[PC] Recurring DEVICE_HUNG — same crash hash {823c07cc} — RTX 3080

Subject: Recurring DXGI_ERROR_DEVICE_HUNG — GPU hang in DX12 submission path, identical failure hash across 5 crash dumps

System specs:

• CPU: AMD Ryzen 7 5800X
• GPU: NVIDIA GeForce RTX 3080 (drivers tested: 59649 and 61047 — crash occurs on both)
• RAM: DDR4 3600 MHz, FCLK 1800 (1:1, verified stable)
• OS: Windows 11 build 26100
• Game versions affected: 1.0.426.51308 (Season 3) and 1.0.424.48767
• Resolution: 2560x1440, DX12

Symptom: The game crashes mid-match with DXGI_ERROR_DEVICE_HUNG (0x887a0006). Typical pattern: stable for long sessions (20+ matches / 1-2 hours), then crashes. Never on launch. The most recent crash showed the engine's own watchdog message:

Hang Detected in waitForFrameFences. Timed out after 8000 ms. Waiting for SignalId: 384740. Last Signaled Fence Value: 384741. Unsignaled Fences: ("Graphics" stopped at 384739) File: D:\dev\TnT\Code\Packages\RenderCore2\RenderCore2\dev\source\platform\Dx12\SubmissionManagerDx12.cpp

Note the fence values: the Graphics queue stopped at 384739 while a later fence (384741) was already signaled — suggesting an async queue completed past the point where the graphics queue hung.

WinDbg analysis (5 dumps, June 5–10): All five Frostbite minidumps produce the identical failure hash: {823c07cc-a67a-e92f-3480-f4d74ce569af} — same as reported by other users in this forum (e.g. thread 13405488), on both NVIDIA and AMD GPUs.

Representative stack (latest dump):

KERNELBASE!RaiseException+0x8a bf6!ffxFsr2ResourceIsNull+0x22df201 bf6!ffxFsr2ResourceIsNull+0x22de389 bf6!ffxFsr2ResourceIsNull+0x23aa9a1 Engine_Render_Core2_PlatformPcDx12_retail!rc2DeviceInit+0x23e76 Engine_Render_Core2_PlatformPcDx12_retail!rc2DeviceInit+0x1433 Engine_Render_Core2_PlatformPcDx12_retail!rc2GetSystemServiceProvider+0x584 bf6!ffxFsr2ResourceIsNull+0x22dd6d1 [... worker thread frames ...]

Key observation: running ~*kc on the dump shows a live thread named "AMD FSR WaitableObject tracking Thread" inside amd_fidelityfx_framegeneration_dx12, even though:

• GPU is NVIDIA (RTX 3080)
• All upscaling disabled: GstRender.NvidiaUpscalingTechnique 0, GstRender.AMDIntelUpscalingTechnique 0, GstRender.FrameGeneration 0, GstRender.AMDFrameGenerationEnabled 0
• Config files set to read-only to prevent overwrites

The FidelityFX module is loaded and active regardless of user settings, consistent with what other users have reported.

Troubleshooting already performed (no effect):

• Clean driver reinstall via DDU (two different driver versions)
• All upscaling/frame generation disabled via in-game settings and config files
• Shader cache cleared
• TDR registry values (TdrDelay) adjusted
• DRAM stability verified (FCLK/UCLK/MCLK 1:1 at 1800 MHz)
• Overlays disabled
• Crash reproduces regardless

Attachments: latest .mdmp crash dump, WinDbg !analyze -v output, WinDbg ~*kc all-threads output, screenshot of the in-game error dialog.

This appears to be the same engine-level bug reported by multiple users across GPU vendors. Happy to provide additional dumps or run DRED-enabled captures if useful to the dev team.