Odd UE4 behaviour?

[Wraith]

So last week we decided to get back into PUBG as games seem to have dried up recently and bordom is starting to set in..

Which is where the problem arises, within 7 minutes my game crashes to desktop with the "D3D11 error hardware can not be found" and this happens every time I try to play. So after many failed google searches for a solution which only turns up results from 4 years back I gave up.

A day later I decide I need to know what's happening, so launch PUBG and watch CAM monitor all hardware which is when I notice a slight issue.. GPU temps are climbing and climbing but the clocks don't do what they are supposed to do with Pascal no throttling is happening in the slightest, so eventually GPU hits 80c and boom CTD!

So this monitored I decide DDU and fresh drivers (431.36) just in case bad drivers are at fault and nope same issue, at this point alarm bells start ringing and my RMA senses start tingling, I ran countless benches and even went as far as trying Furmark with not a single issue.. My loyal 1080Ti is a happy little GPU.

I go into stage 2 install Rime and Fortnite (don't judge) well low and behold they all exhibit the same CTD and GPU Boost fails to do its thing, I then decide to run the GPU in Debug mode with great results no more crashing and the card does its thing.. so I'd like to know if at all are any of you guys experiencing these issues with UE4 titles? I'm getting this problem solely with UE4 games, every other game runs fine.

 

[AlienALX]

My guess is this... It's something like CAM.

I had an issue with BL2 recently where it wouldn't launch at all. No error and annoyingly nothing in task manager to close to stop steam making the most annoying chime sound in the world ever.

Four hours of hair ripping layer and I found out it was Afterburner combined with razer's trash so I stopped both from starting.

Ironically EVGA precision worked fine, even though it's apparently the same.

I couldn't find anything meaningful on Google either. Just crap from 7 years ago.

 

[NeverBackDown]

It's probably CAM. My cousin on almost any game if he has CAM open it will most likely crash his game.

CAM is awful software tbh. Afterburner is lighter weight and can monitor anything CAM does besides system fans.

 

[Wraith]

We can rule CAM out guys it still crashes without it running, I'm talking bare bones no processes running that don't need to.

 

[Warchild]

hmm maybe something with the latest Nvidia update? they did just roll out a new driver.

 

[Wraith]

hmm maybe something with the latest Nvidia update? they did just roll out a new driver.

Ran DDU in safe mode and reinstalled latest driver 431.36 even did it twice trying the last known working drivers, same problem.

Well today ran Malwarebytes, CCleaner, did yet another DDU and fresh drivers and still getting crashes.. odd thing is the GPU is doing precisely what it's supposed too with GPU Boost, Temps, Voltages, Frequencys, Fans all good well within normal acceptable working parameters.. motherboard temps and voltages all check out fine too.

I've removed the vertical GPU mount and gone horizontal again just to rule out a faulty ribbon cable still same old crashes.

Run a single monitor - Crashes!
Uninstalled CAM - Crashes!
Checked memory - Crashes!
Uninstalled and Re installed games - Crashes!
Checked every PSU Cable even swapped out the PCIe - Crashes!

I did however get a different error report from PUBG this time though.

<ErrorMessage>LowLevelFatalError [File:\wk\tsl-client-dev-branch\build\UnrealEngine\Engine\Source\Runtime\RenderCore\Private\RenderingThread.cpp] [Line: 1101] &nl;GameThread timed out waiting for RenderThread after 30.00 secs&nl;</ErrorMessage>

I can also confirm it is not solely UE4 titles as Horizons 4 does it too.

I'm at a loss now aside from a fresh Windows install.

 

[NeverBackDown]

Damn. It seems to be a deep rooted issue and my only guess would be something in the registry maybe got screwed up.

Or it's a hardware issue which doesn't make sense as your benchmark software runs but no games.

 

[Wraith]

I can run benches for days no worries, but games no hope! I'm leaning towards a bad windows update and borked registry namely the TDR side of things just going to try something with Nsight and see if I can't debug this further try and narrow it down. At least find a reason for it so it can be helpful to someone else in the future.

I'll be doing a full Windows install tomorrow.

 

[NeverBackDown]

It might be Windows. My last update for 1809, one of the cumlative ones, instantly ruined everything and it was asking me to repair windows. I didn't of course(just restart and it'll sort itself out without losing data) but yeah I also feel like my OC wasn't holding on anymore since the update and bumping up the vcore .05 seemed to help.

So maybe try that as well? Unstable CPU OC caused by an update?

 

[Wraith]

It might be Windows. My last update for 1809, one of the cumlative ones, instantly ruined everything and it was asking me to repair windows. I didn't of course(just restart and it'll sort itself out without losing data) but yeah I also feel like my OC wasn't holding on anymore since the update and bumping up the vcore .05 seemed to help.

So maybe try that as well? Unstable CPU OC caused by an update?

I'm not getting any BSODs and my CPU isn't OC'd and hasn't been for a while just never got around to re-doing it.

The issue is pretty much nailed down to bad windows update and Timeout Detection Recovery on the NV drivers, PUBG has really good crash reports and with the info in Event Manager it's pointing to a borked Registry.

 

[NeverBackDown]

Bad OC's don't always cause BSODs, just usually. Was just reporting my recent experiences.

Fresh install never hurts. I always end up cloning the drive back again so I wonder why fresh installs are a thing when I just reclone the drive, so to me it's funny it fixes the problem lol

 

[Wraith]

Gentlemen! We have success!


After much searching, reinstalling windows and everything else on top including the usual...

I've finally solved the problem, it was nvidia control panel with the latest drivers the Fast Sync option no longer works right (for me at least) it pushes the card way too hard and the TDR can't keep up with the frame requests. Turned it off and I'm all back to normal.



Shocking to think that one setting caused all this bother, the instant I turned it off clocks dropped, load went from 99% to 45% temps crashed down from 75c to 54c and the fans fell silent. Ah well we live we learn I guess, thanks for all the support and thoughts guys.



Hopefully some one finds this thread useful in the future.

 

[NeverBackDown]

Wth? FastSync? Damn, that's not something I would ever expect.

The fact that reinstalling windows didn't do anything would have puzzled me too.

I've been saying this for a while but Nvidia drivers are kinda crap now and have been for a while. ALWAYS something. People give AMD crap, which makes no sense because nobody uses them and their 7 year old notions on AMD are wrong, but at least AMD doesn't have these massive issues(though they have had some persisting issues for a few years now in specific cases so there is that).

I update my nvidia drivers WHEN I need to support a new game I want to play, if it's not stable then I find a stable one I know works, and then just leave it alone. I went 7 months without updating my drivers, even played a game that technically wasn't supported, still worked fine. That is what I would suggest for everyone else too.

 

[AlienALX]

Always expect the unexpected.

Sorry dude I should have remembered I had issues with both DSR,VSR and Gsync lately. I could have saved you hours there

I had a feeling it would be something small and almost insignificant but I've used my PCs so little since I moved house I forgot.

Mega glad you fixed it. Nothing worse than a problem Google can't solve.

NBD - agree. The issues I had on BL2 didn't happen on my AMD card.

Nvidia driver department are slacking.