My used GTX 1070 crashes during games, even after I removed and reinstalled the graphics drivers, disabled hardware acceleration, switched to a balanced power plan, underclocked the card, and limited its power target. During a crash, the GPU either becomes unusually loud or the display cuts out completely with a "No source detected, entering power saving mode shortly" message. The computer then has to be powered off at the wall, which has already caused Windows file corruption once. The card may have been used for mining before I bought it, and I am considering replacing the thermal paste. Windows only records an unexpected shutdown. I have reseated the card and updated the chipset drivers, but the system works normally when I install my older GPU, so the GTX 1070 seems to be the problem. What else can I test or repair? My system has a GTX 1070, an Intel i3-10105F, and 16 GB of DDR4-2666 memory.
1 Answer
Because the older GPU works normally in the same computer, the GTX 1070 is the main suspect. Still, check the power supply, PCIe power connector, and cable first, especially if the failure only happens when the card is under load. Monitor the GPU core and hotspot temperatures with HWiNFO or GPU-Z. Replacing the thermal paste could improve the core temperature, but worn thermal pads may still be allowing the VRAM or VRM components to overheat. Test the card at completely stock settings and verify that its VBIOS is appropriate. If it continues to black-screen after clean drivers, stable power, and normal temperatures, it may have failing VRAM, VRM circuitry, or another hardware fault.
Thanks, I’ll check the power connections, temperatures, and drivers first, then test it at stock settings.
Also make sure the driver installation is genuinely clean and check whether the card has a newer stable driver available. That is worth trying before assuming the hardware is beyond repair.