RTX 5090 (GB202) + Ryzen 9 9950X3D: “GPU fallen off the bus” / TDR under load and at idle (Windows)
I encountered similar issues on a similar system; maybe additional data helps.
System Information
| Component | Detail |
|---|---|
| GPU | NVIDIA RTX 5090 32GB (ZOTAC SOLID) |
| Motherboard | Gigabyte X870E AORUS ELITE WIFI7, BIOS F11 (F5 also affected) |
| CPU | AMD Ryzen 9 9950X3D |
| RAM | 64GB (2×32) DDR5 @ 4800 MT/s |
| PSU | Seasonic PRIME TX-1600 Noctua Edition, native 12V-2x6 cable |
| OS | Windows 11 (Build 26200) |
Symptom
The failure presents in two ways:
Variant A — GPU drops off the bus, system stays up:
- Monitors lose signal
- GPU fans ramp up aggressively
- GPU output never recovers
- GPU is no longer visible to the system (only the motherboard HDMI output works)
- Requires a reboot to restore GPU functionality
Variant B — black screen + automatic reboot.
In both cases the only reliable recovery is a full reboot / cold power cycle.
When it occurs
- Crashes happened during gaming (low GPU load, e.g. FPS locked CS2 and HOMM:OE).
- Crashes have also occurred after the system has idled for a long time.
Crash Signature
Every occurrence logs a driver timeout/watchdog event, similar to:
Problem Event Name : LiveKernelEvent
Code : 141
Parameter 1 : ffffb801ae578010
Parameter 2 : fffff80e76997e60
Parameter 3 : 0
Parameter 4 : ffffb801ecf2c080
Associated events that appear sporadically around the crash:
| Event ID | Source | Details |
|---|---|---|
| 153 | nvlddmkm | \Device\Video3 — GpuRcReset TDR occurred on GPUID:100 |
| 14 | nvlddmkm | \Device\000000de CMDre 00000026 0000fffc ffffffff 0000000f ffffffff |
| 10 | Realtek | Realtek PCIe 2.5GbE Family Controller is reset by ESD timer |
| 15 | HAL | IOMMU error (details below) |
IOMMU error detail:
The IOMMU has detected an error.
Device: 0x7900
Fault Information: 0xFEF341100
FaultReason: 0x6
ExtendedData: 0x0
Bugcheck (captured occurrence)
One of the automatic-reboot occurrences produced a 0x116 VIDEO_TDR_FAILURE bugcheck. Minidump: C:\Windows\Minidump\070226-22828-01.dmp.
VIDEO_TDR_FAILURE (116)
Attempt to reset the display driver and recover from timeout failed.
Arg1: ffffd50a3b9dc010 (pointer to TDR_RECOVERY_CONTEXT)
Arg2: fffff8043d2054c0 (responsible device driver module — nvlddmkm)
Arg3: ffffffffc000009a (NTSTATUS of last failed operation — 0xC000009A = STATUS_INSUFFICIENT_RESOURCES)
Arg4: 0000000000000004
FAILURE_BUCKET_ID: 0x116_IMAGE_nvlddmkm.sys
MODULE_NAME: nvlddmkm
IMAGE_NAME: nvlddmkm.sys
Stack (TDR reset path):
nt!KeBugCheckEx
dxgkrnl!TdrBugcheckOnTimeout+0x101
dxgkrnl!ADAPTER_RENDER::Reset+0x220
dxgkrnl!DXGADAPTER::Reset+0x58a
dxgkrnl!TdrResetFromTimeout+0x15
dxgkrnl!TdrResetFromTimeoutWorkItem+0x22
nt!ExpWorkerThread+0x4bb
nt!PspSystemThreadStartup+0x5a
nt!KiStartSystemThread+0x34
Variables Tested
| Variable | Result |
|---|---|
| GPU / CPU / chipset driver updates | No change |
| BIOS updates (F5 → F11) | No change |
| PCIe ASPM disabled | No change |
| PCIe forced to Gen4 | No change |
| Disabling CPU C-states | ~4 weeks without issues, then the issue returned |
Notes
- Not reliably reproducible on demand.
- GPU issues noticed since May 2026.