Im working with Threadripper 7pro and GTX4090 but my GPU is crashing multiple times while working. even if working on simplest file still GPU is taking too much load.
Steps to Reproduce
Hello,
From the logs it appears this is a GPU Hang with a GPU Pagefault. This could be because a movie frame render has completed and r.ResetRenderTargetsExtent was issued to resize the buffers for the next frame’s lower resolution and somehow the translucency volume lighting is using the old buffer or trying to index outside the buffer size. Try reproducing the crash with r.TranslucentLightingVolume=0 to see if Translucency Volumes are related.
`[2025.04.03-16.54.49:054][669]LogD3D12RHI: Error: GPU crash detected:
- Device 0 Removed: DXGI_ERROR_DEVICE_HUNG
[2025.04.03-16.54.49:054][669]LogRHI: Error: Active GPU breadcrumbs:
Device 0, Pipeline Graphics: (In: 0x8061d7f9, Out: 0x8061d7f5)
(ID: 0x8061d7a9) [ Active] Frame 44669
(ID: 0x8061d82d) [ Active] FRDGBuilder::Execute
(ID: 0x8061d7b6) [ Active] Scene
(ID: 0x8061d7ee) [ Active] RenderDeferredLighting
(ID: 0x8061d7ef) [ Finished] DiffuseIndirectAndAO
(ID: 0x8061d7f0) [ Finished] LumenScreenProbeGather
(ID: 0x8061d7f1) [ Finished] UpdateRadianceCaches
(ID: 0x8061d7f2) [ Finished] Integrate
(ID: 0x8061d7f3) [ Finished] TranslucencyVolumeLighting
(ID: 0x8061d7f4) [ Finished] LumenReflections
(ID: 0x8061d7f5) [ Active] InitTranslucencyLightingVolumeTextures
(ID: 0x8061d7f6) [ Active] Lights
(ID: 0x8061d7f7) [ Active] DirectLighting
(ID: 0x8061d7f8) [ Active] InjectTranslucencyLightingVolume
(ID: 0x8061d7f9) [ Active] InjectTranslucencyLightingVolume(View=0)
Device 0, Pipeline AsyncCompute: (In: 0x8061d819, Out: 0x8061d792)
(ID: 0x8061d7a9) [ Active] Frame 44669
(ID: 0x8061d82d) [ Active] FRDGBuilder::Execute
(ID: 0x8061d7b6) [ Active] Scene
(ID: 0x8061d818) [ Active] PostProcessing
(ID: 0x8061d819) [ Active] TemporalSuperResolution(sg.AntiAliasingQuality=3 AlphaChannel) 1280x720 → 1280x720
Decoding Aftermath GPU Crash:
Device Info:
Status : PageFault (Error_DMA_PageFault)
LogD3D12RHI: Error: PageFault: PageFault at VA GPUAddress “0x1813FA0003F800” (GPU 0)`There are also a number of errors about a high number of temporal samples.
MoviePipelinePIEExecutor: Error: Too many temporal samples for the given shutter angle/tick rate combination. Temporal Samples: 16 Shutter Angle: 0.000000 TicksPerOutputFrame: Frame: 800 Subframe: 0.000000 TicksPerSample: Frame: 0 Subframe: 0.138889. Consider converting to Spatial Samples instead!I’m also noticing you have an error in the logs related to your Lumen settings.
LogOutputDevice: Error: CreateTexture Lumen.ScreenProbeGather.ScreenProbeFilteredRadianceWithBorder Y size too large: 17730, Max: 16384, clampingYou may need to lower r.Lumen.ScreenProbeGather.TracingOctahedronResolution before that frame is rendered to not exceed the supported maximum on your GPU. This doesn’t appear to be related to the GPU hang though.
If you are able to reproduce the crash in a sample project and attach it that would help us further investigate the issue.
Just an idea, can you try reproducing the crash with `-nosound` command-line argument? I’m currently investigating a similar crash, and it seems to be happening only when there’s sound in the game (through we’re using Wwise).
I’m using Path Tracing and i would like to know if you have any idea for that as well. that will be grateful.
Without a sample project to reproduce the crash I don’t have additional suggestions. Occasionally for situations where the GPU hangs from too much work we suggest increasing the TDRDelay, but in your case it looks like a DMA fault which may be driver related. You may want to try rolling your driver back to 553.24
Hello Friend. thanks for help. also I’m using Path Tracing and i would like to know if you have any idea for that as well. that will be grateful.