BSOD and Skyrim

[ripple]

I purchased a XFX HD 7850 2GB in January to replace my old EVGA 9800 GTX+ 512MB GPU that was burnt out while playing Skyrim, and I've experienced BSOD issues with Skyrim since then. I've been trying to resolve the issue for 4 months and now I am at my wits end, because while I use to think BSODs were always driver or hardware related, in this case it seems to be Skyrim-specific. I'll document what I've done so far to try and 'troubleshoot' this problem and how I arrived at this conclusion, in the hopes that someone (MShoap, I am looking at you) might have come across something that might help shed light on this. It's a long one, so here goes.

My specs:
CPU: Core 2 Duo E8400 3.0 GHz
Mobo: EVGA 750i SLI FTW
PSU: OCZ GXS 600W
GPU: XFX HD 7850 2GB
OS: Windows 7 SP1 64-bit
RAM: 4GB

The BSODs started shortly after I installed the new GPU, and it's always the same TDRx116 error:

Attempt to reset the display driver and recover from timeout failed

FAILURE_BUCKET_ID: X64_0x116_IMAGE_atikmpag.sys

It would happen:
1) during Skyrim, after 10 - 45 minutes of play.
2) after I quit Skyrim, and launch Firefox.
3) I could go days without seeing a BSOD, then it would happen persistently for a couple of days, then I could again go days to a week without see a BSOD before getting hit with persistent BSODs. The cycle continues.

The symptom: I would see artifacting on the entire screen for about 2 - 5 seconds, sometimes I can still pull up the game menu (which, strangely, was perfectly visible) and make a save, before the BSOD occurs (in even rarer cases, the save would actually work).

At first, I naturally thought it was an issue with the Catalyst driver or a bad GPU. So I (clean) installed and tested other Catalyst driver versions, stress tested the GPU with Furmark and MSI Kombuster, and used MSI Afterburner/HWInfo to monitor the GPU and CPU temperature in game to see if they were overheating. I even posted my Windows crash dump to a few sites to solicit help in reading them. I uploaded a bunch of logs, was told to update my BIOS, to which I could only reply that my BIOS is in fact the latest version even though it's dated from 2009, because my motherboard is so old. Many 'general suggestions' were offered, but none of them helped to resolve the BSODs.

Finally, when I couldn't resolve the issue, I RMA the card to XFX. They tested it, and said they found no problems with the card, but that they would send me a replacement GPU regardless. But after I installed the replacement GPU, I started experiencing the same BSOD (with the same error message) again.

Here is where the story takes a strange turn: I got really frustrated, so I decided I would install another game to try and see if the BSOD was 'Skyrim-specific.' I installed 'Assassin's Creed 3' (which was actually stressing my system more than Skyrim), and sure enough, I finished it over the span of 2 weeks and did not see one single BSOD. As soon as I resumed playing Skyrim after 2 weeks--bang, same cycle of BSOD frequency and error message. I was completely mind-boggled.

Just to recap, here are all the things I've done to try and address the BSODs:

- tried every version of Catalyst driver ('stable' release as well as beta) from 13.1 to 13.5 beta 2 (that's the latest).
- flashed BIOS to the latest I could find for the 750i SLI (from 2009) and updated all drivers where necessary (chipset and audio).
- installed MSI Afterburner/HWInfo to monitor my GPU and CPU temperature during Skyrim (all within normal parameters).
- Furmark 3D and MSI Kombuster to stress test the GPU (both the one I purchased and the replacement from XFX).
- tested RAM sticks using MemTest (8 passes with no errors).
- got a power supply tester from a friend to test my PSU to make sure it was not failing.
- RMA the GPU, tested by manufacturer who said there was nothing wrong with the card, but they sent me a replacement GPU anyway.
- completely reinstalled Skyrim/mods.
- took an air duster and vacuumed my PC case (several times in the span of the 4 months).
- constructed a make-shift shrine to the Nine Divines on a shelf in my bedroom closet.

I think I've pretty much exhausted everything short of reinstalling my OS (and I think I'd rather just stop playing Skyrim than reinstall the OS). If this was a driver/hardware issue, I can't see how BSODs with TDR errors would be specific to Skyrim but not other games (but apparently it is). I just don't get it.

Edited May 6, 2013 by ripple

[Georgiegril]

This may be off, but is there any chance that there are remnants of your old driver left (from your old GPU) and that Windows is trying to boot those instead to your new GPU/drivers?

I don't have your drivers, but there is a known issue with NVidia drivers that requires users to turn off Windows update features in order to properly clean install drivers. On the off chance, maybe do a complete manual clean uninstall/reinstalll of your drivers?

 

Also, when you pull up your system info on Steam, is it properly referenceing your new drivers/GPU?

 

Man that sucks. I totally am with you on not wanting to reformat your OS!

[ripple]

The driver issue was something I spent a great deal of time looking into. I always do clean installs of drivers, so basically, uninstalling the previous drivers, using 'Driver Sweeper' to remove remnants in the Windows registry in safe mode, before installing a new driver. When AMD came out with a driver cleaner utility, I gave that one a try too (it still left registry data, so needed Driver Sweeper). At some point, I stopped installing the junk that came with the Catalyst driver package (transcoder, etc.), and it didn't help. I've also tried installing just the display driver and not even the CCC, which didn't help either. Everytime I 'fiddle' with something, the BSODs would disappear for a few days, giving the illusion the problem was 'solved.' But then the cycle would start after a few days.

There's a sort of 'cycle' to the occurrence of the BSODs where when they happen, they will continue to happen for a few days, and when they don't happen, I can do all sorts of stuff without seeing the BSOD (play Skyrim for hours, launch Firefox after quitting, etc.)

