PC power cycling occurs unpredictably while switching the GPU's TDP during a stress test.
PC power cycling occurs unpredictably while switching the GPU's TDP during a stress test.
Over the past two weeks my computer has begun shutting down unexpectedly, without any blue screen of death. It appears to power off abruptly, as if you simply cut the power and then restarted automatically. Occasionally it occurs twice daily, other times it runs smoothly for several days without issues. These interruptions typically happen while I was streaming videos on YouTube, not during gameplay. The first error logged in the event record at the time of failure was a kernel power 41 issue. During troubleshooting, I've observed that adjusting my GPU's thermal design power (TDP) during stress tests often leads to crashes when the TDP jumps from around 58% (with performance throttled as per gpu-z) to higher values like 80%. Conversely, lowering it back to 60% doesn't prevent the issue, but going up to 80% can trigger a failure. While the system remains stable at around 100°C during GPU stress without problems—even when the GPU is under load—I've noticed that repeatedly changing the TDP during stress tests eventually leads to shutdowns. I also ran Prime95 to test my CPU and it maintained normal temperatures, even while stressing the GPU without altering TDP. Additionally, I checked my boot drive and RAM using the command prompt and found no errors. There are no loose cables detected. It's worth noting that my PC is likely 8-9 years old. My hypothesis is that increasing the TDP during stress tests might destabilize the PSU, causing it to supply insufficient voltage to other components, resulting in shutdowns. This seems more likely when the system is idle while connected to the internet, though I can't rule out a failing power supply entirely. Specifications: OS – Windows 10 64-bit CPU – Intel i7 4790K GPU – Gigabyte RTX 2070 Super PSU – Seasonic 650W SSR650-RT RAM – 4x 8GB Kingston HyperX Fury