Random GPU crashes and timeouts on Windows with D3D12 mostly on app startup

Development Rendering
, ,
1

Hello! We’re seeing a large number of random GPU crashes and GPU timeouts on Windows in our latest test and would like to ask for your guidance.

During this test we noticed a significant increase in GPU crashes. These issues may have existed in previous tests as well, but we likely didn’t see them because the application was exiting via FPlatformMisc::RequestExit(). Recently we added FPlatformMisc::RaiseException() so these cases are reported to Sentry, and now they are visible.

Many crashes and hangs happen very early at startup. Around 40% of all crashes occur during application launch, before any 3D scene rendering starts. We also observe similar startup hangs in our automated tests, but so far we haven’t been able to identify a clear pattern or dependency.

For crashes there are more diagnostic data than for hangs. In some cases the logs contain Aftermath crash dump decoding information and GPU breadcrumbs. I attached an archive with all collected logs and dumps. Driver versions and relevant settings used in these tests are also included in the attached data.

Here is the distribution of crashes across GPUs:

[Image Removed]

I’d like to ask:

  • Is this a known issue on Windows (UE / D3D12 / GPU drivers) that could explain frequent GPU crashes or timeouts, especially at startup?
  • Are there any recommended settings, workarounds, or engine options we should try?
  • Could you suggest the best way to debug and investigate such startup GPU crashes and timeouts? What additional data or tooling would you recommend collecting?

Any advice or direction would be very helpful.

[Attachment Removed]

2

Hello,

Thank you for reaching out.

I’ve been assigned this issue, and we will be looking into these GPU crashes for you.

[Attachment Removed]

3

Hello,

The presence of debug information in the logs you provided is inconsistent. Can you please collect similar logs with the following command-line options, and share them with us?

-gpucrashdebugging -gpuvalidation -d3ddebugThis should enable Breadcrumbs and DRED on all logs, and Aftermath for NVIDIA GPUs.

Both the Breadcrumb data and DRED will help narrow down what pass is causing the crash.

This video from Unreal Fest 2023 discusses this tool:

https://www.youtube.com/watch?v=CyrGLMmVUAI

NVIDIA Aftermath can be used to see exactly what shader and what part of that shader triggered a crash, in the case of a Page Fault. This does require shader debug symbols. For more information on this, please see these two tickets:

[Content removed]

[Content removed]

This document has more details about the data provided by Breadcrumbs, DRED, and Aftermath, and how it helps with debugging:

https://dev.epicgames.com/documentation/en-us/unreal-engine/dealing-with-a-gpu-crash-when-using-unreal-engine?application_version=5.6

[Attachment Removed]

4

Hi!

Thanks for your response. We’ve tried to reproduce the issue using the command-line arguments (with Graphics Tools installed, as required - according to the logs) on our devices, including both our automated testing setups and our work machines, but so far we haven’t been able to reproduce it.

We’ll get back to you as soon as we manage to collect more data.

Best regards,

Daniil

[Attachment Removed]

5

Hello Stephen,

We’re still facing the GPU timeout and crash issues, but unfortunately, we haven’t been able to reproduce them consistently. The only pattern we see in our dev environment is GPU Timeouts which happen during autotests, right at app startup. Even so, it’s pretty irregular: sometimes the crash happens every second run when I’m doing 20 tests where the test just opens a level and that’s it, other times, I can run up to 50 gameplay scenario tests without a single issue.

Since these particular case happens so early in the process, before any app-specific rendering starts, I’m wondering if it’s something to do with the engine or the GPU driver. Do you know if there are any known issues in this area, or if anything in recent releases (5.6, 5.7, 5.8) might have addressed it?

Here are a few things I’d appreciate your input on:

  1. Is this a known issue on Windows (with D3D12 or GPU drivers) that might explain the frequent GPU timeouts or crashes, especially on startup?
  2. Any engine settings, workarounds, or options you’d recommend to try out?

Regarding debugging, unfortunately -gpucrashdebugging doesn’t provide any useful information when the GPU timeout happens. We don’t see any DRED, Aftermath, or Breadcrumb data in those cases.

The same applies to -gpuvalidation. On some devices (for example RTX 3060), enabling it even causes a crash in D3D12CreateDevice with an E_INVALIDARG error, despite Graphics Tools being installed.

Also, I’d like to clarify one more thing. Initially, we didn’t see these issues in our crash statistics at all, because the engine was just exiting without triggering any exception. We only started seeing them in Sentry after we explicitly added an RaiseException call before exit.