The last thing I did was to actually migrate my mod installation from the Mod Organizer to the NMM, which is basically a complete re-installation of my mods (I had like...180...so it took a while...), because I thought the BSODs might be related to MO in some way and I needed to test this. I didn't see the BSODs for a week, then it started happening again a few days ago.

I didn't have any of these issues with 'Assassin's Creed 3', which played just perfectly fine with no BSOD or issues at all over the span of 2 weeks+, and is how I clued into the possibility that the BSODs could be Skyrim specific.

Everything checks out fine on Steam, and also on all the third party utilities I started using to monitor my system (like Speedfan, HWInfo, etc.).

This is all particularly ironic because I have not seen a CTD in Skyrim for...what, 3 months now? In spite of the BSODs, I've managed to log quite a few hours in Skyrim, especially when I am in the 'safe cycle' where I've gone for a few '4 hour' sessions. My Skyrim mod setup is as stable as a rock, if only I didn't have system BSOD.....just, ironic. :/

Edited May 6, 2013 by ripple

[Georgiegril]

I agree that is really strange.

Obviously you know your way around a computer (and Skyrim)... I fear anything I might think of will be something you thought of already. :confused:

If anything else pops in my head though, I'll post back.

 

Meanwhile I hope someone else has had your problem and can tell your how they solved it!

[ripple]

Thanks. At this point, I'll try anything that I haven't already tried. I would reinstall the OS if I could verify this was a 'system-wide' problem, but if it's just Skyrim.....I am thinking about buying another new game and trying that, just to make sure this is tied to Skyrim. I can't think of any new games that float my boat right now....maybe XCOM, but I don't think it will stress my system like Skyrim and AC3.

Edited May 6, 2013 by ripple

[bben46]

First guess - did you allow the game to regenerate a default ini after you replaced the GPU? Some of the entries in the ini are specific to certain registers in the GPU. And a different GPU, may not use the exact same register addresses for the exact same thing.

 

Simple way - first back up the existing ini files. (back up the entire data folder while you are at it - it takes a while but may save you from having to redo the entire game.) Then delete the ini files. (both sets) You may have to allow steam to verify your game - it will see the missing ini files and replace them with a default ini set. then when you start the game, it will run you through the setting up the video sequence - Play with default for a while to see if that makes any difference before making any changes to the ini files.

[prod80]

I feel driver compatibility issues.....

 

Have you tried like old drivers? Like these: http://www.guru3d.com/files_details/amd_catalyst_12_10_whql_download.html (12.10 WHQL)

 

I read more things of people having issues with 13.x drivers (various issues - but instability nonetheless in Skyrim)

[ripple]

Yup. I generated new ini files after installing the new GPU to register the new GPU, and again when I did a complete 'reinstall' of the game and mods by verifying files via Steam and switching to NMM from MO. Since I installed the new GPU, I've made virtually no edits to the ini files other than to increase the number of auto-saves and to reposition the third-person camera view. I didn't have to make any memory tweaks in the ini files since I was no longer on a 512MB GPU where I had to 'stretch' the game engine to sustain playable framerates, and I found with my new 2GB GPU, the game performed much better and was far more stable without fiddling with those tweaks and just going with the default ini settings.

I am also not using any of those '4K texture' mods, and just have the Hi-Res DLC and aMidianborn's 2k texture replacers. Game is very smooth, the darn BSOD notwithstanding.

@prod80
I think the oldest Catalyst driver I rolled back to was 12.8 beta...something (and pretty sure ever stable release versions after that, so 12.10 stable release version for sure)? I experienced minor artifacting issues in Skyrim (random lines appearing, not the full blown entire screen artifacting just before I BSOD), but I put up with it hoping it would at least solve the BSOD issue, but it didn't. :/ I am weary about going back to driver versions further than that, because I *think* I'd be looking at drivers that were made before Skyrim was release (or at least only shortly after it was released).

Edited May 7, 2013 by ripple

[grtim5]

Yup. I generated new ini files after installing the new GPU to register the new GPU, and again when I did a complete 'reinstall' of the game and mods by verifying files via Steam and switching to NMM from MO. Since I installed the new GPU, I've made virtually no edits to the ini files other than to increase the number of auto-saves and to reposition the third-person camera view. I didn't have to make any memory tweaks in the ini files since I was no longer on a 512MB GPU where I had to 'stretch' the game engine to sustain playable framerates, and I found with my new 2GB GPU, the game performed much better and was far more stable without fiddling with those tweaks and just going with the default ini settings.

 

I am also not using any of those '4K texture' mods, and just have the Hi-Res DLC and aMidianborn's 2k texture replacers. Game is very smooth, the darn BSOD notwithstanding.

 

@prod80

I think the oldest Catalyst driver I rolled back to was 12.8 beta...something (and pretty sure ever stable release versions after that, so 12.10 stable release version for sure)? I experienced minor artifacting issues in Skyrim (random lines appearing, not the full blown entire screen artifacting just before I BSOD), but I put up with it hoping it would at least solve the BSOD issue, but it didn't. :/ I am weary about going back to driver versions further than that, because I *think* I'd be looking at drivers that were made before Skyrim was release (or at least only shortly after it was released).

 

 

 

 

Well this probably is preaching to the choir... but it sounds like data corruption issues that have happened over time...

The nV chipset that is in your motherboard is famous for this kind of crap.... This is why the green team no longer supports Intel CPU's.

And they are entirely out of the Core Logic chipset buisness....

 

Your board is EOL.,.... so BIOS will never happen... which o/s are you on ?

Edited May 7, 2013 by grtim5

[ripple]

Yeah, latest BIOS is from 2009. I am on Win7 64-bit. You know of a 'work around'?