GFX Card Crashing + BSOD due to gfx drivers?

[mankvill]

Hey all,

I'm at my wits end here. This is a problem I've been having for about a week now.

I have a Nvidia GeForce GTX 560 Ti and it keeps on constantly crashing + recovering while I'm just browsing the web or whatever. It says "display driver nvidia windows kernel mode driver, version 314.22 stopped responding" and sometimes it crashes Firefox.

When I go to play a game I'll get more of those, and then sometimes a BSOD. I have WhoCrashed, and this is what it has for the latest BSOD:

[FONT=Segoe UI, Arial]On Sun 12/22/2013 1:09:00 AM GMT your computer crashed
crash dump file: C:\Windows\Minidump\122113-27783-01.dmp
This was probably caused by the following module: [FONT=Segoe UI, Arial]ntoskrnl.exe[/FONT] (nt+0x75BC0)
Bugcheck code: 0xC4 (0xF6, 0x1B8, 0xFFFFF980D5D9CB30, 0xFFFFF8000333E1A5)
Error: [FONT=Segoe UI, Arial]DRIVER_VERIFIER_DETECTED_VIOLATION[/FONT]
file path: C:\Windows\system32\ntoskrnl.exe
product: [FONT=Segoe UI, Arial]Microsoft® Windows® Operating System[/FONT]
company: [FONT=Segoe UI, Arial]Microsoft Corporation[/FONT]
description: NT Kernel & System
Bug check description: This is the general bug check code for fatal errors found by Driver Verifier.
A driver references a user-mode handle as kernel mode. This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
The crash took place in the Windows kernel. Possibly this problem is caused by another driver that cannot be identified at this time. [/FONT]

I have uninstalled drivers and reinstalled them and rolled them back to previous versions and updated them to the latest, but it keeps happening. I really don't know what's going on. I have never had a problem like this with this computer in the 2+ years I've had it.

Help? If I need to provide more info, let me know.

 

[SenileAnimal]

First, I'll admit I have next to no idea how to read a dump file.

However, I have had similar GPU driver issues in the past, all of which seemed to be voltage related. If it is overclocked, it likely requires more voltage for stability. If it is not overclocked, it is possible that it is unstable at stock voltage.

My GTX 580 requires a bump in voltage for stability at stock clocks in GPU intensive games. I've read about many others with the same issue. I guess Nvidia set the stock voltage a bit on the low side for marketing reasons and to be "greener"? Anyway, it is possible your 560 wants more voltage as well.

Just my 2 cents. I hope that it may help you.

 

[mankvill]

I don't overclock it, so that's not the issue. Why would it suddenly need more voltage after 2+ years of use with no problems? I don't even know how I'd change that.

 

[Solarstarshines]

if it does it every 10 to 15 seconds GPU has gone bad mine did that and i had to send it in for a replacement

 

[SenileAnimal]

As it is getting BSODs with (presumably) low GPU usage, voltage does seem very unlikely, though it is certainly possible.

I'd see what the GPU manufacturer has to say about the issue, as it seems like the GTX 560 could need an RMA, as Solarstarshines suggested.

 

[mankvill]

if it does it every 10 to 15 seconds GPU has gone bad mine did that and i had to send it in for a replacement

well it's only a constant of "the driver has crashed and recovered" when I'm watching videos or playing a game or something. I don't really get constant BSOD's anymore.

 

[mankvill]

As it is getting BSODs with (presumably) low GPU usage, voltage does seem very unlikely, though it is certainly possible.

I'd see what the GPU manufacturer has to say about the issue, as it seems like the GTX 560 could need an RMA, as Solarstarshines suggested.

Yeah, I might do that. Do you think I might just need a new graphics card altogether? Like, is there an easy way to tell if it's actually failing and not just a driver issue?

 

[SenileAnimal]

AFAIK there is no easy way; it would have to be determined via process of elimination.

Though fairly unlikely to be contributing factors, I'd rule out:
1. Malware: Malwarebytes is widely regarded as the best of the best. Also do a full scan (and memory and boottime scans, as well as rootkit scans if possible) with your chosen antivirus with the detection sensitivity on max.
2. OS corruption: run an elevated command prompt and run "sfc /scannow". Also run checkdisk.
3. Other errors: check the windows event viewer to see if any other major issues are occuring. I'd run memtest overnight to rule out memory as a contributing factor.
4. Drivers: try a clean installation of all GPU drivers instead of updating them or rolling them back.
5. Temperature: Download a temperature monitoring program and keep an eye on it. Generally ~90C/190F is the absolute max. I recommend Realtemp. You can set an over temperature alarm within the program.
6. Instability at Stock voltage: I'd download either MSI Afterburner or Nvidia Inspector (which is portable) and bump the voltage up (probably to ~1025mv). EVGA Precision is the best IMO, but only available if you have an EVGA card. If you do have an EVGA card, their customer service is second to none, so I'd forgo further troubleshooting and just contact them, as they are phenomenal. If the GPU is unstable at stock voltage, but stabilizes with additional voltage, I would think you are entitled to an RMA, if desired.

You could also try using system restore and going back to a known good Windows configuration. If that doesnt work you could wipe the HDD or SSD and reinstall Windows or access a clean Windows install on another HDD/SSD.

If none of the above helps, I'd be contacting the GPU manufacturer to see what they have to say. It does sound a lot like hardware failure to me, so if you don't wish to do extensive troubleshooting, you could circumvent the troubleshooting and just contact the manufacturer.

PS: I notice you have your graphics listed as onboard + GTX 560. Your not running a program that uses both the IGPU and the GTX 560 at the same time (like Lucid Virtu), are you? I had nothing but issues with that software (BSODs included), and would delete it, if you are using it. I'd use something like IOBit uninstaller, as it will delete related registry entries that the standard Windows uninstaller will miss.

 

[mankvill]

Yeah, I emailed them to see what they have to say. Thanks for all the help, I'll try this stuff tomorrow.

 

[jonnyhotchkiss]

gfx issues, BSOD - how to verify/eliminate voltage as the culprit

Been researching this for a while now, running Geforce gt540m on (dell) xps l502x. Problems with optimus technology (that handles the Intel/NVID GPU mgmt) fixed with clean install, but when running an extended HDMI display, everything goes wrong.

There is a kind of minecraft-esque pixellation effect on the extended display: sometimes system-tray-notification reports "NVIDIA display driver service" restarted, usually the OS just becomes (completely) unresponsive. The problem is that it's NOT crashing - no BSOD, so no memory dump.

I've set the reg for a complete dump

, and tried to setup "Crashonctrl". Trying to configure dump1/2 keys to allow crashonctrl command execution (noscrolllockkey on l502x)


Back to the "is it the voltage" conundrum - I believe the voltage cannot be set (it's value is dictated by what you set other values to? or maybe it's a BIOS-mod thing) - but, when looking at the stats (GPU-Z) the GT540M is showing voltage @ .98V.

Is this too high?
for a while I thought my power adapter / psu might be the cause...

 

[jonnyhotchkiss]

Solved my issues (no extended display, let alone executing high-performance games on gt540m) by watching lots of videos on youtube... about to flash bios cry when (it seems!) the culmination of my efforts finally paid off.

In nvidia CPL, set power to max performance.
In PhysX, set to Geforce.
Personally, I "underclock" to 550,850 -stable settings for me.

also using latest nvidia driver (347.09) which seems to show diff voltage on NVIDIA (previously .98V)

Think my mains adaptor may be degraded, resulting in power supply issues, but I can't prove it (my battery is cr@p, so can't test without mains).

the important thing is it finally works! (for now)