My VR unity experience keeps crashing with a ntdll.dll error, help!

Hey everyone,

I'm working on a VR social world called Convrge where people can meet up and hang out with others. I thought I would lead with this since our build is public so if anyone is able to reproduce my error and knows a solution please let me know! Download is at www.convrge.co

Our latest build runs fine on my Macbook pro and most every other PC I've loaded it onto or that our users have. This seems to be a very niche problem / something specific to my machine. I can stay in the experience for about 5 - 10 mins on average before being kicked and seeing the below ntdll.dll error message. Sometimes it's instant as soon as I load the world, sometimes it takes 15 mins, there doesn't seem to be a pattern:

Error message I see after the game crashes: http://i.imgur.com/ZTjbCtL.jpg

Full text of the error:
---------
Problem signature:
Problem Event Name: APPCRASH
Application Name: Convrge_Win_0.10.exe
Application Version: 4.6.1.51269
Application Timestamp: 54819edb
Fault Module Name: ntdll.dll
Fault Module Version: 6.3.9600.17031
Fault Module Timestamp: 5308893d
Exception Code: c0000005
Exception Offset: 0003fd12
OS Version: 6.3.9600.2.0.0.768.101
Locale ID: 1033
Additional Information 1: 46d1
Additional Information 2: 46d1c12205ae881d9878aa5b56fc9648
Additional Information 3: 3eba
Additional Information 4: 3ebaf8d2848cbfbab57d22308d930cdb
---------

I've also seen people include their dxdiag.txt file for reports like this so here's mine (attached)

And here's the laptop itself: http://www.amazon.com/G750JM-DS71-17-3-inch-Gaming-GeForce-Graphics/dp/B00IKF2H12

At one point I thought this error was being caused by other clients connecting to the game, but I've been testing it tonight with no one but myself in the world and the error still occurs. I've tried rolling back my Nvidia drivers one version at a time for about 4 versions with no success. I'm back on the latest Nvidia drivers now. I've installed all available win 8.1 updates. Unity is version 4.6.1 due to a plugin were using that isn't supported on any further versions.

This bug seems to occur randomly and not associated with any particular part of the world, other users, character movement, etc.

Please let me know if you need any more info. We're heading to GDC in two days and would really like to show our stuff on this machine without it crashing all the time

Thanks so much!

 

Anything in the output log? (stack trace)

Not really related to this, but we are having similar random crash problem with a win8.1+kinect v2 game,
only happens few times a month though..(pc is on running ~10hrs per day)

 

My buddy just checked and it looks like there's no Kinect code in there anymore.

I'm on the design side, so I'm not sure how to read the output log, but here it is. (attached)

Also FYI someone asked me on another forum if my Rift was plugged in at the time of generating the dxdiag, which it wasn't. We allow desktop users / full screen mode, which is what I have been doing recently / when I generated the dxdiag. But I get the same error with the rift plugged in as well.

Thanks so much for the help

 

My friend thinks it looks like a direct X issue but we're not sure what we can do about it since I'm 100% updated on win 8.1 and seem to have the latest version of direct X.

 

Still no luck on our end, any ideas for things we could test to resolve this?

 

Tried it few times on win7, no oculus.. no crashing here.

Probably you could add some debug.logs in different places, and then hope that it crashes between some of them.

 

Issue is still occurring and now happening on other users machines. Mostly seems to be people with win 8.1 and Nvidia cards.

Any ideas please? This is causing 5 - 6 of us to crash constantly and we don't know what we can do to fix it.

Thanks!

 

Here's the most recent output log. Can anyone please confirm if this looks like a direct X issue?

 

Vague NTDLL errors when using something that requires a lot of power *could* be power supply issues. I've had one computer where the voltage output from the 12V rail wasn't quite up to scratch, and it had similar crashes until I, in my desperation, rebooted into the BIOS menu and stared at numbers till I was enlightened. At 11.something volts, it was enough to boot, but not actually do anything fancy with the GPU

It could also be heat. I'm not sure what the expected operating temperature range is on that GPU/motherboard combination (same GPU from different device manufacturers add all sorts of variables), and I'm sure things are supposed to throttle down if they get too hot (it's recent enough for that). But it could be a sensor malfunction. Is any silicon dripping out of the computer somewhere?

Give it a torture test with the Unigine benchmark demo.

 

Thanks for the idea. I don't believe its a heat issue since our world is very low poly and not tasking at all, but I'll give that benchmark a try.

 

Had about 6 people having this issue last night. Seemed like most of them were on win 8.1 with Nvidia cards. Any update on what we might do about these crashes?

Also I'll add that the crashes start slowly for most people and then escalate in frequency, which makes me think it might be a memory leak issue?

 

Same thing here.

Working on integrating a few different devices into one application - using the following devices: Oculus, zSpace, IDesk, and the leap motion as the Oculus Input. We have a bunch of the devices working/networked with no issues, but the moment we began including/building with the Oculus we started getting random crash reports. Initially I thought it was the Leap Motion because there is some documentation out there suggesting that the leap has some issues with some of the newer Unity Builds, so I downgraded back to 4.6.1 to 'fix' this issue but the more I've adjusted our build the more I've realized I think it is something to do with Windows 8.1 - as we are running a few machines on 8.1 with NVidia Quadro Cards.

The application I'm building uses a standalone server approach - one of my machines does all of the hosting, while the other machines connect to that one. In order to have all of these different devices load without issues we have some Network Instantiating that is done when the user loads based on their setup.
This crash log is from the server side, and as you will see it just crashes while in the middle of nothing going on, haven't even had any one connect yet.
The server machine is also not using the Oculus, nor has it plugged in, just using the default equipment.

It crashes randomly, but generally between 5-20 minutes in. I have an older Windows 7 machine that I am going to test it on tomorrow when I'm back on campus to see if I get the same issues.
I've included our crash report, not sure if it helps but maybe there's something similar to what you guys are getting - just search for 'Crash!!!' /Line 179

 

So I had a chance to do some other configurations with some of my hardware, by forcing my build to run in OpenGL I didn't end up getting these random crashes, was able to keep a build of mine up and running with another PC networked for well longer than my standard 15-20 minutes, so I'm thinking that it might have something to do with Windows 8.1 and Directx as you guys pointed out. Going to do some more tests tomorrow with a variety of machines we have in our lab and will try and post our logs if we have a catastrophic failure.

 

Oh hey! Sorry, I lost hope and hadn't checked this thread in a while haha. Did you end up finding anything out via testing on more machines?

If this issue only arises once the Oculus code is added perhaps we should take this to their forums. Although I already made a post here with lackluster results: https://forums.oculus.com/viewtopic.php?t=20838

 

same happening to us now....when baking lightmap its occurring,...two days back it was working fine..I reverted from 5.3.1 to 5.3.0 and 5.2.4f1...though same happening to all versions