Game Crash - SwapChain1->ResizeBuffers - GPUBreadCrumb Warnings

Hello, my current games crashes sometimes when people use shortcuts to switch from game full screen to windows desktop.
It randomly occurs, on some machines more on some less. But in general on all machines.

It’s not related to old driver, because I know more machines are affected and some are my own. Just some PC are more often affected, some less.
I thought for a time running OBS in background is doing the crash more often. But it seems it occurs nearly the same often without.

Can somebody advise me how to dive deeper into such a case and how to get the pin point in my game code (or causing ue4 code, or causing feature) what causes the issue?

Here is the GPU example Crash Report of my last crash.

[2020.11.05-19.41.40:898][ 95]LogD3D12RHI: **Error**: SwapChain1->ResizeBuffers(NumBackBuffers, SizeX, SizeY, GetRenderTargetFormat(PixelFormat), SwapChainFlags) failed 
 at G:/UE4_data/UnrealEngine-4.25.4-release/Engine/Source/Runtime/D3D12RHI/Private/Windows/WindowsD3D12Viewport.cpp:372 
 with error DXGI_ERROR_DEVICE_REMOVED with Reason: 887A002B
Num=3, Size=(1600,900), PF=18, DXGIFormat=0x18, Flags=0x802
[2020.11.05-19.41.40:898][ 95]LogD3D12RHI: Error: [GPUBreadCrumb] Last tracked GPU operations:
[2020.11.05-19.41.40:898][ 95]LogD3D12RHI: Error: No Valid GPU Breadcrumb data found. Use -gpucrashdebugging to collect GPU progress when debugging GPU crashes.
[2020.11.05-19.41.44:916][ 95]LogWindows: FPlatformMisc::RequestExit(1)
[2020.11.05-19.41.44:916][ 95]LogCore: Engine exit requested (reason: Win RequestExit)

I also get warnings when I start the game:

[2020.11.05-19.40.52:099][  0]LogD3D12RHI: Adapter has 0MB of dedicated video memory, 0MB of dedicated system memory, and 8152MB of shared system memory, 0 output[s]
[2020.11.05-19.40.52:099][  0]LogD3D12RHI: Chosen D3D12 Adapter Id = 0
[2020.11.05-19.40.52:107][  0]LogD3D12RHI: Emitting draw events for PIX profiling.
[2020.11.05-19.40.52:127][  0]LogD3D12RHI: The system supports ID3D12Device1.
[2020.11.05-19.40.52:127][  0]LogD3D12RHI: The system supports ID3D12Device2.
[2020.11.05-19.40.52:222][  0]LogD3D12RHI: Warning: [GPUBreadCrumb] Failed to OpenExistingHeapFromAddress, error: 80070057
[2020.11.05-19.40.52:238][  0]LogD3D12RHI: Warning: [GPUBreadCrumb] Failed to OpenExistingHeapFromAddress, error: 80070057
[2020.11.05-19.40.52:251][  0]LogD3D12RHI: Warning: [GPUBreadCrumb] Failed to OpenExistingHeapFromAddress, error: 80070057
[2020.11.05-19.40.52:256][  0]LogD3D12RHI: Display: Not using pipeline state disk cache per r.D3D12.PSO.DiskCache=0
[2020.11.05-19.40.52:256][  0]LogD3D12RHI: Display: Not using driver-optimized pipeline state disk cache per r.D3D12.PSO.DriverOptimizedDiskCache=0
[2020.11.05-19.40.52:256][  0]LogD3D12RHI:     GPU DeviceId: 0x1916 (for the marketing name, search the web for "GPU Device Id")

Of course I search forums, answers and web for the error messages. It seems they often occure, but reasons are quiet different. So I need something how to get to the causing code for this (or causing efeature).

Help very welcome!

Thanks a lot!

Some advises would be very appreciated.

Hi… Have you enabled shader caching in the Nvidia control panel?
And have you set a pagefile in windows for you SSD drive?



image

image

The Nvidia control panel shader cache size is something that is good to set to help on loading textures and stuff in heavy games. And also set the Cuba system fallback to prefer system fallback to avoid out of memory crash:
image