Question Black screen, fans at full speed, GPU driver disabled by Windows. OCCT PSU Test Pushes CPU to 100°C leading to crashes.

BorrisD1010

Reputable
Sep 16, 2019
5
0
4,510
Hello everyone!

I've been facing an issue with my computer for a few weeks now, where it crashes during gaming sessions, especially on Warzone and The Finals. The crashes happen very randomly, and I can't identify the problem.

In practical situations, I launch my game, play for an hour or more, and then, between games, the screen goes black, the fans spin at full speed, and nothing is accessible anymore. I have to completely shut down the computer. The crashes are quite random, occurring at the beginning of a game, at the end of a game, or in the middle of a battle. I've never experienced freezes in the game menus or during purely desktop activities. It's also worth noting that during a crash (black screen), I can hear sounds for a very short period before the system completely crashes, for example, a YouTube video playing in the background, in-game communications, or the game sound.

So far, I've updated my motherboard, reset the BIOS, wiped the contents of the SSD completely, reinstalled Windows 10, updated packages, and installed all drivers. I've also performed an AMD driver installation with only the GPU driver and another time with the AMD Adrenalin software (it didn't change anything). I've monitored temperatures, power, usage percentages, and capacities of all my components with HWMonitor and OCCT (CPU, GPU, RAM, PSU, etc.). No anomalies were found during gaming sessions. Regarding temperatures, the GPU performs well with an average of 50°C during stress tests and gaming sessions. A bit warmer on the CPU side, ranging between 50-75°C during tests and gaming sessions. During desktop activities, the GPU is at 32°C, and the CPU is at 35°C.

When I conducted stress tests with OCCT, I didn't have any errors except for one time: Errors found on physical cores 3 and 4. I repeated the test later, and there were no more errors.
No crashes occurred during different INDIVIDUAL stress tests. However, I noticed a significant anomaly with OCCT as follows:

  • GPU-only stress tests: 100% usage, 50°C max, correct frequencies. No crash.
  • CPU-only stress tests: 100% usage, 70°C max, correct frequencies. No crash.
  • PSU stress test (GPU + CPU at the same time): the GPU displays the same values as the first stress test. However, the CPU rises to 100-101°C. And there, the PC crashes as in a gaming session.

Another useful piece of information is that when I restarted the computer after a crash and checked the device manager, the system had disabled the graphics card driver: "This device is disabled. (Code 22)." In the display properties, it indicated being connected to the "Microsoft Basic Display Driver," and the screen was at 60 Hz instead of 144 Hz. I followed a tutorial suggesting to create a DWORD32 key in the registry to increase the GPU's response time: TdrDelay in Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers. This modification prevented the GPU from being disabled on restart, but the crashes persist.

I extracted events from the Windows Event Viewer, but I found nothing interesting/conclusive. The most frequent error is: "The description for Event ID 56 from source Application Popup cannot be found. - The following information was included with the event: ACPI 2."

Some clues make me think it could be a GPU problem, others suggest the CPU, and I'm also considering a power supply failure.
That's all I can tell you about my problem. If you have any ideas, I'm open to suggestions! I can provide Windows Event Viewer logs if necessary or a video of OCCT monitoring during a crash in the middle of a game.

Here is my PC configuration:
  • Intel I7 10700K (Installed in 2020)
  • Gigabyte Aorus Z490i (BIOS Version: F22, June 2023) (Installed in 2020)
  • ASUS TUF Gaming AMD Radeon RX 7900 XT OC Edition 20GB (Installed in 2023)
  • CORSAIR Vengeance LPX 16 (2x8) 3000Mhz - C15 (XMP enabled) (Installed in 2019)
  • CORSAIR CX750 - 750W - PLUS Bronze (Installed in 2020)
  • CORSAIR HYDRO Series H100x (Installed in 2019)
  • SAMSUNG EVO 970 Plus - 1TB (Installed in 2023)
  • Arctic MX-4 (Installed in 2020)
  • Cooler Master MasterBox MB520 (Installed in 2023)

Thanks in advance for your help! Have a great day, everyone!
 
Last edited:
CORSAIR CX750 - 750W - PLUS Bronze
How old is the PSU?

Cooler Master MasterBox MB520
I'm assuming you have fans at the front of the case, if you do, take the front panel off and see if the temps on your components change.
PSU is 3,5 years. Bought it new in June 2020.

Indeed, I have two fans at the front, two on top, and one at the rear.
The front fans are drawing in fresh air into the radiator to cool down the CPU and then pushing it inside the case.
 
Being able to get a 10700K to 100C while using an AIO seems high (unless it's overclocked) especially with it getting clean air from the front. The rest of your temps seem to be completely fine though so I'm not sure if this isn't just an anomaly.

As for the random crashes you're describing the only time I've ever seen that personally was a power problem. While the wattage might be enough I'm not sure I'd trust that PSU to hold up to transient power draw from a modern high power GPU (meaning AMD 6000/7000 NV 30/40). If you have a way to swap this out it would be the place I would start with.
 
For any future readers out there... I had similar problems as the OP:
Screen goes black, fans rev high, still hear sound for a little bit, then need to do the hard rest. Crashes occurred mostly while gaming but seemingly at completely random intervals. Could be two minutes or two hours into a game. Also, power on button would work intermittently. Would often need to flip off the power switch on the PSU, drain the power, then flip back on before the power button would work again.
Build: Darkhero VIII MB, 5900x, 4090 Suprim Liquid, 850w PSU
Based on the advice of many forums I stumbled onto, things I tried include:
  • Upgrading PSU to 1000w - didn't help
  • Bought one of those L bend psu adapters for my GPU - didn't help
  • Bought a TPM module - didn't help
  • Bought new ram - didn't help
  • Various cmds like sfc /scannow, etc. Sometimes it would find and repair errors but didn't help with problem
  • Removed drivers via DDU, installed older ones I thought might be more stable - Kinda helped?
  • Updated all the chipset drivers. BIOS was already current. - didn't help
  • Messed with a bunch of BIOS settings like ERP, performance settings, overclocking, etc. - didn't help
  • Moved save location of games I was playng to my M2 windows main drive - didn't help
  • Reinstalled Windows 11 - This actually made it worse somewhat. Instead of Black screen, high fan rev, I got BSOD with "Critical Process has died" error almost immediately after starting Diablo IV. Despite crash/error posting, would never produce a minidump file.

WHAT DID HELP - Making sure my GPU's 4x8 12vhpwr pin connector actually had 4 independent cables (each plugged into their own port on the PSU) powering it. Originally, I believe I was using three cables with one of the heads splitting into two 8-pin connectors. Haven't had any problems since. Hope this helps someone in the future! Good luck!