BSOD video_tdr_failure after applying an nvidia driver [Hasn't been solved yet]

cscrille

Honorable
Dec 23, 2013
46
0
10,530
Due to a game's issues, I've decided to apply an older driver than the currently newest one. After the process has finished, the display started as if switching between displays. Once being a whole black screen and then attempting to display my desktop (most of it was a shade of black, however at some time some features were distinguishable, such as the computer folder or the start button, etc.). After the constant switching between displays, I receive a BSOD, which includes the error "video_tdr_failure".

I've reinstalled my OS to the same version and applied the newest NVidia driver (straight after the fresh install, therefore, no graphics drivers at presence) and I've received the same issue again. General process felt slower than before applying the driver and after some time of flashing, I've received the same BSOD error message.

Does anyone have a solution?

Specs:
CPU:i5 3330
GPU: Asus 660 DirectCU 2 OC Edition
PSU: Evga 600W Bronze
Mboard: GA-H61M-S1
Memory: 8gb Corsair Vengeance LP
 
Windows asked the video card to do something and did not get a response, then it attempted to reset the hardware but also did not get a response. So windows called a bugcheck. The end solution is going to be update the driver and hope for the best.

Most likely your video card driver is deadlocked with several processes waiting for the other to finish. These problems are often timing related and you may be able to change the timing to be a very small time frame, the faster the graphics card the more chance to hit these timing issues.

work around the problem:
Try this: go to control panel, device manager, find your high definition audio devices and turn off any that you do not use.
in particular, turn off you graphics adapter sound support for HDMI and display port and use the one for from your motherboard.
now your graphics driver does not have to worry about sound and it does not have another thread that can cause problems for it.

-second, turn off your screen saver. This will prevent windows from trying to grab some graphics memory for itself and hang when it does not get it. (I have seen systems hang when the screen saver is getting ready to display a second background.)

I guess you might even go as far as turn off hardware acceleration in your browser. (but I don't do that)

 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


I've reinstalled windows once more. Currently don't have a Nvidia driver installed as I assume same will repeat. I am considering on trying out the 314 driver, although it's also risk.

Before the current reinstall, It would slowly process at the start menu and if I switched to Desktop it'd start flashing, although it would even in the start menu. Therefore, utilizing the desktop was useless. No access whatsoever. Didn't bother with Safe mode.
 
after you install windows run a chkdsk /f /r on your boot drive
and run the system file checker
sfc.exe /scannow
it will confirm that your copy of the core windows files have not been corrupted during the install. (it happens, often depending how you install the OS)

I am not sure about your description of the problem. Booting into safe mode can tell you if it is the graphics processor itself or if it is the graphics driver that has the issue. (different graphics drivers used for safe mode),

Does the screen work ok while in BIOS? kind of sounds like a hardware problem. Standard things apply for that:
update BIOS (or reset it to defaults), confirm that there is no over clock on the PCI bus, no over clock on the GPU, confirm the fans in the CPU are working, confirm the power supply connections are good and the power supply is ok. Things like that.

even isolating the problem form windows can be useful. running memtest86 from a cd for example to confirm your basic function of your machine by checking your memory. then run a test of the drive and graphics.
 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


It appears that the nvidia drivers might not be causing the issue. After I've reinstalled Windows, I've used it fine straight after the install. Next day as I boot up, the symptoms were appearing after I've signed in, as I got to the start menu, switched to Desktop and I assume it started flashing again ( I went afk) and then the BSOD came up with the same error message. And I haven't installed any Nvidia drivers like before.

Update: After the installation, I had black borders around my desktop and checked the nvidia control panel plugin. It stated that I already had an nvidia driver installed even if I haven't. As I boot it next day, the black borders are gone (this is the result of installing graphics drivers, but I haven't done so) and the original issue persists.
 
There are many reasons why a video card will not respond, some are hard caused by hardware
(thermal issues, overclock issues, Card overclocked, or PCI bus overclocked, and power issues)
The hardware problems will still show up when you boot with out windows. for example, you can boot memtest86 on its own CD and run tests to help Isolate problem from problems with windows.

if the machine works as expected when booted on a test CD, then you need to work on why windows is having a problem. There are many causes for this and it is best to start with the machines BIOS, reset the defaults and reboot. The BIOS will pass a database list of all the hardware to windows, and windows will try to use that info, in needs to be correct but the BIOS does not rebuild is database until you change hardware, or update the BIOS or reset the BIOS back to defaults.

Next, you look at why a video card stops responding. PCI bus speed issue, the BIOS reset should have fixed that. Cooling: you get graphics corruption on your screen, lines and such most of the time and. Low power, you machine will bugcheck with certain special error codes or it will simply reset and start back in the BIOS.

