BSOD, hard locks, driver crashes and video output drops

[DrSNAFU]

Ever since the end of last month I've been having some issues on my PC. While playing games my computer can lock up tight, the game can crash THEN lock up tight, my video drivers crash, recover and then lock up tight. The window can go a single color then lock up tight. There's also an interesting phenomena where input to both of my monitors dies but I can still shut down windows with the power button. In most cases on reboot everything is just fine again.


I have hunches but I'd like some opinions before I do anything really invasive or start replacing parts. I've also tried to take notes on each occurrence but I'll throw that into post 2 because it's so wordy.

 

[DrSNAFU]

Crash 1: Saturday 30th 11:09pm
Playing vermintide 2. Picture froze, sound stuttered and stopped. After a few seconds the signal from video card cut. Went to no signal and both screens went to standby. Unplugging replugging cables did nothing. Pushed power button to initiate shutdown, successful. Turned power off at supply, back on
and rebooted, functioned normally.

Checking logs afterward I found a BSOD citing the directx along with nvidia drivers. Unsure how BSOD occurred since PC managed to shut itself down.


Crash 2: Sunday July 1st
Was playing SPAZ2, game abruptly froze up. Mouse got jerky and unresponsive, but still moved slightly, got progressively worse. Unable to kill game, had to reboot PC.

Disconnected PC and cleaned dust out of case.


Crash 3: Monday?
Shortly after startup the PC became largely unresponsive. High CPU cycles despite nothing apparently using them. Monitors froze up, had to reboot.

Checked for system restore points. Only one, from July 1st, after the issue started presenting itself. Ran Bitdefender quick scan, nothing found. Uninstalled Bitdefender antivirus, cleaned remainder with removal tool. (Have had issues with antivirus being destabilized before) Uninstalled video drivers, cleaned geforce 391.35 drivers with removal tool and installed 398.36. Ran Memtest86 for 9 hours, no problems found. Ran prime95 for 2 hours, no problems found. Ran Malwarebytes full scan, nothing found.


Crash 4: Friday, 1:40am
Playing Dying Light. Game froze, given dialog box that game crashed, option to check for fix or close. Used spacebar to select close. No responsiveness after that. Seeing windows update listed ready to install on tray. Also seeing blue I notification on systray. (Later found out that was an alert that the game wasn't compatible with aero and it had been disabled. Though I've also had times where the same game gets backgrounded to display a warning that system had become slow, giving me the option to disable desktop composition, which I was assuming was the same thing.) Had to to hard reset.

Event viewer recorded nothing other than an unexpected shutdown of the kernel at 1:40, and reboot at 1:46. Let raid verify itself for a while, then shut down.


Crash 5: Saturday 12:48pm
Turned on PC, raid verification resumed at 40% from last night. Waited for OS use HD activity to die down. Started Steam and began playing Dying Light. About 5 minutes in the game froze up. Attempted to alt-tab, mouse cursor appeared but nothing else. PC completely locked, capslock unresponsive. Held power button to hard reset

Instead of shutting down, PC restarted. Opted to start windows normally from diag screen. Screen went black before going to login screen. Logged in, screen went completely white, then black, then recovered to windows desktop. Nvidia tray item says that Nvidia kernel recovered from an error.

Ran prime95 for another half hour, no issues, temps within norms. Ran rkill to test for malicious programs, and ran malwarebytes rootkit scan for good measure, nothing found. Malwarebytes junk removal tool removed a couple registry keys. Windows defender scan found nothing.

Attempted to use video ram testers. Most were limited at 1 or 2 gb of ram. Memtest G80 could go up to 3gb of ram testable, but after that it errors out and won't run or completes all tests in 0ms with millions of errors.


Sunday: 12:55am
After running a video stress test with OCCTPT for an hour with GPU temp looking normal, then played some Dying Light. About 20m in both screens turned orange and then shut off, no signal. (Same as the first event on saturday) No responsiveness from PC hdd or capslock for approx 2 minutes, then resumed activity. Able to press power button to initiate shutdown, which succeeded.

Event viewer shows nvlddmkm stopped responding and recovered, but by then signal to both monitors had been lost and hadn't returned. Application log shows information event for windows error reporting, LiveKernelEvent, no signatures, dump files available. Also registered “Device action request for device [device ID of my video card] was vetoed by [Device ID of my video card] with veto type 8. No BSOD recorded. The only one I have is from Saturday's initial instance. Ran windows file integrity scan SFC /SCANNOW, everything checked out. Disconnected HDMI out to TV, disconnected Corsair USB headset to see if those were cause.


Tuesday: 1:31am
Playing Vermintide 2 approx 50 minutes. Game crashed to desktop, Nvidia tray icon saying nvidia driver kernel crashed and recovered. About 10 seconds later the entire screen went blue-gray, all responsiveness lost, no capslock, power button control or hdd access. Hard shut down with power button, automatically restarted. On restart, (didn't notice it post) no input to screen, no hdd activity, no capslock. Single press of power button turned everything off near-instantly. Powered down at switch on pow supply 1minute, turned on and powered up. Booted to windows.

Only errors recorded in event viewer was display driver nvlddmkm stopped responding and successfully recovered, and alert that kernel lost power unexpectedly. Only other log after display drivers recovered was diagnostic system host service entering a running state then the logs stop.

Also observed: 1:30:26, request made by process (4) to disable Desktop Window Manager, followed by DWM unable to start because composition was disabled by running application. Display driver stopped responding at 1:30:27. Possible correlation.

 

[DrSNAFU]

Disabled desktop composition. Uninstalled video drivers from safe mode with DDU, reinstalled freshly downloaded drivers, same version. Started Vermintide 2 and looked inside case to observe fans on GPU do function.


Tuesday July 10th, 5:30:07pm

Attempted to join a game of Vermintide 2 after checking the fans at the hub area. Almost instantaneously upon load-in, picture froze, sound stuttered then turned into a buzz and screen turned mono-color. Went to check fans again and could hear a squealing noise from inside the computer's case. (sounds like capacitor noise but worryingly loud) Held power button to shut down.

Attempting to restart, opted to load windows normally. Windows loaded but instead of login screen, black screen. No HDD activity, no capslock no mouse. Held power button to hard shut down again. Rebooted again and got into windows. Event viewer shows nothing other than a loss of power to kernel, no other logs from that time period. No bluescreen log.

 

[DrSNAFU]

After further TSing, resorted to installing my old GTX 560. That seems to have worked, no more crashes. Purchased a new video card.

 

[DrSNAFU]

So after using my old video card with no problems and purchasing a new video card it's happened again. Same thing as the first time too. Just playing a game and the video feed to my monitor goes dead. Registers as a bluescreen. Attached the log. Appears to be directx and nvidia, just like before. I'm out of ideas.

 

[philc43]

The automated analysis of your crash dump shows the following:

BugCheck 116, {fffffa8015fd2010, fffff880100a6230, ffffffffc000009a, 4}

*** WARNING: Unable to verify timestamp for nvlddmkm.sys
*** ERROR: Module load completed but symbols could not be loaded for nvlddmkm.sys
Probably caused by : nvlddmkm.sys ( nvlddmkm+a89230 )

Then driver seems to be up to date but for some reason the graphics card is timing out. Is it overheating? I suggest you run a temperature monitor while doing a graphics test to check this out. In the tutorial below you will find the instructions. Also note that you can start the GPU-Z monitor to check on the status of temperatures etc.

Video Card - Stress Test with Furmark