My PC suddenly loses all power while gaming. There is no game crash or blue screen—the system instantly turns off and must be started again with the power button. The hardware is a Ryzen 3 3100, RX 6600, 16GB DDR4, Gigabyte A520M DS3H motherboard, Corsair CX650 PSU, and a 500W 220V AVR. The PSU and AVR are about a year old, while the motherboard and CPU are roughly six years old.
Event Viewer only shows Kernel-Power 41 with BugcheckCode 0, no WHEA errors, and no obvious display or AMD driver errors immediately beforehand. In RDR2, the GPU reached about 98% utilization, 67°C, and 100W while the CPU was around 55°C. The PC shut down after five minutes. After reseating the motherboard, CPU, GPU power cables, and RAM, it lasted about 31 minutes before shutting down again.
Reducing the GPU power limit by 6% allowed RDR2 to run for an hour, with the GPU using around 94W and staying near 55°C. However, the system later shut down while playing League of Legends, where GPU usage was only around 40–50% and power draw was roughly 30–50W. CPU and GPU temperatures appear normal, so overheating does not seem obvious.
The issue began shortly after I updated the AMD graphics driver, cleaned the PC, and repasted the CPU. I have also noticed corrosion around some motherboard rear I/O connectors and what may be corrosion on the PSU fan grille. When the shutdown happens, the AVR makes a clicking sound.
I am considering reseating the GPU, trying the PC directly from a wall outlet or with a basic surge protector, and testing with a known-good PSU. Could the AVR, PSU, motherboard corrosion, or cooler mounting after the repaste cause this kind of instantaneous shutdown? What should I test next?
5 Answers
Inspect the corrosion closely. Surface discoloration on an external metal bracket may be harmless, but corrosion on motherboard components, connectors, or traces can create intermittent faults. Power the system off, disconnect it completely, and check for moisture, residue, damaged contacts, or burnt areas. If corrosion is substantial, replacing the motherboard is safer than trying to clean and trust it.
The reported temperatures are not high enough to explain the shutdown, and a CPU repaste problem would usually show rapidly rising temperatures or throttling. Still, verify that the CPU cooler is firmly mounted, the fan is spinning, and no cable was disturbed during cleaning. I would not repaste the GPU as a first step, especially since the card is relatively new.
Test one variable at a time: first bypass the AVR, then run a separate CPU-only and GPU-only stress test while monitoring behavior. Reseat the GPU and inspect its power connector, but avoid repeatedly stressing the machine if the AVR is clicking or the PSU and motherboard show corrosion. If the shutdown continues directly from stable wall power, test with a verified PSU; if it still happens with that PSU, the motherboard or GPU becomes more likely.
The strongest clues point to power delivery rather than a normal software crash. Kernel-Power 41 only means Windows noticed that power disappeared unexpectedly; it does not identify the failed component. Since the AVR clicks at the exact time of the shutdown, test the computer without that AVR if the wall outlet is known to be safe. A faulty or overloaded AVR can disconnect its output. Do not keep using hardware that shows significant corrosion or signs of moisture damage.
A known-good PSU is the most useful next test. Even though a CX650 has enough rated wattage for this system, an aging or failing unit can shut down instantly under changing GPU loads without leaving a useful Windows error. Use the replacement PSU's own PCIe cables rather than mixing modular cables, and inspect the GPU connector, motherboard power sockets, and PSU cables for heat damage or looseness.
The reduced GPU power limit may simply be lowering transient power spikes enough to keep the PSU or AVR from tripping. That does not prove the graphics card is defective.