• Welcome to our Forum! Ask PC-Build Questions, discuss Tech-News, Content Creation & Gaming Workloads or get to know the CGDirector Community off-topic. Feel free to chime in with insight or questions on any existing topic too! :)

Lately, my dual 1080ti configuration gives me lots of BSOD

V

verysame

New member
Joined
Dec 25, 2020
Messages
11
Reaction score
5
Points
3
Hi there,

Not sure what happened, maybe a Windows update: the problem is that lately, I'm getting a lot of BSOD.

Here's the scenario: it happens when I'm using Redshift's IPR or when I'm previewing scenes in Blender through Cycles in the viewport.
The feeling I have is that GPUs run out of memory and the system simply stops. Apparently, it happens in situations where I also have a video playing in Microsoft Edge (that maybe uses GPU acceleration?). But not always, my guess is that sometimes it just runs out of memory and it freezes.

I know it sounds vague but I'm hoping someone has some tips to share.

The PC used to work smoothly with C4D and Blender running side by side. I notice a total GPU memory loss if I'm using Davinci Resolve or Fusion, I guess they both make full usage of the GPU memory. But Davinci/Fusion aside, I wouldn't expect BSOD only because the GPU runs out of memory. In fact, the way it used to work was that when there was no more memory available, Redshift would simply give a memory error and I would need to restart Cinema. That would make sense, but BSOD??

Here's my config:

Alienware Area 51 R-3
AMD Ryzen Threadripper 1950X 16-Core Processor
(2) GeForce GTX 1080 Ti
64 GB ram
NVMe 512 + SSD 2 TB
 
V

verysame

New member
Joined
Dec 25, 2020
Messages
11
Reaction score
5
Points
3
OK, apparently this is happening with most of the drivers after 441.28
The thing is, some software now doesn't properly function without the latest drivers.
 
Jerry James

Jerry James

Hardware Nerd @ CGDirector
Staff member
Joined
Jun 19, 2020
Messages
260
Reaction score
88
Points
28
Oh that does sound like a driver issue. Could you try using DDU and removing the display driver completely, and then installing the NVIDIA Studio drivers instead?
 
  • Like
Reactions: verysame
V

verysame

New member
Joined
Dec 25, 2020
Messages
11
Reaction score
5
Points
3
Thanks, @Jerry James

Today I think I did it probably 10 times 😭

So, it seems older drivers might work better but then, due to a job I'm working on, I need to export H265 from Davinci, and without the newest driver that option is not available.

Then, I thought to check the event log and I found something quite ugly: a lot of system errors.
They look like this:

The description for Event ID 56 from source Application Popup cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.

or this

DCOM got error "1084" attempting to start the service ShellHWDetection with arguments "Unavailable" in order to run the server:
{DD522ACC-F821-461A-A407-50B198B896DC}

DCOM got error "1084" attempting to start the service camsvc with arguments "Unavailable" in order to run the server:
Windows.Internal.CapabilityAccess.CapabilityAccess

DCOM got error "1084" attempting to start the service netprofm with arguments "Unavailable" in order to run the server:
{A47979D2-C419-11D9-A5B4-001185AD2B89}


and many, many more. A lot of DCOM 1084...
 
V

verysame

New member
Joined
Dec 25, 2020
Messages
11
Reaction score
5
Points
3
Yes, the latest drivers are the ones I have to use in order to get H265 and that's the problem because the latests also are the reason for BSOD.
 
V

verysame

New member
Joined
Dec 25, 2020
Messages
11
Reaction score
5
Points
3
Oh, try following these steps too - https://answers.microsoft.com/en-us...em-files/bc609315-da1f-4775-812c-695b60477a93

Also, the latest NVIDIA studio drivers don’t give you H265 export on DaVinci?

I did run both with these results

scannow
Windows Resource Protection did not find any integrity violations.
---
DISM /Online /Cleanup-Image /RestoreHealth
Deployment Image Servicing and Management tool
Version: 10.0.19041.844

Image Version: 10.0.19041.867

[==========================100.0%==========================] The restore operation completed successfully.
The operation completed successfully.
 
Jerry James

Jerry James

Hardware Nerd @ CGDirector
Staff member
Joined
Jun 19, 2020
Messages
260
Reaction score
88
Points
28
I did run both with these results

scannow
Windows Resource Protection did not find any integrity violations.
---
DISM /Online /Cleanup-Image /RestoreHealth
Deployment Image Servicing and Management tool
Version: 10.0.19041.844

Image Version: 10.0.19041.867

[==========================100.0%==========================] The restore operation completed successfully.
The operation completed successfully.
Oh, so can rule that out then.
Well, that's a tough one. Will have to wait for a driver fix from NVIDIA if rolling back the driver does fix the issue :/
You should post the issue on here - GeForce 461.92 Game Ready Driver Feedback Thread | NVIDIA GeForce Forums
 
  • Like
Reactions: verysame
V

verysame

New member
Joined
Dec 25, 2020
Messages
11
Reaction score
5
Points
3
Just an update on this issue, in case someone else may experience issues similar to this.
Not sure this is the solution yet, still testing. So far I haven't had any BSOD or crashes, but it needs more testing.

So, I knew already the TDR is quite important when it comes to GPU renders and I had already tweaked it. The most common value (which is also found in the Substance Painter documentation), is to have the delay equal to 60.

I had it like that and yet I was getting BSOD and crashes.

I've also found other values around. So, now this is how my Regedit looks like:

1616280155354.png
Again, still in testing mode. I'll keep this thread updated in case things work out and, hopefully, this can help others who have issues with their GPUs.

Cheers,
Andrew

P.S. almost forgot: I also read that some users are experiencing problems when using two monitors, most likely due to a different refresh rate.
Since I'm in production, and above all, I'm trying to get the machine working, I also have turned my 2nd monitor off for the time being. I know it's not ideal when trying to troubleshoot errors as if things work out it would be difficult to say whether it was the keys in the regedit or the monitor to cause the problem. But, if it works, I'll proceed with the 2nd monitor on and that should give me a conclusive indication.
 
  • Like
Reactions: Jerry James and Alex Glawion
Top