So I’m trying to understand whether this was something specific to our setup, or if this is just how UE handles GPU crashes/timeouts by default - does it silently exit without generating a crash report?

[Attachment Removed]

6

Hello,

Thank you for sharing the tickets and CLs!

Unfortunately, we have not been able to reproduce these crashes or timeouts on our side. All cases come from our public test crash reports, so we cannot check if disabling Lumen Reflections or GPU Skin Cache changes the behavior.

We will apply the CLs you shared and watch the results during the next public test.

The only issue we can reproduce from time to time in our environment is a GPU timeout during application startup. I slightly changed the code and managed to capture GPU breadcrumbs for these cases. For now, since the code update, I have seen this on two devices (GTX 1050 and RTX 3060). Eight other test devices (with Intel UHD Graphics, AMD Graphics, GTX 1650, GTX 1660, RTX 3050 Ti), do not show this problem in more than 100 launches each.

The logs with GPU Breadcrumbs are attached. We would really appreciate any thoughts or suggestions about this issue.

Thank you.

[Attachment Removed]

7

Hello Kevin,

I’ve tried increasing r.D3D12.SubmissionTimeout from 5 to 50 seconds, but GPU timeouts still occur. You can find the log attached.

Regarding -d3ddebug, I’ve also tried this option, but on the device that fails with gpuvalidation (RTX 3060), it also fails with d3ddebug. I’m attaching the log as well.

On another device (GTX 1050), where there is no failure at startup with either of these two options enabled, I wasn’t able to reproduce any GPU hang or crash while these options were active. However, it’s worth mentioning that in my latest iterations I wasn’t able to reproduce the timeout on this device even without these options, so I will continue testing.

Kind regards.

[Attachment Removed]

8

Hello Stephen,

Could you please clarify the exact steps you’re recommending?

As I understand it, the “Engine\Binaries\Win64\D3D12” folder is located in UE repository. Are you suggesting that we delete this folder, run the Setup.bat script to re-download the files, and then repackage the game build?

[Attachment Removed]

9

Hello,

For this case, can you start by deleting the D3D12 binaries in your project folder version so it can copy re-copy the engine version in and potentially fix SDK mismatches in the dlls?

This would be under “<ProjectName>\Binaries\Win64\D3D12”.

UBT will stage the appropriate dlls during the build.

If this does not pass I would recommend resyncing the D3D12 binaries on the engine side too to ensure you do not have a local version issue.

[Attachment Removed]

10

Hello, the builds where I see the crashes are assembled using the TeamCity infrastructure, and we do a full clean of both the game and engine repositories before each build, so if there are no UE specific caching mechanisms, these files are refetched each time.

[Attachment Removed]

11

Hello,

From looking at the collected logs, I noticed some of the GPU timeout logs come from GPUs with low VRAM and outdated drivers. Additionally, some of the GPU timeouts are preceded by PSO creation failures, which could indicate driver issues.

Can you suggest players to update to latest drivers?

[Attachment Removed]

12

Hi Stern,

Thank you for taking a look, and sorry for the delayed response.

As I understand it, the minimal and recommended driver versions can be specified via the Hardware.ini config.

[GPU_NVIDIA Windows]
SuggestedDriverVersion="536.40"
 
; The newest driver which supports legacy Nvidia GPUs (700 series and older) is 474.64. We allow this driver for DX11.
+DriverDenyList=(DriverVersion="<474.64", Reason="These driver versions have known stability issues and missing features", RHI="D3D11")
 
; DX12 and Vulkan require recent drivers. The 536 series was released in June 2023.
+DriverDenyList=(DriverVersion="<536.00", Reason="These driver versions have known stability issues and missing features", RHI="D3D12")
+DriverDenyList=(DriverVersion="<536.00", Reason="These driver versions have known stability issues and missing features", RHI="Vulkan")

Do you suggest changing the settings there, or just communicating this to users in the public release notes?

If it’s worth changing SuggestedDriverVersion or DriverDenyList, are there any specific driver versions you would recommend specifying?

[Attachment Removed]

13

Hello,

I would recommend suggesting users should update to their latest driver via public release notes since a specific problematic driver version has not been identified.

On a separate note than the GPU timeout, the “Blitz (35) (aftermath)” log has Aftermath information indicating a page fault during ray tracing acceleration structure build.

