My PC was stable while playing Marvel Rivals for several hours, but after waking it from hibernation the game began crashing. I removed the riser cable, reset the BIOS and Afterburner settings to stock, and even tried underclocking the GPU, but the crashes continued. The main monitor is connected to an Asus TUF RTX 3080 over DisplayPort, while a second monitor uses the integrated graphics. When the game crashes, the main display disconnects and only reconnects after restarting the PC. 3DMark Time Spy can complete both GPU tests and the CPU test, but OCCT's VRAM test reproduces the crash and reports numerous errors. GPU temperatures are around 65–70°C, with memory temperatures around 70–75°C. I also performed a clean NVIDIA driver reinstall with DDU recently. Is the graphics card likely failing, or are there other steps worth trying?
1 Answer
The temperatures look normal, so overheating is not the obvious cause. However, repeatable errors during an OCCT VRAM test are concerning and can point to defective or unstable video memory, even if a benchmark like Time Spy completes successfully. Since you have already reset the tuning settings and used DDU, try the VRAM test with a different DisplayPort cable, reseat the card and its power connectors, and test with a clean driver installation again. If the errors follow the card, it is probably failing and may need warranty service or replacement.
I already ran DDU and removed the GPU tuning. The system was stable the previous night, but the crashes started after waking from hibernation, so I’m going to retest the card and connections before assuming it is permanently dead.