For the past few weeks, my PC has been completely black-screening during games, especially when GPU usage spikes. The display cuts out, the GPU fans ramp up, and the system becomes unresponsive, although I can still hear the game briefly in the background. I have to hold the power button to reboot. It has happened in Path of Exile, Dota 2, and World of Warcraft, but never during normal desktop use.
I have already used DDU to reinstall both current and older graphics drivers, updated the BIOS, reseated the GPU, checked CPU and GPU temperatures, run Windows Memory Diagnostics, and tested the system with FurMark 2 without problems. Lowering the GPU power limit from 100% to 83% reduces how often it crashes but does not eliminate the issue. Event Viewer only shows Kernel-Power Event 41 after the forced restart.
My parts are relatively new: ASRock B650M Pro RS WiFi motherboard, Ryzen 7 7800X3D, MSI MPG A1000G PCIe 5 1000W power supply, 32GB Kingston DDR5-6000 CL30 memory, and a PNY GeForce RTX 5060 Ti 16GB OC running Windows 11. When I reinstall my older RTX 2060, I can play for an entire evening without a crash. Could the newer GPU itself or its power delivery be faulty, or should I still investigate the motherboard or RAM?
3 Answers
Kernel-Power Event 41 is usually just a record that Windows lost power or was rebooted unexpectedly; it does not identify the original fault. The fact that the older RTX 2060 works normally while the newer card crashes across several games makes the newer GPU, its power connection, or a power-transient issue more suspicious than a particular game. Try separate PSU cables rather than a shared daisy-chained cable, confirm every connector is fully seated, and test the newer card in another system if possible. If the problem follows the card, an RMA or replacement is the logical next step.
A FurMark pass does not completely clear the GPU or power delivery. Some games create different rapid load changes than a steady synthetic benchmark. Since reducing the power limit lowers the crash frequency and the older card is stable, test with the GPU at stock settings, then with a modest underclock or undervolt. Also check whether the graphics card has a newer VBIOS and try disabling overlays or hardware acceleration temporarily. If it still black-screens at stock settings, avoid treating the power-limit workaround as a fix and pursue hardware testing or warranty service.
RAM or the motherboard cannot be ruled out solely because the RTX 2060 works, but they become less likely when the failure follows the newer GPU and only appears under heavier graphics loads. If you want to eliminate memory as a variable, run a proper bootable memory test and temporarily use conservative DDR5 settings instead of the advertised overclocked profile. Testing the 5060 Ti in another known-good PC would provide the clearest answer.
The crashes happening in multiple games also make a single-game software bug unlikely, although checking the Windows and game logs for graphics-driver errors can still provide useful clues.
That makes sense. I had assumed the power supply was automatically ruled out because it is rated at 1000W, but the difference in behavior between the two cards is still concerning.