This well-known "nvlddmkm.sys" issue is really taking a toll on your mental state!
This well-known "nvlddmkm.sys" issue is really taking a toll on your mental state!
Evening all,
I need to vent a bit and share some background so you can skip ahead if you wish. Feel free to jump ahead—I'll provide the full system specifics along with the troubleshooting steps and links to my minidump files.
As for the context:
Over the past weeks, I've been gathering new components to upgrade or downsize my case into an MFF Lian Li A3 build. I planned to upgrade to AM5 using some existing parts but also added several brand new ones.
Here’s what I have so far:
- CPU: 7800x3d (new)
- Motherboard: Asrock B650M Pro RS (new)
- GPU: 3090 FE (existing since launch)
- RAM: CORSAIR VENGENCE DDR5 32GB (2x16GB, DDR45, 6000MHZ CL30)
- NVMe: SAMSUNG 980 2TB (from previous system, but with a fresh Windows install)
- Power Supply: Thermatake Toughpower SFX1000 ATX 3.1 Platinum (new)
The problem:
I keep experiencing black screens that sometimes last an hour or around 5 minutes during booting, without any BSOD. Occasionally it appears in Event Viewer and sometimes as a nvlddmkm.sys error, with a bug report pointing to the same issue. This error is well documented, and I've spent several days searching forums, Reddit, and videos trying to find a solution, but nothing seems to work.
I've followed all the standard troubleshooting steps: running DDU in safe mode, updating drivers, disabling auto driver updates during installation, changing power settings, granting full admin rights on the actual file in sys32. Despite these efforts, it still doesn’t resolve. I’ve also adjusted the CPU overclocking and cleaned up settings in MSI afterburner, even slightly lowering the 3090’s clock speed. However, it remains unresolved.
After a thorough clean Windows installation, I haven’t found any other issues. I’m leaning toward hardware problems now.
Details:
- PSU – Initially suspected, but replacing didn’t fix the issue. Used a Corsair HX1000 from previous systems, which still triggered the problem.
- Motherboard – Tested both this model and a Gigabyte board; it still fails.
- NVMe – Still encounters the same error with both fresh installs and as a clone of another system drive.
- RAM – No issues detected even when testing one stick at a time.
From a hardware standpoint, the 3090 works perfectly in my older AM4 system, so the problem isn’t there. The PSU seems unrelated, the motherboard hasn’t been ruled out yet, and NVMe is still a concern.
RAM and NVMe have been tested thoroughly, and I’m ready to try additional RAM and another NVMe drive tomorrow. I’m still stuck and would appreciate your help in reviewing my dump files for any overlooked issues.
Also, I’ve tried standard fixes like disabling PBO, turning off the GPU, running at stock speeds, updating BIOS, and even rebuilding the system multiple times. Everything seems to be seated and connected properly.
Attached are today’s compressed Mini Dump files for your reference:
Compressed Mini Dump.zip
Thank you!
When you start Safe Mode, do the problems show up there too? Or are they only visible when you boot normally into the OS?
Remove the GPU from the system, connect the monitor to MoBo, and check if the issues disappear.
In a few uncommon cases, an older GPU might have problems in a newer system.
To verify, use a known working GPU; preferably an RTX 40-series, and see whether the system functions properly.
Safe mode didn’t cause problems during driver removals, though I haven’t confirmed it long enough to be certain. I’ll include it in my upcoming tests.
Running it with MOBO for setup initially worked without issues, but again, maybe not thoroughly tested.
The older GPU is more challenging to test since I don’t have another one available.
EDIT: Just placed an order for a 4080 super—didn’t want to upgrade my GPU as it’s really not necessary, but I’m desperate to get this working. I’ve spent too much time and money on it. Plus, my old rig includes everything a friend is buying, like case, fans, PSU, cables, CPU, RAM, so it could handle the 3090 too. Thanks for the advice; I have a few more things to check.
The guaranteed RAM speed your CPU can handle is 5200MHz (with DDR5). This doesn't confirm you can reach 6000MHz, but it indicates compatibility isn't assured at that higher rate. It's recommended to avoid overclocking via DOCP/XMP and stick to the native SPD speed of around 4800MHz. If stable at that speed, try increasing to 5200MHz and observe stability. If stable, you're on the right track; continue adjusting until instability occurs.
Thank you for the note, I have confirmed testing with standard speeds and no overclocking.
So fresh RAM has been added. I ran OCCT for an hour just to test it while working. It lasted about 20 minutes so far without any crashes (previously with old RAM it crashed after 15 minutes with the same errors in the minidump).
I’m hoping for a better outcome!
Tomorrow I’ll get a 4080 Super GPU, but I’m doing everything possible to avoid using it and keep it unopened.
UPDATE: The new RAM didn’t help; crashes continued with the same error. It’s running at normal speeds, not expo.
I’m growing more anxious that this problem might persist, especially since similar issues keep appearing in my research about 40 series cards.
The error message indicates a resource limitation when trying to run the API. Globmerger.sys and Kerb3961Kernel.sys are related kernel files, possibly involved in system operations or encryption processes.
thanks, just did a bit of googling on that one and got a bit lost. All i could see was make sure have enough space on hard drive and maybe expand pagefile sizes?
Then saw quite a few threads still talking about being related to graphics drivers.
Do you have anything to add? Given you spotted it you already know more than me
😀
Thank you
Sorry, I noticed you made changes now. I don't understand those others—this is based on a clean Windows setup with only essential programs installed because of all my troubleshooting. I have no idea what's causing them.
I plan to do another fresh install tomorrow using a different NVMe drive and try with an AMD card. I've seen this error too many times, which makes me even more concerned about upgrading to a 40-series model.
(For reference, I just removed my 3090 from the new system and reinstalled it in the old one; it's been running smoothly.)
note: Microsoft released an update to the SymCryptk library on September 5, 2024, to enhance its resistance against quantum computing attacks. The process requires precise memory allocation, which may cause compatibility issues if older versions are still in use.
Microsoft Strengthens Essential Encryption Library with Quantum-Safe Algorithms
The previous week, tech giant Microsoft enhanced SymCrypt, their custom encryption library, to include quantum-resistant algorithms.
digitalchew.com
I noticed the SymCryptk.dll appears in your driver list. After 8 seconds of testing, the system reported a resource exhaustion error. Clearing the pagefile.sys and restarting might resolve it.
For deeper troubleshooting, you’d need to submit a kernel memory dump file (memory.dmp) containing detailed debug information.
More details: https://github.com/microsoft/SymCrypt