Question BSOD DRIVER_OVERRAN_STACK_BUFFER nvlddmkm.sys

[l0r]

Hello!

I have been experiencing this BSOD amongst others for the past couple of weeks/months now and have not been able to fix it since.

The BSOD happens seemingly out of nowhere, either while idling, having only a Twitch stream opened, while browsing in Chrome or opening a random image file.
The symptoms are always the same: the mouse cursor freezes, background sound is looping and the system locks up (sometimes with the video output glitching) for a few seconds until it restarts on its own.
I checked WinDbg and it always shows "nvlddmkm.sys" as the culprit.

System specifications are as follows:

• PSU: Corsair RM850x
• MB: ASUS ROG Strix Z490-E Gaming
• CPU: Intel i7-10700K
• RAM: Corsair Vengeance LPX 64GB (2x32GB) DDR4 3600MHz C18 (QVL verified)
• GPU: Gigabyte NVIDIA GeForce RTX 4060 GAMING OC (top PCIe x16 slot)
• SSD: Samsung 980 PRO NVMe M.2 SSD (bottom M.2_2 slot)

Some things to note:

• BIOS is up to date (Ver. 3001)
• GPU is brand new (around 2 months old)
• I also experienced other BSODs in the past, namely "SYSTEM_ THREAD_EXCEPTION_NOT_HANDLED", "DPC_WATCHDOG_VIOLATION" and "VIDEO_TDR_FAILURE" and using WinDbg all dump files pointed to "nvlddmkm.sys".

So I thought: case clear, must be a bad GPU or/and driver, right? Well... these BSODs all happened with another (older) GPU and non-qvl RAM and on Windows 10 before.

Since then, I have replaced the GPU with a new one (RTX 4060), explicitly new QVL RAM and wiped my drive, installing a fresh copy of Windows 11. And still: DRIVER_OVERRAN_STACK_BUFFER with nvlddmkm.sys from time to time.

Also at some point I thought the reason might be having XMP 1 enabled, which could be a trigger for those BSODs so I disabled it and indeed haven't had any crashes for a week since then. But could this just be a coincidence? Everything else is running on stock, BIOS is completely on default besides XMP.
Right now I am running my RAM with XMP 2 enabled, which puts the DRAM voltage to 1.361 (instead of 1.35). So far haven't had any crashes, but still testing.
Would it be possible that for some reason enabling XMP 1 could be the culprit after all? But if so, why should it?

Besides, system performance overall is very good, I can run Prime95 + Furmark simultaneously without any crashes and temps are also good.
Only this pesky DRIVER_OVERRAN_STACK_BUFFER BSOD out of nowhere sometimes...

I am really at a loss here. Please help.

 

[JohnBonhamsGhost]

I thought: case clear, must be a bad GPU or/and driver, right? Well... these BSODs all happened with another (older) GPU and non-qvl RAM and on Windows 10 before.

Since then, I have replaced the GPU with a new one

just changing out the physical card is not going to bring new drivers with it.
you would have to manually remove/replace the existing driver package.

run DDU from Windows' Safe Mode and choose to remove all optional data.
then reinstall the latest available driver package directly from Nvidia, without GeForce Experience.

 

[l0r]

Hello and thanks for your reply!

just changing out the physical card is not going to bring new drivers with it.
you would have to manually remove/replace the existing driver package.

run DDU from Windows' Safe Mode and choose to remove all optional data.
then reinstall the latest available driver package directly from Nvidia, without GeForce Experience.

Yes, I am aware but as I said, I did a full reinstall of Windows even to make sure I have no messed up drivers.

The last overran stack buffer BSOD was a couple days ago on a new Win 11 OS and I just let Windows install the driver for my GPU (which was an old one from 2023 though). I am now testing with the most recent driver from Nvidia (no GeForce Experience).

And yes, I already tried DDU several times (in safe mode) when I was on my older Win 10 OS.

Edit 1: There is still the question about XMP hanging in the room, whether or not it would be related to that problem in any way at all.
Also: is it normal behaviour that the DRAM voltage goes up to 1.361V when XMP 2 is enabled although it is set to 1.35V?

Edit 2: I just checked with other RAM settings in BIOS: when set to AUTO (= default settings), the RAM runs @ 2133Mhz and DRAM voltage is fluctuating between 1.200V and 1.217V. Is that also normal?

 

[JohnBonhamsGhost]

is it normal behaviour that the DRAM voltage goes up to 1.361V when XMP 2 is enabled

the XMP profiles are configured and tested by the manufacturer for those specific RAM kits.
i would leave it @ XMP I unless you notice other strange issue(s) arising.

if you are getting more heat generated than you like, or think necessary, just use manual settings and input the RAM manufacturer's rated specs. but .011v is probably not enough to make any difference anywhere.

many modern motherboard manufacturer use built-in profiles that tend to boost voltages for their 'auto-overclock' or other 'enhanced performance' modes.

 

[l0r]

Thank you again for your reply!
I should note that I checked and XMP 1 also puts the DRAM voltage to 1.3500V and the BIOS reads a voltage of 1.361V, so no difference between XMP 1 and 2 there.

I conclude then that everything works as intended, right?

Another question I have: wouldn't a non-functioning XMP profile OOB with explicitly using a QVL RAM kit be a reason to suspect a hardware issue somewhere?

 

[l0r]

I just got another BSOD "VIDEO_SCHEDULER_INTERNAL_ERROR" pointing to "nvlddmkm.sys" while doing nothing but browsing through Windows explorer files.
What can I do now?