My RTX 3080 Founders Edition and Corsair SF750 worked perfectly for years, including during demanding games such as Cyberpunk 2077. Recently, though, the computer powers off after about 5–10 minutes of gaming. There is no blue screen or error message; Windows only reports Event 41 Kernel-Power after reboot.
I originally suspected the power supply, so it was replaced under warranty, but the replacement behaves the same way. I have also reseated the cables and memory, cleaned out the dust, and tried reducing the GPU power limit in the NVIDIA app. A 90% limit helped temporarily, but the crashes returned, and even 85% or 80% no longer prevents them.
During the latest crash, the display froze, the fans ramped up loudly for roughly 10 seconds, and then the system shut off. Just beforehand, the GPU was around 78°C at 98% usage and 275W, while the CPU was around 61°C, 22% usage, and 48W.
The system uses an ASUS ROG STRIX B560-I Gaming WiFi motherboard, an Intel i5-11400F, and 32GB of Corsair Vengeance LPX DDR4 memory. Could the graphics card be failing, or is there another likely cause?
3 Answers
A 750W SF unit should normally be sufficient for an RTX 3080 and this CPU, so the wattage rating alone is not the obvious problem. The fact that lowering the power limit helped for a while does suggest the card may be becoming unstable when it reaches higher transient loads. Try a strict frame-rate cap and a conservative undervolt; if it still shuts down in multiple demanding games, test the card in another system or with another known-good GPU.
Event 41 only confirms that Windows lost power unexpectedly; it does not identify the faulty component. Since the replacement PSU has the same symptom, I would focus on isolating the GPU rather than buying an even larger power supply. Also make sure the card is powered with separate PCIe cables where applicable, and test with an FPS limit to reduce power spikes.
The freeze followed by fans going to maximum speed and then a hard power-off is more concerning than the temperature readings. That pattern can happen when the GPU crashes at a hardware or driver level, although it is not absolute proof that the card is dead. Completely remove and reinstall the graphics driver, check every PCIe power connection, and see whether a different graphics card runs the same games normally.
I already reseated the cables and tried lowering the power limit, but I’ll test with a strict frame-rate cap and another GPU before replacing anything.