RTX 5080 Crashing at idle

[mongoosemango]

RTX 5080 Crash Investigation Summary

 RTX 5080 crashes during gaming (especially Stalker 2) with Event ID 41 and 0x00000133 (DPC_WATCHDOG_VIOLATION) errors, causing black screens and system reboots during extended gaming sessions.

 

Ryzen 9 9950x3d

G.SKILL Trident Z5 Royal Neo Series (AMD Expo) DDR5 RAM 64GB (2X 32GB) 6000MT/s

Running at 5600mhz now.

Running A asrock taichi 1300W PSU

Switch rammed out put in one stick of another brand same issue 

on the newest bios update

1. Power Issues (RULED OUT) - Tested triple 8-pin cables, also tried 600w single cable confirmed 356W power delivery is stable in FurMark
2. PCIe Settings - Set all PCIe slots to Gen4, unified PCIe configuration, disabled Fast Boot/ERP
3. Memory Stability - Reduced DDR5 from 6000MHz to 5600MHz (kept EXPO enabled) based on 9950X3D compatibility issues
4. FCH Spread Spectrum - Tried both enabled and disabled  noticed it made things worse when enabled)

Current Status: Still getting the same DPC_WATCHDOG_VIOLATION crashes despite all BIOS optimizations. The issue persists with  HYTE Y70 riser cable setup.

Riser Cable Testing - tried multiple riser cables:

• Original HYTE 4.0 riser cable → crashes
• LINKUP 5.0 riser cable → still crashes
• Both failed with the same DPC_WATCHDOG_VIOLATION errors

6. BIOS Optimizations Beyond Basic Settings:

• Disabled Fast Boot and ERP
• Set all PCIe slots to Gen4 (not just the GPU slot)
• Unified PCIe configuration to eliminate timing conflicts
• Attempted FCH Spread Spectrum tweaks

7. Stress Testing Patterns:

• FurMark runs fine (short bursts, different driver code path)
• Crashes specifically during extended gaming sessions
• "Nap test" pattern - crashes during 2-3 hour unattended gaming
• Crashes happen during sustained load, not synthetic benchmarks

8. Driver Version Confirmation:

• Running latest NVIDIA 576.40 drivers also tired rolling back to the three oldest drivers. STILL have issues with display cutting out then turns into me having to hard reboot or i come back to the windows enter your pin screen seems like its only when left Nap test" pattern - crashes during 2-3 hour unattended gamin does anyone have any other ideas of what to possibly do next?

[DoctorNick]

Watchdog is usually not CPU GPU related. Reset CMOS and run everything on stock apart from rebar enabled. Make sure ram is in the right slots A2 and B2 (2+4). Edit: if you disable fast boot in bios, remember to do the same in windows.

Edit: CPU to gpu..

[mongoosemango]

37 minutes ago, DoctorNick said:

Watchdog is usually not CPU related. Reset CMOS and run everything on stock apart from rebar enabled. Make sure ram is in the right slots A2 and B2 (2+4). Edit: if you disable fast boot in bios, remember to do the same in windows.

if you disable fast boot in bios, remember to do the same in windows both are off ram is in the right slots A2 and B2

[DoctorNick]

1 hour ago, mongoosemango said:

if you disable fast boot in bios, remember to do the same in windows both are off ram is in the right slots A2 and B2

Ok then do the rest and test more. What board are you using?

[Poinkachu]

6 hours ago, mongoosemango said:

Riser Cable Testing - tried multiple riser cables:

• Original HYTE 4.0 riser cable → crashes
• LINKUP 5.0 riser cable → still crashes
• Both failed with the same DPC_WATCHDOG_VIOLATION errors

What happens without riser cable.

[Mark Kaine]

Did you DDU*in safe mode*?

 

Also try installing 572.83 "Studio drivers" ... that worked for me just fine on 5xxx besides some small annoyances and bugs admittedly, but no crashing, black screens, etc.