PC fails to start correctly following a GPU overvolt error.
PC fails to start correctly following a GPU overvolt error.
CK OKAY so in short I didn’t go wrong...
I was using an MSI R9 390 with a RM 550 Corsair board. The setup was tight because the RM series isn’t very reliable (like CrapXon), so I lowered the power and voltage settings on my GPU to keep things stable until I could upgrade. After switching to an RM 750X, I restored everything.
But during the upgrade, I accidentally set the core voltage too high (+40 instead of +0). I realized it was too late—just before accepting the changes in afterburner—and the system crashed immediately. Now only Safe Mode on the 390 boots and works, or normally on my iGPU.
I tried uninstalling Afterburner and disabling OC at startup from both states, but nothing worked. I can still reach the login screen, but it shows a rainbow light show before going completely black.
I’m convinced the card is damaged and have already started the RMA process. But I have two doubts:
1. Why does the card still show something during early boot or Safe Mode? If it’s broken, would this still happen? Could it be a total driver failure?
2. Is 40mV really enough to cause damage, or is it just a rare silicon issue?
I don't believe +40 is excessively high and it's unlikely to cause harm on a desktop. For example, my 980 runs at 1008mV on the desktop and adding +40 brings it to +44, reaching 1052mV. This is significantly lower than the voltage the card uses for 3D applications, which I've observed as up to 1212mV.
I would attempt this change, boot into Safe Mode or with integrated graphics, and completely remove Afterburner. When asked about preserving settings, confirm a "No." Then restart in normal mode. If the issue persists, uninstall drivers via DDU and reinstall them. The first step should likely resolve the problem, but I'm confident both actions are needed.
if possible, you can test the driver theory by starting from a Linux USB drive. download an iso (puppy for a small one, mint for a bigger one, suggestions only) and use rufus to convert it into a bootable USB. restart, and if Linux boots successfully then the v-card is probably fine. if not, you'll need more info.
I don't believe +40 is excessively high and it's unlikely to cause harm on a desktop. For example, my 980 runs at 1008mV on the desktop and adding +40 brings it to +44, reaching 1052mV. This is significantly lower than the voltage the card uses for 3D applications, which I've observed as up to 1212mV.
I would attempt this change, boot into Safe Mode or with integrated graphics, and completely remove Afterburner. When asked about preserving settings, confirm a "No." Then restart in normal mode. If the issue persists, uninstall drivers via DDU and reinstall them. The first step should likely resolve the problem, but I'm confident both actions are needed.
I tried the Afterburner solution already-no luck. DDU was my next port of call, will try tomorrow when it's not midnight. I remain cautiously optimistic, and idk why I didn't do that sooner. I guess I was preoccupied imagining all the excuses eBuyer could throw up about how "the card was tested but works perfectly" or "a direct replacement cannot be issued", as has happened in the past. Will try the Linux fix if still no luck. Good god, you know you're desperate if your last hope is Linux.
It seems the screen resolution is likely higher than what the monitor supports. Bios and safe mode appear at reduced sizes.
I appreciate your perspective, R_1, though it seems the card will actually show the lock screen wallpaper on Win10 for a short time before the sh*t starts. I should have been more precise about that. I’ll look for any proof to back up what you’re saying, but ultimately it might still point to a driver problem. The setting in Control Panel did seem correct at first glance.
Think in text more often—it feels like it fits what you expect.
While I’m here, maybe I should share a spec sheet.
Intel Core i5 4690k (I haven’t managed to overclock it, but it still runs at 3.9ghz)
MSI Z97A Gaming 7
Kingston HyperX low profile 1866mhz DDR3 2x8 GB (I think it’s called Vengeance, hard to remember right now. It works with XMP.)
Cooler Master Hyper 412S cooler
Fractal Design Define R5, with an extra front 140mm fan (that should provide enough cooling)
Samsung SSD 840 Pro 256GB
WD Blue 7200RPM 1TB HDD
Corsair RM750X.
IT WORKS!! It turned out all I had to do was DDU the drivers and then reinstall. I've been running Unigine Valley on Extreme HD for an hour, and it's completely stable with no impact on performance. DDU is quickly becoming my favorite tool. Also, it feels great to finally have my PC fully finished and upgraded to my preferences—another three years until everything needs replacement again. Thanks for the help, everyone!
Good to hear everything is working now. I wasn't expecting any damage from such a minor voltage shift. It's odd you didn't experience problems even with heavy 3D usage. I've faced similar crashes during overclocking, usually after using Afterburner, though sometimes the driver is the culprit. Corruption happens when it crashes, and only reinstalling either fixes the issue.