My Windows 11 gaming PC intermittently loses display and immediately restarts while launching or playing games. There is usually no BSOD or crash-to-desktop; Windows only records Kernel-Power Event 41 and volmgr Event 161. Earlier troubleshooting also found repeated WHEA-Logger Event 19 entries reporting a corrected processor-core Cache Hierarchy Error on APIC ID 0, although those errors have recently stopped.
The system uses a Ryzen 9 7950X, RTX 4090, ASUS PRIME X670E-PRO WIFI motherboard, ASUS ROG Strix 1000W Gold PSU, 32GB of DDR5-6000 memory, a Samsung 990 Pro SSD, and a 360mm Corsair AIO. The crashes occur in several games and can happen during loading, even after the PC passes synthetic testing.
The BIOS, chipset drivers, and graphics drivers are up to date. I tested BIOS defaults, disabled DOCP, tried a second memory kit at stock settings, ran MemTest86 overnight, tested the CPU and individual cores with OCCT, tested the GPU and combined power load, reduced the GPU power limit to 60%, disabled Global C-State Control, and reseated the major power connections. Temperatures look normal and all tests passed.
There is another symptom: occasionally after a normal shutdown, pressing the case power button does nothing even though the motherboard RGB remains lit. Turning the PSU switch off, waiting, and turning it back on restores normal startup. Does this point more strongly to the PSU, motherboard power delivery, CPU, GPU, or another issue? I am considering an RMA, but would prefer to identify the most likely component first.
3 Answers
The PSU is the first component I would investigate. The system refusing to start until the PSU is switched off and back on sounds like a protection circuit has latched. Passing steady OCCT loads does not rule out a PSU problem because a game can create very fast GPU or combined load transitions that a synthetic test may not reproduce.
Before replacing it, try separate PCIe cables for the 4090 rather than a daisy-chained lead, verify every connector is fully seated, and test from another wall outlet without a surge strip. If possible, temporarily test with a known-good, high-quality PSU. Given that the unit is under warranty, an RMA is reasonable if the behavior continues.
The earlier WHEA Cache Hierarchy errors should not be ignored, even though the recent resets no longer generate them. Test every CPU core individually or with a tool that cycles through all cores; testing only core 0 is not enough to identify intermittent instability. Also check BIOS for any Curve Optimizer or voltage offset and temporarily set it to neutral values. On Ryzen systems, setting Power Supply Idle Control to Typical Current Idle can sometimes help with shutdown and restart behavior.
I ran an OCCT test that cycled through all cores for several minutes each and it passed without errors. I will also verify that there are no remaining voltage or Curve Optimizer settings enabled.
If a known-good PSU does not change anything, I would move the motherboard ahead of the CPU. The combination of hard resets, occasional failure to respond to the power button, and power-state-related behavior can point to motherboard power management or VRM circuitry. The 4090 is still worth testing in another system or with another GPU, but the normal GPU test and reduced power limit make a simple graphics overheating problem less likely.
The fact that the failure can happen during game launch is consistent with a transient load rather than overheating. A power-limit reduction not fixing it does not completely clear the PSU, since the startup spike or connection could still be the issue.