Bounded official defect research — 2026-09-16
Outcome: no confirmed matching Unity/HDRP or NVIDIA defect with an applicable fixed version or documented workaround was identified in this bounded search. This is not evidence that no engine or driver defect exists. No downloads, installation, setting changes, or GPU tests were performed.
Local comparison: Unity 6000.3.24f1 / changeset 4e7b9b5b6244, HDRP 17.3.0 pinned package 700710090fa9, RTX 4060 Laptop, NVIDIA 32.0.16.1692 (616.92). Actual failure occurs in D3D11 and D3D12, also after one freshly loaded level with Low quality and Clear weather. Driver events confirm TDRs; low memory use and DRED page-fault VA 0 do not identify a specific defect.
| Official item | Date / version | Preconditions and relevance |
|---|---|---|
| UUM-140181: D3D12 device crash finalizing lightmap generation | Reported 2026-04-17; reproduced 6000.3.13f1, 6000.4.4f1, 6000.5.0b4, 6000.6.0a3; Won't Fix | Similar CheckDeviceStatus / commandlist path and NVIDIA-only reproduction. But it occurs in the Editor during lightmap baking, on RTX 2070S/3080Ti Laptop, and reportedly not with debug arguments. Unity's resolution suspects a driver issue and names no corrected driver. Our runtime D3D11 failure and active-debug D3D12 failure do not match those conditions. No endorsed workaround. |
| 6000.3.24f1 known issue UUM-149781 / issue | Current release dated 2026-09-10 | D3D12 use-after-free when a shader is unloaded immediately after ShaderWarmup.WarmupShaderFromCollection. Local search in Rampage and pinned HDRP runtime found no ShaderWarmup/WarmupShaderFromCollection/WarmupAllShaders/WarmUp call. Only unrelated UnloadUnusedAssets in CampaignMemoryQA. Required warmup/unload sequence unestablished; cannot explain D3D11. Listed as known, no applicable fixed version obtained. |
| UUM-137629: D3D12 DLL loaded despite removed graphics API | Reported 2026-03-20; fixed 6000.3.16f1, 6000.4.8f1, 6000.5.0b10, 6000.6.0a5 | Can crash specific hardware during undesired D3D12 loading. Already fixed before our 6000.3.24f1; no matching regression evidence. |
| UUM-115025 in Unity 6000.3.0f1 notes | Already fixed in 6000.3.0f1 | CPU crash while logging a DRED report after a GPU crash with -force-d3d12-debug. Post-crash diagnostic failure rather than the initial GPU reset; already included in our version. |
NVIDIA's official 615.71.09 / Windows 616.92 Data Center notes, dated 2026-09-09, describe Blackwell/CUDA, server, Linux, and other specific issues. They are not GeForce release notes and provide no established match for Ada RTX4060 Laptop using Unity Direct3D. In particular, a rare CUDA process hang after upgrading branches is not evidence for this already-rebooted Direct3D workload.
The current GeForce 616.92 known-issue document could not be retrieved from official NVIDIA sources within the search limit; candidate official PDF endpoints failed. Therefore no claim is made that its complete known-issue list is clean. Search results mentioning the exact NV_UFLUSH_FB_FLUSH wording were Linux user forum reports on different GPU/driver branches, not a confirmed matching Windows 616.92 defect or official fix, and were excluded from conclusions. Unverified third-party rollback recommendations were also excluded.
Practical implication: continue the existing controlled feature isolation and preserve native driver evidence. These sources do not justify promising that a particular Unity/driver update, rollback, or settings change will fix the observed reset.