My PC sometimes suddenly shuts down while playing PUBG, without showing an error or warning. The first time, it powered off after about an hour of play and then shut down again when I tried to launch the game. After that, it would not start properly: the motherboard showed a red DRAM light, the CPU fan did not spin, and only the other case fans were running. After being unplugged overnight, it started normally the next morning.
I have since limited the RTX 3080 to 85% power in MSI Afterburner and capped PUBG at 230 FPS. The GPU is connected with a daisy-chained PCIe power cable. While playing at 1080p, HWMonitor showed roughly 220W GPU power usage and temperatures around 60–63°C. I played for several more hours before the PC crashed again, and this time I saw black spots or pixel artifacts on the screen. Everything looked normal after restarting.
My system has a Ryzen 5 7600X3D, 16GB of DDR5 RAM, an RTX 3080, and a 750W PowerSpec PSU. The CPU had been thermal throttling for about two weeks before I fixed it, although current temperatures appear normal. Could a power spike from the RTX 3080 or the daisy-chained cable be overwhelming the PSU even if monitoring shows only around 220W during gameplay? I no longer have the original PSU cables, so I also need to know whether replacement PCIe cables must be specifically made for this PSU. What should I test first, and is it safe to keep gaming on this system?
1 Answer
Overheating is one possibility, so check the CPU heatsink and GPU for dust and confirm temperatures while the system is under load. A 60–63°C GPU reading is not overheating, though, and monitoring software may not capture short power spikes. Since the PC is shutting off completely and you saw display artifacts, also test the PSU, use separate PCIe power leads if the PSU supports them, and avoid using cables from another PSU model. Replacement modular cables must match the exact PSU and pinout; do not buy a generic cable just because the connector fits.
The CPU, motherboard, and memory were installed only about three weeks ago, and the current readings look normal. I will still inspect the heatsink and check temperatures again under sustained load.