Please see this ticket that discusses a related issue:

[Content removed]

[Attachment Removed]

14

Hello Stern,

Thanks, we’ll include this recommendation in our release notes.

Regarding the ticket you shared: as I understand it, the fix proposed there is already included in UE 5.5, which means we are already using it on our side.

[Attachment Removed]

15

Hello Stephen,

Thank you for providing this. Are there any recommended values we should set?

In the meantime, we’ve disabled the ray tracing feature we were using (Lumen Reflections) to check if it makes a difference in terms of crash rate.

We are about to begin our next public test, so I will be able to provide additional data in a couple of weeks.

[Attachment Removed]

16

Hi,

Apologies for the delay. We don’t have a recommended size for r.RHICmd.MaxAccelerationStructureBuildScratchSize, but the default/max is 2GB so you might do some local testing with the card to see what is an acceptable floor.

Were you able to test whether disabling Lumen Reflections lowered the GPU crash rate?

[Attachment Removed]

17

Hello Alex,

Yes, the test has finished, and it appears that Lumen Reflections did not significantly impact GPU crashes.

This time, we also have the driver version and date included in the crash statistics. The variability is quite wide, and it shows that even the most recent drivers still experience many GPU crashes.

The most unusual and at the same time the most common case occurs on app startup. We also see this in our regular autotests in the form of GPU hangs.

I ran an experiment where I progressively disabled different parts of the rendering that occur before the 5-7th frame, when the crash happens. In the end, almost nothing was running on the GPU, yet GPU hangs were still occurring.

I disabled the following:

  1. FDeferredShadingSceneRenderer::Render
  2. FSlateRHIRenderer::DrawWindow_RenderThread
  3. FScene::Update
  4. USE_BUFFER_POOL_ALLOCATOR & USE_TEXTURE_POOL_ALLOCATOR
  5. FD3D12DynamicRHI::RHIUpdateTexture3D
  6. FMediaTextureResource::InitRHI & FMediaTextureResource::Render

So from these RHI Breadcrumbs:

2026-04-23 02:37:19: [ 7]LogD3D12RHI: Warning: GPU timeout: A payload (0x000002606A426A60) on the [0x0000026069276840, 3D] queue has not completed after 5.000472 seconds.

2026-04-23 02:37:19: [ 7]LogD3D12RHI: Error: GPU crash detected:

2026-04-23 02:37:19: - Device 0 OK (no device removed reason)

2026-04-23 02:37:19:

2026-04-23 02:37:19: [ 7]LogRHI: Error: Active GPU breadcrumbs:

2026-04-23 02:37:19:

2026-04-23 02:37:19: Device 0, Pipeline Graphics: (In: 0x80000290, Out: 0x8000028f)

2026-04-23 02:37:19: (ID: 0x80000256) [ Active] Frame 4

2026-04-23 02:37:19: (ID: 0x800002a2) [ Active] FRDGBuilder::Execute

2026-04-23 02:37:19: (ID: 0x8000025e) [ Active] Scene

2026-04-23 02:37:19: (ID: 0x8000028e) [ Active] PostProcessing

2026-04-23 02:37:19: (ID: 0x8000028f) [ Active] ThirdParty FDLSSSceneViewFamilyUpscaler 2560x1440 -> 2560x1440

2026-04-23 02:37:19: (ID: 0x80000290) [ Active] DLSS

2026-04-23 02:37:19: (ID: 0x800002a9) [Not Started] FRDGBuilder::Execute

2026-04-23 02:37:19: (ID: 0x800002a6) [Not Started] SlateUI Title = Blitz (64-bit Development PCD3D_SM6)

2026-04-23 02:37:19: (ID: 0x800002a7) [Not Started] DrawDebugCanvas

2026-04-23 02:37:19: (ID: 0x800002a8) [Not Started] CanvasFlush

2026-04-23 02:37:19: (ID: 0x800002ae) [Not Started] BufferPoolCopyOps

2026-04-23 02:37:19: (ID: 0x800002af) [Not Started] TexturePoolCopyOps

2026-04-23 02:37:19: (ID: 0x800002ac) [Not Started] Frame 5

2026-04-23 02:37:19: (ID: 0x800002ad) [Not Started] WorldTick

2026-04-23 02:37:19: (ID: 0x800002b0) [Not Started] SendAllEndOfFrameUpdates

2026-04-23 02:37:19: (ID: 0x800002b1) [Not Started] RayTracingGeometry

2026-04-23 02:37:19: (ID: 0x800002f9) [Not Started] FRDGBuilder::Execute

2026-04-23 02:37:19: (ID: 0x800002b5) [Not Started] Scene

2026-04-23 02:37:19: (ID: 0x800002b7) [Not Started] FXSystemPreRender

2026-04-23 02:37:19: (ID: 0x800002fa) [Not Started] GPUParticles_PreRender

2026-04-23 02:37:19: (ID: 0x800002b8) [Not Started] GPUSceneUpdate

2026-04-23 02:37:19: (ID: 0x800002ba) [Not Started] BuildRenderingCommandsDeferred(Culling=On)

