[Help] Frequent BSOD with NVIDIA RTX 3060 on Windows 11 – Need Advice

[superadmin]

ChatGPT replied:

What the Error Suggests

• Code:

  VIDEO_DXGI_FATAL_ERROR
  usually points to DirectX graphics subsystem problems, most often linked to:
  
  • GPU drivers
    
  • VRAM instability
    
  • Power delivery issues (GPU not getting clean/stable power under load)
    
  • Rarely, a Windows 11 update–driver conflict
    
• The Event Viewer entry “nvlddmkm stopped responding and has successfully recovered” is a TDR (Timeout Detection and Recovery) error, meaning the GPU didn’t respond in time.
  

---

⚡ Why It’s Probably Not Overheating

• You reported 78°C max under load, which is well within safe limits (RTX 3060 is usually fine up to ~83–85°C before thermal throttling).
  
• If it was strictly thermal, FurMark would have crashed the card too (since FurMark typically pushes GPU temps higher than games).
  

So overheating is unlikely the root cause here.

---

✅ Other Possible Causes

1. Driver Conflict / Windows 11 Update
   
   • Sometimes NVIDIA drivers + certain Win 11 builds cause instability.
     
   • Rolling back to a slightly older stable driver (e.g., 537.58 or 546.33) sometimes resolves this.
     
2. Power Delivery Issue
   
   • Your RM650x is a solid PSU, but if the PCIe cable or slot is loose or if the GPU is drawing unstable power under gaming load (which differs from FurMark), it could trigger BSODs.
     
3. VRAM Fault / GPU Degradation
   
   • Stress tests like FurMark mainly load shaders, but some games stress VRAM differently.
     
   • A slightly faulty VRAM chip could pass FurMark but crash in real-world workloads.
     
4. PCIe Bus / BIOS Settings
   
   • Sometimes forcing PCIe Gen 3 in BIOS (instead of Auto/Gen 4) improves stability on B550 boards with RTX 30 series cards.
     

---

? Recommended Next Steps

1. Clean Driver Reinstall (DDU)
   
   • Use Display Driver Uninstaller (DDU) in Safe Mode.
     
   • Reinstall NVIDIA drivers (try 546.33 or 537.58 — not just the latest).
     
2. Check BIOS + Chipset
   
   • Update your ASUS B550 BIOS to the latest.
     
   • Install latest AMD chipset drivers.
     
3. Force PCIe Gen 3
   
   • In BIOS: Advanced → PCIe/Link Speed → Set GPU slot to Gen 3.
     
   • Many users report crashes disappearing with this setting.
     
4. Test with Underclock
   
   • Use MSI Afterburner to reduce GPU core clock by ~100 MHz and VRAM by ~300 MHz.
     
   • If BSODs stop, the GPU silicon or VRAM is unstable → RMA recommended.
     
5. Try Different PSU Cable
   
   • Use a different PCIe power cable (separate rail if modular).
     
   • Reseat GPU in slot.
     

---

? Answering Your Questions

1. Could it be Windows 11 driver conflict?
   → Yes, it’s possible. Try an older NVIDIA driver after a DDU clean install.
   
2. Deeper GPU diagnostics?
   
   • Tools: OCCT GPU VRAM test, 3DMark stress test, HWInfo64 (voltage/power draw logs).
     
   • These often reveal VRAM or power instability missed by FurMark.
     
3. Replace GPU?
   
   • Only if underclocking + driver rollback + BIOS tweaks don’t help.
     
   • Within RM 2000 (~$400), a RTX 4060 Ti 16GB or RX 7700 XT would be good upgrades.
     
4. RMA in Malaysia
   
   • If your GPU is still under warranty, check with your ASUS/NVIDIA authorized distributor (e.g., IdealTech, Tech Armory, or directly through ASUS Malaysia support).
     

---

? My gut feeling: not overheating, but likely driver/PCIe compatibility or VRAM instability.
Try the DDU clean reinstall + PCIe Gen 3 setting first — those solve 70% of these RTX 30-series BSOD cases.
Would you like me to walk you step-by-step on how to set your BIOS to PCIe Gen 3 and run a VRAM-specific stress test? That way we can confirm if it’s hardware or software before thinking about RMA.