Dying 2080 Super?
Dying 2080 Super?
Hey there, I'm facing some issues with my 2-year-old GPU (Gigabyte RX 2080 Super Windforce OC 8G). Under heavy use—like playing most of Final Fantasy 14 at 1440p—I sometimes see random blackouts and the fans crank up to maximum. The system becomes unresponsive until I power it off or shut down the PSU. I can still hear voice chat and game sounds during these moments. My thought is it's probably related to overheating. I've tried opening the case, changing the thermal paste (which dropped the GPU temperature by about 10°C under load), reinstalling drivers, and using different cables. Still nothing fixed the problem. I haven't found any solutions online for this specific card, except swapping thermal pads. I'm really stuck and hoping someone can help. Replacing it would be a big expense right now—thanks for your support! Covid thanks! More details: I own Windows 10 Home x64, everything new, built myself two years ago. It worked perfectly until recently with the Ryzen 7 2700 GTX RX 2080 Super Windforce OC 8G from ASUS Prime X470 Pro and Corsair RM750X.
You might consider setting up MSI afterburner and adjusting your power limits to their minimum possible values (remember to save changes by dragging the toggle left). This helps check your card's VRM performance. If it stabilizes once you lower the voltage, the issue likely lies with the VRM. Your graphics card may not perform optimally but will continue running until you locate a replacement. Alternatively, you can run DDU to fully reset your display drivers and reinstall them after a restart. Make sure to download the updated drivers before turning off your network adapter, as Windows might attempt to install them automatically. Review the Event Log > Windows Logs > System > Filter Current Log > Search for critical events near when your computer crashed to get more details about the problem.
These crashes often stem from an unstable OC setup. Your card includes a default OC, which could be contributing to the problems. I concur with @Founders—try using Afterburner, reduce your power limit and/or core clock, and observe if it improves performance.
I've experimented with all the suggestions you gave: lowering the power limit in afterburner helped restore playability. However, I'm seeing odd memory clock readings (refer to the attached screenshot). It seems the correct frequency should be near 1900 MHz, but CPUID HWMonitor shows 1938 MHz and GPU-Z reports 1938 MHz as well. DDU didn't help either, with persistent crashes at full power limits and strange readings. Windows logs only capture a sudden power loss after I shut down the PC following a crash. It's unclear how to manage these clocks—I'm hesitant to try further without more guidance. Thanks for all your support! You've definitely saved my evenings :3
Hello, I realize I arrived late to the discussion. I was checking if this resolved your problem smoothly. I’ve followed all the steps GIGABYTE suggested. The power going to the GPU bus rail and PCI-E lights is fine, which means there’s enough power available. Still, the GPU keeps failing due to a Kernel-Power issue, and I’m unsure what to do next. It seems unnecessary to restrict the card’s capabilities for it to work properly. Thanks.