GPU experiences a sudden jump to 100% usage upon launching games, causing the entire system to freeze.
GPU experiences a sudden jump to 100% usage upon launching games, causing the entire system to freeze.
Hello, first of all: DxDiag + GPU-Z logs I'm running: 3 monitors (1920x1080, one of them 144 Hz G-Sync, rest 60 Hz) PSU: AeroCool gt-sg Series 1050W Graphics card: Asus ROG Strix 1080 Ti OC (didn't do the OC myself, bought the "OC-Version" CPU: i7 8700k @ 4,7 Ghz (OC'd myself) Mainoard: Z370 Extreme 4 5 HDDs, 2 SSDs Mouse: Corsair Ironclaw RGB Keyboard: Magic Eagle blue switches Audio interface: Scarlett 2i4 When I'm about to play games (The Division 2, Battlefield V e.g.) the system freezes when loading into them or shortly after. I could figure out that it might have to do with a spike regarding my GPU (it jumps to ~1V and 100% usage). The strange thing: everything worked fine for like 2 years. I didn't change anything, no new hardware or software. But 3 days ago, while gaming, the screens suddenly went black and the GPU fans ran at max speed. That happened a few more times when trying to play games (which wasn't successfull since that happened) but now it just changed to freezing the whole system. Another interesting thing: I randomly had the "nvlddmkm.sys stopped responding and has recovered" error after that first crash but somehow resolved it by something (I know it still occurred after getting new / old drivers). I already experimented with over- and undervolting my graphics card to no success (used Afterburner, EVGA Precision and Asus GPU Tweak II). Ofc I also reseated the card, switched PCIe-Slot and used other PCIe cables. Appreciate your help! I've been troubleshooting for the past two days and can definitely say that there's no problem with: Windows (reinstalled, updated) RAM (tested it, tried using only one stick at a time) CPU Overheating (GPU ~70°C under full load, CPU ~60°C) Drivers (updated everything, used old NVIDIA drivers) I tried a few more games that run on dx12 or dx11. When I enable dx11 it doesn't freeze but crashes with this error from windows: Application has been blocked from accessing graphics hardware I have no other graphics card or PSU to test, unfortunately: If I had to guess, I'd nail down the problem to either the PSU or GPU. But my 1080 Ti is only about 1,5 years old now, the PSU about 2,5 or 3. But I've heard that other people (especially with a 1080 Ti) have the same problem. Is this just and NVIDIA thing? Do you have any clues? EDIT: SOLVED. I even bought a new PSU today and the same error was still there. What solved it was the removal of the old (dry) thermal paste and I applied a new one. I've found that solution in a similar, age-old thread and it really worked. Really strange since my card didn't even overheat (60°C under full load)
I faced this problem with my 1070. I tried a registry change to give the driver extra time before it crashes. This worked on my 1080ti, but not on mine, so I used ddu and reinstalled the drivers again.
Thank you for your assistance. The regedit didn't work at all. The updated drivers seem to require a 5th reinstallation now.
I resolved the issue by replacing the old thermal paste and using a fresh one. I discovered this fix in a comparable, long-standing discussion, and it effectively resolved the problem. It's odd that my card didn't overheat even at 60°C under full load.
Hello from the distant future! I faced a similar problem and after seeing your repair method, I went to the GPU at position "50." It was almost burned by touch. Essentially, the heatsink wasn't properly connecting to the GPU die. In my case, I only had two screws since I was testing or preparing for disassembly. I added more paste to fill the gaps and installed the remaining screws, and it's now working.