Long story short:
Starfield is crashing to desktop. Playtime can range from 1 minute to roughly an hour, but it does inevitably crash, but not to any one thing specifically. No located fix is working after days of searching and implementing. There is a nvlddmkm error event that triggers on each crash, which seems to be related to possible hardware issues. I tested my RAM and it seems to have passed several tests, but I am having trouble truly testing VRAM. I did noticed that every crash report has the same memory address at the point of failure (0x000000000401FE4D).
The only other game I really play that is graphically advanced is cyberpunk 2077, which is also crashing to desktop. However, highly stress-testing the GPU with various tools for dozens of minutes at 100% usage doesn’t crash.
I was looking into blacklisting memory addresses and located talk of ECC and dynamic page retirement that seems to tell the driver to reserve the faulty addresses and to skip them when processing. From what I’ve located though I don’t think GeForce devices have this available
I was wondering if anyone knows of any possible way of doing that manually with a GeForce driver? As far as I can tell, it’s just the one address, and I would really rather not RMA my laptop if it’s just one address on VRAM.
Perhaps my assessment of the issue is incorrect, but after everything I’ve attempted, it seems to be one of the few remaining possibilities. I appreciate any insight.

  • BigMan7o0@alien.topBEnglish
    1·
    1 year ago

    I want to preface this by saying I could very well could be wrong, but to the best of my knowledge event viewer memory addresses are system memory not vram.

    If I am correct about that, ive had so so SO many gpu driver crashes related to unstable ram so it’s possible either A: you’ve applied an oc that’s not stable or B: your memory is straight up faulty

    In my experience this can be the case even while passing memory tests. ive had mem ocs that are memtest and aida64 stable but when I game my gpu drivers absolutely WILL crash at some point.

    If you have control over your memory speed/settings I would recommend either turning the speed down a notch, or try turning up the dram voltage by 0.010v until it’s stable or you reach 1.4v (you can go higher but it might get too toasty in a laptop. If you push voltage farther than that pay close attention to mem temps)

    If that doesn’t solve it, it could be faulty ram. I would definitely exhaust every other option before RMA though.

    If your memory is SODIMM and you’re comfortable opening the laptop you could try the ram sticks individually as well