My PC randomly restarts while gaming with no BSOD: the screen goes black and the system reboots. Event Viewer reports Kernel-Power Event 41 and bugcheck 278 (VIDEO_TDR_FAILURE).
The system uses a Ryzen 7 7800X3D with a long-stable Curve Optimizer setting, a Gigabyte RTX 5070 Ti Windforce, and a Corsair RM850e 850W power supply. The crashes began in Cyberpunk and Deadlock, then also occurred in Genshin and Overwatch, including during game startup. Updating to the newest Nvidia driver and disabling the Nvidia app's beta features did not help.
HWiNFO logs from two crashes show normal temperatures: the GPU core reached roughly 78–81°C, memory junction stayed around 76–78°C, and motherboard VRM temperatures remained near 42–51°C. Thermal throttling, VRAM exhaustion, obvious PCIe errors, and a DLSS or ray-tracing-specific issue seem unlikely.
What stood out was the GPU's Reliability Voltage performance-limit flag activating immediately before each crash. The 12V-2x6 connector reading normally sits near 11.9V, but under load it dropped as low as 11.105V and generally stayed between 11.1V and 11.6V. The second log started after a reboot and showed the voltage already low, around 11.25–11.5V. Just before that crash, GPU power fell from approximately 284W to 245W, the voltage-limit flag appeared, the readings briefly stopped changing, and the frame rate dropped to zero.
The card is being taken to a repair shop so they can inspect the 12V-2x6 connector and cable, test both ends, check the PCIe slot, and reseat the GPU. Has anyone seen this pattern before? Should the PSU itself be the main suspect, or could a damaged cable, connector, or graphics card cause this kind of voltage sag?
3 Answers
A cable can cause voltage drop if a terminal is loose, damaged, contaminated, or making poor contact, but the PSU is the larger suspect here. The fact that the voltage remains low after a reboot points to a persistent power-delivery problem rather than a one-off driver crash. I would avoid adapters or mixed modular cables and replace the cable or PSU if either shows any sign of damage.
The power supply is the first thing I would suspect. A bad PSU can produce instability under load even if its wattage rating looks more than sufficient. Have the shop test the system with a known-good PSU, preferably using that PSU's own native 12V-2x6 cable. Don't keep stressing the card if the connector is genuinely dropping below spec.
Test the GPU and PSU separately if possible. Put this graphics card in another known-stable computer, or try a different PSU in this one. That should quickly narrow it down to the power delivery path versus the card itself. Also have them inspect the connector for heat damage, loose terminals, discoloration, or poor contact, and verify that the plug is fully inserted.
It has since started rebooting even while browsing, and at one point it got stuck in a restart loop before showing blue and red error screens. That makes a general PSU or hardware problem seem more likely than a game-specific issue.