Now, most of the time if you don't have a physical problem you have a problem with device drivers. You update them and you get blocked because there are no more updates to be had. If your system hangs, the screen locks up, it is not corrupted, but does not refresh you most likely have your software for your video card in a software deadlock. one process locks access to some data, while the other process wait there turn to get access. They never get access and your system hangs. Graphics cards run games, and the driver has to be as fast as it can get. One way they do this is by having the minimum amount of error checking. Windows may detect that the graphics driver stopped responding but the graphic software is just waiting for another part of its own software to release its lock so it can continue.

What you can do: these errors are timing dependent, you can change the timing and change the rate at which you hit the error in the code. One way to change the timing is by going to the control panel device manager and disabling the high definition audio for your graphics card. this audio is provided for you HDMI or display port cable to allow your monitor to have speakers. if your monitor does not have speaker the graphics card is processing sound for no reason and you can actually speed up your graphics by telling the card not to process sound. And you don't have to worry about and sound driver conflicts with other sound sources.
(web cam sound, motherboard sound, pci sound card,...)

anyway, hope this helps.






 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


Am certain it isn't a physical issue. The GPU is new and has been in use for around 2 weeks if not less. Regarding cooling issues and overclocking, the name of the gpu is "Asus 660 DirectCU II OC Edition", therefore, physical retail upgrades shouldn't also be causing any issues. I myself haven't done any additional upgrading (overclocking, etc).

When I've installed an OS on this current build for the first time and applied the newest driver at the time, it was working fine until the creation of this thread, when I applied an older driver.

I've recently launched in safe mode, uninstalled my gpu driver and then was able to access my desktop normally, however, without an nvidia driver on board and running on standard vga. The black borders around my desktop indicated that there was no video driver at presence. The issue did not occur. Nevertheless, the device manager states that there is a video driver installed, even if I have uninstalled it in safe mode. Assuming the gpu has a default driver pre-installed and every time you uninstall it, next boot it will already be there.

Towards my conclusion, it seems that the issue is caused by the presence of nvidia drivers (although, it's weird that it was function fine for some time at the very beginning) and not by gpu's performance.

I also assume that switching to Win7 (what I intend to do in the near future) would possibly solve this issue. I am on Windows 8.1.
 
I would recommend that you use the current windows 8.x and the current windows 8.x graphics driver and use control panel device manager to disable the graphics driver sound support in device manager. Do not uninstall the sound support, the PNP system will just reinstall for you. If your problems continue, I would disable any low power modes for the graphics processor in the control panel power management. If that does not help, I would reduce the the graphics processors and memory overclock by 25Mhz or to the default reference card values. This is what the makers of the driver will have tested against. If you really want to know the exact cause of the problem with the driver you will want to run verifier.exe and turn on deadlock detection on the graphics driver. The system will bugcheck when a deadlock condition is encounter. It does not fix your problem but it does confirm that is its a software timing problem with the driver.

The last time I took the effort to really look at this, it looked like windows attempting to preload a screen saver into graphics memory and not getting permission to do so by the graphics adapter. Kind of brought the system to its knees for a stupid reason. (disable screen changer to avoid the condition in that case)


- If you do go back to windows 7 Make sure you reboot before and after you use the OEM graphics adapter update program. (just a heads up, to prevent a common problem)

- Also, the new graphics drivers hit known bugs in the old windows 7 system, these bugs require that you install the service pack and the windows update to prevent getting bugchecks every few hours with the new graphics drivers. (again just a heads up to prevent a common problem)



 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


Could my '09 HDD influence towards the issue?
 
I normally say no, but now I have to say maybe. That is because I have had seen a video_tdr_failure caused by a SSD firmware bug. The SSD firmware took too long to respond and windows attempted to reset the driver but then got a video_tdr_failure bugcheck. (it was a mix of two or three problems)
- if your old HDD is involved, you should see a the windows disk subsystem attempting to reset the port that the drive is connected to. (you would see this in the eventlog, only if the reset worked, otherwise the log would be in a buffer when the machine bugchecked and would not have been written to disk)

most of the video_tdr_failure problems are just bugs in the video drivers that cause a deadlock. (one process waiting for permission to continue but never getting the permission) Video drivers tend to have poor error checking because they want to work very fast. This can lead one piece of the driver hitting a error condition and blocking the other 2 threads. It ends up causing windows to panic and attempt to recover by reseting the card. When that does not work all windows can do is bugcheck.

