Mar 11, 2025
9
0
10
Hi experts,

3 years ago I build a new PC and it keeps crashing. I can´t figure out the reason.

Mostly when gaming, my pc will crash and restart. The screen turns black and windows will boot again. I can not find anything specific in the event logger other than "The system has rebooted without cleanly shutting down first....". Recently I was seeing this with Path of Exile 2.


The things I have checked and tried.
  • checked temperatures - all within 50-60 C
  • changed PSU
  • changed GPU
  • removed all drives except the one windows runs on
  • clean windows install
  • downgrade to win 10 with fresh install and disabled TPM
  • flashed latest BIOS
  • installed current GPU drivers
  • ran stability test on RAM - no errors found
  • enabled/disabled XMP
  • used a smart socket to measure watts used - when gaming this went to 600W roughly

My configuration:

CvjYIqz.png


PSU:
The PSU is a be quiet Straight Power 11 Platinum 750W.
I know the 7900XTX is recommended for 850W and I can give a different PSU a shot as well, I recently upgraded to this card and had a 3060Ti before (the crashed were also present whit the old card). However i measured my power consumption and only went to around 600 W.
The card has 3 8-Pin power connectors and I used the PCIe 1 with both connections and PCIe 2 with only one connection. I assume this is correct but am not sure.

CPU Cooler:
AIO EK Water Blocks AIO 360 D-RGB

Fans: Corsair iCUE SP120 RGB Elite x 6




I found a thread where people were having the same issue and linked it to Gigabyte Mobos. I am tempted to give a change of motherboard a shot.

The things I know the least about if they give me trouble are Motherboard, RAM and the NVMe drive. Are there ways to rule these out?

I am open to any suggestions. Thank you all!
 
Thanks for the quick reply!
I just ran the OCCT stress test for CPU, GPU, memory etc. As soon as I clicked start, the system crashed and rebooted.
I will run the recommended benchmark as well.

I already tried 3 different GPUs over time. 3060Ti, 4070 and now the RX7900XTX. The issues was always persistent. I also already tried a 850 W GPU with my 3060Ti.

I am open to trying a stronger PSU again.
 
Multiple GPUs and multiple PSUs? Before you do anything else, lets sort through a few things.

If you are overclocking or undervolting anything...stop. Revert to stock speeds. This applies to the CPU, GPU, and memory. Everything at default. Does this change anything?

Have you ever tested your memory? If so, with what app?

What version of Windows are you currently using?
 
  • Like
Reactions: kaemo
The original configuration was the 750 W PSU and the 3060Ti.
To rule out the gpu, I changed to a 4070 but the problem persisted and I gave back the gpu. Afterwards I tried a 850 W psu to rule out the error coming from there. The problem still persisted and I gave back the psu.

I upgraded to the RX7900XTX and an AOC 180hz 1440p monitor a week ago.

I did not overclock or undervolt. All settings are stock.
Only change I made was enabling XMP, however it is disabled again.

I am currently running windows 10, as I read, downgrading from windows 11 and disabling TPM might be something to look into. Did not work. Currently running win 10 22H2 19045

I did in the past run the memory test through prime95 and it passed. However, I did not run it recently.
Yesterday I ran the native memory diagnosis tool without errors.
 
Yes, I did not change anything regarding the cpu or its cooling.
Since the temps are fine I did not suspect anything wrong.

Could there be an issue with the aio?
 
thank you for the hint. I could not notice a big difference. The CPU package temp under stress was 48C, however the case was open during that. It goes to 74C with the case closed.
 
Last edited:
I ran GPU stress test through OCCT and there was no issue. Also only testing CPU + RAM was fine.
When I tried the "power" stress test, the system shuts off.
The difference between the tubes of the AiO is noticable but not that big. In case the pump was not working, the temperature of the CPU should really be higher, shouldn´t it?
 
What do you mean by:
" I used the PCIe 1 with both connections and PCIe 2 with only one connection " ?

From left to right, there are three PCIe 8pin power slot on the 7900 XTX, I name them 1, 2 and 3.
It seems that you use a PCIe power cable divided into 2x 8pins and connect them to 1 and 2, and plugged another single PCIe power cable into 3.
If you are doing this, immediately stop! The three slots don't simply take average electricity. For these three slots, the more closer to the GPU core(=left) it is, the more electricity it takes.
So, plug the simple PCIe 8pin cable into 1, and use the 1-divided-into-2 8pin cable into 2 and 3.
 
  • Like
Reactions: kaemo
I indeed did this wrong, thank you! I got a Corsair RM850x in the meantime. Just arrived today. This has the 12VHPWR 16pin connector. I would use this for the slots 1&2 and plug a regular 8 pin into slot 3, correct?
Or should I rather just use 3 separate 8 pin connectors?
 
Last edited: