I recently upgraded from an RTX 3080 10GB to a Radeon RX 9070 XT. Games run normally at first, but after a few minutes—never more than about 20 minutes—they crash. It happens in Alan Wake 2, Dying Light 2, Cyberpunk 2077, and Dragon's Dogma 2, both with and without ray tracing at 1440p and 4K. I used DDU in Safe Mode before reinstalling the graphics drivers, and I also tried reverting from the newest driver to version 25.12.1, but the crashes continued. My system has a Ryzen 7 5800X, 32GB of DDR4 RAM, and an 850W Gigabyte P850GM power supply. How can I determine whether this is a driver, temperature, memory, power supply, or defective-GPU problem?
3 Answers
Check the GPU’s core and hotspot temperatures with HWiNFO64 while gaming. A bad cooler mount or uneven thermal paste could cause crashes after the card heats up. Since you already tried different driver versions and a clean installation, monitor temperatures and other sensor readings before deciding whether to return the card.
The exact power-supply model matters, not just the wattage. An 850W unit should generally provide enough power, but the Gigabyte P850GM has a poor reliability reputation. Check the cables and connections, avoid using a single daisy-chained PCIe cable if separate cables are available, and test with a known-good power supply if possible. If temperatures, memory, drivers, and power checks don’t reveal anything, an RMA or replacement GPU is reasonable.
Don’t overlook system memory. Unstable RAM can show up as game crashes, even if the GPU appears to be the obvious cause. Try temporarily disabling XMP/DOCP or running a memory test, then see whether the crashes continue at stock settings.
The problem only began after I installed the new graphics card, so I initially assumed the GPU was responsible, but I’ll still test the RAM and other components.
I already tried the newest driver and then reverted to version 25.12.1 after using DDU. I’ll monitor the hotspot and other temperatures with HWiNFO64.