Sign in

stress tests pass, but an open-world rpg still dies right after loading a save. what am i missing?

crashdirectxgpu10 replies · 194 views · started
Reply
  1. Arjun
    @cache_loop

    This is driving me up the wall lmao an open-world RPG runs beautifully on my RTX 3080 until right after loading a save, then black screen, then the crash dialog:

    Fatal error

    Fatal error
The game has crashed and will now close.

DXGI_ERROR_DEVICE_REMOVED
The GPU device instance has been suspended.

    ive reinstalled the driver, verified game files and even turned off my overlay apps. benchmarks and other games are completely fine, which makes it even more confusing. I just want to finish one session without losing progress.

    for context, teh PC is otherwise rock solid. i can run stress tests for an hour without a single error :/ that's what makes this so confusing. thx in advance!!

    cry

    3
  2. Alba
    @anolan

    as a diagnostic you can raise teh GPU timeout: in regedit go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers, create a DWORD TdrDelay and set it to 8. Reboot. If teh crashes stop, you know it was a timeout issue, not hardware. Remove it again once youve found the real cause.

    4 0 helpful
  3. Nina Sato
    @nina_signal

    been there man make a backup before u touch the firmware, leraned that teh hard way. came here to say exactly this lmk how it goes

    4 2 helpful
  4. Camila
    @camila.park4

    i come back to this thread every week just for the drama :)

    1 4 helpful
  5. Ivan S.
    @ivans526
    in reply to @camila.park4

    ok so check ur power cabling. If your card needs two 8-pin connectors, use two separate cables from the PSU, not one cable with a daisy-chained pigtail. Transient spikes on a single cable can cause exactly this kinda sudden drop. lmk how it goes

    14 0 helpful
  6. Camila
    @camila.park4

    oof pro tip: write down what u change. i didnt and had to redo after from scratch 🤷 exact same thing happened to my desktop last month

    12 2 helpful
  7. Nina Ziegler
    @nina_kernel

    hmm idk everyone keeps saying this but nobody posts proof

    0 2 helpful
  8. Alba
    @anolan
    in reply to @nina_kernel

    yeahhh do a proper clean driver install: boot into Safe Mode, run DDU (Display Driver Uninstaller) to wipe the old driver, then install teh current stable driver with networking unplugged sooo Windows Update doesnt push its own version in between.

    1 1 helpful
  9. Esme Jansen
    @regexrider
    in reply to @anolan

    lol same thing happened to me try a small negative offset before anything else. In MSI Afterburner drop the core clock by 50 to 100 MHz and test again. Many factory-overclocked cards boost a bit too aggressivley in specific engines, and that shows up exactly like this.

    4 0 helpful
  10. Esme Jansen
    @regexrider

    hmmmm not sure about that id be careful with the TdrDelay tweak as a fix. it just hides teh timeout, and if something is actually unstable youll get freezes instead of crashes.

    lol

    6 0 helpful
  11. Arjun
    @cache_loop
    in reply to @regexrider

    @regexrider dude it was teh driver after all. DDU in Safe Mode, fresh install, no crashes since. I wasted a whole weekend on this, sooo hopefully this thread saves someone elses.

    6 2 helpful

Reply to this topic

Sign in or create an account to reply.