I recently upgraded to a Corsair HX1200i SHIFT power supply and an ASUS ROG Astral GeForce RTX 5090 OC. A few times while gaming—usually when sitting in a static menu—the system's power draw and fan noise suddenly spike, the displays lose signal, and the screen stays black while audio continues. The computer does not appear to shut down, but the reset button does not restore video. I have to hold the power button, switch the PSU off, and then start the system again before display output returns.
The GPU temperature seemed to be around 80°C, while my Ryzen 7 9800X3D package temperature reached 100°C even with Eco Mode set to 65 W. The problem has only happened in menus, despite having the frame rate capped at 116 FPS through the NVIDIA App. I'm wondering whether some menus ignore that cap and drive the GPU or CPU much harder than expected.
Windows 11 also makes me reset my PIN after these incidents. I'm using both CPU power connectors on my ASUS ROG STRIX X870E-E GAMING WIFI, along with a ThermalProtect PCIe 5.1 600 W 12V-2x6 cable. Could the cable's protection mechanism be cutting power to the GPU, or is this more likely a driver, cooling, or frame-rate issue?
3 Answers
The symptoms sound consistent with the ThermalProtect cable reacting to an overheating or poorly seated GPU power connection: the GPU loses display output while the rest of the system continues running. The 80°C reading is the GPU core temperature, so it does not rule out a problem at the connector. Shut the system down, unplug it, and carefully inspect and reseat the 12V-2x6 cable. Look for deformation, discoloration, or melting, and make sure it is fully inserted without a sharp bend near the plug. The cable or GPU connector may need to be replaced if there is any damage.
The 9800X3D reaching 100°C is a separate issue that should also be investigated. Check the cooler mounting, fan operation, airflow, and thermal paste. A Dark Rock 4 may be marginal for the platform, especially with a very hot GPU inside the case. If you remove the cooler, warming the system first and twisting the cooler gently can help prevent the paste from acting like adhesive.
Check the individual core temperatures, actual menu FPS, GPU power draw, and CPU/GPU utilization instead of relying only on the frame-rate limit. Some game menus can ignore an external cap or briefly render at extremely high FPS, causing a sudden power spike. Try enabling an in-game limiter or a driver-level limiter with a lower test value, such as 60 FPS, and see whether the behavior changes.
I would also test a clean installation of a different NVIDIA driver version. Since the CPU temperature is unusually high even at a low power limit, verify the cooler installation and airflow before assuming the black screen is purely a GPU problem.
A black screen with continuing audio usually means the system is still operating but the graphics card or its driver has stopped responding. I would prioritize the physical 12V-2x6 connection and GPU power delivery rather than replacing the PSU based only on brand opinions. Avoid repeatedly running the system if the connector shows heat damage. After checking the cable, use monitoring software or logging to record GPU hotspot temperature, connector-related warnings, power draw, CPU core/package temperatures, and the real FPS while entering a menu.
I didn’t see any shader compilation happening. The power spike occurs whenever I switch to certain menus, and the fans ramp up while my UPS makes a loud click. I haven’t tested another NVIDIA driver yet. The cooler is a be quiet! Dark Rock 4, and it does seem to run hotter than it used to.