2026-04-23 02:37:19: (ID: 0x800002bb) [Not Started] PrePass DDM_AllOpaqueNoVelocity (Forced by DBuffer)

2026-04-23 02:37:19: (ID: 0x800002bd) [Not Started] HZB

2026-04-23 02:37:19: (ID: 0x800002be) [Not Started] BuildHZB(ViewId=0)

2026-04-23 02:37:19: (ID: 0x800002bf) [Not Started] BuildHZB

2026-04-23 02:37:19: (ID: 0x800002c0) [Not Started] SortLights

2026-04-23 02:37:19: (ID: 0x800002c1) [Not Started] ComputeLightGrid

2026-04-23 02:37:19: (ID: 0x800002c2) [Not Started] CullLights 40x23x32 NumLights 0 NumCaptures 0

2026-04-23 02:37:19: (ID: 0x800002c3) [Not Started] LightFunctionAtlasGeneration

2026-04-23 02:37:19: (ID: 0x800002c5) [Not Started] BasePass

2026-04-23 02:37:19: (ID: 0x800002c6) [Not Started] EditorPrimitives

2026-04-23 02:37:19: (ID: 0x800002c7) [Not Started] ShadowDepths

2026-04-23 02:37:19: (ID: 0x800002c8) [Not Started] BuildRenderingCommandsDeferred(Culling=On)

2026-04-23 02:37:19: (ID: 0x800002ca) [Not Started] RenderDeferredLighting

2026-04-23 02:37:19: (ID: 0x800002cc) [Not Started] InitTranslucencyLightingVolumeTextures

2026-04-23 02:37:19: (ID: 0x800002d4) [Not Started] FilterTranslucentVolume 64x64x64 Cascades:2

2026-04-23 02:37:19: (ID: 0x800002d6) [Not Started] ReflectionIndirect

2026-04-23 02:37:19: (ID: 0x800002d7) [Not Started] ScreenSpaceReflections(Quality=2)

2026-04-23 02:37:19: (ID: 0x800002d8) [Not Started] ScreenSpaceReflections

2026-04-23 02:37:19: (ID: 0x800002d9) [Not Started] ReflectionEnvironment

2026-04-23 02:37:19: (ID: 0x800002dc) [Not Started] PostRenderOpsFX

2026-04-23 02:37:19: (ID: 0x800002dd) [Not Started] FXSystemPostRenderOpaque

2026-04-23 02:37:19: (ID: 0x800002fb) [Not Started] GPUParticles_PostRenderOpaque

2026-04-23 02:37:19: (ID: 0x800002fc) [Not Started] NiagaraUpdateCollisionGroupsMap

2026-04-23 02:37:19: (ID: 0x800002de) [Not Started] Translucency

2026-04-23 02:37:19: (ID: 0x800002e0) [Not Started] RenderTranslucency

2026-04-23 02:37:19: (ID: 0x800002e5) [Not Started] PostProcessing

2026-04-23 02:37:19: (ID: 0x800002e6) [Not Started] ThirdParty FDLSSSceneViewFamilyUpscaler 2560x1440 -> 2560x1440

2026-04-23 02:37:19: (ID: 0x800002e7) [Not Started] DLSS

2026-04-23 02:37:19: (ID: 0x80000300) [Not Started] FRDGBuilder::Execute

2026-04-23 02:37:19: (ID: 0x800002fd) [Not Started] SlateUI Title = Blitz (64-bit Development PCD3D_SM6)

2026-04-23 02:37:19: (ID: 0x800002fe) [Not Started] DrawDebugCanvas

2026-04-23 02:37:19: (ID: 0x800002ff) [Not Started] CanvasFlush

2026-04-23 02:37:19:

2026-04-23 02:37:19: Device 0, Pipeline AsyncCompute: (In: 0x800002b5, Out: 0x80000256)

2026-04-23 02:37:19: (ID: 0x800002ac) [ Active] Frame 5

2026-04-23 02:37:19: (ID: 0x800002f9) [ Active] FRDGBuilder::Execute

2026-04-23 02:37:19: (ID: 0x800002b5) [ Active] Scene

I got these:

[2026.04.20-12.23.20:281][ 6]LogD3D12RHI: Warning: GPU timeout: A payload (0x0000020E5BB38E00) on the [0x0000020E5BDE6840, 3D] queue has not completed after 5.000000 seconds.