Windows 8.x will attempt to check your drive and make repairs where it can. It will find places on the drive that are still readable but produce errors on the read, it will read the data off, move it to a new place and mark the old cluster as bad. But a spinning hard drive has a known failure rate that just adds up over time, I think it is something like 25% during the first year and 16% failure per year until it dies after about 5 years the failure rate goes way up.
Windows 8 should help extend the life of these drives but they are doomed to fail just based on how they are designed. (they will just fail faster on windows 7 because people don't check them until they system actually fails to boot, windows 8 checks them a little bit at a time when the system goes idle, and moves your data before the cluster is totally unreadable)

I have retired my spinning drives on my systems. They still work and I have plugged them in to make backups but they are on a shelf now. I found that the only thing i really care about is my photos and I have backups on DVD and the cloud. (and on my old hard drive, if i ever get around to doing the photoshop work on them)
I now have various SSDs in my system at least when they wear out I can still read the data off them even if I can not write to them.




 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


Nevertheless, I'm in doubt that the issue is caused by hardware, as mentioned before, the fact that I was able to run this build for more than a few days without issues. My main concern are the drivers and I have no idea how to fix this. Whenever they're utilized, they simply initate this slow process and then the flashing proceeds and the bsod. All this, caused by applying an older driver. My HDD does bottleneck, but it's functional for the time being until it's replaced. Do you have any specific solutions regarding the drivers themselves, apart from reducing overclock speeds, deleting sound support for the gpu, editing power rationing, etc.

Regarding the check via verifier.exe, I assume I have to be booted up in normal and a nvidia driver utilized for it to be checked?
 
I would think this is a driver problem in you graphics driver. I you want to prove it you would run verifier.exe on the graphics driver, select a bunch of the options (most likely deadlock detetion) set your memory dump to be a full or kernel memory dump (not mini dump) and boot your system with these setting and go until you bugcheck. At that time the state of your machine will be saved and with a few commands you can see what the cause of the hang was.

then you can look for a driver update or change the clocking of your card, maybe disable your screen saver or some other windows service to reduce the how often you hit the problem. But at least it will be clear that you are looking in the correct subsystem for the problem.

If you get a good bugcheck and memory dump, put it up on a cloud server and maybe i can see what the cause is if you want.
when you find the cause you can also report it and give the memory dump to the vendor of the driver if they will take it.
(generally, they won't actually look at it if it is a known problem or is on a outdated driver, or a OS without proper updates)
 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


I have to boot in normal in order for the graphics driver to be displayed in the list via verifier. Unless it's displayed by default and I can't recognize it. The issue prevents me from accessing my desktop via normal. Safe mode doesn't display the drivers I need, only audio and hdmi nvidia.
 
it sounds like you really just need to boot in safe mode, disable plug and play in control panel and uninstall the nvidia software. And remove the oem.inf files so it does not get automatically reinstalled, then boot into normal mode and install a clean copy of the nvidia drivers. reboot and turn your plug and play system back on.

you can use the pnputil.exe -e command in safe mode to get a list of your oem drivers, look at the list and find the nvidia driver and its oem.inf file. Then type
pnputil.exe -d oemxx.inf to remove the package, be sure to remove the sound package for the graphics card as well. The oemxx.oinf where the xx = the number of the file shown in the enum command (pnputil -e)

the pnputil.exe command will take out the hidden backup copy of the driver from your driver store so it will not automatically get reinstalled a few seconds after you attempt to remove it. (Windows trying to save you by installing a broken driver you just tried to remove.)




 

cscrille

Honorable
Dec 23, 2013
46
0
10,530


To confirm that my hardware isn't causing the issue, specifically the gpu, I'll replace my current GPU with an older unit, 9800GTX+. Is this a practical troubleshoot?
 
it just opens another can of worms. where you have to reset your BIOS again to get the DMI pool updated, then remove and update drivers inside windows.

It would be better to get the configuration you want to get working to work. The problem you are having is a pretty common driver problem with the newer cards. Overclocked card hit the problem the most often. It is one of the reasons that i no longer buy attempt to run a over clocked card until the driver issues are worked out on someone other peoples systems. The last overclocked card I got took almost a year of driver updates to get stable an the overclock. Now i just look for a good cooler, and reset the clock to near the reference standard and let the people play with a good card with out timing problems in the drivers.



 

cscrille

Honorable
Dec 23, 2013
46
0
10,530

I've installed the known "stable" driver - 314.22 and during installation it utilized the driver and It did flash a few times (indicating the symptoms of my issue), however the driver has recovered after a few flashes (less than usual) and has been working fine for 15mins after the driver installation. Then as I went onto youtube, the screen just froze and had to restart and am facing the same issue once more.
 

TRENDING THREADS