[SOLVED] Most games suddenly causing fans 100% and black screen crash. Sounds still working. GTX 1080 Ti. Need help!

[Deeling]

Need some help figuring out the next diagnostic step. Is it a GPU issue? PSU issue? Or software issue? Or some other issue?

Symptoms:
Current setup was working fine until a couple days ago. No changes to hardware, software or driver configuration which could have triggered it. Computer remains 100% stable when using Windows, Chrome and Microsoft Word etc.

Computer only crashes when starting most games, such as DCS (Digital Combat Simulator), Titanfall 2 and Cyberpunk. Crashes usually occur when loading the game engine. Display goes black, often (but not always) accompanied by fans revving to max. Sounds playing indicate that the computer is still working, just unable to output display (e.g. Youtube audio or game audio still working). Temperatures of GPU and CPU remain at or around 80 deg at most, per CPUID HWMonitor. Because of the nature of the crash, it produces no crash logs. Event Viewer has a whole bunch of errors near the time of each crash, but none seem to consistently precede the crashes.

System specs:
Windows 10 Pro 19042.804
Processor: Intel Core i7-8700 CPU
Graphics card: GTX 1080 ti
Driver version: NVDIA GeForce Experience 3.21.036
RAM: 32GB DDR4
Motherboard: ASUS Prime B360M-A, bios version 2811 (date: 2020/07/27)
Hard drive: Samsung SSD 860 EVO 1TB SSD
PSU: Cooler Master EX2 625

What I've tried:
-Redoing the thermal paste on GPU
-Stress tested the GPU on another computer on Furmark and 3DMark (did this at the GPU's warranty service center). Warranty replacement request was rejected as the GPU passed the stress tests.
(Note: Running Furmark on my own rig sometimes results in crash after many minutes, but this is not a reliable way to trigger it)
-Ran MEMTEST86 (passed)
-Ran GpuMemTest (passed)
-Upgraded the cooler on the CPU
-Pointed a gigantic fan at my open case
-Used DDU to remove the display drivers and reinstalled the latest NVIDA drivers
-Rolled back all Windows updates prior to the start of the crashes
-Updated BIOS to latest version
-Reseated both RAM chips
-Tried command prompt: sfc /scannow
-Tried command prompt: dism /online /cleanup-image /restorehealth

What I have not tried:

1. Swapping out the GPU for an identical model and seeing if crash can be triggered
2. Replacing PSU
3. Reinstalling Windows
4. Replacing MoBo

Spent a whole week on troubleshooting, to no avail. Any help as to my next steps would be greatly appreciated!

 

[iTRiP]

If I where you I would format and delete partitions and re-install latest windows and drivers and go from there, might get lucky with a clean system, if not look into the main parts PSU, MB, GPU.

 

[geofelt]

My guess is that the psu has started to fail.
See if you can't test with a known good psu.
Gaming causes the graphics card to work much harder and it will draw more power.
Cooler master units have variable quality, I don't think yours was one of the best.
If you can't borrow one, buy a quality replacement from a shop with a good return policy.
a 14% restocking charge is fair if you want to return it.

You are probably on the edge with a 650w unit that might have deteriorated over time.
http://www.realhardtechx.com/index_archivos/Page362.htm
Consider the psu as a good long term investment.
Find a quality unit with a 10 year warranty.

 

[dorsai]

Have you tested your SSD with Samsung Magician ?

 

[--SID--]

Replace the psu. Cooler Master Extreme 2 625w is a poor groupregulated psu. It isn't even a 625w. It's a 500w because that's what available at 12v.

 

[Deeling]

Have you tested your SSD with Samsung Magician ?

Nope, but the crashes happen with games that are installed on my other drives as well. I'll give it a test though.

Thanks for the comments, I will see if I can procure a 700W PSU and report back on the issue. Any recommended brands and models?

 

[Deeling]

Could I check, if it is indeed a PSU issue, why wouldn't Furmark or 3DMark trigger it though? While they are not games, they should be equally taxing, if not more so given that they are running the same processes.

 

[--SID--]

Furmark and 3D Mark are static loads. While gaming the load is highly dynamic. That's the difference.

 

[Deeling]

Furmark and 3D Mark are static loads. While gaming the load is highly dynamic. That's the difference.

Thanks. FWIW, it's now crashing upon INITIALIZING Furmark. As in, when I click the "GPU stress test" button, it crashes within 1 second of the Furmark render window popping up. It's certainly not enough time for the GPU to heat up at all.

Would this support any of the above hypotheses?

I'm probably going down to a repair shop in the next day or two. To minimise my repair bill, hopefully I'll have the problem dialed in by then. TIA!

 

[Deeling]

Thanks all. I replaced the PSU and no more issues! Fingers crossed that's the end of my woes.... took a whole week to diagnose this!

 

[geofelt]