Deadlock causing severe crashes that fully reboot the computer

[Guggleywubbins]

Hmmmm. Assuming event viewer shows the same thing?

I'll keep following this and keep digging to see if I can help over the next few days. I have time for 1 game then bed lol

Good luck man, hopefully it's something stupid.

Event Viewer showing the same thing. I decided to download a program to view the minidump files that are being generated, and it is implicating the dxgkrnl.sys and nvlddmkm.sys drivers. While I'm more technically proficient than the average bear, interpreting memory addresses and hexadecimal parameters is a bit beyond me. I'm going to just keep googling some of these strings and see if anything useful pops up.

 

[prfl]

Just had the exact same thing happen to me on Steam Deck (this happens with Deadlock only), i don't think this is a driver issue specifically

 

[matimaster07]

There was an update to deadlock where I was prompted to install "Microsoft Visual C++ 2015-2022 Redistributable (x86) 14.40.33810"... I also have an Asrock motherboard, after installing this I been running into nothing but hardware issues. for whatever reason my pc will randomly turn off its display output and started revving up the graphics card. I rolled back the install and my pc is back to normal. I am actually able to play the game without having to install it. Maybe try it this out....

 

[Guggleywubbins]

Unfortunately this is still happening. I tried some troubleshooting where I gave all users Full Access permissiom to the nvlddmkm.sys driver in system32 based on some reddit thread I found that seemed to work for people, but it makes no difference here. Installing different Nvidia driver versions after DDU wiping hasn’t helped either.

The picture above it what it looks like while it’s frozen by the way. Usually stays that way for about 10 - 15 seconds with the music still going before the whole thing goes down.

 

[wheretheirisawilltheirisaway]

Your crash dump indicates a stability issue with the video card driver, in essence the GPU is taking more time than permitted to display graphics to the display. Some of the key things for troubleshooting you have already done:

install the latest updates for your display driver
Over-clocked components
Insufficient system power

what you could try:
Change the timing on your RAM, are you using an XMP profile? Try it without it see if you have better stability
I would be curious if during the event if you press shift+CTL+B if the thing recovers, this is a keyboard shortcut to reset the video card driver.

Otherwise you might be looking at a failing video card.

Other options is to change how the card is called, are you running the game DirectX or Vulcan? I would suggest swapping, this can make a pretty big difference as the calls to the card are quite a bit different.

 

[chojo1125]

Hey, I just resolved this issue by going into my bios and enabling EXPO. Apparently, my memory was running at a slower speed causing me to get nvlddmkm errors in other games as well. I was getting constant crashes in deadlock (even on the lowest settings). Now it runs fine, not a single crash. Hope this helps.

 

[Guggleywubbins]

what you could try:
Change the timing on your RAM, are you using an XMP profile? Try it without it see if you have better stability
I would be curious if during the event if you press shift+CTL+B if the thing recovers, this is a keyboard shortcut to reset the video card driver.

I have tried this with the XMP profile both off and on. Same issue occurs regardless, so I keep it on for now. As for the shortcut, I'll try it next time.

Otherwise you might be looking at a failing video card.

I thought this too, so I actually RMA'd my 3080 about two months ago and got another one from Nvidia directly. Unfortunately this one experiences the exact same issue, so unless I got really unlucky and got two failing video cards in a row, I think it's an OS/software issue.

Other options is to change how the card is called, are you running the game DirectX or Vulcan? I would suggest swapping, this can make a pretty big difference as the calls to the card are quite a bit different.

Currently running on Direct X, but I can switch to Vulkan to see if that fixes it. If it is caused by certain API calls, then maybe this will help. Thanks for the suggestion.

 

[wheretheirisawilltheirisaway]

if you can upload the minidumps I can analyze them, based on the type of error your getting there may be some hidden nuggets of information I can gleam from them.

 

[Guggleywubbins]

if you can upload the minidumps I can analyze them, based on the type of error your getting there may be some hidden nuggets of information I can gleam from them.

Sure thing! I've just DMed you one. Let me know if you find anything interesting.

 

[Delmerik]

Someone else said it, I'm assuming your gpu is bad. Literally could have 1 damaged capacitor and only be triggered by specific games.