[2026.04.20-12.23.20:281][ 6]LogD3D12RHI: Error: GPU crash detected:

- Device 0 OK (no device removed reason)

[2026.04.20-12.23.20:281][ 6]LogRHI: Error: Active GPU breadcrumbs:

Device 0, Pipeline Graphics: (In: 0x8000002a, Out: 0x8000002a)

(ID: 0x80000026) [ Active] Frame 5

(ID: 0x8000002a) [ Active] FRDGBuilder::Execute

(ID: 0x8000002b) [Not Started] FRDGBuilder::Execute

(ID: 0x8000002c) [Not Started] Frame 6

(ID: 0x8000002d) [Not Started] WorldTick

(ID: 0x8000002e) [Not Started] SendAllEndOfFrameUpdates

(ID: 0x8000002f) [Not Started] FRDGBuilder::Execute

(ID: 0x80000030) [Not Started] FRDGBuilder::Execute

Device 0, Pipeline AsyncCompute: (In: 0x8000002c, Out: 0x80000026)

No breadcrumb nodes found for this queue.

At some point, I also added additional logs and call stack dumps to the RHI Breadcrumbs macro, and after that the crash rate dropped significantly, almost 0 in my runs. This could indicate that the issue is not directly related to GPU work itself, and may instead involve a CPU-side concurrency problem.

What do you think about this and what would you suggest as the next steps to investigate it further?

Also during my experiments I was able to capture this GPU Hang with d3debug, gpuvalidation, gpucrashdebugging, dred, but there was no additional data, I guess because it is not detected as a GPU crash by the driver. Anyway, I will try enabling lightweight DRED for the next public test to gather more information.

[Attachment Removed]

18

Hi,

Apologies for the delay.

almost nothing was running on the GPU, yet GPU hangs were still occurring.

Have you taken an insights trace or sampling profiler trace when these occur to determine what else is going on? My main suspicions is PSO compilation somehow being involved, but there could also be large resource uploads happening or some kind of allocations that could provide a clue as to the source. We have seen issues where PSO compilation will take a really long time, and we’ve seen contention in the driver when multiple PSO compilation tasks are in flight at a time so it might be good to get a picture of what PSO compilation, if any, is occurring. If there is a lot of PSO compilation happening, we could try limiting the number of threads involved.

Do the breadcrumbs in the startup crash always look the same?

2026-04-23 02:37:19: (ID: 0x8000028f) [ Active] ThirdParty FDLSSSceneViewFamilyUpscaler 2560x1440 -> 2560x1440

2026-04-23 02:37:19: (ID: 0x80000290) [ Active] DLSS

This could indicate that the issue is not directly related to GPU work itself, and may instead involve a CPU-side concurrency problem.

The logs provided above in Blitz_RTX3060_increased_timeout.log don’t contain breadcrumbs and suggest a deadlock/concurrency problem on the CPU, but it’s difficult to tell - can you reproduce this with a debugger attached so you can catch it when the deadlock occurs - or maybe the .dmp file has more clues?

I was able to capture this GPU Hang with d3debug, gpuvalidation, gpucrashdebugging, dred, but there was no additional data, I guess because it is not detected as a GPU crash by the driver.

Can you provide more details - or a log sample - showing what was in the log output that looked like a GPU hang but the driver didn’t detect it? Do you mean windows detected the TDR and reset the GPU but no Aftermath dump was created?

Also you mentioned this is happening on a 1050 and 3060 - do you have multiple 1050s and 3060s in your automation farm or does the startup crash only repro on one of the 1050s or 3060s. It’d be good to know if there’s a remote chance of there being a hardware issue.

I also noticed in the startup crash where D3D12CreateDevice failed with E_INVALIDARG, the -d3ddebug startup command was used. If this always fails you should check if the D3D debug layer dlls are properly installed on that device and it’s possible that enabling the debug layer will add overhead that could make the crash more likely to occur.

