My PC randomly restarts while playing demanding games. Event Viewer only shows the generic Kernel-Power 41 entry, with no WHEA errors or other useful information.
The system has a Ryzen 7 7700, Gigabyte RTX 5070 Windforce OC SFF 12 GB, Gigabyte B650 EAGLE AX motherboard, 32 GB of DDR5-5600 CL36 memory, a DeepCool PL750D 750 W ATX 3.1 power supply, a Lexar NM790 1 TB SSD, and an Endorfy Fera 5 Dual Fan cooler.
The computer has already been returned to the service center twice. The GPU was replaced during the first repair, but the restarts continued. During the second inspection, the issue did not happen at all, yet it returned as soon as I started using the PC at home.
I also replaced the PSU with another unit of the exact same model. Initially, the PC survived about five hours of demanding 4K gaming, but it later restarted again. Previously, the restart could happen after only five minutes, or sometimes after 30–60 minutes. Limiting GPU power to 95% seemed to improve things somewhat, occasionally allowing around 90 minutes of play.
The system passes normal stress tests, including a 24-hour test, without crashing. The problem only seems to occur during demanding games, possibly because of rapid changes in power draw. I have also noticed that moving the PC can sometimes cause the GPU to lose display output and trigger the motherboard's red GPU diagnostic light; reseating the graphics card temporarily fixes that. This happened with both GPUs. What should I check next?
1 Answer
The 95% power limit is a useful clue. It may be reducing brief power spikes or rapid load changes that a steady stress test does not reproduce. Try testing with the GPU limited much further, such as 70%, and see whether the restarts stop. If that makes the system stable, inspect the entire GPU power path: remove and firmly reseat the card, check that the card is fully supported and not flexing in the slot, and inspect the power connector and cable for looseness or damage. Use a dedicated PSU cable rather than a daisy-chained one, and make sure the plug is inserted completely.
The fact that moving the case can cause GPU signal loss and the GPU debug LED to appear is especially concerning. That points toward an intermittent connection, card movement, PCIe slot issue, or cable/connector problem rather than a software error. Since the same symptom occurred with both graphics cards, carefully check the motherboard slot and mounting alignment as well. Kernel-Power 41 only records that the system lost power or reset unexpectedly; it does not identify the failed component.
That makes sense. I had mainly been focusing on the stress-test results, but the improvement from lowering the power limit and the display loss after moving the case do suggest an intermittent GPU connection or power-delivery problem. I’ll test with a much lower limit and inspect the slot, mounting, and connector carefully.