Had a similar issue with my gpu when I played ffxiv. No other game triggered it. Tested it on both my own and my wife's pc. Only real way to isolate it is by plugging in the same gpu if you don't have another system to plug it into.

You could test this by buying the same gpu, you will have like 7-15 days to return it depending on the retailer. I think best buy is shorter. Microcenter is 15 I believe.

As the most expensive component, it does suck but it's most likely the culprit at this point.

 

[Guggleywubbins]

You could test this by buying the same gpu, you will have like 7-15 days to return it depending on the retailer. I think best buy is shorter. Microcenter is 15 I believe.

As the most expensive component, it does suck but it's most likely the culprit at this point.

You might be right, but the thing is I already did this. I RMA'd my GPU two months ago to Nvidia and got a new 3080 FE sent to me, which is the one I'm using now. It still has the same issue, and the likelihood of both GPUs being broken in the exact same way seems unlikely to me.

This is part of the reason I think it's software-related, though the fact that it mostly happens when I play Source 2 games may be related to the fact I mostly play Source 2 games anyway. Perhaps it's some other hardware component (CPU? motherboard? power supply?), but if so I can't figure out what it is.

 

[Delmerik]

You might be right, but the thing is I already did this. I RMA'd my GPU two months ago to Nvidia and got a new 3080 FE sent to me, which is the one I'm using now. It still has the same issue, and the likelihood of both GPUs being broken in the exact same way seems unlikely to me.

This is part of the reason I think it's software-related, though the fact that it mostly happens when I play Source 2 games may be related to the fact I mostly play Source 2 games anyway. Perhaps it's some other hardware component (CPU? motherboard? power supply?), but if so I can't figure out what it is.

Are you absolutely sure they sent an entire new GPU? That is normally the worst case scenario for RMAs and I haven't seen that done much.

Sadly, the only real way to test your hardware is to throw each piece into another known working system and recreate the problem.

Also, your card could still be overheating even know the temps are reporting good. The GDDR6X modules on that card are known for getting up to 110 degrees. Your chipset may report nice temps, but if your case doesnt have great pressure/flow it could be the culprit?

I feel like we just keep bouncing around with this lol - what's your cooling setup like? How many in/out, are they directly pushing fresh air into the open sections of the card?

If you truly think it's software, 100% a full reinstall would solve it. a 3080FE isn't going to encounter issues with a fresh windows 10/11 install for any mainstream games especially considering your PC build.

Im gonna go take a look at your specs again and see if there may be anything weird possibly going on(probably wont find anything).

 

[Delmerik]

...Holy shit - I've never seen such a detailed build page! I think I've built like 100+ pcs in my life and have never gone that deep lol. Lots of info there XD

 

[over825hp]

I would swap RAM first, even if tests come out good. 8gb on a clean windows would be enough just to test for a couple of days. Considering you are not a regular user I would assume you monitored your temps, but just in case you didn't - maybe it's GPU overheating, run furmark or heaven. Usually the weirdest problems are caused by psu in my experience, so maybe borrowing it from someone for just a day would be good. Unplug everything but your mouse/keyboard, in some cases problems are caused by faulty cheap USB hubs, etc.

 

[Delmerik]

Ok - so after typing too long of a rant to care....
Likely culprits
-VRAM on mobo
-Bad riser cable
-Hit the reverse power supply lottery meaning its not running at full voltage.

I agree with over825 - most weird problems tend to be PSU.

If your GPU fits without the riser - at least to rule it out.

Also - forget if you said or not - but have you underclocked the gpu?

 

[wheretheirisawilltheirisaway]

I have gone through all the dumps you DM'd me and they all have the exact same stack trace: DirectX render requests, the driver gets stuck so the system reset the driver it fails to recover in time and then boom. Below is the same thing I DM'd you, so unless you follow the phases I outline below it's going to be hard to pin down the exact issue. Some suggestions you could try before attacking the software front: Move the video card to a different PCIe port and or remove any riser cards, this will make sure you don't have issues with a specific port. Pulling the riser card would mean the timing on the PCIe was off causing the delay on the driver. The issue could be related to the RAM, test this by: Pulling one stick of RAM, swapping that stick of RAM to the other port, do it again with the other stick. This will at least tell you if you have a bad RAM slot or if the RAM is bad.

Please try each phase, do not reverse a phase after testing just keep stacking them.

