Hi there,
It seems like Unreal Engine 5.2 is crashing when trying to load the sample scene with Body Tracking using DirectX 12 (which is what it’s set to by default when you load the sample project).
Running a fresh Windows Install with the latest SDK 4.0.6 with CUDA 12.1 on an RTX 4090. Happens across multiple machines. DirectX 11 works fine.
It’s critical for our operations do be able to use DirectX 12 with the ZED camera array that we have, is there a way to fix this?
Thank you!
Hi,
We indeed has some user that reported a similar issue than yours (for example Crash when using Unreal Engine ZEDSamples - #9 by MarkJ).
Unfortunately, we have not found the source of the problem yet. I have almost the same config as yours (RTX4070 + CUDA 12.1) and I don’t experience any crash at all which make this issue quite difficult to fix.
Do you have any crash log you can share that would help us find the problem?
To make things more complicated: my initial issues and solution had me “stuck” at DX11… but I’m now on a new project, integrating the StereoLabs Unreal Plugin and that project is DX12.
Same machine (RTX3080Ti) as in my other post, but now working fine.
What Nvidia Driver do you currently have?
I have a RTX4070 ti and my driver version is 537.13. It is working fine with DX12.
Hey guys,
Thanks for the information,
I ran a few tests and have some interesting results to share,
All of our machines are running the Studio Driver versions (536.99), so my first thought was to install the game-ready drivers and test it with those (537.13 versions as mentioned above).
As a test, before doing so, I made sure that the project was still crashing on the studio version, and it was.
I installed the game-ready drivers, and low and behold everything started working.
Playing around with it a little bit, I decided to go back to the studio drivers and see if that started working again, and unfortunately, it was still crashing.
Went back to game-ready, and guess what, it crashed again.
Re-cap of the events:
Studio driver - crash
installed game ready drivers - worked
Back to studio - crash
re-installed game ready - crash
The project settings have been confirmed every time before the launch and also the machine has been restarted multiple times in-between each test.
This doesn’t really solve the issue for us but I hope it provides some insight into what might be going on.
I’m going to use DDU to completely uninstall our graphics card drivers and install a fresh copy of the game ready drivers and I’ll report back,
Thanks
Also, here are the last few lines from the log file after the crash
[2023.08.28-21.31.00:599][830]PIE: Play in editor total start time 0.05 seconds.
[2023.08.28-21.31.00:719][831]SlCameraProxy: Warning: Waiting for ZED to be connected
[2023.08.28-21.31.00:719][831]SlCameraProxy: Warning: ZED connected
[2023.08.28-21.31.00:719][831]SlCameraProxy: Initializing
[2023.08.28-21.31.04:448][263]SlCameraProxy: Initialized
[2023.08.28-21.31.58:668][342]SlCameraProxy: Error: Can’t enable BodyTracking: “EC_CorruptedSDKInstallation”
[2023.08.28-21.31.58:670][342]SlAIThread: BT Thread started
[2023.08.28-21.31.58:926][346]LogD3D12RHI: Error: CurrentQueue.Fence.D3DFence->GetCompletedValue() failed
at D:\build++UE5\Sync\Engine\Source\Runtime\D3D12RHI\Private\D3D12Submission.cpp:939
with error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_REMOVED
[2023.08.28-21.32.00:955][346]LogRHI: Error: [Aftermath] Status: DmaFault
[2023.08.28-21.32.00:955][346]LogRHI: Error: [Aftermath] Failed to get Aftermath stack data
[2023.08.28-21.32.00:955][346]LogD3D12RHI: Error: [GPUBreadCrumb] Last tracked GPU operations:
[2023.08.28-21.32.00:955][346]LogD3D12RHI: Error: [GPUBreadCrumb] 3D Queue 0 - End: Frame 2344 - Scene - SubmitCommands
[2023.08.28-21.32.00:984][346]LogD3D12RHI: Error: [GPUBreadCrumb] Copy Queue 0 - No Data
[2023.08.28-21.32.00:997][346]LogD3D12RHI: Error: [GPUBreadCrumb] Compute Queue 0 - No Data
[2023.08.28-21.32.01:010][346]LogD3D12RHI: Error: DRED: No breadcrumb head found.
[2023.08.28-21.32.01:010][346]LogD3D12RHI: Error: DRED: No PageFault data.
[2023.08.28-21.32.01:010][346]LogD3D12RHI: Error: Memory Info from frame ID 2345:
[2023.08.28-21.32.01:010][346]LogD3D12RHI: Error: Budget: 23388.00 MB
[2023.08.28-21.32.01:010][346]LogD3D12RHI: Error: Used: 7316.07 MB
[2023.08.28-21.32.03:172][346]LogWindows: Error: Error reentered: Fatal error: [File:D:\build++UE5\Sync\Engine\Source\Runtime\D3D12RHI\Private\D3D12Util.cpp] [Line: 873]
CurrentQueue.Fence.D3DFence->GetCompletedValue() failed
at D:\build++UE5\Sync\Engine\Source\Runtime\D3D12RHI\Private\D3D12Submission.cpp:939
with error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_REMOVED
Seems like there’s an issue with the SDK installation maybe? Yet running ZED Diagnostics tells me that everything is fine
Another quick thing that I wanted to add is the following observation.
If you change the project to DirectX 11, play the project (works), and then go to DirectX 12 (with proper restart and shader compilation) it WILL WORK. However, if you close the project, reboot the machine and launch the same project again (with DirectX 12) it will crash. Go figure.
Another quick thing that I wanted to add is the following observation.
If you change the project to DirectX 11, play the project (works), and then go to DirectX 12 (with proper restart and shader compilation) it WILL WORK. However, if you close the project, reboot the machine and launch the same project again (with DirectX 12) it will crash. Go figure.
[Discourse post]
I wanted to thank you a lot for your help.
From what I found after some research, this error “error DXGI_ERROR_DEVICE_REMOVED with Reason: DXGI_ERROR_DEVICE_REMOVED” is relatively common.
Maybe, you can try the solution available on different website, such as this one : https://www.makeuseof.com/dxgi-error-device-removed-error-windows/#:~:text=The%20DXGI_ERROR_DEVICE_REMOVED%20error%20message%20clearly,a%20GPU’s%20driver%20in%20Windows.
Unfortunately, I can’t test it on my computer.
Let me know how it goes. And thank you again.
Hey everyone, I was running into the same issue earlier this week. It seems to be just very specific hardware and project settings combinations.
After hours of trying everything I could find to attempt to fix this issue I’ve found that just creating a new project and setting it up according to the docs, rather than using the example project worked just fine.
Not sure if that’s a solution we should strive for, but I’m pretty sure it’s something on Unreals end, rather than Stereolabs.
Hi,
Thanks for sharing your solution.
It might be related to a project setting of our project. Maybe, that one parameter set at the non-default value is causing this crash. I still do not explain why it does happen on specific hardware.