Nvlddmkm (display Has Stopped And Recovered) Constantly

Hi LT586,

Sorry to hear you're having problems with playing TS and the display driver constantly stopping and restarting.

I wanted to relate a personal experience with this exact same issue a few years ago in case it helps with your situation. My problem was ultimately traced to a faulty VRM on the GPU, that was running way hotter than it was designed to - ultimately a failed card but was only showing up under very specific circumstances. Numerous games worked perfectly but there was just one that was causing the display driver to stop and restart every time.

This might not be your issue but there is a way to validate whether you have a genuine problem with your GPU (sometimes it comes down to silicon lottery) or whether it's just Windows being overly sensitive when it comes to your particular brand of GPU - and there's a way to make it less sensitive - a trick I use whenever this comes up with other games I play.

Firstly, a bit of background. In case you didn't know, Windows employs a tech called Timeout Detection and Recovery (TDR). This tech determines whether a device has responded in an appropriate amount of time or not and, if not, restarts the device driver to save you having to completely reboot your PC. If you're old enough to remember this from older versions of Windows, it was really annoying to have to do a complete reboot just to get a display working again or even just any device.

Some of the most common causes of the TDR error messages are usually due to having a lot of background applications/process running which are also requiring effort from the GPU (if you use Chrome and you have numerous tabs open, this is a major culprit). It can also be caused due to problems with specific display drivers, an overheating GPU or a failed component on the GPU.

TDR has a preset time period in which it looks to reboot the display drivers. You can change this value by editing your computer’s registry. Although the process is pretty simple, be sure to take extra care when tinkering around with your computer’s registry. I'm only suggesting this as it helped me figure out if it was just Windows being overly sensitive or not:

1. Press the Windows key + R to launch the Run dialog
2. Type regedit into the Run dialog and press Enter to launch the Registry Editor.
3. In the left pane of the Registry Editor, navigate to the following directory: HKEY_LOCAL_MACHINE > SYSTEM > CurrentControlSet > Control
4. Click on GraphicsDrivers in the left pane to have its contents displayed in the right pane.
5. In the right pane, right-click on an empty space and select New to expand the context menu. If you are using a 32-bit version of Windows, click on DWORD (32-bit) Value in the context menu. If you are using a 64-bit version of Windows, click on QWORD (64-bit) Value in the context menu.
6. Name the new registry value TdrDelay and press Enter.
7. Double-click on the TdrDelay registry value to modify it, type 8 into its Value data field and click on OK.
8. Close the Registry Editor, restart your PC and check to see whether or not the problem still persists once it boots up.

This should solve the problem of seeing the error message repeatedly but it will mean you will need to restart your PC if you experience a display driver crash. If you do end up in this situation, it's highly probable that something's not right with your GPU and the fault is only showing up under very specific circumstances.

Hope this helps

Best, Steve

 

Hi LT586,

Been hiding!

Just out of interest, have you tried rolling back to an older driver?

The 15 fps issue is exactly the same as I had with mine. What would happen is the GPU would be working normally and then all of a sudden I'd get like 15 fps and then the driver restarts - there's always a delay between getting 15 fps and Windows restarting the driver. The only way I could get around it until I found the TDR tweak was limiting the frame rate in TS to 30 fps and downclocking the card. If you've not tried it, might be worth seeing if it solves the problem. If you're not sure how to do that, just add -FPSLimit=30 to your Launch Options in Steam.

On the VRM - sadly, not an easy thing to spot. You'll need an oscilloscope to check a VRM is functioning properly and probably a few years of electronics training. If you have no qualms about removing the heatsink, you could always check to see whether TIM has expired (Thermal Interface Material - sometimes it does go hard and is not very efficient at transferring heat) and the thermal pads are properly seated on the RAM and VRMs.

As for what can cause a VRM to fail - typically, can be a multitude of things - component lottery (some are just manufactured with inherent defects that can be more prone to failure than others, even from the same batch); overheating; over-voltage (pushed the card a little too hard with an overclock... normally get this with a factory overclocked card); and generally just age.

One other thing you can try is a slight underclock of the GPU - drop it down on both Core Clock and RAM Clock by the smallest amount you can (some cards can allow you to do 1% and others' minimums are 3%) and see if that makes a difference. If it does, then you'll likely have the same problem I had.

Best, Steve