Phase 1:

Do you have any RGB management software installed? Asrock has ASRock Polychrome RGB but also supports Razor Chroma, if you have this software installed I would remove it.

Try again does the issue persist? if so continue to phase 2

Phase 2:

Try removing all background processes
1. Open msconfig (you can find this by just typing it in search)
2. Click the radio button "selective startup"
3. uncheck "load startup items"
4. Click the tab "services"
5. Click the tab "Hide All Microsoft Services"
6. Click Disable all
7. Apply
8. Reboot

This is basically safe mode with networking without being in safe mode, this can help narrow down if a software on your system or some installed application is causing the issue. After testing to see if you crash, it then becomes a matter of determine what installed software triggers the issue.

Phase 3:

Disable Fast boot
1. go to: Control Panel\Hardware and Sound\Power Options
FYI: you can just copy the above and paste in an explorer window
2. Click "Choose what the power buttons do"
3. Click "Change settings that are currently unavailable"
4. Uncheck "Turn on fast startup"
5. Reboot your system, it will take longer now but it will take time on driver initialization vs caching

test again, if not move to phase 4:

Phase 4:

Deep driver removal, you have already done this already. This is just a more extreme version of what you did.

1. open window search and look for: change device installation settings
2. Click "no" after opening it (this action should be reverted after the driver is reinstalled)
3. Download DDU if you don't already have it
4. Download the latest driver for your video card
4. Reboot into safe mode
5. use msconfig, click the boot tab select safe mode for startup no networking
NOTE: you will have jacked up video after the restart 640 x 480 may be the highest resolution hopefully it's at least 800 by 600 just because it's easier to navigate.
6. Run DDU
7. Reboot normally
8. install the Nvidia driver

you can revert step 2 at this point and test again, this is about as clean an install as you can get for the drivers.

Now if all this still doesn't resolve the issue the last phase is what I call nuke and pave, it's a bit more advanced than just doing a windows reset to reinstall windows. This is a complete wipe of the whole primary disk that hosts your OS. I would recommend you backup any data you wish to keep, if you have a thumb drive on hand create a windows 11 installation media: https://www.microsoft.com/software-download/windows11

I verified that your motherboard includes some secure erase tools for both SSD/HDD and NVME drives, this can be found by going into your UEFI firmware and going to the tab "tools". After backing everything up and having your installation media at the ready, please use the respective tool to the type of disk you have in the UEFI firmware. The one labeled "SSD Secure Erase Tool" will also work on HDD if you happen to be running that with your primary OS.

When the secure erase is done, which may literally take a second start the installation of windows. When complete I would focus on just updating the video card driver installing steam and getting the game installed to test.

if the game crashes after all of this, then I can confidently say the issue lays with the hardware somewhere. Could be power supply, could be ram, could be motherboard and it could be the Video Card. Ram is easy to test for if you have two sticks of ram, pull one stick and see if you crash then swap the RAM if you do and test the other stick. This at least removes that as the culprit. Motherboards and Video cards can't be verified unless you have known good hardware, so this is about as far as I think I can take you.

Hopefully something here helps you out.

 

[Guggleywubbins]

Are you absolutely sure they sent an entire new GPU? That is normally the worst case scenario for RMAs and I haven't seen that done much.

Yes, I checked and the serial numbers embossed into the slot metal are different.

Also, your card could still be overheating even know the temps are reporting good. The GDDR6X modules on that card are known for getting up to 110 degrees. Your chipset may report nice temps, but if your case doesnt have great pressure/flow it could be the culprit? I feel like we just keep bouncing around with this lol - what's your cooling setup like? How many in/out, are they directly pushing fresh air into the open sections of the card?

It's a Meshlicious case, so it stays relatively cool. I don't generally see it get any higher than 88°C or so at full load.

If you truly think it's software, 100% a full reinstall would solve it. a 3080FE isn't going to encounter issues with a fresh windows 10/11 install for any mainstream games especially considering your PC build.

I've already done a full reformat of Windows 11 about a month ago to try this out. What's interesting is I didn't experience this crash at first, but at some point Windows Update forced me to upgrade to 24H2 and new drivers and then it started again. Maybe I'd just been lucky before, but it seemed to start after I was forced to upgrade Windows version and Nvidia driver. Perhaps I should just reformat and install the Windows 10 LTSC next time.

