My RTX 2060 Super began randomly losing display output during gaming shortly after I moved apartments. The PC usually remains running: fans, lighting, audio, and Caps Lock continue to work, but DisplayPort shows "No Signal" until I force a restart. I have also seen occasional white or garbled output, and HDMI is now completely dead on every cable and port.
The system uses a Ryzen 5 3600X, 32GB of DDR4, an MSI B450 Tomahawk Max, the RTX 2060 Super, and a 750W power supply. The card and most of the system were purchased in 2020, while the RAM was added in 2023. The computer saw relatively little use from 2023 to 2025 and had been stable before the move.
I have confirmed that the GPU is running at stock settings, tried different cables and outputs, performed several clean driver removals and reinstalls, cleared shader and DirectX caches, forced the fans to maximum, and tested memory underclocks and core undervolting. Nothing prevented the crashes. They occur at varying temperatures and loads, including one incident while the computer was idle.
Hardware logging shows NVIDIA sensor values freezing while Windows graphics and VRAM counters continue updating. The GPU does not disappear from the PCIe bus, and VRAM usage stayed well below the card's capacity. Event Viewer shows no WHEA or PCIe errors and no successful driver-recovery events. One crash produced a 0x133 watchdog BSOD identifying nvlddmkm.sys in an interrupt-service-routine path, with a DPC queue execution timeout.
Because the problem started immediately after moving, I also checked the electrical environment. A socket had roughly a one-volt neutral-to-earth difference, but the issue continued from another outlet, and mains voltage appeared steady around 225V. I have reseated the GPU and changed the PSU connection as well.
Does this pattern usually indicate failing GPU hardware, or could it still be a driver or power-quality problem? I am planning to test the card in another complete system, but I would appreciate any similar experiences or additional physical tests worth trying.
3 Answers
The physical checks still worth doing are reseating the card and its power connector, inspecting the connector and PCB for scorching or damaged components, trying the other PCIe slot if the board allows it, and checking that the PSU cable is firmly seated. Also avoid repeatedly changing drivers as a diagnostic method: a reinstall can temporarily alter initialization or power-state behavior without repairing a hardware fault. A card that loses multiple outputs and fails under both load and idle is unlikely to be fixed by cache cleanup or an ordinary driver update.
The strongest evidence points to the graphics card or its onboard power and display circuitry rather than a normal Windows driver problem. A driver name in a 0x133 crash does not prove the driver itself is defective; it is often the component that was waiting on, or servicing, hardware when the watchdog expired. The dead HDMI output, corrupted frames, frozen NVIDIA telemetry, and progressively shorter periods between failures are especially concerning for a deteriorating card.
At this point I would stop spending much time on reinstalls and focus on swapping physical components. Test a known-good graphics card in the current PC, then test the RTX 2060 Super in a different system with a reliable power supply and display cable. If the 2060 reproduces the failure elsewhere, especially with HDMI still dead, it is effectively confirmed as faulty. If another GPU also fails in this machine, investigate the motherboard PCIe slot, PSU, power connectors, and the case or outlet.
A different outlet or removing the power strip can rule out a simple connection issue, but it cannot prove that brief electrical transients caused the original damage. Since the card worked for a while after changing the outlet and then worsened, I would treat the timing as circumstantial rather than evidence of an ongoing mains problem.
That matches my suspicion. I am going to test the card in another system, and I will also test a known-good GPU in this one so the card and the rest of the system can be separated.