Whenever I launch a game, the entire computer hard-freezes after one to three minutes and requires a forced restart. After rebooting, the system no longer detects my Radeon RX 5700 and switches to the integrated graphics. I can get the card working again only after repairing the graphics driver installation. The GPU is nearly seven years old. I've already updated the drivers, performed a clean removal and reinstall with DDU, and reseated the card. Temperatures were around 60°C when the crashes occurred, so overheating seems unlikely. What else should I test to determine whether the GPU is dying?
3 Answers
The CPU can cause similar freezes if it overheats, so monitor its temperature as well as the GPU while gaming. A Ryzen 5 3600 reaching only about 60°C at the time of the crash would make CPU overheating unlikely, but checking temperatures, power delivery, and system stability together is still worthwhile.
A hard lock followed by the GPU disappearing after reboot can point to a failing graphics card, unstable power delivery, or a motherboard PCIe issue rather than just a driver problem. Test with another known-good power supply or GPU if you can, check Windows Event Viewer for hardware or display-driver errors, and run a GPU stress test briefly. If the RX 5700 fails in another computer or under multiple power supplies, replacement is probably the practical fix.
Since the card is several years old, it’s worth cleaning the fans and heatsink and checking that the fans ramp up properly under load. Repasting and replacing thermal pads could help if the cooler has degraded, although the reported 60°C core temperature makes a basic overheating problem less likely. Also inspect the PCIe power connectors and try the card in another system if possible.
The GPU and CPU were both around 60°C during the crash, so I’ll focus next on checking the power supply, PCIe connections, and whether the card is detected in another system.