Ok - so after typing too long of a rant to care....
Likely culprits
-VRAM on mobo
-Bad riser cable
-Hit the reverse power supply lottery meaning its not running at full voltage.

I agree with over825 - most weird problems tend to be PSU. If your GPU fits without the riser - at least to rule it out.

Unfortunately in my SFF case the riser is necessary to connect the GPU as it sits behind the motherboard. I could try to extract the whole thing and run it like a test bench, but it would be a huge headache. Besides, I didn't have this problem at all for ~15 months after building it, and then it started all at once. I wish I had an exact date when I first noticed it.

That said, I don't know if PSUs suddenly start going bad after 15 months. Do you know how I would test for PSU failure?

...Holy shit - I've never seen such a detailed build page! I think I've built like 100+ pcs in my life and have never gone that deep lol. Lots of info there XD

Oh stop it you.

 

[Guggleywubbins]

I have gone through all the dumps you DM'd me and they all have the exact same stack trace: DirectX render requests, the driver gets stuck so the system reset the driver it fails to recover in time and then boom. Below is the same thing I DM'd you, so unless you follow the phases I outline below it's going to be hard to pin down the exact issue.

Hey! I appreciate your taking the time to look through those crash logs.

Some suggestions you could try before attacking the software front: Move the video card to a different PCIe port and or remove any riser cards, this will make sure you don't have issues with a specific port. Pulling the riser card would mean the timing on the PCIe was off causing the delay on the driver. The issue could be related to the RAM, test this by: Pulling one stick of RAM, swapping that stick of RAM to the other port, do it again with the other stick. This will at least tell you if you have a bad RAM slot or if the RAM is bad.

I've got a microATX board that only has a single PCIe slot, and the riser is required to connect it inside the PC since it's a Meshlicious SFF and fits parallel behind the motherboard. Here's a link to the build.

Please try each phase, do not reverse a phase after testing just keep stacking them.

[...]

if the game crashes after all of this, then I can confidently say the issue lays with the hardware somewhere. Could be power supply, could be ram, could be motherboard and it could be the Video Card. Ram is easy to test for if you have two sticks of ram, pull one stick and see if you crash then swap the RAM if you do and test the other stick. This at least removes that as the culprit. Motherboards and Video cards can't be verified unless you have known good hardware, so this is about as far as I think I can take you.

Hopefully something here helps you out.

I'll start these troubleshooting phases soon. Thanks for the exhaustive guide! I'll let you know if I ever manage to pin down a culprit.

 

[Guggleywubbins]

I wanted to follow up with you guys (@Delmerik, @wheretheirisawilltheirisaway) after a few weeks of troubleshooting. I ended up doing a full reformat of my computer, but I installed the Windows 10 LTSC 2021 instead of the latest version of Windows 11 like I'd previously used. Long story short, the crashes have basically disappeared.

While I did have a single DirectX / Nvidia driver crash very early on outside the game when I only had Steam, Deadlock, and Firefox installed, it hasn't happened again. I've since installed several more of my typical background programs (Discord, Spotify, PCPanel, &c.) and updated to a more recent Nvidia driver and it hasn't happened again. I've made no hardware changes during this time.

I really think there's something going on with Windows 11 and Nvidia drivers that Deadlock was triggering. One major difference between this installation and the previous one is that I'm using NVCleanInstall to install just the graphic driver with none of the other GeForce Experience bloatware, which means I don't have a bunch of separate Nvidia containers constantly lingering in the background.

At this point, I don't know if this problem is something that Valve can do anything about, but I keep seeing other posts in the Bug Reports forum of people crashing under similar circumstances.

 

[wheretheirisawilltheirisaway]

the LTSC version of windows strips out a bunch of things as well for OS bloat, so it's possible a feature that was apart of that bloat is no longer present. Did you happen to run the secure erase tool built into your motherboard? That could also explain it if you had. Otherwise I would chock it up to random bits being aligned in a way that just conflicted, change of OS made it so those bits didn't align anymore. Can't tell you how many times I have seen some random computer crashes, or poor performance that made zero sense on a clean install of windows just for it to be different after a different OS install or a secure/low level wipe of the disk.

Glad to hear it's all working again though, it certainly didn't seem like failing hardware.