My RX 7900 XTX worked normally for nearly three years, but it began crashing last November and has gradually worsened. Now any game renders roughly one frame before the entire PC hard-resets and boots back into Windows.
FurMark can run for 18 minutes at full power without a crash, with the GPU reaching about 63°C, so this does not appear to be a temperature problem. I have already replaced the power supply, replaced the PCIe riser cable, tested both PCIe Gen 3 and Gen 4 settings, performed several DDU driver cleanups, reinstalled Windows, checked system files with SFC and DISM, and confirmed that my RAM is running at stock settings. My drives also test healthy, and repasting the card did not change anything.
The system has a Ryzen 9 7900X3D, 32 GB of RAM, an MSI X670E Carbon motherboard, a 1200 W PSU, and Windows 11. Does a card surviving a steady FurMark load but crashing immediately when a game begins rendering point to a particular fault? I am planning to test the GPU in another PC, but I would appreciate any other likely causes to check.
1 Answer
A failing power-delivery path can behave exactly like this. FurMark is a fairly steady load, while games constantly change GPU clocks and power draw, creating fast transient spikes that can trip a weak or degrading PSU, cable, connector, or the card's own power circuitry. Testing the GPU in a second system is a good next step. If it crashes there too, the card is the leading suspect; if it works normally, investigate the motherboard, cabling, and power delivery in the original PC. If the PSU is multi-rail, also make sure the GPU's power connectors are using separate appropriate cables rather than one daisy-chained lead.
I initially suspected the PSU too, but it was replaced with a new unit only six months ago. The replacement helped for about two months before the crashes gradually returned, and they have become much more frequent over the last couple of weeks.