My PC suddenly black-screens and restarts while playing modded Cyberpunk 2077. The same thing happens after a few minutes in Heavenly Benchmark 4.0, so it doesn't seem limited to one game. Lowering the GPU power limit to 80% makes the system more stable, but it can still crash under enough load. CPU, memory, and other OCCT tests have completed without crashes, and each restart is logged as Kernel-Power Event 41. The system has an RTX 3080 Ti Founders Edition and a Corsair RM750x power supply, both used for roughly four to five years. Could this be a failing GPU or power supply, and what tests should I run next?
2 Answers
The most useful next step would be to test the graphics card in another known-good PC, or test a different GPU in your system. That can separate a card problem from a power-supply, motherboard, or cabling problem. Also check that the 3080 Ti uses separate PCIe power cables rather than one daisy-chained cable. Since reducing the power limit helps, the PSU or the GPU’s power-delivery circuitry could both be involved, and Kernel-Power Event 41 only confirms that Windows lost power unexpectedly—it doesn’t identify the failing component.
The GPU itself may need fresh thermal paste or thermal pads, especially after several years of use. A degraded thermal interface can cause instability when the card is heavily loaded. Check GPU core, hotspot, and memory temperatures with a hardware monitor while running a benchmark. If the hotspot temperature is dramatically higher than the core temperature, or temperatures climb rapidly before the crash, repasting may be worth investigating. Opening the card will usually void or complicate warranty coverage, so only do it if you’re comfortable servicing it.
I had considered repasting, but I’m not sure how to confirm that it’s necessary. Would abnormal hotspot temperatures be the main thing to look for before opening the card?