[Attachment Removed]

19

Hello Alex,

Have you taken an insights trace or sampling profiler trace

I will try to get this data, but since I can reproduce this issue only in ATs, it will take some time to implement this functionality there.

Do the breadcrumbs in the startup crash always look the same?

No, DLSS appears there only when it’s enabled. By default, we have it enabled on the RTX 3060 and disabled on the GTX 1050, but that was the first thing I disabled and even without it, there are crashes on the RTX 3060 as you can see in Blitz_GPU_Timeout_logs_with_breadcrumbs logs, also, for example, in the case where almost all render work is disabled, there are also no DLSS breadcrumb nodes.

The logs provided above in Blitz_RTX3060_increased_timeout.log don’t contain breadcrumbs and suggest a deadlock/concurrency problem on the CPU

To be honest, I don’t understand this point, in this log there is a typical GPU Timeout problem we’re facing, could you please elaborate on it?

Can you provide more details - or a log sample - showing what was in the log output that looked like a GPU hang but the driver didn’t detect it?

You can see this in Blitz_RTX3060_increased_timeout.log, it’s a GPU Timeout, but no GPU crash was detected: LogNvidiaAftermath: Warning: Aftermath did not detect a GPU crash. Skipping crash dump processing.

Also you mentioned this is happening on a 1050 and 3060 - do you have multiple 1050s and 3060s in your automation farm

On our automation farm, we have only a single GTX 1050 and a single RTX 3060, but the same GPU Timeouts happen on other GPUs as well (e.g. GTX 1650, RX580), and not only on our automation farm, but in public tests as well (e.g. different GPUs of RTX 40 series)

I also noticed in the startup crash where D3D12CreateDevice failed with E_INVALIDARG, the -d3ddebug startup command was used. If this always fails you should check if the D3D debug layer dlls are properly installed on that device and it’s possible that enabling the debug layer will add overhead that could make the crash more likely to occur.

As I’ve already mentioned I finally was able to successfully run the build with -d3ddebug and catch a GPU Timeout with it, but there was no additional data in the logs (I’ve attached the log file)

Also, I just realized that the timeout in Blitz_RTX3060_increased_timeout.log is still 5s. I’ll do another run to test this properly.

[Attachment Removed]

20

Thanks for providing the additional log.

The log file I was referring to (Blitz_RTX3060_increased_timeout.log) that didn’t seem to be a GPU hang has this at the end instead of any GPU crash logs - perhaps there are multiple logs with that name?

2026-02-24 11:45:46: [  7]LogRendererCore: Error: GameThread timed out waiting for RenderThread after 120.00 seconds:
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ffd6ea6bc5f KERNELBASE.dll!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff687cd5606 Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff687758d1c Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff68775939e Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff68c14a8b8 Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff68c15d2f9 Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff687d41058 Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ff687d385bf Blitz.exe!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ffd7105e8d7 KERNEL32.DLL!UnknownFunction []
2026-02-24 11:45:46: [  7]LogRendererCore: Error:   0x00007ffd71a41c44 ntdll.dll!UnknownFunction []
2026-02-24 11:45:46: [  7]LogWwiseMonitor: Error: Voice Starvation
2026-02-24 11:45:46: [  7]LogRHI: Error: Breadcrumbs 'RHIThread'
2026-02-24 11:45:46:  - Frame 5
2026-02-24 11:45:46: 
2026-02-24 11:45:46: [  7]LogWindows: Could not start crash report client using ../../../Engine/Binaries/Win64/CrashReportClient-Win64-Debug.exe
2026-02-24 11:45:46: [  7]LogMemory: Platform Memory Stats for Windows
2026-02-24 11:45:46: [  7]LogMemory: Process Physical Memory: 1435.66 MB used, 1671.73 MB peak
2026-02-24 11:45:46: [  7]LogMemory: Process Virtual Memory: 2227.14 MB used, 2227.82 MB peak
2026-02-24 11:45:46: [  7]LogMemory: Physical Memory: 12418.90 MB used,  20162.53 MB free, 32581.43 MB total
2026-02-24 11:45:46: [  7]LogMemory: Virtual Memory: 13050.54 MB used,  24394.89 MB free, 37445.43 MB total

It could be interesting to see what’s going on in an Insights trace leading up to the crash. I’ve asked a few colleagues for their thoughts/suggestions as well.

[Attachment Removed]